commit 6b53cfd3d896fd176de5625d95cc6a08facc915d Author: ModelHub XC Date: Wed May 13 01:12:25 2026 +0800 初始化项目，由ModelHub XC社区提供模型 Model: Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.01 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..edccbdd --- /dev/null +++ b/README.md @@ -0,0 +1,59 @@ +--- +base_model: meta-llama/Llama-3.1-8B-Instruct +datasets: Neelectric/MoT_science_Llama3_4096toks +library_name: transformers +model_name: Llama-3.1-8B-Instruct_SFT_sciencev00.01 +tags: +- generated_from_trainer +- sft +- open-r1 +- trl +licence: license +--- + +# Model Card for Llama-3.1-8B-Instruct_SFT_sciencev00.01 + +This model is a fine-tuned version of [meta-llama/Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct) on the [Neelectric/MoT_science_Llama3_4096toks](https://huggingface.co/datasets/Neelectric/MoT_science_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/Llama-3.1-8B-Instruct_SFT_sciencev00.01", 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_science/runs/8jnqzhse) + + +This model was trained with SFT. + +### Framework versions + +- TRL: 0.28.0.dev0 +- Transformers: 4.57.6 +- Pytorch: 2.9.0 +- Datasets: 4.5.0 +- 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..9cd4898 --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 3.1829322264618205e+19, + "train_loss": 0.922418457405818, + "train_runtime": 33639.8393, + "train_samples": 145693, + "train_samples_per_second": 12.993, + "train_steps_per_second": 0.812 +} \ No newline at end of file diff --git a/chat_template.jinja b/chat_template.jinja new file mode 100644 index 0000000..0ab931a --- /dev/null +++ b/chat_template.jinja @@ -0,0 +1,121 @@ +{{- bos_token }} +{%- if custom_tools is defined %} + {%- set tools = custom_tools %} +{%- endif %} +{%- if not tools_in_user_message is defined %} + {%- set tools_in_user_message = true %} +{%- endif %} +{%- if not date_string is defined %} + {%- set date_string = "26 Jul 2024" %} +{%- endif %} +{%- if not tools is defined %} + {%- set tools = none %} +{%- endif %} + +{#- This block extracts the system message, so we can slot it into the right place. #} +{%- if messages[0]['role'] == 'system' %} + {%- set system_message = messages[0]['content']|trim %} + {%- set 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: +... + + +... +" %} +{%- endif %} + +{#- System message + builtin tools #} +{{- "<|start_header_id|>system<|end_header_id|>\n\n" }} +{%- if builtin_tools is defined or tools is not none %} + {{- "Environment: ipython\n" }} +{%- endif %} +{%- if builtin_tools is defined %} + {{- "Tools: " + builtin_tools | reject('equalto', 'code_interpreter') | join(", ") + "\n\n"}} +{%- endif %} +{{- "Cutting Knowledge Date: December 2023\n" }} +{{- "Today Date: " + date_string + "\n\n" }} +{%- if tools is not none and not tools_in_user_message %} + {{- "You have access to the following functions. To call a function, please respond with JSON for a function call." }} + {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }} + {{- "Do not use variables.\n\n" }} + {%- for t in tools %} + {{- t | tojson(indent=4) }} + {{- "\n\n" }} + {%- endfor %} +{%- endif %} +{{- system_message }} +{{- "<|eot_id|>" }} + +{#- Custom tools are passed in a user message with some extra guidance #} +{%- if tools_in_user_message and not tools is none %} + {#- Extract the first user message so we can plug it in here #} + {%- if messages | length != 0 %} + {%- set first_user_message = messages[0]['content']|trim %} + {%- set messages = messages[1:] %} + {%- else %} + {{- raise_exception("Cannot put tools in the first user message when there's no first user message!") }} + {%- endif %} + {{- '<|start_header_id|>user<|end_header_id|>\n\n' -}} + {{- "Given the following functions, please respond with a JSON for a function call " }} + {{- "with its proper arguments that best answers the given prompt.\n\n" }} + {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }} + {{- "Do not use variables.\n\n" }} + {%- for t in tools %} + {{- t | tojson(indent=4) }} + {{- "\n\n" }} + {%- endfor %} + {{- first_user_message + "<|eot_id|>"}} +{%- endif %} + +{%- for message in messages %} + {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %} + {%- if message['role'] == 'assistant' %} + {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }} + {% generation %} + {{- message['content'] | trim + '<|eot_id|>' }} + {% endgeneration %} + {%- else %} + {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' }} + {%- endif %} + {%- elif 'tool_calls' in message %} + {%- if not message.tool_calls|length == 1 %} + {{- raise_exception("This model only supports single tool-calls at once!") }} + {%- endif %} + {%- set tool_call = message.tool_calls[0].function %} + {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}} + {% generation %} + {%- if builtin_tools is defined and tool_call.name in builtin_tools %} + {{- "<|python_tag|>" + tool_call.name + ".call(" }} + {%- for arg_name, arg_val in tool_call.arguments | items %} + {{- arg_name + '="' + arg_val + '"' }} + {%- if not loop.last %} + {{- ", " }} + {%- endif %} + {%- endfor %} + {{- ")" }} + {%- else %} + {{- '{"name": "' + tool_call.name + '", ' }} + {{- '"parameters": ' }} + {{- tool_call.arguments | tojson }} + {{- "}" }} + {%- endif %} + {%- if builtin_tools is defined %} + {{- "<|eom_id|>" }} + {%- else %} + {{- "<|eot_id|>" }} + {%- endif %} + {% endgeneration %} + {%- elif message.role == "tool" or message.role == "ipython" %} + {{- "<|start_header_id|>ipython<|end_header_id|>\n\n" }} + {%- if message.content is mapping or message.content is iterable %} + {{- message.content | tojson }} + {%- else %} + {{- message.content }} + {%- endif %} + {{- "<|eot_id|>" }} + {%- endif %} +{%- endfor %} +{%- if add_generation_prompt %} + {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }} +{%- endif %} \ No newline at end of file diff --git a/config.json b/config.json new file mode 100644 index 0000000..e1d9068 --- /dev/null +++ b/config.json @@ -0,0 +1,35 @@ +{ + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "dtype": "bfloat16", + "eos_token_id": 128009, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 14336, + "max_position_embeddings": 131072, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 32, + "num_hidden_layers": 32, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "factor": 8.0, + "high_freq_factor": 4.0, + "low_freq_factor": 1.0, + "original_max_position_embeddings": 8192, + "rope_type": "llama3" + }, + "rope_theta": 500000.0, + "tie_word_embeddings": false, + "transformers_version": "4.57.6", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/generation_config.json b/generation_config.json new file mode 100644 index 0000000..1996dc1 --- /dev/null +++ b/generation_config.json @@ -0,0 +1,8 @@ +{ + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128009, + "temperature": 0.6, + "top_p": 0.9, + "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..17a627d --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:557a4e2edf9340d14bddf1146ceb30b8aad24e402775a77ec8f65e280ba5a258 +size 4976698672 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..f1d75b3 --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cabbd6bce574a880eb54da021a9a321c1701d08d9f753058b7294906030e15dc +size 4999802720 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..0e475fa --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10ed29c9e9e801c7350b3e1445c0455a0f50780517b37c4ddde629d4e265a4b8 +size 4915916176 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..63357c1 --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de30c1f59432c3f34af02ccc0f70ba546635da5e3956abbbed7c615dc1d0de15 +size 1168138808 diff --git a/model.safetensors.index.json b/model.safetensors.index.json new file mode 100644 index 0000000..5c64f1e --- /dev/null +++ b/model.safetensors.index.json @@ -0,0 +1,299 @@ +{ + "metadata": { + "total_parameters": 8030261248, + "total_size": 16060522496 + }, + "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-00002-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-00002-of-00004.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00002-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-00003-of-00004.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00003-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-00003-of-00004.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00003-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-00003-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-00004-of-00004.safetensors", + "model.layers.31.mlp.down_proj.weight": "model-00004-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-00004-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.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-00002-of-00004.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors", + "model.norm.weight": "model-00004-of-00004.safetensors" + } +} diff --git a/special_tokens_map.json b/special_tokens_map.json new file mode 100644 index 0000000..e8f05fa --- /dev/null +++ b/special_tokens_map.json @@ -0,0 +1,10 @@ +{ + "bos_token": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": "<|eot_id|>" +} diff --git a/tokenizer.json b/tokenizer.json new file mode 100644 index 0000000..1c1d8d5 --- /dev/null +++ b/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b +size 17209920 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000..8b0c7c1 --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,2062 @@ +{ + "added_tokens_decoder": { + "128000": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128001": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128002": { + "content": "<|reserved_special_token_0|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128003": { + "content": "<|reserved_special_token_1|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128004": { + "content": "<|finetune_right_pad_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128005": { + "content": "<|reserved_special_token_2|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128006": { + "content": "<|start_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128007": { + "content": "<|end_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128008": { + "content": "<|eom_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128009": { + "content": "<|eot_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128010": { + "content": "<|python_tag|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128011": { + "content": "<|reserved_special_token_3|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128012": { + "content": "<|reserved_special_token_4|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128013": { + "content": "<|reserved_special_token_5|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128014": { + "content": "<|reserved_special_token_6|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128015": { + "content": "<|reserved_special_token_7|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128016": { + "content": "<|reserved_special_token_8|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128017": { + "content": "<|reserved_special_token_9|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128018": { + "content": "<|reserved_special_token_10|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128019": { + "content": "<|reserved_special_token_11|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128020": { + "content": "<|reserved_special_token_12|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128021": { + "content": "<|reserved_special_token_13|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128022": { + "content": "<|reserved_special_token_14|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128023": { + "content": "<|reserved_special_token_15|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128024": { + "content": "<|reserved_special_token_16|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128025": { + "content": "<|reserved_special_token_17|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128026": { + "content": "<|reserved_special_token_18|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128027": { + "content": "<|reserved_special_token_19|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128028": { + "content": "<|reserved_special_token_20|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128029": { + "content": "<|reserved_special_token_21|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128030": { + "content": "<|reserved_special_token_22|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128031": { + "content": "<|reserved_special_token_23|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128032": { + "content": "<|reserved_special_token_24|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128033": { + "content": "<|reserved_special_token_25|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128034": { + "content": "<|reserved_special_token_26|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128035": { + "content": "<|reserved_special_token_27|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128036": { + "content": "<|reserved_special_token_28|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128037": { + "content": "<|reserved_special_token_29|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128038": { + "content": "<|reserved_special_token_30|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128039": { + "content": "<|reserved_special_token_31|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128040": { + "content": "<|reserved_special_token_32|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128041": { + "content": "<|reserved_special_token_33|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128042": { + "content": "<|reserved_special_token_34|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128043": { + "content": "<|reserved_special_token_35|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128044": { + "content": "<|reserved_special_token_36|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128045": { + "content": "<|reserved_special_token_37|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128046": { + "content": "<|reserved_special_token_38|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128047": { + "content": "<|reserved_special_token_39|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128048": { + "content": "<|reserved_special_token_40|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128049": { + "content": "<|reserved_special_token_41|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128050": { + "content": "<|reserved_special_token_42|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128051": { + "content": "<|reserved_special_token_43|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128052": { + "content": "<|reserved_special_token_44|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128053": { + "content": "<|reserved_special_token_45|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128054": { + "content": "<|reserved_special_token_46|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128055": { + "content": "<|reserved_special_token_47|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128056": { + "content": "<|reserved_special_token_48|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128057": { + "content": "<|reserved_special_token_49|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128058": { + "content": "<|reserved_special_token_50|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128059": { + "content": "<|reserved_special_token_51|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128060": { + "content": "<|reserved_special_token_52|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128061": { + "content": "<|reserved_special_token_53|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128062": { + "content": "<|reserved_special_token_54|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128063": { + "content": "<|reserved_special_token_55|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128064": { + "content": "<|reserved_special_token_56|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128065": { + "content": "<|reserved_special_token_57|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128066": { + "content": "<|reserved_special_token_58|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128067": { + "content": "<|reserved_special_token_59|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128068": { + "content": "<|reserved_special_token_60|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128069": { + "content": "<|reserved_special_token_61|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128070": { + "content": "<|reserved_special_token_62|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128071": { + "content": "<|reserved_special_token_63|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128072": { + "content": "<|reserved_special_token_64|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128073": { + "content": "<|reserved_special_token_65|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128074": { + "content": "<|reserved_special_token_66|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128075": { + "content": "<|reserved_special_token_67|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128076": { + "content": "<|reserved_special_token_68|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128077": { + "content": "<|reserved_special_token_69|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128078": { + "content": "<|reserved_special_token_70|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128079": { + "content": "<|reserved_special_token_71|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128080": { + "content": "<|reserved_special_token_72|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128081": { + "content": "<|reserved_special_token_73|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128082": { + "content": "<|reserved_special_token_74|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128083": { + "content": "<|reserved_special_token_75|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128084": { + "content": "<|reserved_special_token_76|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128085": { + "content": "<|reserved_special_token_77|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128086": { + "content": "<|reserved_special_token_78|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128087": { + "content": "<|reserved_special_token_79|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128088": { + "content": "<|reserved_special_token_80|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128089": { + "content": "<|reserved_special_token_81|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128090": { + "content": "<|reserved_special_token_82|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128091": { + "content": "<|reserved_special_token_83|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128092": { + "content": "<|reserved_special_token_84|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128093": { + "content": "<|reserved_special_token_85|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128094": { + "content": "<|reserved_special_token_86|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128095": { + "content": "<|reserved_special_token_87|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128096": { + "content": "<|reserved_special_token_88|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128097": { + "content": "<|reserved_special_token_89|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128098": { + "content": "<|reserved_special_token_90|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128099": { + "content": "<|reserved_special_token_91|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128100": { + "content": "<|reserved_special_token_92|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128101": { + "content": "<|reserved_special_token_93|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128102": { + "content": "<|reserved_special_token_94|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128103": { + "content": "<|reserved_special_token_95|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128104": { + "content": "<|reserved_special_token_96|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128105": { + "content": "<|reserved_special_token_97|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128106": { + "content": "<|reserved_special_token_98|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128107": { + "content": "<|reserved_special_token_99|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128108": { + "content": "<|reserved_special_token_100|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128109": { + "content": "<|reserved_special_token_101|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128110": { + "content": "<|reserved_special_token_102|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128111": { + "content": "<|reserved_special_token_103|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128112": { + "content": "<|reserved_special_token_104|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128113": { + "content": "<|reserved_special_token_105|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128114": { + "content": "<|reserved_special_token_106|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128115": { + "content": "<|reserved_special_token_107|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128116": { + "content": "<|reserved_special_token_108|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128117": { + "content": "<|reserved_special_token_109|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128118": { + "content": "<|reserved_special_token_110|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128119": { + "content": "<|reserved_special_token_111|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128120": { + "content": "<|reserved_special_token_112|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128121": { + "content": "<|reserved_special_token_113|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128122": { + "content": "<|reserved_special_token_114|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128123": { + "content": "<|reserved_special_token_115|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128124": { + "content": "<|reserved_special_token_116|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128125": { + "content": "<|reserved_special_token_117|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128126": { + "content": "<|reserved_special_token_118|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128127": { + "content": "<|reserved_special_token_119|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128128": { + "content": "<|reserved_special_token_120|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128129": { + "content": "<|reserved_special_token_121|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128130": { + "content": "<|reserved_special_token_122|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128131": { + "content": "<|reserved_special_token_123|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128132": { + "content": "<|reserved_special_token_124|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128133": { + "content": "<|reserved_special_token_125|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128134": { + "content": "<|reserved_special_token_126|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128135": { + "content": "<|reserved_special_token_127|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128136": { + "content": "<|reserved_special_token_128|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128137": { + "content": "<|reserved_special_token_129|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128138": { + "content": "<|reserved_special_token_130|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128139": { + "content": "<|reserved_special_token_131|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128140": { + "content": "<|reserved_special_token_132|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128141": { + "content": "<|reserved_special_token_133|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128142": { + "content": "<|reserved_special_token_134|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128143": { + "content": "<|reserved_special_token_135|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128144": { + "content": "<|reserved_special_token_136|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128145": { + "content": "<|reserved_special_token_137|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128146": { + "content": "<|reserved_special_token_138|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128147": { + "content": "<|reserved_special_token_139|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128148": { + "content": "<|reserved_special_token_140|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128149": { + "content": "<|reserved_special_token_141|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128150": { + "content": "<|reserved_special_token_142|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128151": { + "content": "<|reserved_special_token_143|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128152": { + "content": "<|reserved_special_token_144|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128153": { + "content": "<|reserved_special_token_145|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128154": { + "content": "<|reserved_special_token_146|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128155": { + "content": "<|reserved_special_token_147|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128156": { + "content": "<|reserved_special_token_148|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128157": { + "content": "<|reserved_special_token_149|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128158": { + "content": "<|reserved_special_token_150|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128159": { + "content": "<|reserved_special_token_151|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128160": { + "content": "<|reserved_special_token_152|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128161": { + "content": "<|reserved_special_token_153|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128162": { + "content": "<|reserved_special_token_154|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128163": { + "content": "<|reserved_special_token_155|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128164": { + "content": "<|reserved_special_token_156|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128165": { + "content": "<|reserved_special_token_157|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128166": { + "content": "<|reserved_special_token_158|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128167": { + "content": "<|reserved_special_token_159|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128168": { + "content": "<|reserved_special_token_160|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128169": { + "content": "<|reserved_special_token_161|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128170": { + "content": "<|reserved_special_token_162|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128171": { + "content": "<|reserved_special_token_163|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128172": { + "content": "<|reserved_special_token_164|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128173": { + "content": "<|reserved_special_token_165|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128174": { + "content": "<|reserved_special_token_166|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128175": { + "content": "<|reserved_special_token_167|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128176": { + "content": "<|reserved_special_token_168|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128177": { + "content": "<|reserved_special_token_169|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128178": { + "content": "<|reserved_special_token_170|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128179": { + "content": "<|reserved_special_token_171|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128180": { + "content": "<|reserved_special_token_172|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128181": { + "content": "<|reserved_special_token_173|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128182": { + "content": "<|reserved_special_token_174|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128183": { + "content": "<|reserved_special_token_175|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128184": { + "content": "<|reserved_special_token_176|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128185": { + "content": "<|reserved_special_token_177|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128186": { + "content": "<|reserved_special_token_178|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128187": { + "content": "<|reserved_special_token_179|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128188": { + "content": "<|reserved_special_token_180|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128189": { + "content": "<|reserved_special_token_181|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128190": { + "content": "<|reserved_special_token_182|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128191": { + "content": "<|reserved_special_token_183|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128192": { + "content": "<|reserved_special_token_184|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128193": { + "content": "<|reserved_special_token_185|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128194": { + "content": "<|reserved_special_token_186|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128195": { + "content": "<|reserved_special_token_187|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128196": { + "content": "<|reserved_special_token_188|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128197": { + "content": "<|reserved_special_token_189|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128198": { + "content": "<|reserved_special_token_190|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128199": { + "content": "<|reserved_special_token_191|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128200": { + "content": "<|reserved_special_token_192|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128201": { + "content": "<|reserved_special_token_193|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128202": { + "content": "<|reserved_special_token_194|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128203": { + "content": "<|reserved_special_token_195|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128204": { + "content": "<|reserved_special_token_196|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128205": { + "content": "<|reserved_special_token_197|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128206": { + "content": "<|reserved_special_token_198|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128207": { + "content": "<|reserved_special_token_199|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128208": { + "content": "<|reserved_special_token_200|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128209": { + "content": "<|reserved_special_token_201|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128210": { + "content": "<|reserved_special_token_202|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128211": { + "content": "<|reserved_special_token_203|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128212": { + "content": "<|reserved_special_token_204|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128213": { + "content": "<|reserved_special_token_205|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128214": { + "content": "<|reserved_special_token_206|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128215": { + "content": "<|reserved_special_token_207|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128216": { + "content": "<|reserved_special_token_208|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128217": { + "content": "<|reserved_special_token_209|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128218": { + "content": "<|reserved_special_token_210|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128219": { + "content": "<|reserved_special_token_211|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128220": { + "content": "<|reserved_special_token_212|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128221": { + "content": "<|reserved_special_token_213|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128222": { + "content": "<|reserved_special_token_214|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128223": { + "content": "<|reserved_special_token_215|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128224": { + "content": "<|reserved_special_token_216|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128225": { + "content": "<|reserved_special_token_217|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128226": { + "content": "<|reserved_special_token_218|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128227": { + "content": "<|reserved_special_token_219|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128228": { + "content": "<|reserved_special_token_220|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128229": { + "content": "<|reserved_special_token_221|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128230": { + "content": "<|reserved_special_token_222|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128231": { + "content": "<|reserved_special_token_223|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128232": { + "content": "<|reserved_special_token_224|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128233": { + "content": "<|reserved_special_token_225|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128234": { + "content": "<|reserved_special_token_226|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128235": { + "content": "<|reserved_special_token_227|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128236": { + "content": "<|reserved_special_token_228|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128237": { + "content": "<|reserved_special_token_229|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128238": { + "content": "<|reserved_special_token_230|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128239": { + "content": "<|reserved_special_token_231|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128240": { + "content": "<|reserved_special_token_232|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128241": { + "content": "<|reserved_special_token_233|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128242": { + "content": "<|reserved_special_token_234|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128243": { + "content": "<|reserved_special_token_235|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128244": { + "content": "<|reserved_special_token_236|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128245": { + "content": "<|reserved_special_token_237|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128246": { + "content": "<|reserved_special_token_238|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128247": { + "content": "<|reserved_special_token_239|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128248": { + "content": "<|reserved_special_token_240|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128249": { + "content": "<|reserved_special_token_241|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128250": { + "content": "<|reserved_special_token_242|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128251": { + "content": "<|reserved_special_token_243|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128252": { + "content": "<|reserved_special_token_244|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128253": { + "content": "<|reserved_special_token_245|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128254": { + "content": "<|reserved_special_token_246|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128255": { + "content": "<|reserved_special_token_247|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "<|begin_of_text|>", + "clean_up_tokenization_spaces": true, + "eos_token": "<|eot_id|>", + "extra_special_tokens": {}, + "model_input_names": [ + "input_ids", + "attention_mask" + ], + "model_max_length": 131072, + "tokenizer_class": "PreTrainedTokenizerFast" +} diff --git a/train_results.json b/train_results.json new file mode 100644 index 0000000..9cd4898 --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 3.1829322264618205e+19, + "train_loss": 0.922418457405818, + "train_runtime": 33639.8393, + "train_samples": 145693, + "train_samples_per_second": 12.993, + "train_steps_per_second": 0.812 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..36e5613 --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,245905 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.0, + "eval_steps": 500, + "global_step": 27318, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.00010981770261366133, + "grad_norm": 6.513667106628418, + "learning_rate": 0.0, + "loss": 1.5235, + "mean_token_accuracy": 0.6157128810882568, + "num_tokens": 30018.0, + "step": 1 + }, + { + "epoch": 0.00021963540522732265, + "grad_norm": 6.408206939697266, + "learning_rate": 3.6603221083455343e-10, + "loss": 1.4062, + "mean_token_accuracy": 0.628243088722229, + "num_tokens": 58744.0, + "step": 2 + }, + { + "epoch": 0.000329453107840984, + "grad_norm": 7.284219741821289, + "learning_rate": 7.320644216691069e-10, + "loss": 1.4743, + "mean_token_accuracy": 0.6152434945106506, + "num_tokens": 84293.0, + "step": 3 + }, + { + "epoch": 0.0004392708104546453, + "grad_norm": 7.03470516204834, + "learning_rate": 1.0980966325036603e-09, + "loss": 1.5504, + "mean_token_accuracy": 0.5983664989471436, + "num_tokens": 113639.0, + "step": 4 + }, + { + "epoch": 0.0005490885130683066, + "grad_norm": 7.088189601898193, + "learning_rate": 1.4641288433382137e-09, + "loss": 1.515, + "mean_token_accuracy": 0.6073604822158813, + "num_tokens": 139615.0, + "step": 5 + }, + { + "epoch": 0.000658906215681968, + "grad_norm": 6.859297752380371, + "learning_rate": 1.8301610541727671e-09, + "loss": 1.6127, + "mean_token_accuracy": 0.5814723372459412, + "num_tokens": 168855.0, + "step": 6 + }, + { + "epoch": 0.0007687239182956292, + "grad_norm": 7.64864444732666, + "learning_rate": 2.1961932650073206e-09, + "loss": 1.5226, + "mean_token_accuracy": 0.6033573150634766, + "num_tokens": 192650.0, + "step": 7 + }, + { + "epoch": 0.0008785416209092906, + "grad_norm": 6.30922794342041, + "learning_rate": 2.562225475841874e-09, + "loss": 1.5185, + "mean_token_accuracy": 0.6044977903366089, + "num_tokens": 225812.0, + "step": 8 + }, + { + "epoch": 0.000988359323522952, + "grad_norm": 8.92734146118164, + "learning_rate": 2.9282576866764274e-09, + "loss": 1.5179, + "mean_token_accuracy": 0.6050848960876465, + "num_tokens": 245223.0, + "step": 9 + }, + { + "epoch": 0.0010981770261366132, + "grad_norm": 6.8748908042907715, + "learning_rate": 3.294289897510981e-09, + "loss": 1.5658, + "mean_token_accuracy": 0.5966095924377441, + "num_tokens": 273619.0, + "step": 10 + }, + { + "epoch": 0.0012079947287502745, + "grad_norm": 7.1734089851379395, + "learning_rate": 3.6603221083455343e-09, + "loss": 1.3784, + "mean_token_accuracy": 0.6416056752204895, + "num_tokens": 297418.0, + "step": 11 + }, + { + "epoch": 0.001317812431363936, + "grad_norm": 6.0631632804870605, + "learning_rate": 4.026354319180088e-09, + "loss": 1.3931, + "mean_token_accuracy": 0.6354196667671204, + "num_tokens": 328966.0, + "step": 12 + }, + { + "epoch": 0.0014276301339775973, + "grad_norm": 7.570906162261963, + "learning_rate": 4.392386530014641e-09, + "loss": 1.3888, + "mean_token_accuracy": 0.6385470032691956, + "num_tokens": 353003.0, + "step": 13 + }, + { + "epoch": 0.0015374478365912585, + "grad_norm": 7.241986274719238, + "learning_rate": 4.758418740849194e-09, + "loss": 1.4651, + "mean_token_accuracy": 0.6214442849159241, + "num_tokens": 377957.0, + "step": 14 + }, + { + "epoch": 0.0016472655392049198, + "grad_norm": 8.12817096710205, + "learning_rate": 5.124450951683748e-09, + "loss": 1.4742, + "mean_token_accuracy": 0.6148027181625366, + "num_tokens": 398957.0, + "step": 15 + }, + { + "epoch": 0.0017570832418185812, + "grad_norm": 7.185974597930908, + "learning_rate": 5.490483162518302e-09, + "loss": 1.5622, + "mean_token_accuracy": 0.5949515104293823, + "num_tokens": 424867.0, + "step": 16 + }, + { + "epoch": 0.0018669009444322424, + "grad_norm": 6.976659774780273, + "learning_rate": 5.856515373352855e-09, + "loss": 1.4467, + "mean_token_accuracy": 0.6199159026145935, + "num_tokens": 450753.0, + "step": 17 + }, + { + "epoch": 0.001976718647045904, + "grad_norm": 6.868077754974365, + "learning_rate": 6.222547584187408e-09, + "loss": 1.4528, + "mean_token_accuracy": 0.6171982884407043, + "num_tokens": 478721.0, + "step": 18 + }, + { + "epoch": 0.002086536349659565, + "grad_norm": 8.583649635314941, + "learning_rate": 6.588579795021962e-09, + "loss": 1.5912, + "mean_token_accuracy": 0.5928547978401184, + "num_tokens": 499326.0, + "step": 19 + }, + { + "epoch": 0.0021963540522732263, + "grad_norm": 7.625557899475098, + "learning_rate": 6.954612005856515e-09, + "loss": 1.4427, + "mean_token_accuracy": 0.6247740983963013, + "num_tokens": 522720.0, + "step": 20 + }, + { + "epoch": 0.002306171754886888, + "grad_norm": 6.300516605377197, + "learning_rate": 7.3206442166910685e-09, + "loss": 1.4216, + "mean_token_accuracy": 0.6313163042068481, + "num_tokens": 551422.0, + "step": 21 + }, + { + "epoch": 0.002415989457500549, + "grad_norm": 8.081571578979492, + "learning_rate": 7.686676427525622e-09, + "loss": 1.5622, + "mean_token_accuracy": 0.6011921167373657, + "num_tokens": 572912.0, + "step": 22 + }, + { + "epoch": 0.0025258071601142102, + "grad_norm": 5.994649410247803, + "learning_rate": 8.052708638360176e-09, + "loss": 1.411, + "mean_token_accuracy": 0.6290097236633301, + "num_tokens": 604172.0, + "step": 23 + }, + { + "epoch": 0.002635624862727872, + "grad_norm": 7.52415132522583, + "learning_rate": 8.418740849194728e-09, + "loss": 1.4469, + "mean_token_accuracy": 0.6210207939147949, + "num_tokens": 627461.0, + "step": 24 + }, + { + "epoch": 0.002745442565341533, + "grad_norm": 6.378818511962891, + "learning_rate": 8.784773060029282e-09, + "loss": 1.4482, + "mean_token_accuracy": 0.627587080001831, + "num_tokens": 657516.0, + "step": 25 + }, + { + "epoch": 0.0028552602679551946, + "grad_norm": 6.290201663970947, + "learning_rate": 9.150805270863836e-09, + "loss": 1.5437, + "mean_token_accuracy": 0.5954060554504395, + "num_tokens": 691937.0, + "step": 26 + }, + { + "epoch": 0.0029650779705688557, + "grad_norm": 8.45878791809082, + "learning_rate": 9.516837481698389e-09, + "loss": 1.4643, + "mean_token_accuracy": 0.6265501379966736, + "num_tokens": 711900.0, + "step": 27 + }, + { + "epoch": 0.003074895673182517, + "grad_norm": 7.246087074279785, + "learning_rate": 9.882869692532943e-09, + "loss": 1.4915, + "mean_token_accuracy": 0.6122739315032959, + "num_tokens": 737085.0, + "step": 28 + }, + { + "epoch": 0.0031847133757961785, + "grad_norm": 8.06440544128418, + "learning_rate": 1.0248901903367497e-08, + "loss": 1.5752, + "mean_token_accuracy": 0.602922797203064, + "num_tokens": 759817.0, + "step": 29 + }, + { + "epoch": 0.0032945310784098397, + "grad_norm": 7.5938591957092285, + "learning_rate": 1.0614934114202049e-08, + "loss": 1.4186, + "mean_token_accuracy": 0.6260017156600952, + "num_tokens": 783353.0, + "step": 30 + }, + { + "epoch": 0.003404348781023501, + "grad_norm": 8.365952491760254, + "learning_rate": 1.0980966325036603e-08, + "loss": 1.4707, + "mean_token_accuracy": 0.6186453700065613, + "num_tokens": 804197.0, + "step": 31 + }, + { + "epoch": 0.0035141664836371624, + "grad_norm": 6.885355472564697, + "learning_rate": 1.1346998535871157e-08, + "loss": 1.5128, + "mean_token_accuracy": 0.6067238450050354, + "num_tokens": 830796.0, + "step": 32 + }, + { + "epoch": 0.0036239841862508236, + "grad_norm": 6.580489635467529, + "learning_rate": 1.171303074670571e-08, + "loss": 1.4161, + "mean_token_accuracy": 0.640745222568512, + "num_tokens": 858092.0, + "step": 33 + }, + { + "epoch": 0.0037338018888644848, + "grad_norm": 7.3038482666015625, + "learning_rate": 1.2079062957540264e-08, + "loss": 1.4552, + "mean_token_accuracy": 0.6134449243545532, + "num_tokens": 882474.0, + "step": 34 + }, + { + "epoch": 0.0038436195914781464, + "grad_norm": 8.551104545593262, + "learning_rate": 1.2445095168374816e-08, + "loss": 1.4569, + "mean_token_accuracy": 0.6246424317359924, + "num_tokens": 902361.0, + "step": 35 + }, + { + "epoch": 0.003953437294091808, + "grad_norm": 7.468267440795898, + "learning_rate": 1.281112737920937e-08, + "loss": 1.5078, + "mean_token_accuracy": 0.6057623624801636, + "num_tokens": 927171.0, + "step": 36 + }, + { + "epoch": 0.004063254996705469, + "grad_norm": 7.075837135314941, + "learning_rate": 1.3177159590043924e-08, + "loss": 1.472, + "mean_token_accuracy": 0.6128988265991211, + "num_tokens": 952048.0, + "step": 37 + }, + { + "epoch": 0.00417307269931913, + "grad_norm": 6.8620171546936035, + "learning_rate": 1.3543191800878477e-08, + "loss": 1.5741, + "mean_token_accuracy": 0.596381425857544, + "num_tokens": 981040.0, + "step": 38 + }, + { + "epoch": 0.004282890401932792, + "grad_norm": 6.912696838378906, + "learning_rate": 1.390922401171303e-08, + "loss": 1.4835, + "mean_token_accuracy": 0.6071129441261292, + "num_tokens": 1008756.0, + "step": 39 + }, + { + "epoch": 0.004392708104546453, + "grad_norm": 7.114793300628662, + "learning_rate": 1.4275256222547585e-08, + "loss": 1.4981, + "mean_token_accuracy": 0.6013438105583191, + "num_tokens": 1035140.0, + "step": 40 + }, + { + "epoch": 0.004502525807160114, + "grad_norm": 7.799258708953857, + "learning_rate": 1.4641288433382137e-08, + "loss": 1.5249, + "mean_token_accuracy": 0.6023052930831909, + "num_tokens": 1058792.0, + "step": 41 + }, + { + "epoch": 0.004612343509773776, + "grad_norm": 5.6752214431762695, + "learning_rate": 1.500732064421669e-08, + "loss": 1.4631, + "mean_token_accuracy": 0.6143511533737183, + "num_tokens": 1096383.0, + "step": 42 + }, + { + "epoch": 0.0047221612123874365, + "grad_norm": 6.7811079025268555, + "learning_rate": 1.5373352855051243e-08, + "loss": 1.5742, + "mean_token_accuracy": 0.5920353531837463, + "num_tokens": 1125313.0, + "step": 43 + }, + { + "epoch": 0.004831978915001098, + "grad_norm": 6.306768894195557, + "learning_rate": 1.57393850658858e-08, + "loss": 1.4455, + "mean_token_accuracy": 0.6223504543304443, + "num_tokens": 1155650.0, + "step": 44 + }, + { + "epoch": 0.00494179661761476, + "grad_norm": 7.594419479370117, + "learning_rate": 1.610541727672035e-08, + "loss": 1.4677, + "mean_token_accuracy": 0.6196602582931519, + "num_tokens": 1178843.0, + "step": 45 + }, + { + "epoch": 0.0050516143202284204, + "grad_norm": 6.937159061431885, + "learning_rate": 1.6471449487554904e-08, + "loss": 1.5241, + "mean_token_accuracy": 0.6023076772689819, + "num_tokens": 1206271.0, + "step": 46 + }, + { + "epoch": 0.005161432022842082, + "grad_norm": 5.846798419952393, + "learning_rate": 1.6837481698389456e-08, + "loss": 1.4705, + "mean_token_accuracy": 0.6165028810501099, + "num_tokens": 1241401.0, + "step": 47 + }, + { + "epoch": 0.005271249725455744, + "grad_norm": 7.201146602630615, + "learning_rate": 1.7203513909224012e-08, + "loss": 1.4964, + "mean_token_accuracy": 0.6110290884971619, + "num_tokens": 1267455.0, + "step": 48 + }, + { + "epoch": 0.005381067428069404, + "grad_norm": 6.806131839752197, + "learning_rate": 1.7569546120058564e-08, + "loss": 1.3771, + "mean_token_accuracy": 0.636490523815155, + "num_tokens": 1293955.0, + "step": 49 + }, + { + "epoch": 0.005490885130683066, + "grad_norm": 7.764354228973389, + "learning_rate": 1.7935578330893117e-08, + "loss": 1.4395, + "mean_token_accuracy": 0.6282283067703247, + "num_tokens": 1316419.0, + "step": 50 + }, + { + "epoch": 0.0056007028332967276, + "grad_norm": 7.795925617218018, + "learning_rate": 1.8301610541727673e-08, + "loss": 1.5094, + "mean_token_accuracy": 0.6069018244743347, + "num_tokens": 1338607.0, + "step": 51 + }, + { + "epoch": 0.005710520535910389, + "grad_norm": 6.920229911804199, + "learning_rate": 1.8667642752562225e-08, + "loss": 1.4856, + "mean_token_accuracy": 0.6095184087753296, + "num_tokens": 1366081.0, + "step": 52 + }, + { + "epoch": 0.00582033823852405, + "grad_norm": 7.4536662101745605, + "learning_rate": 1.9033674963396777e-08, + "loss": 1.4827, + "mean_token_accuracy": 0.614389955997467, + "num_tokens": 1390090.0, + "step": 53 + }, + { + "epoch": 0.0059301559411377115, + "grad_norm": 7.532850742340088, + "learning_rate": 1.9399707174231333e-08, + "loss": 1.4923, + "mean_token_accuracy": 0.6115027666091919, + "num_tokens": 1413964.0, + "step": 54 + }, + { + "epoch": 0.006039973643751373, + "grad_norm": 6.608367919921875, + "learning_rate": 1.9765739385065885e-08, + "loss": 1.515, + "mean_token_accuracy": 0.6088537573814392, + "num_tokens": 1442849.0, + "step": 55 + }, + { + "epoch": 0.006149791346365034, + "grad_norm": 6.2248101234436035, + "learning_rate": 2.0131771595900438e-08, + "loss": 1.5721, + "mean_token_accuracy": 0.5892795920372009, + "num_tokens": 1478493.0, + "step": 56 + }, + { + "epoch": 0.006259609048978695, + "grad_norm": 8.224326133728027, + "learning_rate": 2.0497803806734994e-08, + "loss": 1.525, + "mean_token_accuracy": 0.6162484288215637, + "num_tokens": 1498811.0, + "step": 57 + }, + { + "epoch": 0.006369426751592357, + "grad_norm": 6.8067946434021, + "learning_rate": 2.0863836017569546e-08, + "loss": 1.4464, + "mean_token_accuracy": 0.6250715255737305, + "num_tokens": 1525932.0, + "step": 58 + }, + { + "epoch": 0.006479244454206018, + "grad_norm": 7.379829406738281, + "learning_rate": 2.1229868228404098e-08, + "loss": 1.4907, + "mean_token_accuracy": 0.6096035838127136, + "num_tokens": 1550438.0, + "step": 59 + }, + { + "epoch": 0.006589062156819679, + "grad_norm": 8.665096282958984, + "learning_rate": 2.1595900439238654e-08, + "loss": 1.5014, + "mean_token_accuracy": 0.6132106781005859, + "num_tokens": 1569961.0, + "step": 60 + }, + { + "epoch": 0.006698879859433341, + "grad_norm": 6.071230411529541, + "learning_rate": 2.1961932650073206e-08, + "loss": 1.4964, + "mean_token_accuracy": 0.6062522530555725, + "num_tokens": 1604631.0, + "step": 61 + }, + { + "epoch": 0.006808697562047002, + "grad_norm": 7.669257640838623, + "learning_rate": 2.232796486090776e-08, + "loss": 1.4848, + "mean_token_accuracy": 0.6141104102134705, + "num_tokens": 1628098.0, + "step": 62 + }, + { + "epoch": 0.006918515264660663, + "grad_norm": 6.652416229248047, + "learning_rate": 2.2693997071742314e-08, + "loss": 1.48, + "mean_token_accuracy": 0.6208204030990601, + "num_tokens": 1656281.0, + "step": 63 + }, + { + "epoch": 0.007028332967274325, + "grad_norm": 7.4147491455078125, + "learning_rate": 2.3060029282576867e-08, + "loss": 1.5126, + "mean_token_accuracy": 0.6161652207374573, + "num_tokens": 1680754.0, + "step": 64 + }, + { + "epoch": 0.007138150669887986, + "grad_norm": 7.650397300720215, + "learning_rate": 2.342606149341142e-08, + "loss": 1.5118, + "mean_token_accuracy": 0.6084631681442261, + "num_tokens": 1704446.0, + "step": 65 + }, + { + "epoch": 0.007247968372501647, + "grad_norm": 6.903392791748047, + "learning_rate": 2.3792093704245972e-08, + "loss": 1.2964, + "mean_token_accuracy": 0.6483873128890991, + "num_tokens": 1728711.0, + "step": 66 + }, + { + "epoch": 0.007357786075115309, + "grad_norm": 7.930106163024902, + "learning_rate": 2.4158125915080527e-08, + "loss": 1.4215, + "mean_token_accuracy": 0.6310405731201172, + "num_tokens": 1750331.0, + "step": 67 + }, + { + "epoch": 0.0074676037777289695, + "grad_norm": 6.341794013977051, + "learning_rate": 2.452415812591508e-08, + "loss": 1.4364, + "mean_token_accuracy": 0.6183528900146484, + "num_tokens": 1781737.0, + "step": 68 + }, + { + "epoch": 0.007577421480342631, + "grad_norm": 6.564031600952148, + "learning_rate": 2.4890190336749632e-08, + "loss": 1.4812, + "mean_token_accuracy": 0.6119531393051147, + "num_tokens": 1811021.0, + "step": 69 + }, + { + "epoch": 0.007687239182956293, + "grad_norm": 7.517025470733643, + "learning_rate": 2.5256222547584188e-08, + "loss": 1.5342, + "mean_token_accuracy": 0.5973517894744873, + "num_tokens": 1835564.0, + "step": 70 + }, + { + "epoch": 0.007797056885569954, + "grad_norm": 10.665511131286621, + "learning_rate": 2.562225475841874e-08, + "loss": 1.5249, + "mean_token_accuracy": 0.613762617111206, + "num_tokens": 1850951.0, + "step": 71 + }, + { + "epoch": 0.007906874588183616, + "grad_norm": 7.102219581604004, + "learning_rate": 2.5988286969253293e-08, + "loss": 1.5369, + "mean_token_accuracy": 0.6028988361358643, + "num_tokens": 1877690.0, + "step": 72 + }, + { + "epoch": 0.008016692290797276, + "grad_norm": 7.003612041473389, + "learning_rate": 2.6354319180087848e-08, + "loss": 1.5771, + "mean_token_accuracy": 0.592017412185669, + "num_tokens": 1904455.0, + "step": 73 + }, + { + "epoch": 0.008126509993410937, + "grad_norm": 7.816945552825928, + "learning_rate": 2.67203513909224e-08, + "loss": 1.608, + "mean_token_accuracy": 0.5886359810829163, + "num_tokens": 1928049.0, + "step": 74 + }, + { + "epoch": 0.008236327696024599, + "grad_norm": 8.1840181350708, + "learning_rate": 2.7086383601756953e-08, + "loss": 1.4994, + "mean_token_accuracy": 0.61424320936203, + "num_tokens": 1948422.0, + "step": 75 + }, + { + "epoch": 0.00834614539863826, + "grad_norm": 7.051959037780762, + "learning_rate": 2.745241581259151e-08, + "loss": 1.4679, + "mean_token_accuracy": 0.619006872177124, + "num_tokens": 1974329.0, + "step": 76 + }, + { + "epoch": 0.008455963101251922, + "grad_norm": 6.84087610244751, + "learning_rate": 2.781844802342606e-08, + "loss": 1.4233, + "mean_token_accuracy": 0.630859375, + "num_tokens": 2000531.0, + "step": 77 + }, + { + "epoch": 0.008565780803865584, + "grad_norm": 7.152336597442627, + "learning_rate": 2.8184480234260614e-08, + "loss": 1.4783, + "mean_token_accuracy": 0.6134531497955322, + "num_tokens": 2023823.0, + "step": 78 + }, + { + "epoch": 0.008675598506479244, + "grad_norm": 7.374361515045166, + "learning_rate": 2.855051244509517e-08, + "loss": 1.5135, + "mean_token_accuracy": 0.6030427813529968, + "num_tokens": 2048734.0, + "step": 79 + }, + { + "epoch": 0.008785416209092905, + "grad_norm": 6.683512210845947, + "learning_rate": 2.8916544655929722e-08, + "loss": 1.4143, + "mean_token_accuracy": 0.6265289783477783, + "num_tokens": 2075741.0, + "step": 80 + }, + { + "epoch": 0.008895233911706567, + "grad_norm": 7.396485328674316, + "learning_rate": 2.9282576866764274e-08, + "loss": 1.5167, + "mean_token_accuracy": 0.6025339365005493, + "num_tokens": 2100247.0, + "step": 81 + }, + { + "epoch": 0.009005051614320228, + "grad_norm": 7.769872188568115, + "learning_rate": 2.964860907759883e-08, + "loss": 1.5996, + "mean_token_accuracy": 0.6024358868598938, + "num_tokens": 2122939.0, + "step": 82 + }, + { + "epoch": 0.00911486931693389, + "grad_norm": 8.888970375061035, + "learning_rate": 3.001464128843338e-08, + "loss": 1.4996, + "mean_token_accuracy": 0.6033248901367188, + "num_tokens": 2141424.0, + "step": 83 + }, + { + "epoch": 0.009224687019547552, + "grad_norm": 6.2641921043396, + "learning_rate": 3.038067349926794e-08, + "loss": 1.398, + "mean_token_accuracy": 0.6277154684066772, + "num_tokens": 2171269.0, + "step": 84 + }, + { + "epoch": 0.009334504722161213, + "grad_norm": 6.798359394073486, + "learning_rate": 3.074670571010249e-08, + "loss": 1.3789, + "mean_token_accuracy": 0.6455862522125244, + "num_tokens": 2195892.0, + "step": 85 + }, + { + "epoch": 0.009444322424774873, + "grad_norm": 6.117819309234619, + "learning_rate": 3.111273792093704e-08, + "loss": 1.4785, + "mean_token_accuracy": 0.6196902394294739, + "num_tokens": 2227992.0, + "step": 86 + }, + { + "epoch": 0.009554140127388535, + "grad_norm": 6.5824995040893555, + "learning_rate": 3.14787701317716e-08, + "loss": 1.5353, + "mean_token_accuracy": 0.6106098890304565, + "num_tokens": 2256995.0, + "step": 87 + }, + { + "epoch": 0.009663957830002196, + "grad_norm": 6.198845386505127, + "learning_rate": 3.184480234260615e-08, + "loss": 1.4481, + "mean_token_accuracy": 0.6219071745872498, + "num_tokens": 2286492.0, + "step": 88 + }, + { + "epoch": 0.009773775532615858, + "grad_norm": 7.748722076416016, + "learning_rate": 3.22108345534407e-08, + "loss": 1.5044, + "mean_token_accuracy": 0.6075475215911865, + "num_tokens": 2309004.0, + "step": 89 + }, + { + "epoch": 0.00988359323522952, + "grad_norm": 6.335278511047363, + "learning_rate": 3.257686676427526e-08, + "loss": 1.4033, + "mean_token_accuracy": 0.6211801767349243, + "num_tokens": 2336480.0, + "step": 90 + }, + { + "epoch": 0.009993410937843181, + "grad_norm": 6.656136989593506, + "learning_rate": 3.294289897510981e-08, + "loss": 1.4479, + "mean_token_accuracy": 0.6151189208030701, + "num_tokens": 2363526.0, + "step": 91 + }, + { + "epoch": 0.010103228640456841, + "grad_norm": 6.249757766723633, + "learning_rate": 3.3308931185944364e-08, + "loss": 1.3985, + "mean_token_accuracy": 0.6286534070968628, + "num_tokens": 2393847.0, + "step": 92 + }, + { + "epoch": 0.010213046343070502, + "grad_norm": 6.646218776702881, + "learning_rate": 3.367496339677891e-08, + "loss": 1.4983, + "mean_token_accuracy": 0.6057878136634827, + "num_tokens": 2421639.0, + "step": 93 + }, + { + "epoch": 0.010322864045684164, + "grad_norm": 7.011570453643799, + "learning_rate": 3.404099560761347e-08, + "loss": 1.4514, + "mean_token_accuracy": 0.6237108707427979, + "num_tokens": 2445896.0, + "step": 94 + }, + { + "epoch": 0.010432681748297826, + "grad_norm": 6.568840980529785, + "learning_rate": 3.4407027818448024e-08, + "loss": 1.5576, + "mean_token_accuracy": 0.6004482507705688, + "num_tokens": 2475814.0, + "step": 95 + }, + { + "epoch": 0.010542499450911487, + "grad_norm": 7.219907283782959, + "learning_rate": 3.477306002928257e-08, + "loss": 1.3844, + "mean_token_accuracy": 0.6398393511772156, + "num_tokens": 2499512.0, + "step": 96 + }, + { + "epoch": 0.010652317153525149, + "grad_norm": 6.588298320770264, + "learning_rate": 3.513909224011713e-08, + "loss": 1.56, + "mean_token_accuracy": 0.5938717126846313, + "num_tokens": 2528412.0, + "step": 97 + }, + { + "epoch": 0.010762134856138809, + "grad_norm": 7.161757469177246, + "learning_rate": 3.5505124450951685e-08, + "loss": 1.5046, + "mean_token_accuracy": 0.6099551320075989, + "num_tokens": 2553855.0, + "step": 98 + }, + { + "epoch": 0.01087195255875247, + "grad_norm": 8.16589641571045, + "learning_rate": 3.5871156661786234e-08, + "loss": 1.5378, + "mean_token_accuracy": 0.6075024604797363, + "num_tokens": 2574996.0, + "step": 99 + }, + { + "epoch": 0.010981770261366132, + "grad_norm": 6.765076160430908, + "learning_rate": 3.623718887262079e-08, + "loss": 1.4906, + "mean_token_accuracy": 0.6121035814285278, + "num_tokens": 2601684.0, + "step": 100 + }, + { + "epoch": 0.011091587963979794, + "grad_norm": 7.04382848739624, + "learning_rate": 3.6603221083455345e-08, + "loss": 1.538, + "mean_token_accuracy": 0.6018838882446289, + "num_tokens": 2628612.0, + "step": 101 + }, + { + "epoch": 0.011201405666593455, + "grad_norm": 6.904319763183594, + "learning_rate": 3.6969253294289894e-08, + "loss": 1.4373, + "mean_token_accuracy": 0.6371680498123169, + "num_tokens": 2652776.0, + "step": 102 + }, + { + "epoch": 0.011311223369207117, + "grad_norm": 6.944467067718506, + "learning_rate": 3.733528550512445e-08, + "loss": 1.4063, + "mean_token_accuracy": 0.6311491131782532, + "num_tokens": 2677807.0, + "step": 103 + }, + { + "epoch": 0.011421041071820778, + "grad_norm": 8.447418212890625, + "learning_rate": 3.7701317715959006e-08, + "loss": 1.5547, + "mean_token_accuracy": 0.5999448299407959, + "num_tokens": 2697989.0, + "step": 104 + }, + { + "epoch": 0.011530858774434438, + "grad_norm": 7.071354866027832, + "learning_rate": 3.8067349926793555e-08, + "loss": 1.5408, + "mean_token_accuracy": 0.6021215915679932, + "num_tokens": 2722341.0, + "step": 105 + }, + { + "epoch": 0.0116406764770481, + "grad_norm": 6.161715984344482, + "learning_rate": 3.843338213762811e-08, + "loss": 1.5184, + "mean_token_accuracy": 0.6020721197128296, + "num_tokens": 2753675.0, + "step": 106 + }, + { + "epoch": 0.011750494179661761, + "grad_norm": 6.807699203491211, + "learning_rate": 3.8799414348462666e-08, + "loss": 1.4658, + "mean_token_accuracy": 0.6138491034507751, + "num_tokens": 2778974.0, + "step": 107 + }, + { + "epoch": 0.011860311882275423, + "grad_norm": 6.235250949859619, + "learning_rate": 3.9165446559297215e-08, + "loss": 1.5174, + "mean_token_accuracy": 0.6038211584091187, + "num_tokens": 2808967.0, + "step": 108 + }, + { + "epoch": 0.011970129584889085, + "grad_norm": 6.626330852508545, + "learning_rate": 3.953147877013177e-08, + "loss": 1.5118, + "mean_token_accuracy": 0.600393533706665, + "num_tokens": 2835393.0, + "step": 109 + }, + { + "epoch": 0.012079947287502746, + "grad_norm": 6.030815124511719, + "learning_rate": 3.9897510980966327e-08, + "loss": 1.4723, + "mean_token_accuracy": 0.6125008463859558, + "num_tokens": 2863678.0, + "step": 110 + }, + { + "epoch": 0.012189764990116406, + "grad_norm": 6.504580497741699, + "learning_rate": 4.0263543191800876e-08, + "loss": 1.4447, + "mean_token_accuracy": 0.6228165626525879, + "num_tokens": 2890376.0, + "step": 111 + }, + { + "epoch": 0.012299582692730068, + "grad_norm": 6.39000129699707, + "learning_rate": 4.062957540263543e-08, + "loss": 1.4415, + "mean_token_accuracy": 0.6234945058822632, + "num_tokens": 2917128.0, + "step": 112 + }, + { + "epoch": 0.01240940039534373, + "grad_norm": 6.949817180633545, + "learning_rate": 4.099560761346999e-08, + "loss": 1.476, + "mean_token_accuracy": 0.6090654134750366, + "num_tokens": 2940812.0, + "step": 113 + }, + { + "epoch": 0.01251921809795739, + "grad_norm": 6.283231258392334, + "learning_rate": 4.1361639824304536e-08, + "loss": 1.502, + "mean_token_accuracy": 0.6025228500366211, + "num_tokens": 2969881.0, + "step": 114 + }, + { + "epoch": 0.012629035800571052, + "grad_norm": 6.7002763748168945, + "learning_rate": 4.172767203513909e-08, + "loss": 1.5018, + "mean_token_accuracy": 0.6079060435295105, + "num_tokens": 2996049.0, + "step": 115 + }, + { + "epoch": 0.012738853503184714, + "grad_norm": 5.76038932800293, + "learning_rate": 4.209370424597365e-08, + "loss": 1.3716, + "mean_token_accuracy": 0.6320133209228516, + "num_tokens": 3025829.0, + "step": 116 + }, + { + "epoch": 0.012848671205798374, + "grad_norm": 6.829304218292236, + "learning_rate": 4.2459736456808197e-08, + "loss": 1.4885, + "mean_token_accuracy": 0.6103053092956543, + "num_tokens": 3051012.0, + "step": 117 + }, + { + "epoch": 0.012958488908412035, + "grad_norm": 6.464592933654785, + "learning_rate": 4.282576866764275e-08, + "loss": 1.5299, + "mean_token_accuracy": 0.6053322553634644, + "num_tokens": 3077750.0, + "step": 118 + }, + { + "epoch": 0.013068306611025697, + "grad_norm": 6.691104888916016, + "learning_rate": 4.319180087847731e-08, + "loss": 1.4794, + "mean_token_accuracy": 0.608944833278656, + "num_tokens": 3103271.0, + "step": 119 + }, + { + "epoch": 0.013178124313639359, + "grad_norm": 7.447789192199707, + "learning_rate": 4.355783308931186e-08, + "loss": 1.5066, + "mean_token_accuracy": 0.6019332408905029, + "num_tokens": 3124511.0, + "step": 120 + }, + { + "epoch": 0.01328794201625302, + "grad_norm": 6.487663269042969, + "learning_rate": 4.392386530014641e-08, + "loss": 1.402, + "mean_token_accuracy": 0.6358776688575745, + "num_tokens": 3150451.0, + "step": 121 + }, + { + "epoch": 0.013397759718866682, + "grad_norm": 6.140832901000977, + "learning_rate": 4.428989751098097e-08, + "loss": 1.5197, + "mean_token_accuracy": 0.6021989583969116, + "num_tokens": 3180151.0, + "step": 122 + }, + { + "epoch": 0.013507577421480343, + "grad_norm": 6.412320137023926, + "learning_rate": 4.465592972181552e-08, + "loss": 1.5134, + "mean_token_accuracy": 0.6061690449714661, + "num_tokens": 3207702.0, + "step": 123 + }, + { + "epoch": 0.013617395124094003, + "grad_norm": 8.089348793029785, + "learning_rate": 4.502196193265007e-08, + "loss": 1.3938, + "mean_token_accuracy": 0.6373574733734131, + "num_tokens": 3226168.0, + "step": 124 + }, + { + "epoch": 0.013727212826707665, + "grad_norm": 6.268423080444336, + "learning_rate": 4.538799414348463e-08, + "loss": 1.4773, + "mean_token_accuracy": 0.6083278656005859, + "num_tokens": 3255714.0, + "step": 125 + }, + { + "epoch": 0.013837030529321327, + "grad_norm": 6.3952741622924805, + "learning_rate": 4.575402635431918e-08, + "loss": 1.5151, + "mean_token_accuracy": 0.6079039573669434, + "num_tokens": 3283574.0, + "step": 126 + }, + { + "epoch": 0.013946848231934988, + "grad_norm": 6.550262451171875, + "learning_rate": 4.6120058565153734e-08, + "loss": 1.459, + "mean_token_accuracy": 0.6193329095840454, + "num_tokens": 3309128.0, + "step": 127 + }, + { + "epoch": 0.01405666593454865, + "grad_norm": 7.406651020050049, + "learning_rate": 4.648609077598828e-08, + "loss": 1.5124, + "mean_token_accuracy": 0.6050333976745605, + "num_tokens": 3331577.0, + "step": 128 + }, + { + "epoch": 0.014166483637162311, + "grad_norm": 6.535178184509277, + "learning_rate": 4.685212298682284e-08, + "loss": 1.5673, + "mean_token_accuracy": 0.5893464088439941, + "num_tokens": 3358884.0, + "step": 129 + }, + { + "epoch": 0.014276301339775971, + "grad_norm": 8.306598663330078, + "learning_rate": 4.7218155197657394e-08, + "loss": 1.5932, + "mean_token_accuracy": 0.5886971950531006, + "num_tokens": 3378122.0, + "step": 130 + }, + { + "epoch": 0.014386119042389633, + "grad_norm": 6.871947288513184, + "learning_rate": 4.7584187408491943e-08, + "loss": 1.5472, + "mean_token_accuracy": 0.6019794940948486, + "num_tokens": 3404642.0, + "step": 131 + }, + { + "epoch": 0.014495936745003294, + "grad_norm": 6.809688568115234, + "learning_rate": 4.79502196193265e-08, + "loss": 1.4417, + "mean_token_accuracy": 0.6156807541847229, + "num_tokens": 3429186.0, + "step": 132 + }, + { + "epoch": 0.014605754447616956, + "grad_norm": 5.958655834197998, + "learning_rate": 4.8316251830161055e-08, + "loss": 1.4777, + "mean_token_accuracy": 0.6180393695831299, + "num_tokens": 3460810.0, + "step": 133 + }, + { + "epoch": 0.014715572150230618, + "grad_norm": 5.928071022033691, + "learning_rate": 4.8682284040995604e-08, + "loss": 1.4439, + "mean_token_accuracy": 0.6186989545822144, + "num_tokens": 3491285.0, + "step": 134 + }, + { + "epoch": 0.01482538985284428, + "grad_norm": 6.408712863922119, + "learning_rate": 4.904831625183016e-08, + "loss": 1.4387, + "mean_token_accuracy": 0.62099289894104, + "num_tokens": 3517893.0, + "step": 135 + }, + { + "epoch": 0.014935207555457939, + "grad_norm": 6.466504096984863, + "learning_rate": 4.9414348462664715e-08, + "loss": 1.4898, + "mean_token_accuracy": 0.6063642501831055, + "num_tokens": 3545614.0, + "step": 136 + }, + { + "epoch": 0.0150450252580716, + "grad_norm": 6.936202526092529, + "learning_rate": 4.9780380673499264e-08, + "loss": 1.5334, + "mean_token_accuracy": 0.5977823138237, + "num_tokens": 3571156.0, + "step": 137 + }, + { + "epoch": 0.015154842960685262, + "grad_norm": 6.95085334777832, + "learning_rate": 5.014641288433382e-08, + "loss": 1.4288, + "mean_token_accuracy": 0.6228137016296387, + "num_tokens": 3595596.0, + "step": 138 + }, + { + "epoch": 0.015264660663298924, + "grad_norm": 6.052608013153076, + "learning_rate": 5.0512445095168376e-08, + "loss": 1.4425, + "mean_token_accuracy": 0.6328141689300537, + "num_tokens": 3625562.0, + "step": 139 + }, + { + "epoch": 0.015374478365912585, + "grad_norm": 7.0913190841674805, + "learning_rate": 5.0878477306002925e-08, + "loss": 1.4961, + "mean_token_accuracy": 0.6199423670768738, + "num_tokens": 3651717.0, + "step": 140 + }, + { + "epoch": 0.015484296068526247, + "grad_norm": 7.214831352233887, + "learning_rate": 5.124450951683748e-08, + "loss": 1.3792, + "mean_token_accuracy": 0.6370892524719238, + "num_tokens": 3674773.0, + "step": 141 + }, + { + "epoch": 0.015594113771139909, + "grad_norm": 7.009411334991455, + "learning_rate": 5.1610541727672036e-08, + "loss": 1.5632, + "mean_token_accuracy": 0.5885941982269287, + "num_tokens": 3701717.0, + "step": 142 + }, + { + "epoch": 0.01570393147375357, + "grad_norm": 6.169476509094238, + "learning_rate": 5.1976573938506585e-08, + "loss": 1.4105, + "mean_token_accuracy": 0.6262778639793396, + "num_tokens": 3729901.0, + "step": 143 + }, + { + "epoch": 0.015813749176367232, + "grad_norm": 6.259419918060303, + "learning_rate": 5.234260614934114e-08, + "loss": 1.511, + "mean_token_accuracy": 0.5940279960632324, + "num_tokens": 3757405.0, + "step": 144 + }, + { + "epoch": 0.01592356687898089, + "grad_norm": 5.944194793701172, + "learning_rate": 5.2708638360175697e-08, + "loss": 1.3733, + "mean_token_accuracy": 0.6361892819404602, + "num_tokens": 3787289.0, + "step": 145 + }, + { + "epoch": 0.01603338458159455, + "grad_norm": 6.421790599822998, + "learning_rate": 5.3074670571010246e-08, + "loss": 1.4092, + "mean_token_accuracy": 0.6321500539779663, + "num_tokens": 3811919.0, + "step": 146 + }, + { + "epoch": 0.016143202284208215, + "grad_norm": 6.366399765014648, + "learning_rate": 5.34407027818448e-08, + "loss": 1.3511, + "mean_token_accuracy": 0.6395090818405151, + "num_tokens": 3835967.0, + "step": 147 + }, + { + "epoch": 0.016253019986821875, + "grad_norm": 6.437642574310303, + "learning_rate": 5.380673499267936e-08, + "loss": 1.3968, + "mean_token_accuracy": 0.627730131149292, + "num_tokens": 3859268.0, + "step": 148 + }, + { + "epoch": 0.016362837689435538, + "grad_norm": 5.185274124145508, + "learning_rate": 5.4172767203513906e-08, + "loss": 1.4468, + "mean_token_accuracy": 0.6131585836410522, + "num_tokens": 3892417.0, + "step": 149 + }, + { + "epoch": 0.016472655392049198, + "grad_norm": 7.06660270690918, + "learning_rate": 5.453879941434846e-08, + "loss": 1.3738, + "mean_token_accuracy": 0.6372190713882446, + "num_tokens": 3913323.0, + "step": 150 + }, + { + "epoch": 0.01658247309466286, + "grad_norm": 5.065357208251953, + "learning_rate": 5.490483162518302e-08, + "loss": 1.2672, + "mean_token_accuracy": 0.6548689007759094, + "num_tokens": 3942390.0, + "step": 151 + }, + { + "epoch": 0.01669229079727652, + "grad_norm": 5.8899149894714355, + "learning_rate": 5.527086383601757e-08, + "loss": 1.4519, + "mean_token_accuracy": 0.6126886606216431, + "num_tokens": 3968060.0, + "step": 152 + }, + { + "epoch": 0.01680210849989018, + "grad_norm": 5.197132587432861, + "learning_rate": 5.563689604685212e-08, + "loss": 1.3292, + "mean_token_accuracy": 0.645074188709259, + "num_tokens": 3996149.0, + "step": 153 + }, + { + "epoch": 0.016911926202503844, + "grad_norm": 5.529796600341797, + "learning_rate": 5.600292825768668e-08, + "loss": 1.463, + "mean_token_accuracy": 0.6079778075218201, + "num_tokens": 4022708.0, + "step": 154 + }, + { + "epoch": 0.017021743905117504, + "grad_norm": 5.873564720153809, + "learning_rate": 5.636896046852123e-08, + "loss": 1.4984, + "mean_token_accuracy": 0.6096013188362122, + "num_tokens": 4048984.0, + "step": 155 + }, + { + "epoch": 0.017131561607731167, + "grad_norm": 5.365451812744141, + "learning_rate": 5.673499267935578e-08, + "loss": 1.3851, + "mean_token_accuracy": 0.6306523084640503, + "num_tokens": 4075654.0, + "step": 156 + }, + { + "epoch": 0.017241379310344827, + "grad_norm": 5.342105388641357, + "learning_rate": 5.710102489019034e-08, + "loss": 1.5403, + "mean_token_accuracy": 0.5871748924255371, + "num_tokens": 4106959.0, + "step": 157 + }, + { + "epoch": 0.017351197012958487, + "grad_norm": 5.259492874145508, + "learning_rate": 5.746705710102489e-08, + "loss": 1.4297, + "mean_token_accuracy": 0.62519371509552, + "num_tokens": 4135143.0, + "step": 158 + }, + { + "epoch": 0.01746101471557215, + "grad_norm": 5.130731105804443, + "learning_rate": 5.7833089311859443e-08, + "loss": 1.3717, + "mean_token_accuracy": 0.6367549300193787, + "num_tokens": 4162641.0, + "step": 159 + }, + { + "epoch": 0.01757083241818581, + "grad_norm": 5.19627046585083, + "learning_rate": 5.8199121522694e-08, + "loss": 1.3299, + "mean_token_accuracy": 0.6348840594291687, + "num_tokens": 4189265.0, + "step": 160 + }, + { + "epoch": 0.017680650120799474, + "grad_norm": 6.213639259338379, + "learning_rate": 5.856515373352855e-08, + "loss": 1.4974, + "mean_token_accuracy": 0.6050631999969482, + "num_tokens": 4209765.0, + "step": 161 + }, + { + "epoch": 0.017790467823413134, + "grad_norm": 5.690797805786133, + "learning_rate": 5.8931185944363104e-08, + "loss": 1.3919, + "mean_token_accuracy": 0.6260151863098145, + "num_tokens": 4232444.0, + "step": 162 + }, + { + "epoch": 0.017900285526026797, + "grad_norm": 4.780436038970947, + "learning_rate": 5.929721815519766e-08, + "loss": 1.3953, + "mean_token_accuracy": 0.6203501224517822, + "num_tokens": 4263314.0, + "step": 163 + }, + { + "epoch": 0.018010103228640457, + "grad_norm": 4.824879169464111, + "learning_rate": 5.966325036603222e-08, + "loss": 1.4034, + "mean_token_accuracy": 0.6269404292106628, + "num_tokens": 4292268.0, + "step": 164 + }, + { + "epoch": 0.018119920931254117, + "grad_norm": 5.312188625335693, + "learning_rate": 6.002928257686676e-08, + "loss": 1.47, + "mean_token_accuracy": 0.6153057813644409, + "num_tokens": 4317038.0, + "step": 165 + }, + { + "epoch": 0.01822973863386778, + "grad_norm": 5.097341060638428, + "learning_rate": 6.039531478770131e-08, + "loss": 1.4737, + "mean_token_accuracy": 0.6049293279647827, + "num_tokens": 4344439.0, + "step": 166 + }, + { + "epoch": 0.01833955633648144, + "grad_norm": 5.045886039733887, + "learning_rate": 6.076134699853588e-08, + "loss": 1.3815, + "mean_token_accuracy": 0.628003716468811, + "num_tokens": 4371828.0, + "step": 167 + }, + { + "epoch": 0.018449374039095103, + "grad_norm": 5.59626579284668, + "learning_rate": 6.112737920937042e-08, + "loss": 1.3012, + "mean_token_accuracy": 0.6416448354721069, + "num_tokens": 4393611.0, + "step": 168 + }, + { + "epoch": 0.018559191741708763, + "grad_norm": 5.149595260620117, + "learning_rate": 6.149341142020497e-08, + "loss": 1.3815, + "mean_token_accuracy": 0.6267452239990234, + "num_tokens": 4419458.0, + "step": 169 + }, + { + "epoch": 0.018669009444322426, + "grad_norm": 5.072066307067871, + "learning_rate": 6.185944363103954e-08, + "loss": 1.3351, + "mean_token_accuracy": 0.6413905620574951, + "num_tokens": 4444110.0, + "step": 170 + }, + { + "epoch": 0.018778827146936086, + "grad_norm": 5.700331687927246, + "learning_rate": 6.222547584187409e-08, + "loss": 1.3718, + "mean_token_accuracy": 0.6262304782867432, + "num_tokens": 4465140.0, + "step": 171 + }, + { + "epoch": 0.018888644849549746, + "grad_norm": 5.259968280792236, + "learning_rate": 6.259150805270863e-08, + "loss": 1.393, + "mean_token_accuracy": 0.6226049065589905, + "num_tokens": 4489605.0, + "step": 172 + }, + { + "epoch": 0.01899846255216341, + "grad_norm": 4.88452672958374, + "learning_rate": 6.29575402635432e-08, + "loss": 1.3404, + "mean_token_accuracy": 0.6358234286308289, + "num_tokens": 4515095.0, + "step": 173 + }, + { + "epoch": 0.01910828025477707, + "grad_norm": 4.755145072937012, + "learning_rate": 6.332357247437773e-08, + "loss": 1.3754, + "mean_token_accuracy": 0.6243820786476135, + "num_tokens": 4541160.0, + "step": 174 + }, + { + "epoch": 0.019218097957390733, + "grad_norm": 5.558831691741943, + "learning_rate": 6.36896046852123e-08, + "loss": 1.3628, + "mean_token_accuracy": 0.6260701417922974, + "num_tokens": 4563441.0, + "step": 175 + }, + { + "epoch": 0.019327915660004392, + "grad_norm": 5.101380348205566, + "learning_rate": 6.405563689604684e-08, + "loss": 1.3105, + "mean_token_accuracy": 0.6393787860870361, + "num_tokens": 4586881.0, + "step": 176 + }, + { + "epoch": 0.019437733362618052, + "grad_norm": 4.636373519897461, + "learning_rate": 6.44216691068814e-08, + "loss": 1.3935, + "mean_token_accuracy": 0.6223844289779663, + "num_tokens": 4616811.0, + "step": 177 + }, + { + "epoch": 0.019547551065231716, + "grad_norm": 5.092663288116455, + "learning_rate": 6.478770131771596e-08, + "loss": 1.2359, + "mean_token_accuracy": 0.658771276473999, + "num_tokens": 4639573.0, + "step": 178 + }, + { + "epoch": 0.019657368767845376, + "grad_norm": 4.823307514190674, + "learning_rate": 6.515373352855052e-08, + "loss": 1.423, + "mean_token_accuracy": 0.6183316707611084, + "num_tokens": 4666392.0, + "step": 179 + }, + { + "epoch": 0.01976718647045904, + "grad_norm": 5.053501129150391, + "learning_rate": 6.551976573938505e-08, + "loss": 1.3213, + "mean_token_accuracy": 0.6356775760650635, + "num_tokens": 4690288.0, + "step": 180 + }, + { + "epoch": 0.0198770041730727, + "grad_norm": 4.577595233917236, + "learning_rate": 6.588579795021962e-08, + "loss": 1.361, + "mean_token_accuracy": 0.6242838501930237, + "num_tokens": 4716583.0, + "step": 181 + }, + { + "epoch": 0.019986821875686362, + "grad_norm": 4.973421573638916, + "learning_rate": 6.625183016105416e-08, + "loss": 1.2924, + "mean_token_accuracy": 0.6485874056816101, + "num_tokens": 4741315.0, + "step": 182 + }, + { + "epoch": 0.020096639578300022, + "grad_norm": 4.575862407684326, + "learning_rate": 6.661786237188873e-08, + "loss": 1.4544, + "mean_token_accuracy": 0.6115446090698242, + "num_tokens": 4771862.0, + "step": 183 + }, + { + "epoch": 0.020206457280913682, + "grad_norm": 4.635465145111084, + "learning_rate": 6.698389458272328e-08, + "loss": 1.4036, + "mean_token_accuracy": 0.6215978860855103, + "num_tokens": 4798609.0, + "step": 184 + }, + { + "epoch": 0.020316274983527345, + "grad_norm": 4.897657871246338, + "learning_rate": 6.734992679355783e-08, + "loss": 1.3454, + "mean_token_accuracy": 0.6302629709243774, + "num_tokens": 4821516.0, + "step": 185 + }, + { + "epoch": 0.020426092686141005, + "grad_norm": 4.872125148773193, + "learning_rate": 6.771595900439237e-08, + "loss": 1.3257, + "mean_token_accuracy": 0.6419086456298828, + "num_tokens": 4845487.0, + "step": 186 + }, + { + "epoch": 0.02053591038875467, + "grad_norm": 4.9699482917785645, + "learning_rate": 6.808199121522694e-08, + "loss": 1.3925, + "mean_token_accuracy": 0.6282047629356384, + "num_tokens": 4868981.0, + "step": 187 + }, + { + "epoch": 0.020645728091368328, + "grad_norm": 4.46227502822876, + "learning_rate": 6.844802342606149e-08, + "loss": 1.3659, + "mean_token_accuracy": 0.6244748830795288, + "num_tokens": 4895095.0, + "step": 188 + }, + { + "epoch": 0.02075554579398199, + "grad_norm": 4.745460510253906, + "learning_rate": 6.881405563689605e-08, + "loss": 1.4304, + "mean_token_accuracy": 0.6124464273452759, + "num_tokens": 4920215.0, + "step": 189 + }, + { + "epoch": 0.02086536349659565, + "grad_norm": 4.294967174530029, + "learning_rate": 6.91800878477306e-08, + "loss": 1.3944, + "mean_token_accuracy": 0.6188781261444092, + "num_tokens": 4950113.0, + "step": 190 + }, + { + "epoch": 0.02097518119920931, + "grad_norm": 5.316690921783447, + "learning_rate": 6.954612005856515e-08, + "loss": 1.3937, + "mean_token_accuracy": 0.6190366744995117, + "num_tokens": 4970901.0, + "step": 191 + }, + { + "epoch": 0.021084998901822975, + "grad_norm": 4.488965034484863, + "learning_rate": 6.99121522693997e-08, + "loss": 1.332, + "mean_token_accuracy": 0.6372287273406982, + "num_tokens": 4996117.0, + "step": 192 + }, + { + "epoch": 0.021194816604436634, + "grad_norm": 4.539755344390869, + "learning_rate": 7.027818448023426e-08, + "loss": 1.3672, + "mean_token_accuracy": 0.625229001045227, + "num_tokens": 5020994.0, + "step": 193 + }, + { + "epoch": 0.021304634307050298, + "grad_norm": 4.334243297576904, + "learning_rate": 7.064421669106881e-08, + "loss": 1.4246, + "mean_token_accuracy": 0.617673397064209, + "num_tokens": 5053438.0, + "step": 194 + }, + { + "epoch": 0.021414452009663958, + "grad_norm": 4.969696521759033, + "learning_rate": 7.101024890190337e-08, + "loss": 1.4084, + "mean_token_accuracy": 0.6204274296760559, + "num_tokens": 5077854.0, + "step": 195 + }, + { + "epoch": 0.021524269712277617, + "grad_norm": 4.594948768615723, + "learning_rate": 7.137628111273792e-08, + "loss": 1.4043, + "mean_token_accuracy": 0.613621711730957, + "num_tokens": 5103447.0, + "step": 196 + }, + { + "epoch": 0.02163408741489128, + "grad_norm": 4.607162952423096, + "learning_rate": 7.174231332357247e-08, + "loss": 1.4106, + "mean_token_accuracy": 0.6137940883636475, + "num_tokens": 5129474.0, + "step": 197 + }, + { + "epoch": 0.02174390511750494, + "grad_norm": 5.412375450134277, + "learning_rate": 7.210834553440702e-08, + "loss": 1.3772, + "mean_token_accuracy": 0.623616099357605, + "num_tokens": 5154331.0, + "step": 198 + }, + { + "epoch": 0.021853722820118604, + "grad_norm": 4.221991062164307, + "learning_rate": 7.247437774524158e-08, + "loss": 1.3714, + "mean_token_accuracy": 0.6293060183525085, + "num_tokens": 5182997.0, + "step": 199 + }, + { + "epoch": 0.021963540522732264, + "grad_norm": 4.07595682144165, + "learning_rate": 7.284040995607613e-08, + "loss": 1.3077, + "mean_token_accuracy": 0.6627575755119324, + "num_tokens": 5209941.0, + "step": 200 + }, + { + "epoch": 0.022073358225345927, + "grad_norm": 4.657059669494629, + "learning_rate": 7.320644216691069e-08, + "loss": 1.2481, + "mean_token_accuracy": 0.6567584276199341, + "num_tokens": 5235749.0, + "step": 201 + }, + { + "epoch": 0.022183175927959587, + "grad_norm": 4.522562026977539, + "learning_rate": 7.357247437774524e-08, + "loss": 1.3228, + "mean_token_accuracy": 0.641194224357605, + "num_tokens": 5259439.0, + "step": 202 + }, + { + "epoch": 0.022292993630573247, + "grad_norm": 4.619617938995361, + "learning_rate": 7.393850658857979e-08, + "loss": 1.3346, + "mean_token_accuracy": 0.6317143440246582, + "num_tokens": 5283571.0, + "step": 203 + }, + { + "epoch": 0.02240281133318691, + "grad_norm": 4.753673076629639, + "learning_rate": 7.430453879941434e-08, + "loss": 1.2663, + "mean_token_accuracy": 0.6507281064987183, + "num_tokens": 5306606.0, + "step": 204 + }, + { + "epoch": 0.02251262903580057, + "grad_norm": 4.2808051109313965, + "learning_rate": 7.46705710102489e-08, + "loss": 1.3644, + "mean_token_accuracy": 0.6409133672714233, + "num_tokens": 5333475.0, + "step": 205 + }, + { + "epoch": 0.022622446738414233, + "grad_norm": 4.39204216003418, + "learning_rate": 7.503660322108345e-08, + "loss": 1.4125, + "mean_token_accuracy": 0.6133631467819214, + "num_tokens": 5360729.0, + "step": 206 + }, + { + "epoch": 0.022732264441027893, + "grad_norm": 4.874942302703857, + "learning_rate": 7.540263543191801e-08, + "loss": 1.3015, + "mean_token_accuracy": 0.6474302411079407, + "num_tokens": 5384083.0, + "step": 207 + }, + { + "epoch": 0.022842082143641557, + "grad_norm": 4.465903282165527, + "learning_rate": 7.576866764275256e-08, + "loss": 1.3198, + "mean_token_accuracy": 0.6347403526306152, + "num_tokens": 5407198.0, + "step": 208 + }, + { + "epoch": 0.022951899846255217, + "grad_norm": 4.655255317687988, + "learning_rate": 7.613469985358711e-08, + "loss": 1.3239, + "mean_token_accuracy": 0.6419397592544556, + "num_tokens": 5433236.0, + "step": 209 + }, + { + "epoch": 0.023061717548868876, + "grad_norm": 5.38681697845459, + "learning_rate": 7.650073206442166e-08, + "loss": 1.3849, + "mean_token_accuracy": 0.6279805898666382, + "num_tokens": 5457306.0, + "step": 210 + }, + { + "epoch": 0.02317153525148254, + "grad_norm": 4.973285675048828, + "learning_rate": 7.686676427525622e-08, + "loss": 1.3099, + "mean_token_accuracy": 0.640975832939148, + "num_tokens": 5477673.0, + "step": 211 + }, + { + "epoch": 0.0232813529540962, + "grad_norm": 3.9803457260131836, + "learning_rate": 7.723279648609077e-08, + "loss": 1.3774, + "mean_token_accuracy": 0.627934455871582, + "num_tokens": 5506648.0, + "step": 212 + }, + { + "epoch": 0.023391170656709863, + "grad_norm": 3.809999704360962, + "learning_rate": 7.759882869692533e-08, + "loss": 1.443, + "mean_token_accuracy": 0.6058349609375, + "num_tokens": 5537209.0, + "step": 213 + }, + { + "epoch": 0.023500988359323523, + "grad_norm": 5.128610610961914, + "learning_rate": 7.796486090775987e-08, + "loss": 1.3905, + "mean_token_accuracy": 0.6314697861671448, + "num_tokens": 5559612.0, + "step": 214 + }, + { + "epoch": 0.023610806061937183, + "grad_norm": 4.707645416259766, + "learning_rate": 7.833089311859443e-08, + "loss": 1.301, + "mean_token_accuracy": 0.6482571363449097, + "num_tokens": 5581853.0, + "step": 215 + }, + { + "epoch": 0.023720623764550846, + "grad_norm": 4.379669666290283, + "learning_rate": 7.869692532942898e-08, + "loss": 1.3131, + "mean_token_accuracy": 0.6434907913208008, + "num_tokens": 5605429.0, + "step": 216 + }, + { + "epoch": 0.023830441467164506, + "grad_norm": 3.9512503147125244, + "learning_rate": 7.906295754026354e-08, + "loss": 1.2859, + "mean_token_accuracy": 0.6424117088317871, + "num_tokens": 5635630.0, + "step": 217 + }, + { + "epoch": 0.02394025916977817, + "grad_norm": 4.467423439025879, + "learning_rate": 7.942898975109809e-08, + "loss": 1.3954, + "mean_token_accuracy": 0.6182119250297546, + "num_tokens": 5661702.0, + "step": 218 + }, + { + "epoch": 0.02405007687239183, + "grad_norm": 4.984197616577148, + "learning_rate": 7.979502196193265e-08, + "loss": 1.2158, + "mean_token_accuracy": 0.6563767790794373, + "num_tokens": 5682274.0, + "step": 219 + }, + { + "epoch": 0.024159894575005492, + "grad_norm": 3.7574613094329834, + "learning_rate": 8.016105417276719e-08, + "loss": 1.3131, + "mean_token_accuracy": 0.6431689262390137, + "num_tokens": 5715819.0, + "step": 220 + }, + { + "epoch": 0.024269712277619152, + "grad_norm": 4.452854633331299, + "learning_rate": 8.052708638360175e-08, + "loss": 1.3466, + "mean_token_accuracy": 0.6319365501403809, + "num_tokens": 5740534.0, + "step": 221 + }, + { + "epoch": 0.024379529980232812, + "grad_norm": 5.830199241638184, + "learning_rate": 8.08931185944363e-08, + "loss": 1.3923, + "mean_token_accuracy": 0.621737003326416, + "num_tokens": 5762357.0, + "step": 222 + }, + { + "epoch": 0.024489347682846475, + "grad_norm": 4.366752624511719, + "learning_rate": 8.125915080527086e-08, + "loss": 1.2186, + "mean_token_accuracy": 0.65312659740448, + "num_tokens": 5787211.0, + "step": 223 + }, + { + "epoch": 0.024599165385460135, + "grad_norm": 4.024194240570068, + "learning_rate": 8.162518301610541e-08, + "loss": 1.3456, + "mean_token_accuracy": 0.6391408443450928, + "num_tokens": 5820989.0, + "step": 224 + }, + { + "epoch": 0.0247089830880738, + "grad_norm": 5.033559322357178, + "learning_rate": 8.199121522693997e-08, + "loss": 1.2987, + "mean_token_accuracy": 0.6343330144882202, + "num_tokens": 5844903.0, + "step": 225 + }, + { + "epoch": 0.02481880079068746, + "grad_norm": 5.151925086975098, + "learning_rate": 8.235724743777451e-08, + "loss": 1.2991, + "mean_token_accuracy": 0.6442680358886719, + "num_tokens": 5867365.0, + "step": 226 + }, + { + "epoch": 0.024928618493301122, + "grad_norm": 4.576485633850098, + "learning_rate": 8.272327964860907e-08, + "loss": 1.3751, + "mean_token_accuracy": 0.6218098998069763, + "num_tokens": 5895675.0, + "step": 227 + }, + { + "epoch": 0.02503843619591478, + "grad_norm": 4.885293006896973, + "learning_rate": 8.308931185944362e-08, + "loss": 1.3312, + "mean_token_accuracy": 0.6392228603363037, + "num_tokens": 5920138.0, + "step": 228 + }, + { + "epoch": 0.02514825389852844, + "grad_norm": 5.022825717926025, + "learning_rate": 8.345534407027818e-08, + "loss": 1.3092, + "mean_token_accuracy": 0.633712887763977, + "num_tokens": 5945763.0, + "step": 229 + }, + { + "epoch": 0.025258071601142105, + "grad_norm": 4.622105598449707, + "learning_rate": 8.382137628111273e-08, + "loss": 1.226, + "mean_token_accuracy": 0.66202712059021, + "num_tokens": 5971811.0, + "step": 230 + }, + { + "epoch": 0.025367889303755765, + "grad_norm": 4.661039352416992, + "learning_rate": 8.41874084919473e-08, + "loss": 1.361, + "mean_token_accuracy": 0.6321172714233398, + "num_tokens": 5995852.0, + "step": 231 + }, + { + "epoch": 0.025477707006369428, + "grad_norm": 5.327050685882568, + "learning_rate": 8.455344070278183e-08, + "loss": 1.2912, + "mean_token_accuracy": 0.6417997479438782, + "num_tokens": 6018357.0, + "step": 232 + }, + { + "epoch": 0.025587524708983088, + "grad_norm": 4.69469690322876, + "learning_rate": 8.491947291361639e-08, + "loss": 1.3396, + "mean_token_accuracy": 0.6353356242179871, + "num_tokens": 6046503.0, + "step": 233 + }, + { + "epoch": 0.025697342411596748, + "grad_norm": 5.04901647567749, + "learning_rate": 8.528550512445094e-08, + "loss": 1.2717, + "mean_token_accuracy": 0.6457571983337402, + "num_tokens": 6067210.0, + "step": 234 + }, + { + "epoch": 0.02580716011421041, + "grad_norm": 4.610098361968994, + "learning_rate": 8.56515373352855e-08, + "loss": 1.2489, + "mean_token_accuracy": 0.6514872312545776, + "num_tokens": 6091649.0, + "step": 235 + }, + { + "epoch": 0.02591697781682407, + "grad_norm": 5.2161335945129395, + "learning_rate": 8.601756954612005e-08, + "loss": 1.3088, + "mean_token_accuracy": 0.6430120468139648, + "num_tokens": 6119052.0, + "step": 236 + }, + { + "epoch": 0.026026795519437734, + "grad_norm": 5.2057294845581055, + "learning_rate": 8.638360175695462e-08, + "loss": 1.2517, + "mean_token_accuracy": 0.6569432020187378, + "num_tokens": 6141688.0, + "step": 237 + }, + { + "epoch": 0.026136613222051394, + "grad_norm": 4.517301559448242, + "learning_rate": 8.674963396778915e-08, + "loss": 1.3317, + "mean_token_accuracy": 0.6297456622123718, + "num_tokens": 6168591.0, + "step": 238 + }, + { + "epoch": 0.026246430924665057, + "grad_norm": 6.287395000457764, + "learning_rate": 8.711566617862371e-08, + "loss": 1.2037, + "mean_token_accuracy": 0.6599892377853394, + "num_tokens": 6186942.0, + "step": 239 + }, + { + "epoch": 0.026356248627278717, + "grad_norm": 4.759090900421143, + "learning_rate": 8.748169838945826e-08, + "loss": 1.2994, + "mean_token_accuracy": 0.6412351131439209, + "num_tokens": 6214324.0, + "step": 240 + }, + { + "epoch": 0.026466066329892377, + "grad_norm": 5.542819976806641, + "learning_rate": 8.784773060029283e-08, + "loss": 1.3172, + "mean_token_accuracy": 0.6369988322257996, + "num_tokens": 6240342.0, + "step": 241 + }, + { + "epoch": 0.02657588403250604, + "grad_norm": 4.985560894012451, + "learning_rate": 8.821376281112737e-08, + "loss": 1.3684, + "mean_token_accuracy": 0.6248915195465088, + "num_tokens": 6266070.0, + "step": 242 + }, + { + "epoch": 0.0266857017351197, + "grad_norm": 5.0811662673950195, + "learning_rate": 8.857979502196194e-08, + "loss": 1.32, + "mean_token_accuracy": 0.6364986896514893, + "num_tokens": 6294087.0, + "step": 243 + }, + { + "epoch": 0.026795519437733364, + "grad_norm": 5.024812698364258, + "learning_rate": 8.894582723279647e-08, + "loss": 1.285, + "mean_token_accuracy": 0.6399350762367249, + "num_tokens": 6318429.0, + "step": 244 + }, + { + "epoch": 0.026905337140347024, + "grad_norm": 4.857804775238037, + "learning_rate": 8.931185944363104e-08, + "loss": 1.317, + "mean_token_accuracy": 0.6409685611724854, + "num_tokens": 6348101.0, + "step": 245 + }, + { + "epoch": 0.027015154842960687, + "grad_norm": 4.563735008239746, + "learning_rate": 8.967789165446558e-08, + "loss": 1.2912, + "mean_token_accuracy": 0.6475382447242737, + "num_tokens": 6373529.0, + "step": 246 + }, + { + "epoch": 0.027124972545574347, + "grad_norm": 5.066126346588135, + "learning_rate": 9.004392386530015e-08, + "loss": 1.3995, + "mean_token_accuracy": 0.6170679926872253, + "num_tokens": 6400776.0, + "step": 247 + }, + { + "epoch": 0.027234790248188007, + "grad_norm": 4.571796894073486, + "learning_rate": 9.04099560761347e-08, + "loss": 1.2654, + "mean_token_accuracy": 0.6419854760169983, + "num_tokens": 6425702.0, + "step": 248 + }, + { + "epoch": 0.02734460795080167, + "grad_norm": 4.5754241943359375, + "learning_rate": 9.077598828696926e-08, + "loss": 1.2711, + "mean_token_accuracy": 0.649328351020813, + "num_tokens": 6447754.0, + "step": 249 + }, + { + "epoch": 0.02745442565341533, + "grad_norm": 3.913433790206909, + "learning_rate": 9.11420204978038e-08, + "loss": 1.2151, + "mean_token_accuracy": 0.6673629283905029, + "num_tokens": 6477535.0, + "step": 250 + }, + { + "epoch": 0.027564243356028993, + "grad_norm": 4.981286525726318, + "learning_rate": 9.150805270863836e-08, + "loss": 1.2572, + "mean_token_accuracy": 0.6519600749015808, + "num_tokens": 6504584.0, + "step": 251 + }, + { + "epoch": 0.027674061058642653, + "grad_norm": 5.393273830413818, + "learning_rate": 9.18740849194729e-08, + "loss": 1.1513, + "mean_token_accuracy": 0.6798288822174072, + "num_tokens": 6529726.0, + "step": 252 + }, + { + "epoch": 0.027783878761256313, + "grad_norm": 6.032054901123047, + "learning_rate": 9.224011713030747e-08, + "loss": 1.3494, + "mean_token_accuracy": 0.6326768398284912, + "num_tokens": 6550265.0, + "step": 253 + }, + { + "epoch": 0.027893696463869976, + "grad_norm": 4.732560157775879, + "learning_rate": 9.260614934114202e-08, + "loss": 1.4211, + "mean_token_accuracy": 0.6164727210998535, + "num_tokens": 6579733.0, + "step": 254 + }, + { + "epoch": 0.028003514166483636, + "grad_norm": 3.824777126312256, + "learning_rate": 9.297218155197657e-08, + "loss": 1.3095, + "mean_token_accuracy": 0.6404423713684082, + "num_tokens": 6613781.0, + "step": 255 + }, + { + "epoch": 0.0281133318690973, + "grad_norm": 5.892072677612305, + "learning_rate": 9.333821376281111e-08, + "loss": 1.2218, + "mean_token_accuracy": 0.6498498320579529, + "num_tokens": 6635982.0, + "step": 256 + }, + { + "epoch": 0.02822314957171096, + "grad_norm": 5.187712669372559, + "learning_rate": 9.370424597364568e-08, + "loss": 1.2866, + "mean_token_accuracy": 0.6370984315872192, + "num_tokens": 6660180.0, + "step": 257 + }, + { + "epoch": 0.028332967274324623, + "grad_norm": 6.233212471008301, + "learning_rate": 9.407027818448023e-08, + "loss": 1.2869, + "mean_token_accuracy": 0.6389632225036621, + "num_tokens": 6687798.0, + "step": 258 + }, + { + "epoch": 0.028442784976938282, + "grad_norm": 5.36053466796875, + "learning_rate": 9.443631039531479e-08, + "loss": 1.3113, + "mean_token_accuracy": 0.641214907169342, + "num_tokens": 6716005.0, + "step": 259 + }, + { + "epoch": 0.028552602679551942, + "grad_norm": 4.727835178375244, + "learning_rate": 9.480234260614934e-08, + "loss": 1.2904, + "mean_token_accuracy": 0.6438925266265869, + "num_tokens": 6744026.0, + "step": 260 + }, + { + "epoch": 0.028662420382165606, + "grad_norm": 6.969109535217285, + "learning_rate": 9.516837481698389e-08, + "loss": 1.26, + "mean_token_accuracy": 0.6442088484764099, + "num_tokens": 6765799.0, + "step": 261 + }, + { + "epoch": 0.028772238084779266, + "grad_norm": 6.038732528686523, + "learning_rate": 9.553440702781844e-08, + "loss": 1.3254, + "mean_token_accuracy": 0.6290289163589478, + "num_tokens": 6792636.0, + "step": 262 + }, + { + "epoch": 0.02888205578739293, + "grad_norm": 7.0465087890625, + "learning_rate": 9.5900439238653e-08, + "loss": 1.1266, + "mean_token_accuracy": 0.6828658580780029, + "num_tokens": 6814122.0, + "step": 263 + }, + { + "epoch": 0.02899187349000659, + "grad_norm": 5.449962139129639, + "learning_rate": 9.626647144948755e-08, + "loss": 1.3381, + "mean_token_accuracy": 0.6278667449951172, + "num_tokens": 6844162.0, + "step": 264 + }, + { + "epoch": 0.029101691192620252, + "grad_norm": 5.895636558532715, + "learning_rate": 9.663250366032211e-08, + "loss": 1.134, + "mean_token_accuracy": 0.6783056259155273, + "num_tokens": 6867856.0, + "step": 265 + }, + { + "epoch": 0.029211508895233912, + "grad_norm": 6.228119850158691, + "learning_rate": 9.699853587115666e-08, + "loss": 1.2347, + "mean_token_accuracy": 0.6629482507705688, + "num_tokens": 6894300.0, + "step": 266 + }, + { + "epoch": 0.029321326597847572, + "grad_norm": 6.515552520751953, + "learning_rate": 9.736456808199121e-08, + "loss": 1.2494, + "mean_token_accuracy": 0.651382327079773, + "num_tokens": 6919880.0, + "step": 267 + }, + { + "epoch": 0.029431144300461235, + "grad_norm": 5.533699989318848, + "learning_rate": 9.773060029282576e-08, + "loss": 1.2934, + "mean_token_accuracy": 0.6452720165252686, + "num_tokens": 6948610.0, + "step": 268 + }, + { + "epoch": 0.029540962003074895, + "grad_norm": 5.5324201583862305, + "learning_rate": 9.809663250366032e-08, + "loss": 1.272, + "mean_token_accuracy": 0.64338219165802, + "num_tokens": 6974058.0, + "step": 269 + }, + { + "epoch": 0.02965077970568856, + "grad_norm": 6.319957256317139, + "learning_rate": 9.846266471449487e-08, + "loss": 1.3246, + "mean_token_accuracy": 0.6340358853340149, + "num_tokens": 7001527.0, + "step": 270 + }, + { + "epoch": 0.029760597408302218, + "grad_norm": 6.336348533630371, + "learning_rate": 9.882869692532943e-08, + "loss": 1.3008, + "mean_token_accuracy": 0.6331230401992798, + "num_tokens": 7025908.0, + "step": 271 + }, + { + "epoch": 0.029870415110915878, + "grad_norm": 6.363037109375, + "learning_rate": 9.919472913616398e-08, + "loss": 1.1979, + "mean_token_accuracy": 0.6574256420135498, + "num_tokens": 7048417.0, + "step": 272 + }, + { + "epoch": 0.02998023281352954, + "grad_norm": 5.515096187591553, + "learning_rate": 9.956076134699853e-08, + "loss": 1.2338, + "mean_token_accuracy": 0.6516445279121399, + "num_tokens": 7072239.0, + "step": 273 + }, + { + "epoch": 0.0300900505161432, + "grad_norm": 4.485733509063721, + "learning_rate": 9.992679355783308e-08, + "loss": 1.297, + "mean_token_accuracy": 0.6396150588989258, + "num_tokens": 7097960.0, + "step": 274 + }, + { + "epoch": 0.030199868218756865, + "grad_norm": 4.286195755004883, + "learning_rate": 1.0029282576866764e-07, + "loss": 1.277, + "mean_token_accuracy": 0.6487964391708374, + "num_tokens": 7129347.0, + "step": 275 + }, + { + "epoch": 0.030309685921370524, + "grad_norm": 5.7115254402160645, + "learning_rate": 1.0065885797950219e-07, + "loss": 1.2684, + "mean_token_accuracy": 0.6435976624488831, + "num_tokens": 7155811.0, + "step": 276 + }, + { + "epoch": 0.030419503623984188, + "grad_norm": 5.898364543914795, + "learning_rate": 1.0102489019033675e-07, + "loss": 1.1608, + "mean_token_accuracy": 0.6726564168930054, + "num_tokens": 7175864.0, + "step": 277 + }, + { + "epoch": 0.030529321326597848, + "grad_norm": 5.539271831512451, + "learning_rate": 1.013909224011713e-07, + "loss": 1.1905, + "mean_token_accuracy": 0.6597288250923157, + "num_tokens": 7197905.0, + "step": 278 + }, + { + "epoch": 0.030639139029211507, + "grad_norm": 5.161451816558838, + "learning_rate": 1.0175695461200585e-07, + "loss": 1.2756, + "mean_token_accuracy": 0.645797610282898, + "num_tokens": 7223710.0, + "step": 279 + }, + { + "epoch": 0.03074895673182517, + "grad_norm": 5.4977617263793945, + "learning_rate": 1.021229868228404e-07, + "loss": 1.229, + "mean_token_accuracy": 0.653950035572052, + "num_tokens": 7245870.0, + "step": 280 + }, + { + "epoch": 0.03085877443443883, + "grad_norm": 7.03364896774292, + "learning_rate": 1.0248901903367496e-07, + "loss": 1.2679, + "mean_token_accuracy": 0.6459632515907288, + "num_tokens": 7271138.0, + "step": 281 + }, + { + "epoch": 0.030968592137052494, + "grad_norm": 6.663631439208984, + "learning_rate": 1.0285505124450951e-07, + "loss": 1.2846, + "mean_token_accuracy": 0.6488204002380371, + "num_tokens": 7297052.0, + "step": 282 + }, + { + "epoch": 0.031078409839666154, + "grad_norm": 5.3051838874816895, + "learning_rate": 1.0322108345534407e-07, + "loss": 1.3524, + "mean_token_accuracy": 0.6288659572601318, + "num_tokens": 7326733.0, + "step": 283 + }, + { + "epoch": 0.031188227542279817, + "grad_norm": 4.182117938995361, + "learning_rate": 1.0358711566617862e-07, + "loss": 1.2983, + "mean_token_accuracy": 0.636183500289917, + "num_tokens": 7358626.0, + "step": 284 + }, + { + "epoch": 0.03129804524489348, + "grad_norm": 5.40377140045166, + "learning_rate": 1.0395314787701317e-07, + "loss": 1.2985, + "mean_token_accuracy": 0.6380600929260254, + "num_tokens": 7385058.0, + "step": 285 + }, + { + "epoch": 0.03140786294750714, + "grad_norm": 4.80905818939209, + "learning_rate": 1.0431918008784772e-07, + "loss": 1.2545, + "mean_token_accuracy": 0.6521672010421753, + "num_tokens": 7414433.0, + "step": 286 + }, + { + "epoch": 0.0315176806501208, + "grad_norm": 5.24266242980957, + "learning_rate": 1.0468521229868228e-07, + "loss": 1.3699, + "mean_token_accuracy": 0.6207088232040405, + "num_tokens": 7446196.0, + "step": 287 + }, + { + "epoch": 0.031627498352734464, + "grad_norm": 7.499354839324951, + "learning_rate": 1.0505124450951683e-07, + "loss": 1.2292, + "mean_token_accuracy": 0.6476700305938721, + "num_tokens": 7469003.0, + "step": 288 + }, + { + "epoch": 0.03173731605534812, + "grad_norm": 7.000964641571045, + "learning_rate": 1.0541727672035139e-07, + "loss": 1.2931, + "mean_token_accuracy": 0.6412399411201477, + "num_tokens": 7494794.0, + "step": 289 + }, + { + "epoch": 0.03184713375796178, + "grad_norm": 4.615601539611816, + "learning_rate": 1.0578330893118593e-07, + "loss": 1.29, + "mean_token_accuracy": 0.6384649276733398, + "num_tokens": 7531151.0, + "step": 290 + }, + { + "epoch": 0.03195695146057544, + "grad_norm": 5.7534499168396, + "learning_rate": 1.0614934114202049e-07, + "loss": 1.1969, + "mean_token_accuracy": 0.6600664854049683, + "num_tokens": 7556362.0, + "step": 291 + }, + { + "epoch": 0.0320667691631891, + "grad_norm": 5.4185075759887695, + "learning_rate": 1.0651537335285504e-07, + "loss": 1.2857, + "mean_token_accuracy": 0.6390218734741211, + "num_tokens": 7580209.0, + "step": 292 + }, + { + "epoch": 0.03217658686580277, + "grad_norm": 5.184093475341797, + "learning_rate": 1.068814055636896e-07, + "loss": 1.3425, + "mean_token_accuracy": 0.6291554570198059, + "num_tokens": 7607823.0, + "step": 293 + }, + { + "epoch": 0.03228640456841643, + "grad_norm": 6.595862865447998, + "learning_rate": 1.0724743777452415e-07, + "loss": 1.3531, + "mean_token_accuracy": 0.6287281513214111, + "num_tokens": 7630083.0, + "step": 294 + }, + { + "epoch": 0.03239622227103009, + "grad_norm": 5.114391803741455, + "learning_rate": 1.0761346998535871e-07, + "loss": 1.3418, + "mean_token_accuracy": 0.6321122050285339, + "num_tokens": 7657339.0, + "step": 295 + }, + { + "epoch": 0.03250603997364375, + "grad_norm": 5.870703220367432, + "learning_rate": 1.0797950219619325e-07, + "loss": 1.1301, + "mean_token_accuracy": 0.6684784889221191, + "num_tokens": 7678789.0, + "step": 296 + }, + { + "epoch": 0.03261585767625741, + "grad_norm": 6.176059246063232, + "learning_rate": 1.0834553440702781e-07, + "loss": 1.2115, + "mean_token_accuracy": 0.6633741855621338, + "num_tokens": 7705531.0, + "step": 297 + }, + { + "epoch": 0.032725675378871076, + "grad_norm": 6.619204521179199, + "learning_rate": 1.0871156661786236e-07, + "loss": 1.2353, + "mean_token_accuracy": 0.6509170532226562, + "num_tokens": 7726907.0, + "step": 298 + }, + { + "epoch": 0.032835493081484736, + "grad_norm": 5.903625965118408, + "learning_rate": 1.0907759882869692e-07, + "loss": 1.402, + "mean_token_accuracy": 0.6286425590515137, + "num_tokens": 7754917.0, + "step": 299 + }, + { + "epoch": 0.032945310784098396, + "grad_norm": 6.54695463180542, + "learning_rate": 1.0944363103953147e-07, + "loss": 1.2556, + "mean_token_accuracy": 0.6439557075500488, + "num_tokens": 7780922.0, + "step": 300 + }, + { + "epoch": 0.033055128486712056, + "grad_norm": 6.303719520568848, + "learning_rate": 1.0980966325036604e-07, + "loss": 1.2759, + "mean_token_accuracy": 0.6457486748695374, + "num_tokens": 7807004.0, + "step": 301 + }, + { + "epoch": 0.03316494618932572, + "grad_norm": 9.123823165893555, + "learning_rate": 1.1017569546120057e-07, + "loss": 1.156, + "mean_token_accuracy": 0.6658633947372437, + "num_tokens": 7830480.0, + "step": 302 + }, + { + "epoch": 0.03327476389193938, + "grad_norm": 6.729947090148926, + "learning_rate": 1.1054172767203513e-07, + "loss": 1.2603, + "mean_token_accuracy": 0.6440127491950989, + "num_tokens": 7856537.0, + "step": 303 + }, + { + "epoch": 0.03338458159455304, + "grad_norm": 9.222710609436035, + "learning_rate": 1.1090775988286968e-07, + "loss": 1.1883, + "mean_token_accuracy": 0.663453221321106, + "num_tokens": 7875428.0, + "step": 304 + }, + { + "epoch": 0.0334943992971667, + "grad_norm": 5.9194183349609375, + "learning_rate": 1.1127379209370424e-07, + "loss": 1.2192, + "mean_token_accuracy": 0.6516638398170471, + "num_tokens": 7899765.0, + "step": 305 + }, + { + "epoch": 0.03360421699978036, + "grad_norm": 7.476350784301758, + "learning_rate": 1.116398243045388e-07, + "loss": 1.2519, + "mean_token_accuracy": 0.6480233669281006, + "num_tokens": 7922002.0, + "step": 306 + }, + { + "epoch": 0.03371403470239403, + "grad_norm": 6.697109222412109, + "learning_rate": 1.1200585651537336e-07, + "loss": 1.2601, + "mean_token_accuracy": 0.6529358625411987, + "num_tokens": 7945165.0, + "step": 307 + }, + { + "epoch": 0.03382385240500769, + "grad_norm": 7.881735801696777, + "learning_rate": 1.1237188872620789e-07, + "loss": 1.1825, + "mean_token_accuracy": 0.6713967323303223, + "num_tokens": 7967060.0, + "step": 308 + }, + { + "epoch": 0.03393367010762135, + "grad_norm": 6.3075175285339355, + "learning_rate": 1.1273792093704245e-07, + "loss": 1.2019, + "mean_token_accuracy": 0.6555215716362, + "num_tokens": 7988205.0, + "step": 309 + }, + { + "epoch": 0.03404348781023501, + "grad_norm": 5.404557704925537, + "learning_rate": 1.13103953147877e-07, + "loss": 1.1446, + "mean_token_accuracy": 0.6646592617034912, + "num_tokens": 8018516.0, + "step": 310 + }, + { + "epoch": 0.03415330551284867, + "grad_norm": 4.585735321044922, + "learning_rate": 1.1346998535871157e-07, + "loss": 1.2407, + "mean_token_accuracy": 0.6554210782051086, + "num_tokens": 8049713.0, + "step": 311 + }, + { + "epoch": 0.034263123215462335, + "grad_norm": 4.931488037109375, + "learning_rate": 1.1383601756954611e-07, + "loss": 1.2395, + "mean_token_accuracy": 0.6534940600395203, + "num_tokens": 8077271.0, + "step": 312 + }, + { + "epoch": 0.034372940918075995, + "grad_norm": 5.343038082122803, + "learning_rate": 1.1420204978038068e-07, + "loss": 1.2482, + "mean_token_accuracy": 0.655802845954895, + "num_tokens": 8100334.0, + "step": 313 + }, + { + "epoch": 0.034482758620689655, + "grad_norm": 6.249810695648193, + "learning_rate": 1.1456808199121521e-07, + "loss": 1.2684, + "mean_token_accuracy": 0.6626524329185486, + "num_tokens": 8127291.0, + "step": 314 + }, + { + "epoch": 0.034592576323303315, + "grad_norm": 5.193748474121094, + "learning_rate": 1.1493411420204978e-07, + "loss": 1.3003, + "mean_token_accuracy": 0.6378147006034851, + "num_tokens": 8158273.0, + "step": 315 + }, + { + "epoch": 0.034702394025916974, + "grad_norm": 5.711767196655273, + "learning_rate": 1.1530014641288432e-07, + "loss": 1.1665, + "mean_token_accuracy": 0.6636372208595276, + "num_tokens": 8180056.0, + "step": 316 + }, + { + "epoch": 0.03481221172853064, + "grad_norm": 7.483816146850586, + "learning_rate": 1.1566617862371889e-07, + "loss": 1.2418, + "mean_token_accuracy": 0.6545781493186951, + "num_tokens": 8203267.0, + "step": 317 + }, + { + "epoch": 0.0349220294311443, + "grad_norm": 6.202247619628906, + "learning_rate": 1.1603221083455344e-07, + "loss": 1.2327, + "mean_token_accuracy": 0.6563759446144104, + "num_tokens": 8225551.0, + "step": 318 + }, + { + "epoch": 0.03503184713375796, + "grad_norm": 5.955916404724121, + "learning_rate": 1.16398243045388e-07, + "loss": 1.284, + "mean_token_accuracy": 0.6339129209518433, + "num_tokens": 8249591.0, + "step": 319 + }, + { + "epoch": 0.03514166483637162, + "grad_norm": 6.81131649017334, + "learning_rate": 1.1676427525622253e-07, + "loss": 1.2525, + "mean_token_accuracy": 0.6447207927703857, + "num_tokens": 8274565.0, + "step": 320 + }, + { + "epoch": 0.03525148253898529, + "grad_norm": 6.4150190353393555, + "learning_rate": 1.171303074670571e-07, + "loss": 1.2711, + "mean_token_accuracy": 0.6427692174911499, + "num_tokens": 8301284.0, + "step": 321 + }, + { + "epoch": 0.03536130024159895, + "grad_norm": 5.879912853240967, + "learning_rate": 1.1749633967789165e-07, + "loss": 1.2434, + "mean_token_accuracy": 0.6454018950462341, + "num_tokens": 8322113.0, + "step": 322 + }, + { + "epoch": 0.03547111794421261, + "grad_norm": 7.387800693511963, + "learning_rate": 1.1786237188872621e-07, + "loss": 1.2788, + "mean_token_accuracy": 0.6380395293235779, + "num_tokens": 8349315.0, + "step": 323 + }, + { + "epoch": 0.03558093564682627, + "grad_norm": 6.698055744171143, + "learning_rate": 1.1822840409956076e-07, + "loss": 1.2375, + "mean_token_accuracy": 0.6480321288108826, + "num_tokens": 8369953.0, + "step": 324 + }, + { + "epoch": 0.03569075334943993, + "grad_norm": 5.760909080505371, + "learning_rate": 1.1859443631039532e-07, + "loss": 1.2692, + "mean_token_accuracy": 0.6454015970230103, + "num_tokens": 8394310.0, + "step": 325 + }, + { + "epoch": 0.035800571052053594, + "grad_norm": 6.67791748046875, + "learning_rate": 1.1896046852122986e-07, + "loss": 1.3203, + "mean_token_accuracy": 0.6317191123962402, + "num_tokens": 8419318.0, + "step": 326 + }, + { + "epoch": 0.035910388754667254, + "grad_norm": 7.103571891784668, + "learning_rate": 1.1932650073206443e-07, + "loss": 1.1002, + "mean_token_accuracy": 0.6802363395690918, + "num_tokens": 8443229.0, + "step": 327 + }, + { + "epoch": 0.036020206457280914, + "grad_norm": 7.889317512512207, + "learning_rate": 1.1969253294289895e-07, + "loss": 1.2522, + "mean_token_accuracy": 0.6516100764274597, + "num_tokens": 8467929.0, + "step": 328 + }, + { + "epoch": 0.03613002415989457, + "grad_norm": 6.139359474182129, + "learning_rate": 1.2005856515373353e-07, + "loss": 1.2147, + "mean_token_accuracy": 0.6557402610778809, + "num_tokens": 8491470.0, + "step": 329 + }, + { + "epoch": 0.03623984186250823, + "grad_norm": 5.574974536895752, + "learning_rate": 1.2042459736456808e-07, + "loss": 1.1827, + "mean_token_accuracy": 0.663699746131897, + "num_tokens": 8519715.0, + "step": 330 + }, + { + "epoch": 0.0363496595651219, + "grad_norm": 7.4387593269348145, + "learning_rate": 1.2079062957540263e-07, + "loss": 1.2419, + "mean_token_accuracy": 0.6472660303115845, + "num_tokens": 8542376.0, + "step": 331 + }, + { + "epoch": 0.03645947726773556, + "grad_norm": 6.903382778167725, + "learning_rate": 1.2115666178623718e-07, + "loss": 1.2525, + "mean_token_accuracy": 0.6519919633865356, + "num_tokens": 8564513.0, + "step": 332 + }, + { + "epoch": 0.03656929497034922, + "grad_norm": 5.2402873039245605, + "learning_rate": 1.2152269399707175e-07, + "loss": 1.1733, + "mean_token_accuracy": 0.6682774424552917, + "num_tokens": 8587957.0, + "step": 333 + }, + { + "epoch": 0.03667911267296288, + "grad_norm": 5.901379585266113, + "learning_rate": 1.2188872620790627e-07, + "loss": 1.1938, + "mean_token_accuracy": 0.6663210391998291, + "num_tokens": 8612155.0, + "step": 334 + }, + { + "epoch": 0.03678893037557654, + "grad_norm": 4.963589191436768, + "learning_rate": 1.2225475841874085e-07, + "loss": 1.2146, + "mean_token_accuracy": 0.6623299717903137, + "num_tokens": 8643677.0, + "step": 335 + }, + { + "epoch": 0.036898748078190206, + "grad_norm": 4.725732803344727, + "learning_rate": 1.226207906295754e-07, + "loss": 1.1711, + "mean_token_accuracy": 0.6642408967018127, + "num_tokens": 8672599.0, + "step": 336 + }, + { + "epoch": 0.037008565780803866, + "grad_norm": 4.753806114196777, + "learning_rate": 1.2298682284040995e-07, + "loss": 1.2625, + "mean_token_accuracy": 0.6557110548019409, + "num_tokens": 8696109.0, + "step": 337 + }, + { + "epoch": 0.037118383483417526, + "grad_norm": 5.358025074005127, + "learning_rate": 1.233528550512445e-07, + "loss": 1.1686, + "mean_token_accuracy": 0.6709367036819458, + "num_tokens": 8725617.0, + "step": 338 + }, + { + "epoch": 0.037228201186031186, + "grad_norm": 5.545661926269531, + "learning_rate": 1.2371888726207907e-07, + "loss": 1.1792, + "mean_token_accuracy": 0.6656195521354675, + "num_tokens": 8753416.0, + "step": 339 + }, + { + "epoch": 0.03733801888864485, + "grad_norm": 5.978397369384766, + "learning_rate": 1.240849194729136e-07, + "loss": 1.2431, + "mean_token_accuracy": 0.6442224979400635, + "num_tokens": 8781610.0, + "step": 340 + }, + { + "epoch": 0.03744783659125851, + "grad_norm": 5.144133567810059, + "learning_rate": 1.2445095168374817e-07, + "loss": 1.1762, + "mean_token_accuracy": 0.6579455137252808, + "num_tokens": 8805386.0, + "step": 341 + }, + { + "epoch": 0.03755765429387217, + "grad_norm": 5.420529842376709, + "learning_rate": 1.2481698389458272e-07, + "loss": 1.3148, + "mean_token_accuracy": 0.6380279660224915, + "num_tokens": 8830009.0, + "step": 342 + }, + { + "epoch": 0.03766747199648583, + "grad_norm": 6.815161228179932, + "learning_rate": 1.2518301610541727e-07, + "loss": 1.1845, + "mean_token_accuracy": 0.6597133874893188, + "num_tokens": 8848499.0, + "step": 343 + }, + { + "epoch": 0.03777728969909949, + "grad_norm": 5.8735833168029785, + "learning_rate": 1.2554904831625182e-07, + "loss": 1.1594, + "mean_token_accuracy": 0.6657552123069763, + "num_tokens": 8869974.0, + "step": 344 + }, + { + "epoch": 0.03788710740171316, + "grad_norm": 5.199695110321045, + "learning_rate": 1.259150805270864e-07, + "loss": 1.2835, + "mean_token_accuracy": 0.6390091180801392, + "num_tokens": 8897952.0, + "step": 345 + }, + { + "epoch": 0.03799692510432682, + "grad_norm": 4.7708420753479, + "learning_rate": 1.2628111273792094e-07, + "loss": 1.2077, + "mean_token_accuracy": 0.6613066792488098, + "num_tokens": 8931394.0, + "step": 346 + }, + { + "epoch": 0.03810674280694048, + "grad_norm": 5.225125789642334, + "learning_rate": 1.2664714494875547e-07, + "loss": 1.2581, + "mean_token_accuracy": 0.6490989923477173, + "num_tokens": 8962484.0, + "step": 347 + }, + { + "epoch": 0.03821656050955414, + "grad_norm": 6.082742691040039, + "learning_rate": 1.2701317715959004e-07, + "loss": 1.2237, + "mean_token_accuracy": 0.6506384015083313, + "num_tokens": 8988810.0, + "step": 348 + }, + { + "epoch": 0.0383263782121678, + "grad_norm": 6.445553779602051, + "learning_rate": 1.273792093704246e-07, + "loss": 1.1684, + "mean_token_accuracy": 0.6684517860412598, + "num_tokens": 9011086.0, + "step": 349 + }, + { + "epoch": 0.038436195914781465, + "grad_norm": 5.982862949371338, + "learning_rate": 1.2774524158125917e-07, + "loss": 1.2453, + "mean_token_accuracy": 0.6504549384117126, + "num_tokens": 9038964.0, + "step": 350 + }, + { + "epoch": 0.038546013617395125, + "grad_norm": 6.643733501434326, + "learning_rate": 1.281112737920937e-07, + "loss": 1.3149, + "mean_token_accuracy": 0.6301469802856445, + "num_tokens": 9065347.0, + "step": 351 + }, + { + "epoch": 0.038655831320008785, + "grad_norm": 6.047738552093506, + "learning_rate": 1.2847730600292824e-07, + "loss": 1.1446, + "mean_token_accuracy": 0.6757219433784485, + "num_tokens": 9089414.0, + "step": 352 + }, + { + "epoch": 0.038765649022622445, + "grad_norm": 6.200734615325928, + "learning_rate": 1.288433382137628e-07, + "loss": 1.3095, + "mean_token_accuracy": 0.6365163922309875, + "num_tokens": 9120551.0, + "step": 353 + }, + { + "epoch": 0.038875466725236105, + "grad_norm": 5.710740566253662, + "learning_rate": 1.2920937042459736e-07, + "loss": 1.2104, + "mean_token_accuracy": 0.6583948135375977, + "num_tokens": 9143936.0, + "step": 354 + }, + { + "epoch": 0.03898528442784977, + "grad_norm": 5.4734649658203125, + "learning_rate": 1.295754026354319e-07, + "loss": 1.2945, + "mean_token_accuracy": 0.6349805593490601, + "num_tokens": 9171765.0, + "step": 355 + }, + { + "epoch": 0.03909510213046343, + "grad_norm": 5.313380718231201, + "learning_rate": 1.2994143484626646e-07, + "loss": 1.249, + "mean_token_accuracy": 0.6465623378753662, + "num_tokens": 9198275.0, + "step": 356 + }, + { + "epoch": 0.03920491983307709, + "grad_norm": 6.378657817840576, + "learning_rate": 1.3030746705710104e-07, + "loss": 1.1998, + "mean_token_accuracy": 0.6507923603057861, + "num_tokens": 9221218.0, + "step": 357 + }, + { + "epoch": 0.03931473753569075, + "grad_norm": 5.889456748962402, + "learning_rate": 1.3067349926793558e-07, + "loss": 1.2006, + "mean_token_accuracy": 0.6505616903305054, + "num_tokens": 9249309.0, + "step": 358 + }, + { + "epoch": 0.03942455523830442, + "grad_norm": 6.909992694854736, + "learning_rate": 1.310395314787701e-07, + "loss": 1.2313, + "mean_token_accuracy": 0.6487318277359009, + "num_tokens": 9270701.0, + "step": 359 + }, + { + "epoch": 0.03953437294091808, + "grad_norm": 5.3145880699157715, + "learning_rate": 1.3140556368960468e-07, + "loss": 1.2631, + "mean_token_accuracy": 0.6477004885673523, + "num_tokens": 9303156.0, + "step": 360 + }, + { + "epoch": 0.03964419064353174, + "grad_norm": 6.824062347412109, + "learning_rate": 1.3177159590043923e-07, + "loss": 1.1599, + "mean_token_accuracy": 0.6740295886993408, + "num_tokens": 9325163.0, + "step": 361 + }, + { + "epoch": 0.0397540083461454, + "grad_norm": 7.882725715637207, + "learning_rate": 1.321376281112738e-07, + "loss": 1.1321, + "mean_token_accuracy": 0.6765903830528259, + "num_tokens": 9349474.0, + "step": 362 + }, + { + "epoch": 0.03986382604875906, + "grad_norm": 5.284257888793945, + "learning_rate": 1.3250366032210833e-07, + "loss": 1.1973, + "mean_token_accuracy": 0.6586986780166626, + "num_tokens": 9376227.0, + "step": 363 + }, + { + "epoch": 0.039973643751372724, + "grad_norm": 5.574831008911133, + "learning_rate": 1.3286969253294288e-07, + "loss": 1.1473, + "mean_token_accuracy": 0.6689085364341736, + "num_tokens": 9398802.0, + "step": 364 + }, + { + "epoch": 0.040083461453986384, + "grad_norm": 5.880070686340332, + "learning_rate": 1.3323572474377745e-07, + "loss": 1.2299, + "mean_token_accuracy": 0.6533167362213135, + "num_tokens": 9425588.0, + "step": 365 + }, + { + "epoch": 0.040193279156600044, + "grad_norm": 3.8336844444274902, + "learning_rate": 1.33601756954612e-07, + "loss": 1.2801, + "mean_token_accuracy": 0.6334208250045776, + "num_tokens": 9457907.0, + "step": 366 + }, + { + "epoch": 0.040303096859213704, + "grad_norm": 6.045873641967773, + "learning_rate": 1.3396778916544655e-07, + "loss": 1.1301, + "mean_token_accuracy": 0.6815659999847412, + "num_tokens": 9482988.0, + "step": 367 + }, + { + "epoch": 0.040412914561827364, + "grad_norm": 6.43412446975708, + "learning_rate": 1.343338213762811e-07, + "loss": 1.2471, + "mean_token_accuracy": 0.6467936038970947, + "num_tokens": 9503553.0, + "step": 368 + }, + { + "epoch": 0.04052273226444103, + "grad_norm": 5.720026969909668, + "learning_rate": 1.3469985358711565e-07, + "loss": 1.2466, + "mean_token_accuracy": 0.6496384739875793, + "num_tokens": 9528274.0, + "step": 369 + }, + { + "epoch": 0.04063254996705469, + "grad_norm": 5.371312141418457, + "learning_rate": 1.3506588579795023e-07, + "loss": 1.2073, + "mean_token_accuracy": 0.6527743339538574, + "num_tokens": 9563887.0, + "step": 370 + }, + { + "epoch": 0.04074236766966835, + "grad_norm": 6.743772506713867, + "learning_rate": 1.3543191800878475e-07, + "loss": 1.2587, + "mean_token_accuracy": 0.6468982696533203, + "num_tokens": 9586556.0, + "step": 371 + }, + { + "epoch": 0.04085218537228201, + "grad_norm": 5.115762233734131, + "learning_rate": 1.3579795021961932e-07, + "loss": 1.2081, + "mean_token_accuracy": 0.6569685935974121, + "num_tokens": 9616303.0, + "step": 372 + }, + { + "epoch": 0.04096200307489567, + "grad_norm": 4.626967906951904, + "learning_rate": 1.3616398243045387e-07, + "loss": 1.2338, + "mean_token_accuracy": 0.6520147919654846, + "num_tokens": 9645693.0, + "step": 373 + }, + { + "epoch": 0.04107182077750934, + "grad_norm": 5.9965128898620605, + "learning_rate": 1.3653001464128845e-07, + "loss": 1.2272, + "mean_token_accuracy": 0.6565917730331421, + "num_tokens": 9672371.0, + "step": 374 + }, + { + "epoch": 0.041181638480122996, + "grad_norm": 5.172848224639893, + "learning_rate": 1.3689604685212297e-07, + "loss": 1.1889, + "mean_token_accuracy": 0.6719295978546143, + "num_tokens": 9705264.0, + "step": 375 + }, + { + "epoch": 0.041291456182736656, + "grad_norm": 6.199886798858643, + "learning_rate": 1.3726207906295752e-07, + "loss": 1.1314, + "mean_token_accuracy": 0.6770509481430054, + "num_tokens": 9729726.0, + "step": 376 + }, + { + "epoch": 0.041401273885350316, + "grad_norm": 5.748904705047607, + "learning_rate": 1.376281112737921e-07, + "loss": 1.2546, + "mean_token_accuracy": 0.6541235446929932, + "num_tokens": 9756079.0, + "step": 377 + }, + { + "epoch": 0.04151109158796398, + "grad_norm": 5.265547752380371, + "learning_rate": 1.3799414348462665e-07, + "loss": 1.1125, + "mean_token_accuracy": 0.6801352500915527, + "num_tokens": 9778317.0, + "step": 378 + }, + { + "epoch": 0.04162090929057764, + "grad_norm": 5.58866024017334, + "learning_rate": 1.383601756954612e-07, + "loss": 1.2446, + "mean_token_accuracy": 0.6496353149414062, + "num_tokens": 9808114.0, + "step": 379 + }, + { + "epoch": 0.0417307269931913, + "grad_norm": 5.06998348236084, + "learning_rate": 1.3872620790629574e-07, + "loss": 1.2238, + "mean_token_accuracy": 0.6435489058494568, + "num_tokens": 9833909.0, + "step": 380 + }, + { + "epoch": 0.04184054469580496, + "grad_norm": 5.413886547088623, + "learning_rate": 1.390922401171303e-07, + "loss": 1.1866, + "mean_token_accuracy": 0.664492666721344, + "num_tokens": 9857091.0, + "step": 381 + }, + { + "epoch": 0.04195036239841862, + "grad_norm": 5.397789001464844, + "learning_rate": 1.3945827232796487e-07, + "loss": 1.1529, + "mean_token_accuracy": 0.6672224402427673, + "num_tokens": 9880395.0, + "step": 382 + }, + { + "epoch": 0.04206018010103229, + "grad_norm": 6.050064563751221, + "learning_rate": 1.398243045387994e-07, + "loss": 1.2757, + "mean_token_accuracy": 0.6415194272994995, + "num_tokens": 9904483.0, + "step": 383 + }, + { + "epoch": 0.04216999780364595, + "grad_norm": 5.3831634521484375, + "learning_rate": 1.4019033674963397e-07, + "loss": 1.1567, + "mean_token_accuracy": 0.6656891107559204, + "num_tokens": 9927865.0, + "step": 384 + }, + { + "epoch": 0.04227981550625961, + "grad_norm": 4.855228424072266, + "learning_rate": 1.4055636896046852e-07, + "loss": 1.2372, + "mean_token_accuracy": 0.6556727290153503, + "num_tokens": 9956712.0, + "step": 385 + }, + { + "epoch": 0.04238963320887327, + "grad_norm": 5.1153645515441895, + "learning_rate": 1.409224011713031e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.688862681388855, + "num_tokens": 9979222.0, + "step": 386 + }, + { + "epoch": 0.04249945091148693, + "grad_norm": 5.605123043060303, + "learning_rate": 1.4128843338213761e-07, + "loss": 1.2208, + "mean_token_accuracy": 0.6476012468338013, + "num_tokens": 10007509.0, + "step": 387 + }, + { + "epoch": 0.042609268614100596, + "grad_norm": 5.025710105895996, + "learning_rate": 1.4165446559297216e-07, + "loss": 1.2566, + "mean_token_accuracy": 0.6492282152175903, + "num_tokens": 10037020.0, + "step": 388 + }, + { + "epoch": 0.042719086316714255, + "grad_norm": 5.115131378173828, + "learning_rate": 1.4202049780380674e-07, + "loss": 1.2782, + "mean_token_accuracy": 0.6393094062805176, + "num_tokens": 10062076.0, + "step": 389 + }, + { + "epoch": 0.042828904019327915, + "grad_norm": 4.833172798156738, + "learning_rate": 1.423865300146413e-07, + "loss": 1.1928, + "mean_token_accuracy": 0.6587375998497009, + "num_tokens": 10093034.0, + "step": 390 + }, + { + "epoch": 0.042938721721941575, + "grad_norm": 5.600562572479248, + "learning_rate": 1.4275256222547584e-07, + "loss": 1.2237, + "mean_token_accuracy": 0.6580433249473572, + "num_tokens": 10117118.0, + "step": 391 + }, + { + "epoch": 0.043048539424555235, + "grad_norm": 4.53516149520874, + "learning_rate": 1.4311859443631039e-07, + "loss": 1.2657, + "mean_token_accuracy": 0.6441454887390137, + "num_tokens": 10151018.0, + "step": 392 + }, + { + "epoch": 0.0431583571271689, + "grad_norm": 7.040074348449707, + "learning_rate": 1.4348462664714493e-07, + "loss": 1.109, + "mean_token_accuracy": 0.67780601978302, + "num_tokens": 10168012.0, + "step": 393 + }, + { + "epoch": 0.04326817482978256, + "grad_norm": 5.3338398933410645, + "learning_rate": 1.438506588579795e-07, + "loss": 1.2402, + "mean_token_accuracy": 0.6462692022323608, + "num_tokens": 10193267.0, + "step": 394 + }, + { + "epoch": 0.04337799253239622, + "grad_norm": 4.56988000869751, + "learning_rate": 1.4421669106881403e-07, + "loss": 1.1467, + "mean_token_accuracy": 0.6629671454429626, + "num_tokens": 10219769.0, + "step": 395 + }, + { + "epoch": 0.04348781023500988, + "grad_norm": 5.54587984085083, + "learning_rate": 1.445827232796486e-07, + "loss": 1.2529, + "mean_token_accuracy": 0.6448350548744202, + "num_tokens": 10244899.0, + "step": 396 + }, + { + "epoch": 0.04359762793762355, + "grad_norm": 6.399552345275879, + "learning_rate": 1.4494875549048316e-07, + "loss": 1.1919, + "mean_token_accuracy": 0.659835934638977, + "num_tokens": 10263877.0, + "step": 397 + }, + { + "epoch": 0.04370744564023721, + "grad_norm": 5.443115711212158, + "learning_rate": 1.4531478770131773e-07, + "loss": 1.2513, + "mean_token_accuracy": 0.6449379920959473, + "num_tokens": 10287450.0, + "step": 398 + }, + { + "epoch": 0.04381726334285087, + "grad_norm": 4.6200714111328125, + "learning_rate": 1.4568081991215226e-07, + "loss": 1.2065, + "mean_token_accuracy": 0.6651197075843811, + "num_tokens": 10314939.0, + "step": 399 + }, + { + "epoch": 0.04392708104546453, + "grad_norm": 4.7281494140625, + "learning_rate": 1.460468521229868e-07, + "loss": 1.0961, + "mean_token_accuracy": 0.6859226822853088, + "num_tokens": 10340075.0, + "step": 400 + }, + { + "epoch": 0.04403689874807819, + "grad_norm": 5.874328136444092, + "learning_rate": 1.4641288433382138e-07, + "loss": 1.2504, + "mean_token_accuracy": 0.6512032747268677, + "num_tokens": 10364116.0, + "step": 401 + }, + { + "epoch": 0.044146716450691854, + "grad_norm": 4.832901477813721, + "learning_rate": 1.4677891654465593e-07, + "loss": 1.2615, + "mean_token_accuracy": 0.6497293710708618, + "num_tokens": 10391839.0, + "step": 402 + }, + { + "epoch": 0.044256534153305514, + "grad_norm": 4.4853692054748535, + "learning_rate": 1.4714494875549048e-07, + "loss": 1.1973, + "mean_token_accuracy": 0.672461211681366, + "num_tokens": 10419954.0, + "step": 403 + }, + { + "epoch": 0.044366351855919174, + "grad_norm": 5.428635597229004, + "learning_rate": 1.4751098096632503e-07, + "loss": 1.2581, + "mean_token_accuracy": 0.6440224647521973, + "num_tokens": 10444143.0, + "step": 404 + }, + { + "epoch": 0.044476169558532834, + "grad_norm": 4.699741363525391, + "learning_rate": 1.4787701317715958e-07, + "loss": 1.1907, + "mean_token_accuracy": 0.6598252654075623, + "num_tokens": 10473697.0, + "step": 405 + }, + { + "epoch": 0.044585987261146494, + "grad_norm": 3.909668445587158, + "learning_rate": 1.4824304538799415e-07, + "loss": 1.1847, + "mean_token_accuracy": 0.664010763168335, + "num_tokens": 10502612.0, + "step": 406 + }, + { + "epoch": 0.04469580496376016, + "grad_norm": 4.703807353973389, + "learning_rate": 1.4860907759882867e-07, + "loss": 1.2285, + "mean_token_accuracy": 0.6496561765670776, + "num_tokens": 10529266.0, + "step": 407 + }, + { + "epoch": 0.04480562266637382, + "grad_norm": 4.976019859313965, + "learning_rate": 1.4897510980966325e-07, + "loss": 1.2078, + "mean_token_accuracy": 0.655895471572876, + "num_tokens": 10550616.0, + "step": 408 + }, + { + "epoch": 0.04491544036898748, + "grad_norm": 4.68854284286499, + "learning_rate": 1.493411420204978e-07, + "loss": 1.172, + "mean_token_accuracy": 0.6656140685081482, + "num_tokens": 10582682.0, + "step": 409 + }, + { + "epoch": 0.04502525807160114, + "grad_norm": 5.0771660804748535, + "learning_rate": 1.4970717423133235e-07, + "loss": 1.1012, + "mean_token_accuracy": 0.6803601384162903, + "num_tokens": 10606229.0, + "step": 410 + }, + { + "epoch": 0.0451350757742148, + "grad_norm": 4.238261699676514, + "learning_rate": 1.500732064421669e-07, + "loss": 1.2491, + "mean_token_accuracy": 0.6450652480125427, + "num_tokens": 10639889.0, + "step": 411 + }, + { + "epoch": 0.04524489347682847, + "grad_norm": 4.738804340362549, + "learning_rate": 1.5043923865300145e-07, + "loss": 1.2552, + "mean_token_accuracy": 0.6391216516494751, + "num_tokens": 10664352.0, + "step": 412 + }, + { + "epoch": 0.04535471117944213, + "grad_norm": 4.1733317375183105, + "learning_rate": 1.5080527086383602e-07, + "loss": 1.1643, + "mean_token_accuracy": 0.6698398590087891, + "num_tokens": 10692255.0, + "step": 413 + }, + { + "epoch": 0.04546452888205579, + "grad_norm": 6.893023490905762, + "learning_rate": 1.5117130307467057e-07, + "loss": 1.1886, + "mean_token_accuracy": 0.6655242443084717, + "num_tokens": 10713360.0, + "step": 414 + }, + { + "epoch": 0.045574346584669446, + "grad_norm": 4.511856555938721, + "learning_rate": 1.5153733528550512e-07, + "loss": 1.1754, + "mean_token_accuracy": 0.6655045747756958, + "num_tokens": 10743290.0, + "step": 415 + }, + { + "epoch": 0.04568416428728311, + "grad_norm": 5.702057838439941, + "learning_rate": 1.5190336749633967e-07, + "loss": 1.1921, + "mean_token_accuracy": 0.6540985107421875, + "num_tokens": 10771567.0, + "step": 416 + }, + { + "epoch": 0.04579398198989677, + "grad_norm": 4.610085487365723, + "learning_rate": 1.5226939970717422e-07, + "loss": 1.1544, + "mean_token_accuracy": 0.6816190481185913, + "num_tokens": 10796435.0, + "step": 417 + }, + { + "epoch": 0.04590379969251043, + "grad_norm": 5.275516986846924, + "learning_rate": 1.526354319180088e-07, + "loss": 1.219, + "mean_token_accuracy": 0.6627026200294495, + "num_tokens": 10826350.0, + "step": 418 + }, + { + "epoch": 0.04601361739512409, + "grad_norm": 4.931993007659912, + "learning_rate": 1.5300146412884332e-07, + "loss": 1.1552, + "mean_token_accuracy": 0.6692383289337158, + "num_tokens": 10851407.0, + "step": 419 + }, + { + "epoch": 0.04612343509773775, + "grad_norm": 3.9152371883392334, + "learning_rate": 1.533674963396779e-07, + "loss": 1.3077, + "mean_token_accuracy": 0.6453163623809814, + "num_tokens": 10885508.0, + "step": 420 + }, + { + "epoch": 0.04623325280035142, + "grad_norm": 4.969761371612549, + "learning_rate": 1.5373352855051244e-07, + "loss": 1.1896, + "mean_token_accuracy": 0.6543772220611572, + "num_tokens": 10911158.0, + "step": 421 + }, + { + "epoch": 0.04634307050296508, + "grad_norm": 5.723127841949463, + "learning_rate": 1.54099560761347e-07, + "loss": 1.124, + "mean_token_accuracy": 0.6750514507293701, + "num_tokens": 10932147.0, + "step": 422 + }, + { + "epoch": 0.04645288820557874, + "grad_norm": 5.151498794555664, + "learning_rate": 1.5446559297218154e-07, + "loss": 1.1776, + "mean_token_accuracy": 0.6641418933868408, + "num_tokens": 10952566.0, + "step": 423 + }, + { + "epoch": 0.0465627059081924, + "grad_norm": 6.211347579956055, + "learning_rate": 1.548316251830161e-07, + "loss": 1.1468, + "mean_token_accuracy": 0.6698039174079895, + "num_tokens": 10971181.0, + "step": 424 + }, + { + "epoch": 0.04667252361080606, + "grad_norm": 5.278365612030029, + "learning_rate": 1.5519765739385066e-07, + "loss": 1.1098, + "mean_token_accuracy": 0.6784648895263672, + "num_tokens": 10991562.0, + "step": 425 + }, + { + "epoch": 0.046782341313419726, + "grad_norm": 5.488775253295898, + "learning_rate": 1.5556368960468521e-07, + "loss": 1.1838, + "mean_token_accuracy": 0.6597262620925903, + "num_tokens": 11016836.0, + "step": 426 + }, + { + "epoch": 0.046892159016033386, + "grad_norm": 4.5734992027282715, + "learning_rate": 1.5592972181551974e-07, + "loss": 1.2262, + "mean_token_accuracy": 0.6506190299987793, + "num_tokens": 11040431.0, + "step": 427 + }, + { + "epoch": 0.047001976718647046, + "grad_norm": 4.152224540710449, + "learning_rate": 1.562957540263543e-07, + "loss": 1.1738, + "mean_token_accuracy": 0.6648803353309631, + "num_tokens": 11065612.0, + "step": 428 + }, + { + "epoch": 0.047111794421260705, + "grad_norm": 5.654637813568115, + "learning_rate": 1.5666178623718886e-07, + "loss": 1.1219, + "mean_token_accuracy": 0.6812989711761475, + "num_tokens": 11090027.0, + "step": 429 + }, + { + "epoch": 0.047221612123874365, + "grad_norm": 3.8061914443969727, + "learning_rate": 1.5702781844802344e-07, + "loss": 1.2874, + "mean_token_accuracy": 0.6334378123283386, + "num_tokens": 11121097.0, + "step": 430 + }, + { + "epoch": 0.04733142982648803, + "grad_norm": 6.542547702789307, + "learning_rate": 1.5739385065885796e-07, + "loss": 1.2032, + "mean_token_accuracy": 0.6530284881591797, + "num_tokens": 11144282.0, + "step": 431 + }, + { + "epoch": 0.04744124752910169, + "grad_norm": 4.87344217300415, + "learning_rate": 1.5775988286969253e-07, + "loss": 1.2053, + "mean_token_accuracy": 0.6613271236419678, + "num_tokens": 11172508.0, + "step": 432 + }, + { + "epoch": 0.04755106523171535, + "grad_norm": 5.089217185974121, + "learning_rate": 1.5812591508052708e-07, + "loss": 1.1561, + "mean_token_accuracy": 0.6714051961898804, + "num_tokens": 11194936.0, + "step": 433 + }, + { + "epoch": 0.04766088293432901, + "grad_norm": 4.151459217071533, + "learning_rate": 1.5849194729136163e-07, + "loss": 1.0578, + "mean_token_accuracy": 0.6909525394439697, + "num_tokens": 11220873.0, + "step": 434 + }, + { + "epoch": 0.04777070063694268, + "grad_norm": 4.630164623260498, + "learning_rate": 1.5885797950219618e-07, + "loss": 1.1999, + "mean_token_accuracy": 0.6560629606246948, + "num_tokens": 11245691.0, + "step": 435 + }, + { + "epoch": 0.04788051833955634, + "grad_norm": 4.7113189697265625, + "learning_rate": 1.5922401171303073e-07, + "loss": 1.2474, + "mean_token_accuracy": 0.6642085313796997, + "num_tokens": 11272386.0, + "step": 436 + }, + { + "epoch": 0.04799033604217, + "grad_norm": 3.9157724380493164, + "learning_rate": 1.595900439238653e-07, + "loss": 1.2524, + "mean_token_accuracy": 0.6487985849380493, + "num_tokens": 11301211.0, + "step": 437 + }, + { + "epoch": 0.04810015374478366, + "grad_norm": 3.9699742794036865, + "learning_rate": 1.5995607613469986e-07, + "loss": 1.2443, + "mean_token_accuracy": 0.6521830558776855, + "num_tokens": 11331253.0, + "step": 438 + }, + { + "epoch": 0.04820997144739732, + "grad_norm": 3.667896032333374, + "learning_rate": 1.6032210834553438e-07, + "loss": 1.167, + "mean_token_accuracy": 0.6742188334465027, + "num_tokens": 11360900.0, + "step": 439 + }, + { + "epoch": 0.048319789150010985, + "grad_norm": 3.980437755584717, + "learning_rate": 1.6068814055636895e-07, + "loss": 1.2798, + "mean_token_accuracy": 0.6360231041908264, + "num_tokens": 11391599.0, + "step": 440 + }, + { + "epoch": 0.048429606852624645, + "grad_norm": 3.807748317718506, + "learning_rate": 1.610541727672035e-07, + "loss": 1.205, + "mean_token_accuracy": 0.659405529499054, + "num_tokens": 11424019.0, + "step": 441 + }, + { + "epoch": 0.048539424555238304, + "grad_norm": 4.815460205078125, + "learning_rate": 1.6142020497803808e-07, + "loss": 1.1771, + "mean_token_accuracy": 0.6564425230026245, + "num_tokens": 11446707.0, + "step": 442 + }, + { + "epoch": 0.048649242257851964, + "grad_norm": 5.105069160461426, + "learning_rate": 1.617862371888726e-07, + "loss": 1.1336, + "mean_token_accuracy": 0.6714131832122803, + "num_tokens": 11467310.0, + "step": 443 + }, + { + "epoch": 0.048759059960465624, + "grad_norm": 4.048864841461182, + "learning_rate": 1.6215226939970718e-07, + "loss": 1.1873, + "mean_token_accuracy": 0.6599280834197998, + "num_tokens": 11493973.0, + "step": 444 + }, + { + "epoch": 0.04886887766307929, + "grad_norm": 4.9202775955200195, + "learning_rate": 1.6251830161054173e-07, + "loss": 1.1263, + "mean_token_accuracy": 0.6839373111724854, + "num_tokens": 11514611.0, + "step": 445 + }, + { + "epoch": 0.04897869536569295, + "grad_norm": 4.331563949584961, + "learning_rate": 1.6288433382137627e-07, + "loss": 1.2696, + "mean_token_accuracy": 0.646397590637207, + "num_tokens": 11538158.0, + "step": 446 + }, + { + "epoch": 0.04908851306830661, + "grad_norm": 4.424675464630127, + "learning_rate": 1.6325036603221082e-07, + "loss": 1.1865, + "mean_token_accuracy": 0.6627397537231445, + "num_tokens": 11570295.0, + "step": 447 + }, + { + "epoch": 0.04919833077092027, + "grad_norm": 4.834066390991211, + "learning_rate": 1.6361639824304537e-07, + "loss": 1.1557, + "mean_token_accuracy": 0.6719046831130981, + "num_tokens": 11597780.0, + "step": 448 + }, + { + "epoch": 0.04930814847353393, + "grad_norm": 3.8743419647216797, + "learning_rate": 1.6398243045387995e-07, + "loss": 1.2115, + "mean_token_accuracy": 0.6611605882644653, + "num_tokens": 11623199.0, + "step": 449 + }, + { + "epoch": 0.0494179661761476, + "grad_norm": 4.574422836303711, + "learning_rate": 1.643484626647145e-07, + "loss": 1.2477, + "mean_token_accuracy": 0.6406362056732178, + "num_tokens": 11651657.0, + "step": 450 + }, + { + "epoch": 0.04952778387876126, + "grad_norm": 3.9517107009887695, + "learning_rate": 1.6471449487554902e-07, + "loss": 1.1862, + "mean_token_accuracy": 0.6576931476593018, + "num_tokens": 11680381.0, + "step": 451 + }, + { + "epoch": 0.04963760158137492, + "grad_norm": 3.950571298599243, + "learning_rate": 1.650805270863836e-07, + "loss": 1.1443, + "mean_token_accuracy": 0.6710270643234253, + "num_tokens": 11705013.0, + "step": 452 + }, + { + "epoch": 0.04974741928398858, + "grad_norm": 3.9882357120513916, + "learning_rate": 1.6544655929721814e-07, + "loss": 1.0862, + "mean_token_accuracy": 0.6854097843170166, + "num_tokens": 11731140.0, + "step": 453 + }, + { + "epoch": 0.049857236986602244, + "grad_norm": 3.9553568363189697, + "learning_rate": 1.6581259150805272e-07, + "loss": 1.2935, + "mean_token_accuracy": 0.6349482536315918, + "num_tokens": 11761261.0, + "step": 454 + }, + { + "epoch": 0.0499670546892159, + "grad_norm": 4.071976661682129, + "learning_rate": 1.6617862371888724e-07, + "loss": 1.1651, + "mean_token_accuracy": 0.6743283271789551, + "num_tokens": 11787241.0, + "step": 455 + }, + { + "epoch": 0.05007687239182956, + "grad_norm": 4.525739669799805, + "learning_rate": 1.6654465592972182e-07, + "loss": 1.1992, + "mean_token_accuracy": 0.6593047380447388, + "num_tokens": 11808564.0, + "step": 456 + }, + { + "epoch": 0.05018669009444322, + "grad_norm": 3.709378480911255, + "learning_rate": 1.6691068814055637e-07, + "loss": 1.1692, + "mean_token_accuracy": 0.6689815521240234, + "num_tokens": 11835170.0, + "step": 457 + }, + { + "epoch": 0.05029650779705688, + "grad_norm": 3.6776235103607178, + "learning_rate": 1.6727672035139092e-07, + "loss": 1.172, + "mean_token_accuracy": 0.6686199903488159, + "num_tokens": 11862838.0, + "step": 458 + }, + { + "epoch": 0.05040632549967055, + "grad_norm": 3.893083333969116, + "learning_rate": 1.6764275256222547e-07, + "loss": 1.3016, + "mean_token_accuracy": 0.6404299736022949, + "num_tokens": 11892865.0, + "step": 459 + }, + { + "epoch": 0.05051614320228421, + "grad_norm": 4.591513156890869, + "learning_rate": 1.6800878477306001e-07, + "loss": 1.118, + "mean_token_accuracy": 0.6807015538215637, + "num_tokens": 11913385.0, + "step": 460 + }, + { + "epoch": 0.05062596090489787, + "grad_norm": 4.365466117858887, + "learning_rate": 1.683748169838946e-07, + "loss": 1.2149, + "mean_token_accuracy": 0.654137134552002, + "num_tokens": 11942433.0, + "step": 461 + }, + { + "epoch": 0.05073577860751153, + "grad_norm": 4.5116400718688965, + "learning_rate": 1.6874084919472914e-07, + "loss": 1.1867, + "mean_token_accuracy": 0.6579225063323975, + "num_tokens": 11968602.0, + "step": 462 + }, + { + "epoch": 0.05084559631012519, + "grad_norm": 4.600101947784424, + "learning_rate": 1.6910688140556366e-07, + "loss": 1.1744, + "mean_token_accuracy": 0.6663768291473389, + "num_tokens": 11989652.0, + "step": 463 + }, + { + "epoch": 0.050955414012738856, + "grad_norm": 4.260495662689209, + "learning_rate": 1.6947291361639824e-07, + "loss": 1.2057, + "mean_token_accuracy": 0.6505403518676758, + "num_tokens": 12014428.0, + "step": 464 + }, + { + "epoch": 0.051065231715352516, + "grad_norm": 4.3982834815979, + "learning_rate": 1.6983894582723279e-07, + "loss": 1.1059, + "mean_token_accuracy": 0.677883505821228, + "num_tokens": 12041211.0, + "step": 465 + }, + { + "epoch": 0.051175049417966176, + "grad_norm": 3.673548936843872, + "learning_rate": 1.7020497803806736e-07, + "loss": 1.2353, + "mean_token_accuracy": 0.6603291630744934, + "num_tokens": 12071813.0, + "step": 466 + }, + { + "epoch": 0.051284867120579836, + "grad_norm": 3.997731924057007, + "learning_rate": 1.7057101024890188e-07, + "loss": 1.2366, + "mean_token_accuracy": 0.6513267755508423, + "num_tokens": 12099495.0, + "step": 467 + }, + { + "epoch": 0.051394684823193496, + "grad_norm": 4.2548675537109375, + "learning_rate": 1.7093704245973643e-07, + "loss": 1.2348, + "mean_token_accuracy": 0.643952488899231, + "num_tokens": 12125207.0, + "step": 468 + }, + { + "epoch": 0.05150450252580716, + "grad_norm": 4.470493316650391, + "learning_rate": 1.71303074670571e-07, + "loss": 1.1908, + "mean_token_accuracy": 0.653472900390625, + "num_tokens": 12146398.0, + "step": 469 + }, + { + "epoch": 0.05161432022842082, + "grad_norm": 3.8488986492156982, + "learning_rate": 1.7166910688140556e-07, + "loss": 1.0905, + "mean_token_accuracy": 0.6800157427787781, + "num_tokens": 12170215.0, + "step": 470 + }, + { + "epoch": 0.05172413793103448, + "grad_norm": 4.695403099060059, + "learning_rate": 1.720351390922401e-07, + "loss": 1.2421, + "mean_token_accuracy": 0.64494788646698, + "num_tokens": 12192610.0, + "step": 471 + }, + { + "epoch": 0.05183395563364814, + "grad_norm": 4.291847229003906, + "learning_rate": 1.7240117130307466e-07, + "loss": 1.1864, + "mean_token_accuracy": 0.6657167673110962, + "num_tokens": 12220138.0, + "step": 472 + }, + { + "epoch": 0.05194377333626181, + "grad_norm": 4.460822582244873, + "learning_rate": 1.7276720351390923e-07, + "loss": 1.1941, + "mean_token_accuracy": 0.6604811549186707, + "num_tokens": 12251326.0, + "step": 473 + }, + { + "epoch": 0.05205359103887547, + "grad_norm": 3.952388048171997, + "learning_rate": 1.7313323572474378e-07, + "loss": 1.2528, + "mean_token_accuracy": 0.647305965423584, + "num_tokens": 12280837.0, + "step": 474 + }, + { + "epoch": 0.05216340874148913, + "grad_norm": 4.393105983734131, + "learning_rate": 1.734992679355783e-07, + "loss": 1.0741, + "mean_token_accuracy": 0.6898907423019409, + "num_tokens": 12304158.0, + "step": 475 + }, + { + "epoch": 0.05227322644410279, + "grad_norm": 3.7554919719696045, + "learning_rate": 1.7386530014641288e-07, + "loss": 1.249, + "mean_token_accuracy": 0.6427255868911743, + "num_tokens": 12331193.0, + "step": 476 + }, + { + "epoch": 0.05238304414671645, + "grad_norm": 3.9891231060028076, + "learning_rate": 1.7423133235724743e-07, + "loss": 1.2299, + "mean_token_accuracy": 0.6536926627159119, + "num_tokens": 12357251.0, + "step": 477 + }, + { + "epoch": 0.052492861849330115, + "grad_norm": 4.111508369445801, + "learning_rate": 1.74597364568082e-07, + "loss": 1.2449, + "mean_token_accuracy": 0.6466517448425293, + "num_tokens": 12384863.0, + "step": 478 + }, + { + "epoch": 0.052602679551943775, + "grad_norm": 3.6291160583496094, + "learning_rate": 1.7496339677891653e-07, + "loss": 1.1044, + "mean_token_accuracy": 0.68010413646698, + "num_tokens": 12415065.0, + "step": 479 + }, + { + "epoch": 0.052712497254557435, + "grad_norm": 4.4170918464660645, + "learning_rate": 1.7532942898975108e-07, + "loss": 1.1183, + "mean_token_accuracy": 0.6758016347885132, + "num_tokens": 12439060.0, + "step": 480 + }, + { + "epoch": 0.052822314957171095, + "grad_norm": 4.159963607788086, + "learning_rate": 1.7569546120058565e-07, + "loss": 1.1268, + "mean_token_accuracy": 0.6698992252349854, + "num_tokens": 12462856.0, + "step": 481 + }, + { + "epoch": 0.052932132659784754, + "grad_norm": 4.7354865074157715, + "learning_rate": 1.760614934114202e-07, + "loss": 1.0931, + "mean_token_accuracy": 0.6814716458320618, + "num_tokens": 12487168.0, + "step": 482 + }, + { + "epoch": 0.05304195036239842, + "grad_norm": 4.423887729644775, + "learning_rate": 1.7642752562225475e-07, + "loss": 1.2515, + "mean_token_accuracy": 0.6439869403839111, + "num_tokens": 12516282.0, + "step": 483 + }, + { + "epoch": 0.05315176806501208, + "grad_norm": 4.106174945831299, + "learning_rate": 1.767935578330893e-07, + "loss": 1.1111, + "mean_token_accuracy": 0.679314136505127, + "num_tokens": 12541665.0, + "step": 484 + }, + { + "epoch": 0.05326158576762574, + "grad_norm": 3.939851999282837, + "learning_rate": 1.7715959004392387e-07, + "loss": 1.2414, + "mean_token_accuracy": 0.6545626521110535, + "num_tokens": 12573589.0, + "step": 485 + }, + { + "epoch": 0.0533714034702394, + "grad_norm": 3.800028085708618, + "learning_rate": 1.7752562225475842e-07, + "loss": 1.135, + "mean_token_accuracy": 0.6736728549003601, + "num_tokens": 12601106.0, + "step": 486 + }, + { + "epoch": 0.05348122117285306, + "grad_norm": 3.72011137008667, + "learning_rate": 1.7789165446559295e-07, + "loss": 1.1587, + "mean_token_accuracy": 0.6698451042175293, + "num_tokens": 12631770.0, + "step": 487 + }, + { + "epoch": 0.05359103887546673, + "grad_norm": 4.585676193237305, + "learning_rate": 1.7825768667642752e-07, + "loss": 1.1432, + "mean_token_accuracy": 0.6629710793495178, + "num_tokens": 12655234.0, + "step": 488 + }, + { + "epoch": 0.05370085657808039, + "grad_norm": 3.610884189605713, + "learning_rate": 1.7862371888726207e-07, + "loss": 1.1922, + "mean_token_accuracy": 0.6597626209259033, + "num_tokens": 12682179.0, + "step": 489 + }, + { + "epoch": 0.05381067428069405, + "grad_norm": 3.761005401611328, + "learning_rate": 1.7898975109809665e-07, + "loss": 1.1527, + "mean_token_accuracy": 0.6683086156845093, + "num_tokens": 12711843.0, + "step": 490 + }, + { + "epoch": 0.05392049198330771, + "grad_norm": 3.9116294384002686, + "learning_rate": 1.7935578330893117e-07, + "loss": 1.1947, + "mean_token_accuracy": 0.6562709212303162, + "num_tokens": 12741567.0, + "step": 491 + }, + { + "epoch": 0.054030309685921374, + "grad_norm": 3.908435821533203, + "learning_rate": 1.7972181551976572e-07, + "loss": 1.1941, + "mean_token_accuracy": 0.6619312763214111, + "num_tokens": 12767598.0, + "step": 492 + }, + { + "epoch": 0.054140127388535034, + "grad_norm": 3.957226037979126, + "learning_rate": 1.800878477306003e-07, + "loss": 1.1478, + "mean_token_accuracy": 0.674636721611023, + "num_tokens": 12790026.0, + "step": 493 + }, + { + "epoch": 0.054249945091148694, + "grad_norm": 3.6180331707000732, + "learning_rate": 1.8045387994143484e-07, + "loss": 1.1537, + "mean_token_accuracy": 0.6724441051483154, + "num_tokens": 12818197.0, + "step": 494 + }, + { + "epoch": 0.05435976279376235, + "grad_norm": 4.176002025604248, + "learning_rate": 1.808199121522694e-07, + "loss": 1.1629, + "mean_token_accuracy": 0.6650805473327637, + "num_tokens": 12840299.0, + "step": 495 + }, + { + "epoch": 0.05446958049637601, + "grad_norm": 4.2049241065979, + "learning_rate": 1.8118594436310394e-07, + "loss": 1.2395, + "mean_token_accuracy": 0.6464589238166809, + "num_tokens": 12865428.0, + "step": 496 + }, + { + "epoch": 0.05457939819898968, + "grad_norm": 3.218015670776367, + "learning_rate": 1.8155197657393852e-07, + "loss": 1.154, + "mean_token_accuracy": 0.6758691072463989, + "num_tokens": 12892978.0, + "step": 497 + }, + { + "epoch": 0.05468921590160334, + "grad_norm": 3.294637441635132, + "learning_rate": 1.8191800878477307e-07, + "loss": 1.1993, + "mean_token_accuracy": 0.6679984331130981, + "num_tokens": 12921515.0, + "step": 498 + }, + { + "epoch": 0.054799033604217, + "grad_norm": 3.6262588500976562, + "learning_rate": 1.822840409956076e-07, + "loss": 1.1827, + "mean_token_accuracy": 0.6644519567489624, + "num_tokens": 12947231.0, + "step": 499 + }, + { + "epoch": 0.05490885130683066, + "grad_norm": 3.7913267612457275, + "learning_rate": 1.8265007320644216e-07, + "loss": 1.1935, + "mean_token_accuracy": 0.6567946076393127, + "num_tokens": 12975796.0, + "step": 500 + }, + { + "epoch": 0.05501866900944432, + "grad_norm": 3.334092617034912, + "learning_rate": 1.830161054172767e-07, + "loss": 1.1295, + "mean_token_accuracy": 0.678063154220581, + "num_tokens": 13005740.0, + "step": 501 + }, + { + "epoch": 0.055128486712057986, + "grad_norm": 3.9542014598846436, + "learning_rate": 1.833821376281113e-07, + "loss": 1.163, + "mean_token_accuracy": 0.6600364446640015, + "num_tokens": 13029608.0, + "step": 502 + }, + { + "epoch": 0.055238304414671646, + "grad_norm": 4.216237545013428, + "learning_rate": 1.837481698389458e-07, + "loss": 1.0723, + "mean_token_accuracy": 0.690470814704895, + "num_tokens": 13050822.0, + "step": 503 + }, + { + "epoch": 0.055348122117285306, + "grad_norm": 3.5601508617401123, + "learning_rate": 1.8411420204978036e-07, + "loss": 0.9964, + "mean_token_accuracy": 0.70522141456604, + "num_tokens": 13076150.0, + "step": 504 + }, + { + "epoch": 0.055457939819898966, + "grad_norm": 3.471226453781128, + "learning_rate": 1.8448023426061494e-07, + "loss": 1.2144, + "mean_token_accuracy": 0.6512569189071655, + "num_tokens": 13103581.0, + "step": 505 + }, + { + "epoch": 0.055567757522512626, + "grad_norm": 3.4171228408813477, + "learning_rate": 1.8484626647144948e-07, + "loss": 1.1464, + "mean_token_accuracy": 0.6628336310386658, + "num_tokens": 13131904.0, + "step": 506 + }, + { + "epoch": 0.05567757522512629, + "grad_norm": 4.48084831237793, + "learning_rate": 1.8521229868228403e-07, + "loss": 1.0537, + "mean_token_accuracy": 0.6911390423774719, + "num_tokens": 13155934.0, + "step": 507 + }, + { + "epoch": 0.05578739292773995, + "grad_norm": 3.885711193084717, + "learning_rate": 1.8557833089311858e-07, + "loss": 1.1092, + "mean_token_accuracy": 0.6789306998252869, + "num_tokens": 13179650.0, + "step": 508 + }, + { + "epoch": 0.05589721063035361, + "grad_norm": 4.139927864074707, + "learning_rate": 1.8594436310395313e-07, + "loss": 1.2008, + "mean_token_accuracy": 0.6536150574684143, + "num_tokens": 13205174.0, + "step": 509 + }, + { + "epoch": 0.05600702833296727, + "grad_norm": 3.4238879680633545, + "learning_rate": 1.863103953147877e-07, + "loss": 1.1421, + "mean_token_accuracy": 0.6659587621688843, + "num_tokens": 13231544.0, + "step": 510 + }, + { + "epoch": 0.05611684603558094, + "grad_norm": 4.162845134735107, + "learning_rate": 1.8667642752562223e-07, + "loss": 1.1708, + "mean_token_accuracy": 0.6617627143859863, + "num_tokens": 13252845.0, + "step": 511 + }, + { + "epoch": 0.0562266637381946, + "grad_norm": 3.504958391189575, + "learning_rate": 1.870424597364568e-07, + "loss": 1.2658, + "mean_token_accuracy": 0.6387331485748291, + "num_tokens": 13285009.0, + "step": 512 + }, + { + "epoch": 0.05633648144080826, + "grad_norm": 3.651442050933838, + "learning_rate": 1.8740849194729135e-07, + "loss": 1.0847, + "mean_token_accuracy": 0.6878920793533325, + "num_tokens": 13312518.0, + "step": 513 + }, + { + "epoch": 0.05644629914342192, + "grad_norm": 3.3906447887420654, + "learning_rate": 1.8777452415812593e-07, + "loss": 1.0845, + "mean_token_accuracy": 0.6889434456825256, + "num_tokens": 13335578.0, + "step": 514 + }, + { + "epoch": 0.05655611684603558, + "grad_norm": 3.0702545642852783, + "learning_rate": 1.8814055636896045e-07, + "loss": 1.1908, + "mean_token_accuracy": 0.666635274887085, + "num_tokens": 13368449.0, + "step": 515 + }, + { + "epoch": 0.056665934548649245, + "grad_norm": 3.737224817276001, + "learning_rate": 1.88506588579795e-07, + "loss": 1.2196, + "mean_token_accuracy": 0.6539262533187866, + "num_tokens": 13397250.0, + "step": 516 + }, + { + "epoch": 0.056775752251262905, + "grad_norm": 4.200257301330566, + "learning_rate": 1.8887262079062958e-07, + "loss": 1.1603, + "mean_token_accuracy": 0.6594668030738831, + "num_tokens": 13422857.0, + "step": 517 + }, + { + "epoch": 0.056885569953876565, + "grad_norm": 3.7983009815216064, + "learning_rate": 1.8923865300146413e-07, + "loss": 1.086, + "mean_token_accuracy": 0.6778309345245361, + "num_tokens": 13447180.0, + "step": 518 + }, + { + "epoch": 0.056995387656490225, + "grad_norm": 4.080022811889648, + "learning_rate": 1.8960468521229868e-07, + "loss": 1.2374, + "mean_token_accuracy": 0.6412807106971741, + "num_tokens": 13469860.0, + "step": 519 + }, + { + "epoch": 0.057105205359103885, + "grad_norm": 3.2981903553009033, + "learning_rate": 1.8997071742313322e-07, + "loss": 1.1854, + "mean_token_accuracy": 0.6682778596878052, + "num_tokens": 13495606.0, + "step": 520 + }, + { + "epoch": 0.05721502306171755, + "grad_norm": 3.401139736175537, + "learning_rate": 1.9033674963396777e-07, + "loss": 1.1666, + "mean_token_accuracy": 0.6655152440071106, + "num_tokens": 13527790.0, + "step": 521 + }, + { + "epoch": 0.05732484076433121, + "grad_norm": 3.5187737941741943, + "learning_rate": 1.9070278184480235e-07, + "loss": 1.1487, + "mean_token_accuracy": 0.6793028712272644, + "num_tokens": 13551986.0, + "step": 522 + }, + { + "epoch": 0.05743465846694487, + "grad_norm": 3.615579605102539, + "learning_rate": 1.9106881405563687e-07, + "loss": 1.195, + "mean_token_accuracy": 0.6575260162353516, + "num_tokens": 13577954.0, + "step": 523 + }, + { + "epoch": 0.05754447616955853, + "grad_norm": 3.353424549102783, + "learning_rate": 1.9143484626647145e-07, + "loss": 1.1944, + "mean_token_accuracy": 0.6535829901695251, + "num_tokens": 13604870.0, + "step": 524 + }, + { + "epoch": 0.05765429387217219, + "grad_norm": 2.8355767726898193, + "learning_rate": 1.91800878477306e-07, + "loss": 1.2104, + "mean_token_accuracy": 0.6593587398529053, + "num_tokens": 13637200.0, + "step": 525 + }, + { + "epoch": 0.05776411157478586, + "grad_norm": 5.038870811462402, + "learning_rate": 1.9216691068814057e-07, + "loss": 1.2033, + "mean_token_accuracy": 0.6508023738861084, + "num_tokens": 13656718.0, + "step": 526 + }, + { + "epoch": 0.05787392927739952, + "grad_norm": 4.514934062957764, + "learning_rate": 1.925329428989751e-07, + "loss": 1.1186, + "mean_token_accuracy": 0.6754990816116333, + "num_tokens": 13680498.0, + "step": 527 + }, + { + "epoch": 0.05798374698001318, + "grad_norm": 4.513001441955566, + "learning_rate": 1.9289897510980964e-07, + "loss": 1.0906, + "mean_token_accuracy": 0.6860166788101196, + "num_tokens": 13701339.0, + "step": 528 + }, + { + "epoch": 0.05809356468262684, + "grad_norm": 4.069243907928467, + "learning_rate": 1.9326500732064422e-07, + "loss": 1.1729, + "mean_token_accuracy": 0.6647647619247437, + "num_tokens": 13724927.0, + "step": 529 + }, + { + "epoch": 0.058203382385240504, + "grad_norm": 3.9002432823181152, + "learning_rate": 1.9363103953147877e-07, + "loss": 1.0784, + "mean_token_accuracy": 0.6872162818908691, + "num_tokens": 13745791.0, + "step": 530 + }, + { + "epoch": 0.058313200087854164, + "grad_norm": 3.640533447265625, + "learning_rate": 1.9399707174231332e-07, + "loss": 1.097, + "mean_token_accuracy": 0.6792428493499756, + "num_tokens": 13770891.0, + "step": 531 + }, + { + "epoch": 0.058423017790467824, + "grad_norm": 4.613619327545166, + "learning_rate": 1.9436310395314787e-07, + "loss": 1.1224, + "mean_token_accuracy": 0.6681752800941467, + "num_tokens": 13790625.0, + "step": 532 + }, + { + "epoch": 0.058532835493081484, + "grad_norm": 3.298290491104126, + "learning_rate": 1.9472913616398242e-07, + "loss": 1.0869, + "mean_token_accuracy": 0.6792846322059631, + "num_tokens": 13813346.0, + "step": 533 + }, + { + "epoch": 0.058642653195695144, + "grad_norm": 3.682076930999756, + "learning_rate": 1.95095168374817e-07, + "loss": 1.2018, + "mean_token_accuracy": 0.655380368232727, + "num_tokens": 13837250.0, + "step": 534 + }, + { + "epoch": 0.05875247089830881, + "grad_norm": 3.4764187335968018, + "learning_rate": 1.9546120058565151e-07, + "loss": 1.1252, + "mean_token_accuracy": 0.6779614090919495, + "num_tokens": 13865730.0, + "step": 535 + }, + { + "epoch": 0.05886228860092247, + "grad_norm": 3.5424602031707764, + "learning_rate": 1.958272327964861e-07, + "loss": 1.1125, + "mean_token_accuracy": 0.6764171719551086, + "num_tokens": 13893215.0, + "step": 536 + }, + { + "epoch": 0.05897210630353613, + "grad_norm": 3.303215980529785, + "learning_rate": 1.9619326500732064e-07, + "loss": 1.1625, + "mean_token_accuracy": 0.6644502282142639, + "num_tokens": 13919232.0, + "step": 537 + }, + { + "epoch": 0.05908192400614979, + "grad_norm": 3.34338641166687, + "learning_rate": 1.9655929721815521e-07, + "loss": 1.1703, + "mean_token_accuracy": 0.6638346910476685, + "num_tokens": 13943661.0, + "step": 538 + }, + { + "epoch": 0.05919174170876345, + "grad_norm": 3.9622814655303955, + "learning_rate": 1.9692532942898974e-07, + "loss": 1.1731, + "mean_token_accuracy": 0.6574737429618835, + "num_tokens": 13967209.0, + "step": 539 + }, + { + "epoch": 0.05930155941137712, + "grad_norm": 3.9625635147094727, + "learning_rate": 1.9729136163982429e-07, + "loss": 1.0371, + "mean_token_accuracy": 0.6957430243492126, + "num_tokens": 13987948.0, + "step": 540 + }, + { + "epoch": 0.059411377113990776, + "grad_norm": 3.1007022857666016, + "learning_rate": 1.9765739385065886e-07, + "loss": 1.2018, + "mean_token_accuracy": 0.659845769405365, + "num_tokens": 14016862.0, + "step": 541 + }, + { + "epoch": 0.059521194816604436, + "grad_norm": 3.2507669925689697, + "learning_rate": 1.980234260614934e-07, + "loss": 1.185, + "mean_token_accuracy": 0.6643625497817993, + "num_tokens": 14041242.0, + "step": 542 + }, + { + "epoch": 0.059631012519218096, + "grad_norm": 3.318122625350952, + "learning_rate": 1.9838945827232796e-07, + "loss": 1.0876, + "mean_token_accuracy": 0.6862042546272278, + "num_tokens": 14064732.0, + "step": 543 + }, + { + "epoch": 0.059740830221831756, + "grad_norm": 3.4818735122680664, + "learning_rate": 1.987554904831625e-07, + "loss": 1.0638, + "mean_token_accuracy": 0.687059760093689, + "num_tokens": 14090186.0, + "step": 544 + }, + { + "epoch": 0.05985064792444542, + "grad_norm": 3.2291789054870605, + "learning_rate": 1.9912152269399706e-07, + "loss": 1.2151, + "mean_token_accuracy": 0.6554352641105652, + "num_tokens": 14121747.0, + "step": 545 + }, + { + "epoch": 0.05996046562705908, + "grad_norm": 2.775606393814087, + "learning_rate": 1.9948755490483163e-07, + "loss": 1.2248, + "mean_token_accuracy": 0.6504952311515808, + "num_tokens": 14152053.0, + "step": 546 + }, + { + "epoch": 0.06007028332967274, + "grad_norm": 4.051575183868408, + "learning_rate": 1.9985358711566616e-07, + "loss": 1.1237, + "mean_token_accuracy": 0.6807409524917603, + "num_tokens": 14170890.0, + "step": 547 + }, + { + "epoch": 0.0601801010322864, + "grad_norm": 2.8794782161712646, + "learning_rate": 2.0021961932650073e-07, + "loss": 1.113, + "mean_token_accuracy": 0.6773687601089478, + "num_tokens": 14200001.0, + "step": 548 + }, + { + "epoch": 0.06028991873490007, + "grad_norm": 3.2080960273742676, + "learning_rate": 2.0058565153733528e-07, + "loss": 1.1528, + "mean_token_accuracy": 0.6700979471206665, + "num_tokens": 14225062.0, + "step": 549 + }, + { + "epoch": 0.06039973643751373, + "grad_norm": 3.366623640060425, + "learning_rate": 2.0095168374816983e-07, + "loss": 1.1246, + "mean_token_accuracy": 0.671647310256958, + "num_tokens": 14248169.0, + "step": 550 + }, + { + "epoch": 0.06050955414012739, + "grad_norm": 3.5410959720611572, + "learning_rate": 2.0131771595900438e-07, + "loss": 1.1457, + "mean_token_accuracy": 0.6749725341796875, + "num_tokens": 14273309.0, + "step": 551 + }, + { + "epoch": 0.06061937184274105, + "grad_norm": 3.223938465118408, + "learning_rate": 2.0168374816983893e-07, + "loss": 1.1049, + "mean_token_accuracy": 0.6805438995361328, + "num_tokens": 14300494.0, + "step": 552 + }, + { + "epoch": 0.06072918954535471, + "grad_norm": 2.994868755340576, + "learning_rate": 2.020497803806735e-07, + "loss": 1.1397, + "mean_token_accuracy": 0.6649997234344482, + "num_tokens": 14327576.0, + "step": 553 + }, + { + "epoch": 0.060839007247968376, + "grad_norm": 2.743821382522583, + "learning_rate": 2.0241581259150805e-07, + "loss": 1.1635, + "mean_token_accuracy": 0.6608244776725769, + "num_tokens": 14359413.0, + "step": 554 + }, + { + "epoch": 0.060948824950582035, + "grad_norm": 3.2971343994140625, + "learning_rate": 2.027818448023426e-07, + "loss": 1.2824, + "mean_token_accuracy": 0.6491294503211975, + "num_tokens": 14389772.0, + "step": 555 + }, + { + "epoch": 0.061058642653195695, + "grad_norm": 3.068410634994507, + "learning_rate": 2.0314787701317715e-07, + "loss": 1.1969, + "mean_token_accuracy": 0.6617318391799927, + "num_tokens": 14416775.0, + "step": 556 + }, + { + "epoch": 0.061168460355809355, + "grad_norm": 3.399911880493164, + "learning_rate": 2.035139092240117e-07, + "loss": 1.2376, + "mean_token_accuracy": 0.644131600856781, + "num_tokens": 14445547.0, + "step": 557 + }, + { + "epoch": 0.061278278058423015, + "grad_norm": 3.39845609664917, + "learning_rate": 2.0387994143484627e-07, + "loss": 1.2089, + "mean_token_accuracy": 0.6513653993606567, + "num_tokens": 14469951.0, + "step": 558 + }, + { + "epoch": 0.06138809576103668, + "grad_norm": 3.332336187362671, + "learning_rate": 2.042459736456808e-07, + "loss": 1.1692, + "mean_token_accuracy": 0.6645593643188477, + "num_tokens": 14492322.0, + "step": 559 + }, + { + "epoch": 0.06149791346365034, + "grad_norm": 2.8993124961853027, + "learning_rate": 2.0461200585651537e-07, + "loss": 1.1599, + "mean_token_accuracy": 0.6662878394126892, + "num_tokens": 14517881.0, + "step": 560 + }, + { + "epoch": 0.061607731166264, + "grad_norm": 3.8086581230163574, + "learning_rate": 2.0497803806734992e-07, + "loss": 1.1896, + "mean_token_accuracy": 0.665078341960907, + "num_tokens": 14543018.0, + "step": 561 + }, + { + "epoch": 0.06171754886887766, + "grad_norm": 3.5531692504882812, + "learning_rate": 2.0534407027818447e-07, + "loss": 1.1694, + "mean_token_accuracy": 0.6600992679595947, + "num_tokens": 14570418.0, + "step": 562 + }, + { + "epoch": 0.06182736657149132, + "grad_norm": 4.001189231872559, + "learning_rate": 2.0571010248901902e-07, + "loss": 1.0265, + "mean_token_accuracy": 0.6966331601142883, + "num_tokens": 14589713.0, + "step": 563 + }, + { + "epoch": 0.06193718427410499, + "grad_norm": 2.993323802947998, + "learning_rate": 2.0607613469985357e-07, + "loss": 1.1699, + "mean_token_accuracy": 0.6615025997161865, + "num_tokens": 14617418.0, + "step": 564 + }, + { + "epoch": 0.06204700197671865, + "grad_norm": 3.1453402042388916, + "learning_rate": 2.0644216691068814e-07, + "loss": 1.1412, + "mean_token_accuracy": 0.6756981611251831, + "num_tokens": 14645074.0, + "step": 565 + }, + { + "epoch": 0.06215681967933231, + "grad_norm": 3.2538177967071533, + "learning_rate": 2.068081991215227e-07, + "loss": 1.2146, + "mean_token_accuracy": 0.653941810131073, + "num_tokens": 14671552.0, + "step": 566 + }, + { + "epoch": 0.06226663738194597, + "grad_norm": 3.4187676906585693, + "learning_rate": 2.0717423133235724e-07, + "loss": 1.1361, + "mean_token_accuracy": 0.6716862916946411, + "num_tokens": 14694223.0, + "step": 567 + }, + { + "epoch": 0.062376455084559634, + "grad_norm": 3.359867572784424, + "learning_rate": 2.075402635431918e-07, + "loss": 1.0828, + "mean_token_accuracy": 0.6904855966567993, + "num_tokens": 14719184.0, + "step": 568 + }, + { + "epoch": 0.062486272787173294, + "grad_norm": 3.0766103267669678, + "learning_rate": 2.0790629575402634e-07, + "loss": 1.1416, + "mean_token_accuracy": 0.6714606285095215, + "num_tokens": 14742235.0, + "step": 569 + }, + { + "epoch": 0.06259609048978695, + "grad_norm": 3.163569211959839, + "learning_rate": 2.0827232796486092e-07, + "loss": 1.2191, + "mean_token_accuracy": 0.6529426574707031, + "num_tokens": 14769348.0, + "step": 570 + }, + { + "epoch": 0.06270590819240061, + "grad_norm": 3.092923879623413, + "learning_rate": 2.0863836017569544e-07, + "loss": 1.0503, + "mean_token_accuracy": 0.6912222504615784, + "num_tokens": 14794555.0, + "step": 571 + }, + { + "epoch": 0.06281572589501427, + "grad_norm": 3.2044427394866943, + "learning_rate": 2.0900439238653001e-07, + "loss": 1.1398, + "mean_token_accuracy": 0.6707562208175659, + "num_tokens": 14817364.0, + "step": 572 + }, + { + "epoch": 0.06292554359762793, + "grad_norm": 3.6153433322906494, + "learning_rate": 2.0937042459736456e-07, + "loss": 1.1731, + "mean_token_accuracy": 0.668265700340271, + "num_tokens": 14837472.0, + "step": 573 + }, + { + "epoch": 0.0630353613002416, + "grad_norm": 2.7976877689361572, + "learning_rate": 2.097364568081991e-07, + "loss": 1.1156, + "mean_token_accuracy": 0.6789260506629944, + "num_tokens": 14865112.0, + "step": 574 + }, + { + "epoch": 0.06314517900285525, + "grad_norm": 3.151637315750122, + "learning_rate": 2.1010248901903366e-07, + "loss": 1.1519, + "mean_token_accuracy": 0.663320779800415, + "num_tokens": 14894808.0, + "step": 575 + }, + { + "epoch": 0.06325499670546893, + "grad_norm": 3.6383278369903564, + "learning_rate": 2.104685212298682e-07, + "loss": 1.0436, + "mean_token_accuracy": 0.6960170865058899, + "num_tokens": 14913195.0, + "step": 576 + }, + { + "epoch": 0.06336481440808259, + "grad_norm": 2.8190557956695557, + "learning_rate": 2.1083455344070279e-07, + "loss": 1.1885, + "mean_token_accuracy": 0.6610988974571228, + "num_tokens": 14943929.0, + "step": 577 + }, + { + "epoch": 0.06347463211069625, + "grad_norm": 3.092632293701172, + "learning_rate": 2.1120058565153734e-07, + "loss": 1.17, + "mean_token_accuracy": 0.6654942035675049, + "num_tokens": 14969696.0, + "step": 578 + }, + { + "epoch": 0.0635844498133099, + "grad_norm": 3.682919979095459, + "learning_rate": 2.1156661786237186e-07, + "loss": 1.012, + "mean_token_accuracy": 0.6977626085281372, + "num_tokens": 14989320.0, + "step": 579 + }, + { + "epoch": 0.06369426751592357, + "grad_norm": 2.8601865768432617, + "learning_rate": 2.1193265007320643e-07, + "loss": 1.077, + "mean_token_accuracy": 0.684740424156189, + "num_tokens": 15014145.0, + "step": 580 + }, + { + "epoch": 0.06380408521853723, + "grad_norm": 2.7264397144317627, + "learning_rate": 2.1229868228404098e-07, + "loss": 1.1568, + "mean_token_accuracy": 0.6619868278503418, + "num_tokens": 15040827.0, + "step": 581 + }, + { + "epoch": 0.06391390292115089, + "grad_norm": 2.7939510345458984, + "learning_rate": 2.1266471449487556e-07, + "loss": 1.2051, + "mean_token_accuracy": 0.6583991646766663, + "num_tokens": 15069707.0, + "step": 582 + }, + { + "epoch": 0.06402372062376455, + "grad_norm": 3.524590253829956, + "learning_rate": 2.1303074670571008e-07, + "loss": 1.1421, + "mean_token_accuracy": 0.6674970388412476, + "num_tokens": 15091567.0, + "step": 583 + }, + { + "epoch": 0.0641335383263782, + "grad_norm": 3.2262349128723145, + "learning_rate": 2.1339677891654466e-07, + "loss": 1.2172, + "mean_token_accuracy": 0.6526905298233032, + "num_tokens": 15117176.0, + "step": 584 + }, + { + "epoch": 0.06424335602899188, + "grad_norm": 2.6839754581451416, + "learning_rate": 2.137628111273792e-07, + "loss": 1.1597, + "mean_token_accuracy": 0.668292224407196, + "num_tokens": 15151656.0, + "step": 585 + }, + { + "epoch": 0.06435317373160554, + "grad_norm": 2.8975577354431152, + "learning_rate": 2.1412884333821375e-07, + "loss": 1.0629, + "mean_token_accuracy": 0.6954790353775024, + "num_tokens": 15176031.0, + "step": 586 + }, + { + "epoch": 0.0644629914342192, + "grad_norm": 3.1390910148620605, + "learning_rate": 2.144948755490483e-07, + "loss": 1.1193, + "mean_token_accuracy": 0.6692193746566772, + "num_tokens": 15197744.0, + "step": 587 + }, + { + "epoch": 0.06457280913683286, + "grad_norm": 3.726433753967285, + "learning_rate": 2.1486090775988285e-07, + "loss": 1.0984, + "mean_token_accuracy": 0.6806284189224243, + "num_tokens": 15219377.0, + "step": 588 + }, + { + "epoch": 0.06468262683944652, + "grad_norm": 2.983743667602539, + "learning_rate": 2.1522693997071743e-07, + "loss": 1.1126, + "mean_token_accuracy": 0.6753726005554199, + "num_tokens": 15244196.0, + "step": 589 + }, + { + "epoch": 0.06479244454206018, + "grad_norm": 2.7894175052642822, + "learning_rate": 2.1559297218155198e-07, + "loss": 1.1649, + "mean_token_accuracy": 0.6742595434188843, + "num_tokens": 15273093.0, + "step": 590 + }, + { + "epoch": 0.06490226224467384, + "grad_norm": 3.023435354232788, + "learning_rate": 2.159590043923865e-07, + "loss": 1.1212, + "mean_token_accuracy": 0.6764988899230957, + "num_tokens": 15299076.0, + "step": 591 + }, + { + "epoch": 0.0650120799472875, + "grad_norm": 3.1504862308502197, + "learning_rate": 2.1632503660322108e-07, + "loss": 1.04, + "mean_token_accuracy": 0.7001880407333374, + "num_tokens": 15323632.0, + "step": 592 + }, + { + "epoch": 0.06512189764990116, + "grad_norm": 2.7018020153045654, + "learning_rate": 2.1669106881405563e-07, + "loss": 1.1371, + "mean_token_accuracy": 0.6725645065307617, + "num_tokens": 15356700.0, + "step": 593 + }, + { + "epoch": 0.06523171535251482, + "grad_norm": 2.8127198219299316, + "learning_rate": 2.170571010248902e-07, + "loss": 1.0448, + "mean_token_accuracy": 0.698154866695404, + "num_tokens": 15382486.0, + "step": 594 + }, + { + "epoch": 0.06534153305512849, + "grad_norm": 3.7363357543945312, + "learning_rate": 2.1742313323572472e-07, + "loss": 1.0708, + "mean_token_accuracy": 0.6920751333236694, + "num_tokens": 15403037.0, + "step": 595 + }, + { + "epoch": 0.06545135075774215, + "grad_norm": 3.1953604221343994, + "learning_rate": 2.177891654465593e-07, + "loss": 1.0661, + "mean_token_accuracy": 0.6901789307594299, + "num_tokens": 15427884.0, + "step": 596 + }, + { + "epoch": 0.06556116846035581, + "grad_norm": 2.5808911323547363, + "learning_rate": 2.1815519765739385e-07, + "loss": 1.1399, + "mean_token_accuracy": 0.6676148176193237, + "num_tokens": 15461669.0, + "step": 597 + }, + { + "epoch": 0.06567098616296947, + "grad_norm": 3.441004753112793, + "learning_rate": 2.185212298682284e-07, + "loss": 1.082, + "mean_token_accuracy": 0.6801406741142273, + "num_tokens": 15484624.0, + "step": 598 + }, + { + "epoch": 0.06578080386558313, + "grad_norm": 2.947899341583252, + "learning_rate": 2.1888726207906295e-07, + "loss": 1.1775, + "mean_token_accuracy": 0.6698120832443237, + "num_tokens": 15508390.0, + "step": 599 + }, + { + "epoch": 0.06589062156819679, + "grad_norm": 2.8820300102233887, + "learning_rate": 2.192532942898975e-07, + "loss": 1.1032, + "mean_token_accuracy": 0.6794074773788452, + "num_tokens": 15535744.0, + "step": 600 + }, + { + "epoch": 0.06600043927081045, + "grad_norm": 2.712113618850708, + "learning_rate": 2.1961932650073207e-07, + "loss": 1.1076, + "mean_token_accuracy": 0.6791846752166748, + "num_tokens": 15565036.0, + "step": 601 + }, + { + "epoch": 0.06611025697342411, + "grad_norm": 2.433363199234009, + "learning_rate": 2.1998535871156662e-07, + "loss": 1.202, + "mean_token_accuracy": 0.6589348316192627, + "num_tokens": 15597806.0, + "step": 602 + }, + { + "epoch": 0.06622007467603777, + "grad_norm": 3.167034864425659, + "learning_rate": 2.2035139092240114e-07, + "loss": 1.1152, + "mean_token_accuracy": 0.6720110774040222, + "num_tokens": 15623941.0, + "step": 603 + }, + { + "epoch": 0.06632989237865144, + "grad_norm": 3.1382527351379395, + "learning_rate": 2.2071742313323572e-07, + "loss": 1.1544, + "mean_token_accuracy": 0.6616818904876709, + "num_tokens": 15649843.0, + "step": 604 + }, + { + "epoch": 0.0664397100812651, + "grad_norm": 2.932135581970215, + "learning_rate": 2.2108345534407027e-07, + "loss": 1.0489, + "mean_token_accuracy": 0.6899476647377014, + "num_tokens": 15677720.0, + "step": 605 + }, + { + "epoch": 0.06654952778387876, + "grad_norm": 2.9723691940307617, + "learning_rate": 2.2144948755490484e-07, + "loss": 1.0579, + "mean_token_accuracy": 0.6884927749633789, + "num_tokens": 15704251.0, + "step": 606 + }, + { + "epoch": 0.06665934548649242, + "grad_norm": 3.325172185897827, + "learning_rate": 2.2181551976573937e-07, + "loss": 1.1377, + "mean_token_accuracy": 0.6684446334838867, + "num_tokens": 15729100.0, + "step": 607 + }, + { + "epoch": 0.06676916318910608, + "grad_norm": 3.087894916534424, + "learning_rate": 2.2218155197657394e-07, + "loss": 1.1639, + "mean_token_accuracy": 0.6569501757621765, + "num_tokens": 15753408.0, + "step": 608 + }, + { + "epoch": 0.06687898089171974, + "grad_norm": 3.219651222229004, + "learning_rate": 2.225475841874085e-07, + "loss": 1.0782, + "mean_token_accuracy": 0.6911170482635498, + "num_tokens": 15777400.0, + "step": 609 + }, + { + "epoch": 0.0669887985943334, + "grad_norm": 2.7011046409606934, + "learning_rate": 2.2291361639824304e-07, + "loss": 1.1919, + "mean_token_accuracy": 0.664283812046051, + "num_tokens": 15808367.0, + "step": 610 + }, + { + "epoch": 0.06709861629694706, + "grad_norm": 3.0773909091949463, + "learning_rate": 2.232796486090776e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.7076832056045532, + "num_tokens": 15831105.0, + "step": 611 + }, + { + "epoch": 0.06720843399956072, + "grad_norm": 3.1118946075439453, + "learning_rate": 2.2364568081991214e-07, + "loss": 1.1299, + "mean_token_accuracy": 0.6740677356719971, + "num_tokens": 15854816.0, + "step": 612 + }, + { + "epoch": 0.06731825170217438, + "grad_norm": 3.418894052505493, + "learning_rate": 2.240117130307467e-07, + "loss": 1.1173, + "mean_token_accuracy": 0.6748814582824707, + "num_tokens": 15873146.0, + "step": 613 + }, + { + "epoch": 0.06742806940478806, + "grad_norm": 2.716681480407715, + "learning_rate": 2.2437774524158126e-07, + "loss": 1.1661, + "mean_token_accuracy": 0.669762134552002, + "num_tokens": 15898651.0, + "step": 614 + }, + { + "epoch": 0.06753788710740172, + "grad_norm": 3.0865671634674072, + "learning_rate": 2.2474377745241578e-07, + "loss": 1.1643, + "mean_token_accuracy": 0.6569715738296509, + "num_tokens": 15921202.0, + "step": 615 + }, + { + "epoch": 0.06764770481001538, + "grad_norm": 3.117743492126465, + "learning_rate": 2.2510980966325036e-07, + "loss": 1.0142, + "mean_token_accuracy": 0.6958410739898682, + "num_tokens": 15940488.0, + "step": 616 + }, + { + "epoch": 0.06775752251262904, + "grad_norm": 2.862530469894409, + "learning_rate": 2.254758418740849e-07, + "loss": 1.0993, + "mean_token_accuracy": 0.6838463544845581, + "num_tokens": 15967477.0, + "step": 617 + }, + { + "epoch": 0.0678673402152427, + "grad_norm": 2.831657648086548, + "learning_rate": 2.2584187408491948e-07, + "loss": 1.1151, + "mean_token_accuracy": 0.6741362810134888, + "num_tokens": 15995250.0, + "step": 618 + }, + { + "epoch": 0.06797715791785636, + "grad_norm": 2.9274516105651855, + "learning_rate": 2.26207906295754e-07, + "loss": 1.226, + "mean_token_accuracy": 0.6544827818870544, + "num_tokens": 16020883.0, + "step": 619 + }, + { + "epoch": 0.06808697562047002, + "grad_norm": 2.7066221237182617, + "learning_rate": 2.2657393850658856e-07, + "loss": 1.2399, + "mean_token_accuracy": 0.6505550146102905, + "num_tokens": 16047600.0, + "step": 620 + }, + { + "epoch": 0.06819679332308368, + "grad_norm": 3.125053644180298, + "learning_rate": 2.2693997071742313e-07, + "loss": 1.1315, + "mean_token_accuracy": 0.6703749299049377, + "num_tokens": 16070115.0, + "step": 621 + }, + { + "epoch": 0.06830661102569734, + "grad_norm": 2.94689679145813, + "learning_rate": 2.2730600292825768e-07, + "loss": 1.1611, + "mean_token_accuracy": 0.6694151163101196, + "num_tokens": 16096102.0, + "step": 622 + }, + { + "epoch": 0.06841642872831101, + "grad_norm": 2.898322820663452, + "learning_rate": 2.2767203513909223e-07, + "loss": 1.196, + "mean_token_accuracy": 0.6625978946685791, + "num_tokens": 16122671.0, + "step": 623 + }, + { + "epoch": 0.06852624643092467, + "grad_norm": 2.8294663429260254, + "learning_rate": 2.2803806734992678e-07, + "loss": 1.0471, + "mean_token_accuracy": 0.6878021955490112, + "num_tokens": 16149145.0, + "step": 624 + }, + { + "epoch": 0.06863606413353833, + "grad_norm": 2.621309995651245, + "learning_rate": 2.2840409956076135e-07, + "loss": 1.1309, + "mean_token_accuracy": 0.6818686127662659, + "num_tokens": 16178336.0, + "step": 625 + }, + { + "epoch": 0.06874588183615199, + "grad_norm": 3.115489959716797, + "learning_rate": 2.287701317715959e-07, + "loss": 1.133, + "mean_token_accuracy": 0.6753160953521729, + "num_tokens": 16200808.0, + "step": 626 + }, + { + "epoch": 0.06885569953876565, + "grad_norm": 2.6366701126098633, + "learning_rate": 2.2913616398243043e-07, + "loss": 1.2537, + "mean_token_accuracy": 0.6429229974746704, + "num_tokens": 16228780.0, + "step": 627 + }, + { + "epoch": 0.06896551724137931, + "grad_norm": 2.896721839904785, + "learning_rate": 2.29502196193265e-07, + "loss": 1.1882, + "mean_token_accuracy": 0.652707040309906, + "num_tokens": 16256783.0, + "step": 628 + }, + { + "epoch": 0.06907533494399297, + "grad_norm": 2.884280204772949, + "learning_rate": 2.2986822840409955e-07, + "loss": 1.163, + "mean_token_accuracy": 0.6583275198936462, + "num_tokens": 16281018.0, + "step": 629 + }, + { + "epoch": 0.06918515264660663, + "grad_norm": 3.0041658878326416, + "learning_rate": 2.3023426061493413e-07, + "loss": 1.0581, + "mean_token_accuracy": 0.6891694068908691, + "num_tokens": 16306575.0, + "step": 630 + }, + { + "epoch": 0.06929497034922029, + "grad_norm": 2.5401973724365234, + "learning_rate": 2.3060029282576865e-07, + "loss": 1.1953, + "mean_token_accuracy": 0.6557273864746094, + "num_tokens": 16337454.0, + "step": 631 + }, + { + "epoch": 0.06940478805183395, + "grad_norm": 2.962475299835205, + "learning_rate": 2.309663250366032e-07, + "loss": 1.1323, + "mean_token_accuracy": 0.6660552024841309, + "num_tokens": 16360648.0, + "step": 632 + }, + { + "epoch": 0.06951460575444762, + "grad_norm": 2.5709869861602783, + "learning_rate": 2.3133235724743777e-07, + "loss": 1.1311, + "mean_token_accuracy": 0.6696734428405762, + "num_tokens": 16389325.0, + "step": 633 + }, + { + "epoch": 0.06962442345706128, + "grad_norm": 3.2216503620147705, + "learning_rate": 2.3169838945827232e-07, + "loss": 1.1623, + "mean_token_accuracy": 0.667812705039978, + "num_tokens": 16412500.0, + "step": 634 + }, + { + "epoch": 0.06973424115967494, + "grad_norm": 2.2427539825439453, + "learning_rate": 2.3206442166910687e-07, + "loss": 1.2094, + "mean_token_accuracy": 0.6468780636787415, + "num_tokens": 16449528.0, + "step": 635 + }, + { + "epoch": 0.0698440588622886, + "grad_norm": 2.6810367107391357, + "learning_rate": 2.3243045387994142e-07, + "loss": 1.0271, + "mean_token_accuracy": 0.705258846282959, + "num_tokens": 16475464.0, + "step": 636 + }, + { + "epoch": 0.06995387656490226, + "grad_norm": 2.654473304748535, + "learning_rate": 2.32796486090776e-07, + "loss": 1.1202, + "mean_token_accuracy": 0.6727931499481201, + "num_tokens": 16503191.0, + "step": 637 + }, + { + "epoch": 0.07006369426751592, + "grad_norm": 2.7126400470733643, + "learning_rate": 2.3316251830161055e-07, + "loss": 1.144, + "mean_token_accuracy": 0.664813220500946, + "num_tokens": 16532650.0, + "step": 638 + }, + { + "epoch": 0.07017351197012958, + "grad_norm": 2.6894094944000244, + "learning_rate": 2.3352855051244507e-07, + "loss": 1.2159, + "mean_token_accuracy": 0.6591262817382812, + "num_tokens": 16560255.0, + "step": 639 + }, + { + "epoch": 0.07028332967274324, + "grad_norm": 2.3951761722564697, + "learning_rate": 2.3389458272327964e-07, + "loss": 1.1731, + "mean_token_accuracy": 0.6591346859931946, + "num_tokens": 16591859.0, + "step": 640 + }, + { + "epoch": 0.0703931473753569, + "grad_norm": 2.6176116466522217, + "learning_rate": 2.342606149341142e-07, + "loss": 1.1648, + "mean_token_accuracy": 0.6656143665313721, + "num_tokens": 16620782.0, + "step": 641 + }, + { + "epoch": 0.07050296507797058, + "grad_norm": 2.888239622116089, + "learning_rate": 2.3462664714494877e-07, + "loss": 1.1725, + "mean_token_accuracy": 0.6545929908752441, + "num_tokens": 16648051.0, + "step": 642 + }, + { + "epoch": 0.07061278278058424, + "grad_norm": 2.603548288345337, + "learning_rate": 2.349926793557833e-07, + "loss": 1.2087, + "mean_token_accuracy": 0.6580303907394409, + "num_tokens": 16677294.0, + "step": 643 + }, + { + "epoch": 0.0707226004831979, + "grad_norm": 2.6245696544647217, + "learning_rate": 2.3535871156661784e-07, + "loss": 1.2419, + "mean_token_accuracy": 0.6458761692047119, + "num_tokens": 16707122.0, + "step": 644 + }, + { + "epoch": 0.07083241818581155, + "grad_norm": 2.505330801010132, + "learning_rate": 2.3572474377745242e-07, + "loss": 1.1447, + "mean_token_accuracy": 0.6655153036117554, + "num_tokens": 16735039.0, + "step": 645 + }, + { + "epoch": 0.07094223588842521, + "grad_norm": 3.085747241973877, + "learning_rate": 2.3609077598828696e-07, + "loss": 1.0863, + "mean_token_accuracy": 0.6867893934249878, + "num_tokens": 16759433.0, + "step": 646 + }, + { + "epoch": 0.07105205359103887, + "grad_norm": 2.6940181255340576, + "learning_rate": 2.3645680819912151e-07, + "loss": 1.0946, + "mean_token_accuracy": 0.6771766543388367, + "num_tokens": 16785559.0, + "step": 647 + }, + { + "epoch": 0.07116187129365253, + "grad_norm": 3.180887460708618, + "learning_rate": 2.3682284040995606e-07, + "loss": 1.079, + "mean_token_accuracy": 0.6845273971557617, + "num_tokens": 16805432.0, + "step": 648 + }, + { + "epoch": 0.0712716889962662, + "grad_norm": 2.897658586502075, + "learning_rate": 2.3718887262079064e-07, + "loss": 1.1008, + "mean_token_accuracy": 0.6799149513244629, + "num_tokens": 16828225.0, + "step": 649 + }, + { + "epoch": 0.07138150669887985, + "grad_norm": 2.8926024436950684, + "learning_rate": 2.375549048316252e-07, + "loss": 1.1388, + "mean_token_accuracy": 0.6768347024917603, + "num_tokens": 16853283.0, + "step": 650 + }, + { + "epoch": 0.07149132440149351, + "grad_norm": 2.742550849914551, + "learning_rate": 2.379209370424597e-07, + "loss": 1.1566, + "mean_token_accuracy": 0.6655136942863464, + "num_tokens": 16880661.0, + "step": 651 + }, + { + "epoch": 0.07160114210410719, + "grad_norm": 2.7705554962158203, + "learning_rate": 2.3828696925329429e-07, + "loss": 1.1948, + "mean_token_accuracy": 0.6557971835136414, + "num_tokens": 16905765.0, + "step": 652 + }, + { + "epoch": 0.07171095980672085, + "grad_norm": 3.2820281982421875, + "learning_rate": 2.3865300146412886e-07, + "loss": 1.1279, + "mean_token_accuracy": 0.6739705801010132, + "num_tokens": 16926151.0, + "step": 653 + }, + { + "epoch": 0.07182077750933451, + "grad_norm": 2.6032018661499023, + "learning_rate": 2.390190336749634e-07, + "loss": 1.1502, + "mean_token_accuracy": 0.6679699420928955, + "num_tokens": 16955098.0, + "step": 654 + }, + { + "epoch": 0.07193059521194817, + "grad_norm": 3.28190016746521, + "learning_rate": 2.393850658857979e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.684471070766449, + "num_tokens": 16975049.0, + "step": 655 + }, + { + "epoch": 0.07204041291456183, + "grad_norm": 3.5436346530914307, + "learning_rate": 2.397510980966325e-07, + "loss": 1.0038, + "mean_token_accuracy": 0.7025696039199829, + "num_tokens": 16993119.0, + "step": 656 + }, + { + "epoch": 0.07215023061717549, + "grad_norm": 2.9045281410217285, + "learning_rate": 2.4011713030746706e-07, + "loss": 1.0823, + "mean_token_accuracy": 0.6827824115753174, + "num_tokens": 17016760.0, + "step": 657 + }, + { + "epoch": 0.07226004831978915, + "grad_norm": 2.2857351303100586, + "learning_rate": 2.404831625183016e-07, + "loss": 1.1437, + "mean_token_accuracy": 0.6639209985733032, + "num_tokens": 17046841.0, + "step": 658 + }, + { + "epoch": 0.0723698660224028, + "grad_norm": 2.66532301902771, + "learning_rate": 2.4084919472913616e-07, + "loss": 1.2292, + "mean_token_accuracy": 0.6481346487998962, + "num_tokens": 17073782.0, + "step": 659 + }, + { + "epoch": 0.07247968372501647, + "grad_norm": 2.5310769081115723, + "learning_rate": 2.412152269399707e-07, + "loss": 1.1659, + "mean_token_accuracy": 0.6679112911224365, + "num_tokens": 17105514.0, + "step": 660 + }, + { + "epoch": 0.07258950142763014, + "grad_norm": 2.8416593074798584, + "learning_rate": 2.4158125915080525e-07, + "loss": 1.1092, + "mean_token_accuracy": 0.6751365065574646, + "num_tokens": 17128183.0, + "step": 661 + }, + { + "epoch": 0.0726993191302438, + "grad_norm": 3.1434035301208496, + "learning_rate": 2.419472913616398e-07, + "loss": 1.1255, + "mean_token_accuracy": 0.668932318687439, + "num_tokens": 17149575.0, + "step": 662 + }, + { + "epoch": 0.07280913683285746, + "grad_norm": 2.7466511726379395, + "learning_rate": 2.4231332357247435e-07, + "loss": 1.0787, + "mean_token_accuracy": 0.6874176263809204, + "num_tokens": 17175446.0, + "step": 663 + }, + { + "epoch": 0.07291895453547112, + "grad_norm": 2.4590165615081787, + "learning_rate": 2.426793557833089e-07, + "loss": 1.0793, + "mean_token_accuracy": 0.6836467385292053, + "num_tokens": 17206437.0, + "step": 664 + }, + { + "epoch": 0.07302877223808478, + "grad_norm": 2.6193344593048096, + "learning_rate": 2.430453879941435e-07, + "loss": 1.1816, + "mean_token_accuracy": 0.6525992155075073, + "num_tokens": 17232640.0, + "step": 665 + }, + { + "epoch": 0.07313858994069844, + "grad_norm": 2.596097230911255, + "learning_rate": 2.4341142020497805e-07, + "loss": 1.1694, + "mean_token_accuracy": 0.6619145274162292, + "num_tokens": 17260421.0, + "step": 666 + }, + { + "epoch": 0.0732484076433121, + "grad_norm": 2.80357027053833, + "learning_rate": 2.4377745241581255e-07, + "loss": 0.9731, + "mean_token_accuracy": 0.7125542759895325, + "num_tokens": 17283253.0, + "step": 667 + }, + { + "epoch": 0.07335822534592576, + "grad_norm": 3.1172444820404053, + "learning_rate": 2.4414348462664715e-07, + "loss": 1.1335, + "mean_token_accuracy": 0.6632641553878784, + "num_tokens": 17303932.0, + "step": 668 + }, + { + "epoch": 0.07346804304853942, + "grad_norm": 2.959700345993042, + "learning_rate": 2.445095168374817e-07, + "loss": 1.0152, + "mean_token_accuracy": 0.6989006996154785, + "num_tokens": 17324634.0, + "step": 669 + }, + { + "epoch": 0.07357786075115308, + "grad_norm": 2.75913405418396, + "learning_rate": 2.4487554904831625e-07, + "loss": 0.9909, + "mean_token_accuracy": 0.7056851387023926, + "num_tokens": 17347927.0, + "step": 670 + }, + { + "epoch": 0.07368767845376675, + "grad_norm": 2.6137778759002686, + "learning_rate": 2.452415812591508e-07, + "loss": 1.0483, + "mean_token_accuracy": 0.6954094767570496, + "num_tokens": 17374596.0, + "step": 671 + }, + { + "epoch": 0.07379749615638041, + "grad_norm": 3.0044474601745605, + "learning_rate": 2.4560761346998535e-07, + "loss": 1.0393, + "mean_token_accuracy": 0.6854168176651001, + "num_tokens": 17397546.0, + "step": 672 + }, + { + "epoch": 0.07390731385899407, + "grad_norm": 2.989506244659424, + "learning_rate": 2.459736456808199e-07, + "loss": 1.2239, + "mean_token_accuracy": 0.6549326181411743, + "num_tokens": 17420746.0, + "step": 673 + }, + { + "epoch": 0.07401713156160773, + "grad_norm": 3.1128745079040527, + "learning_rate": 2.4633967789165444e-07, + "loss": 1.0766, + "mean_token_accuracy": 0.6914466619491577, + "num_tokens": 17441694.0, + "step": 674 + }, + { + "epoch": 0.07412694926422139, + "grad_norm": 2.5937540531158447, + "learning_rate": 2.46705710102489e-07, + "loss": 1.1534, + "mean_token_accuracy": 0.6695252656936646, + "num_tokens": 17468898.0, + "step": 675 + }, + { + "epoch": 0.07423676696683505, + "grad_norm": 3.0947306156158447, + "learning_rate": 2.4707174231332354e-07, + "loss": 1.0355, + "mean_token_accuracy": 0.6929376721382141, + "num_tokens": 17488143.0, + "step": 676 + }, + { + "epoch": 0.07434658466944871, + "grad_norm": 2.6151654720306396, + "learning_rate": 2.4743777452415815e-07, + "loss": 1.0558, + "mean_token_accuracy": 0.6902817487716675, + "num_tokens": 17514042.0, + "step": 677 + }, + { + "epoch": 0.07445640237206237, + "grad_norm": 2.796510934829712, + "learning_rate": 2.478038067349927e-07, + "loss": 1.043, + "mean_token_accuracy": 0.6900087594985962, + "num_tokens": 17538766.0, + "step": 678 + }, + { + "epoch": 0.07456622007467603, + "grad_norm": 2.4958317279815674, + "learning_rate": 2.481698389458272e-07, + "loss": 1.1585, + "mean_token_accuracy": 0.6792969107627869, + "num_tokens": 17567824.0, + "step": 679 + }, + { + "epoch": 0.0746760377772897, + "grad_norm": 2.3869524002075195, + "learning_rate": 2.485358711566618e-07, + "loss": 1.1813, + "mean_token_accuracy": 0.6609454154968262, + "num_tokens": 17599355.0, + "step": 680 + }, + { + "epoch": 0.07478585547990337, + "grad_norm": 2.581305980682373, + "learning_rate": 2.4890190336749634e-07, + "loss": 1.1531, + "mean_token_accuracy": 0.6632917523384094, + "num_tokens": 17625251.0, + "step": 681 + }, + { + "epoch": 0.07489567318251703, + "grad_norm": 2.620534896850586, + "learning_rate": 2.492679355783309e-07, + "loss": 1.1511, + "mean_token_accuracy": 0.6643851399421692, + "num_tokens": 17654107.0, + "step": 682 + }, + { + "epoch": 0.07500549088513069, + "grad_norm": 2.935068368911743, + "learning_rate": 2.4963396778916544e-07, + "loss": 1.053, + "mean_token_accuracy": 0.6945981979370117, + "num_tokens": 17677403.0, + "step": 683 + }, + { + "epoch": 0.07511530858774434, + "grad_norm": 2.4133222103118896, + "learning_rate": 2.5e-07, + "loss": 1.1705, + "mean_token_accuracy": 0.6628766059875488, + "num_tokens": 17706274.0, + "step": 684 + }, + { + "epoch": 0.075225126290358, + "grad_norm": 2.582545280456543, + "learning_rate": 2.5036603221083454e-07, + "loss": 1.1434, + "mean_token_accuracy": 0.6585738658905029, + "num_tokens": 17732923.0, + "step": 685 + }, + { + "epoch": 0.07533494399297166, + "grad_norm": 2.647444725036621, + "learning_rate": 2.507320644216691e-07, + "loss": 1.0877, + "mean_token_accuracy": 0.6865243911743164, + "num_tokens": 17758057.0, + "step": 686 + }, + { + "epoch": 0.07544476169558532, + "grad_norm": 2.424535036087036, + "learning_rate": 2.5109809663250364e-07, + "loss": 1.0872, + "mean_token_accuracy": 0.6795177459716797, + "num_tokens": 17787037.0, + "step": 687 + }, + { + "epoch": 0.07555457939819898, + "grad_norm": 3.290661334991455, + "learning_rate": 2.514641288433382e-07, + "loss": 1.1431, + "mean_token_accuracy": 0.67202228307724, + "num_tokens": 17807531.0, + "step": 688 + }, + { + "epoch": 0.07566439710081264, + "grad_norm": 3.477001190185547, + "learning_rate": 2.518301610541728e-07, + "loss": 1.0343, + "mean_token_accuracy": 0.6912239789962769, + "num_tokens": 17825910.0, + "step": 689 + }, + { + "epoch": 0.07577421480342632, + "grad_norm": 3.0343449115753174, + "learning_rate": 2.521961932650073e-07, + "loss": 1.1926, + "mean_token_accuracy": 0.6511669158935547, + "num_tokens": 17849674.0, + "step": 690 + }, + { + "epoch": 0.07588403250603998, + "grad_norm": 2.3396942615509033, + "learning_rate": 2.525622254758419e-07, + "loss": 1.2061, + "mean_token_accuracy": 0.6598178148269653, + "num_tokens": 17884281.0, + "step": 691 + }, + { + "epoch": 0.07599385020865364, + "grad_norm": 2.6978323459625244, + "learning_rate": 2.5292825768667643e-07, + "loss": 1.1209, + "mean_token_accuracy": 0.6761479377746582, + "num_tokens": 17907327.0, + "step": 692 + }, + { + "epoch": 0.0761036679112673, + "grad_norm": 2.6583118438720703, + "learning_rate": 2.5329428989751093e-07, + "loss": 1.1168, + "mean_token_accuracy": 0.6767540574073792, + "num_tokens": 17931212.0, + "step": 693 + }, + { + "epoch": 0.07621348561388096, + "grad_norm": 2.777357816696167, + "learning_rate": 2.5366032210834553e-07, + "loss": 1.181, + "mean_token_accuracy": 0.6598440408706665, + "num_tokens": 17957452.0, + "step": 694 + }, + { + "epoch": 0.07632330331649462, + "grad_norm": 3.566582679748535, + "learning_rate": 2.540263543191801e-07, + "loss": 1.0588, + "mean_token_accuracy": 0.6821951866149902, + "num_tokens": 17974591.0, + "step": 695 + }, + { + "epoch": 0.07643312101910828, + "grad_norm": 2.710599660873413, + "learning_rate": 2.543923865300146e-07, + "loss": 1.1035, + "mean_token_accuracy": 0.6857475638389587, + "num_tokens": 18002528.0, + "step": 696 + }, + { + "epoch": 0.07654293872172194, + "grad_norm": 2.7524945735931396, + "learning_rate": 2.547584187408492e-07, + "loss": 1.1207, + "mean_token_accuracy": 0.6693429946899414, + "num_tokens": 18027367.0, + "step": 697 + }, + { + "epoch": 0.0766527564243356, + "grad_norm": 2.6640539169311523, + "learning_rate": 2.5512445095168373e-07, + "loss": 1.152, + "mean_token_accuracy": 0.6630476713180542, + "num_tokens": 18052873.0, + "step": 698 + }, + { + "epoch": 0.07676257412694927, + "grad_norm": 2.3868024349212646, + "learning_rate": 2.5549048316251833e-07, + "loss": 1.1389, + "mean_token_accuracy": 0.6663196086883545, + "num_tokens": 18086772.0, + "step": 699 + }, + { + "epoch": 0.07687239182956293, + "grad_norm": 2.5763540267944336, + "learning_rate": 2.5585651537335283e-07, + "loss": 1.1751, + "mean_token_accuracy": 0.6600112318992615, + "num_tokens": 18115267.0, + "step": 700 + }, + { + "epoch": 0.07698220953217659, + "grad_norm": 2.6137657165527344, + "learning_rate": 2.562225475841874e-07, + "loss": 1.1309, + "mean_token_accuracy": 0.6711146831512451, + "num_tokens": 18144689.0, + "step": 701 + }, + { + "epoch": 0.07709202723479025, + "grad_norm": 2.787252426147461, + "learning_rate": 2.56588579795022e-07, + "loss": 1.1415, + "mean_token_accuracy": 0.6738627552986145, + "num_tokens": 18168911.0, + "step": 702 + }, + { + "epoch": 0.07720184493740391, + "grad_norm": 2.905311107635498, + "learning_rate": 2.569546120058565e-07, + "loss": 1.108, + "mean_token_accuracy": 0.6760891675949097, + "num_tokens": 18193758.0, + "step": 703 + }, + { + "epoch": 0.07731166264001757, + "grad_norm": 2.898754119873047, + "learning_rate": 2.57320644216691e-07, + "loss": 1.1012, + "mean_token_accuracy": 0.6738909482955933, + "num_tokens": 18216265.0, + "step": 704 + }, + { + "epoch": 0.07742148034263123, + "grad_norm": 2.6768298149108887, + "learning_rate": 2.576866764275256e-07, + "loss": 1.0708, + "mean_token_accuracy": 0.6823383569717407, + "num_tokens": 18240261.0, + "step": 705 + }, + { + "epoch": 0.07753129804524489, + "grad_norm": 2.5007362365722656, + "learning_rate": 2.580527086383602e-07, + "loss": 1.207, + "mean_token_accuracy": 0.6509036421775818, + "num_tokens": 18267819.0, + "step": 706 + }, + { + "epoch": 0.07764111574785855, + "grad_norm": 2.5534114837646484, + "learning_rate": 2.584187408491947e-07, + "loss": 1.1131, + "mean_token_accuracy": 0.6785297393798828, + "num_tokens": 18295498.0, + "step": 707 + }, + { + "epoch": 0.07775093345047221, + "grad_norm": 2.4959347248077393, + "learning_rate": 2.5878477306002927e-07, + "loss": 1.1473, + "mean_token_accuracy": 0.6616250276565552, + "num_tokens": 18324344.0, + "step": 708 + }, + { + "epoch": 0.07786075115308588, + "grad_norm": 3.1350438594818115, + "learning_rate": 2.591508052708638e-07, + "loss": 1.1138, + "mean_token_accuracy": 0.6757638454437256, + "num_tokens": 18345249.0, + "step": 709 + }, + { + "epoch": 0.07797056885569954, + "grad_norm": 2.7564003467559814, + "learning_rate": 2.5951683748169837e-07, + "loss": 1.2171, + "mean_token_accuracy": 0.6516847014427185, + "num_tokens": 18369976.0, + "step": 710 + }, + { + "epoch": 0.0780803865583132, + "grad_norm": 2.89544415473938, + "learning_rate": 2.598828696925329e-07, + "loss": 1.0485, + "mean_token_accuracy": 0.6891285181045532, + "num_tokens": 18389828.0, + "step": 711 + }, + { + "epoch": 0.07819020426092686, + "grad_norm": 2.438324451446533, + "learning_rate": 2.6024890190336747e-07, + "loss": 1.1243, + "mean_token_accuracy": 0.6795877814292908, + "num_tokens": 18418228.0, + "step": 712 + }, + { + "epoch": 0.07830002196354052, + "grad_norm": 2.6848080158233643, + "learning_rate": 2.6061493411420207e-07, + "loss": 1.123, + "mean_token_accuracy": 0.6640409231185913, + "num_tokens": 18442229.0, + "step": 713 + }, + { + "epoch": 0.07840983966615418, + "grad_norm": 2.502408742904663, + "learning_rate": 2.6098096632503657e-07, + "loss": 1.1541, + "mean_token_accuracy": 0.6648616790771484, + "num_tokens": 18469419.0, + "step": 714 + }, + { + "epoch": 0.07851965736876784, + "grad_norm": 2.568794012069702, + "learning_rate": 2.6134699853587117e-07, + "loss": 0.9561, + "mean_token_accuracy": 0.7132140398025513, + "num_tokens": 18493762.0, + "step": 715 + }, + { + "epoch": 0.0786294750713815, + "grad_norm": 3.000983238220215, + "learning_rate": 2.617130307467057e-07, + "loss": 1.1223, + "mean_token_accuracy": 0.6712412238121033, + "num_tokens": 18516052.0, + "step": 716 + }, + { + "epoch": 0.07873929277399516, + "grad_norm": 2.4317660331726074, + "learning_rate": 2.620790629575402e-07, + "loss": 1.1112, + "mean_token_accuracy": 0.6807543039321899, + "num_tokens": 18542974.0, + "step": 717 + }, + { + "epoch": 0.07884911047660884, + "grad_norm": 3.3216519355773926, + "learning_rate": 2.624450951683748e-07, + "loss": 1.0485, + "mean_token_accuracy": 0.6867311596870422, + "num_tokens": 18561988.0, + "step": 718 + }, + { + "epoch": 0.0789589281792225, + "grad_norm": 2.6449015140533447, + "learning_rate": 2.6281112737920937e-07, + "loss": 1.1676, + "mean_token_accuracy": 0.6630620360374451, + "num_tokens": 18588546.0, + "step": 719 + }, + { + "epoch": 0.07906874588183616, + "grad_norm": 2.743877410888672, + "learning_rate": 2.6317715959004386e-07, + "loss": 1.1952, + "mean_token_accuracy": 0.6523988842964172, + "num_tokens": 18615106.0, + "step": 720 + }, + { + "epoch": 0.07917856358444982, + "grad_norm": 2.4111037254333496, + "learning_rate": 2.6354319180087846e-07, + "loss": 1.1106, + "mean_token_accuracy": 0.6755712032318115, + "num_tokens": 18642433.0, + "step": 721 + }, + { + "epoch": 0.07928838128706348, + "grad_norm": 2.4593231678009033, + "learning_rate": 2.63909224011713e-07, + "loss": 1.1051, + "mean_token_accuracy": 0.673064112663269, + "num_tokens": 18670567.0, + "step": 722 + }, + { + "epoch": 0.07939819898967714, + "grad_norm": 2.9124755859375, + "learning_rate": 2.642752562225476e-07, + "loss": 1.1156, + "mean_token_accuracy": 0.6711128354072571, + "num_tokens": 18691814.0, + "step": 723 + }, + { + "epoch": 0.0795080166922908, + "grad_norm": 2.5653467178344727, + "learning_rate": 2.646412884333821e-07, + "loss": 1.1507, + "mean_token_accuracy": 0.6665278673171997, + "num_tokens": 18717182.0, + "step": 724 + }, + { + "epoch": 0.07961783439490445, + "grad_norm": 2.367548704147339, + "learning_rate": 2.6500732064421666e-07, + "loss": 1.1302, + "mean_token_accuracy": 0.6652135848999023, + "num_tokens": 18746831.0, + "step": 725 + }, + { + "epoch": 0.07972765209751811, + "grad_norm": 2.4081926345825195, + "learning_rate": 2.6537335285505126e-07, + "loss": 1.0717, + "mean_token_accuracy": 0.682959258556366, + "num_tokens": 18777233.0, + "step": 726 + }, + { + "epoch": 0.07983746980013177, + "grad_norm": 2.6321237087249756, + "learning_rate": 2.6573938506588576e-07, + "loss": 1.1228, + "mean_token_accuracy": 0.6772698163986206, + "num_tokens": 18802479.0, + "step": 727 + }, + { + "epoch": 0.07994728750274545, + "grad_norm": 2.3904242515563965, + "learning_rate": 2.661054172767203e-07, + "loss": 1.1458, + "mean_token_accuracy": 0.6626582741737366, + "num_tokens": 18831630.0, + "step": 728 + }, + { + "epoch": 0.08005710520535911, + "grad_norm": 2.8859355449676514, + "learning_rate": 2.664714494875549e-07, + "loss": 0.9643, + "mean_token_accuracy": 0.7153840065002441, + "num_tokens": 18853499.0, + "step": 729 + }, + { + "epoch": 0.08016692290797277, + "grad_norm": 2.827655553817749, + "learning_rate": 2.6683748169838946e-07, + "loss": 1.1802, + "mean_token_accuracy": 0.6641929745674133, + "num_tokens": 18875616.0, + "step": 730 + }, + { + "epoch": 0.08027674061058643, + "grad_norm": 2.874614953994751, + "learning_rate": 2.67203513909224e-07, + "loss": 1.1067, + "mean_token_accuracy": 0.6781864762306213, + "num_tokens": 18895943.0, + "step": 731 + }, + { + "epoch": 0.08038655831320009, + "grad_norm": 2.147780179977417, + "learning_rate": 2.6756954612005856e-07, + "loss": 1.1625, + "mean_token_accuracy": 0.6626900434494019, + "num_tokens": 18930183.0, + "step": 732 + }, + { + "epoch": 0.08049637601581375, + "grad_norm": 2.811235189437866, + "learning_rate": 2.679355783308931e-07, + "loss": 1.1611, + "mean_token_accuracy": 0.6635661125183105, + "num_tokens": 18953193.0, + "step": 733 + }, + { + "epoch": 0.08060619371842741, + "grad_norm": 2.414780616760254, + "learning_rate": 2.6830161054172765e-07, + "loss": 1.1924, + "mean_token_accuracy": 0.6581979393959045, + "num_tokens": 18981251.0, + "step": 734 + }, + { + "epoch": 0.08071601142104107, + "grad_norm": 2.506234884262085, + "learning_rate": 2.686676427525622e-07, + "loss": 1.1416, + "mean_token_accuracy": 0.6783076524734497, + "num_tokens": 19009491.0, + "step": 735 + }, + { + "epoch": 0.08082582912365473, + "grad_norm": 2.4001290798187256, + "learning_rate": 2.6903367496339675e-07, + "loss": 1.0729, + "mean_token_accuracy": 0.6844595670700073, + "num_tokens": 19037182.0, + "step": 736 + }, + { + "epoch": 0.0809356468262684, + "grad_norm": 2.7792272567749023, + "learning_rate": 2.693997071742313e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.6810504198074341, + "num_tokens": 19059901.0, + "step": 737 + }, + { + "epoch": 0.08104546452888206, + "grad_norm": 2.6166558265686035, + "learning_rate": 2.6976573938506585e-07, + "loss": 1.1682, + "mean_token_accuracy": 0.6635751724243164, + "num_tokens": 19085117.0, + "step": 738 + }, + { + "epoch": 0.08115528223149572, + "grad_norm": 2.5093984603881836, + "learning_rate": 2.7013177159590045e-07, + "loss": 1.0753, + "mean_token_accuracy": 0.682453989982605, + "num_tokens": 19111093.0, + "step": 739 + }, + { + "epoch": 0.08126509993410938, + "grad_norm": 2.6781086921691895, + "learning_rate": 2.70497803806735e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6762344837188721, + "num_tokens": 19135550.0, + "step": 740 + }, + { + "epoch": 0.08137491763672304, + "grad_norm": 2.5152628421783447, + "learning_rate": 2.708638360175695e-07, + "loss": 1.0611, + "mean_token_accuracy": 0.6857912540435791, + "num_tokens": 19162633.0, + "step": 741 + }, + { + "epoch": 0.0814847353393367, + "grad_norm": 2.548175811767578, + "learning_rate": 2.712298682284041e-07, + "loss": 1.0538, + "mean_token_accuracy": 0.6962891817092896, + "num_tokens": 19186813.0, + "step": 742 + }, + { + "epoch": 0.08159455304195036, + "grad_norm": 2.353850841522217, + "learning_rate": 2.7159590043923865e-07, + "loss": 1.1021, + "mean_token_accuracy": 0.674694836139679, + "num_tokens": 19217664.0, + "step": 743 + }, + { + "epoch": 0.08170437074456402, + "grad_norm": 3.358088731765747, + "learning_rate": 2.7196193265007315e-07, + "loss": 1.0819, + "mean_token_accuracy": 0.6772592067718506, + "num_tokens": 19239127.0, + "step": 744 + }, + { + "epoch": 0.08181418844717768, + "grad_norm": 2.690098762512207, + "learning_rate": 2.7232796486090775e-07, + "loss": 1.1034, + "mean_token_accuracy": 0.678458571434021, + "num_tokens": 19263028.0, + "step": 745 + }, + { + "epoch": 0.08192400614979134, + "grad_norm": 2.3347890377044678, + "learning_rate": 2.726939970717423e-07, + "loss": 1.1842, + "mean_token_accuracy": 0.6538119316101074, + "num_tokens": 19292808.0, + "step": 746 + }, + { + "epoch": 0.08203382385240501, + "grad_norm": 2.68070387840271, + "learning_rate": 2.730600292825769e-07, + "loss": 1.115, + "mean_token_accuracy": 0.6740881204605103, + "num_tokens": 19314443.0, + "step": 747 + }, + { + "epoch": 0.08214364155501867, + "grad_norm": 2.622652053833008, + "learning_rate": 2.734260614934114e-07, + "loss": 1.0638, + "mean_token_accuracy": 0.6909395456314087, + "num_tokens": 19339945.0, + "step": 748 + }, + { + "epoch": 0.08225345925763233, + "grad_norm": 3.1220016479492188, + "learning_rate": 2.7379209370424594e-07, + "loss": 1.1827, + "mean_token_accuracy": 0.6587932705879211, + "num_tokens": 19370487.0, + "step": 749 + }, + { + "epoch": 0.08236327696024599, + "grad_norm": 2.5361287593841553, + "learning_rate": 2.7415812591508055e-07, + "loss": 1.0839, + "mean_token_accuracy": 0.6803987622261047, + "num_tokens": 19395644.0, + "step": 750 + }, + { + "epoch": 0.08247309466285965, + "grad_norm": 2.2209513187408447, + "learning_rate": 2.7452415812591504e-07, + "loss": 1.165, + "mean_token_accuracy": 0.6594472527503967, + "num_tokens": 19426218.0, + "step": 751 + }, + { + "epoch": 0.08258291236547331, + "grad_norm": 2.5879642963409424, + "learning_rate": 2.748901903367496e-07, + "loss": 1.1621, + "mean_token_accuracy": 0.6662677526473999, + "num_tokens": 19451846.0, + "step": 752 + }, + { + "epoch": 0.08269273006808697, + "grad_norm": 2.5962862968444824, + "learning_rate": 2.752562225475842e-07, + "loss": 1.0851, + "mean_token_accuracy": 0.687134861946106, + "num_tokens": 19476296.0, + "step": 753 + }, + { + "epoch": 0.08280254777070063, + "grad_norm": 2.4098708629608154, + "learning_rate": 2.756222547584187e-07, + "loss": 1.1248, + "mean_token_accuracy": 0.679540753364563, + "num_tokens": 19504200.0, + "step": 754 + }, + { + "epoch": 0.08291236547331429, + "grad_norm": 2.6787304878234863, + "learning_rate": 2.759882869692533e-07, + "loss": 1.2021, + "mean_token_accuracy": 0.656144917011261, + "num_tokens": 19527134.0, + "step": 755 + }, + { + "epoch": 0.08302218317592797, + "grad_norm": 2.3991594314575195, + "learning_rate": 2.7635431918008784e-07, + "loss": 1.0858, + "mean_token_accuracy": 0.682961106300354, + "num_tokens": 19556429.0, + "step": 756 + }, + { + "epoch": 0.08313200087854163, + "grad_norm": 2.203265905380249, + "learning_rate": 2.767203513909224e-07, + "loss": 1.1597, + "mean_token_accuracy": 0.6621212959289551, + "num_tokens": 19587759.0, + "step": 757 + }, + { + "epoch": 0.08324181858115529, + "grad_norm": 3.0797715187072754, + "learning_rate": 2.7708638360175694e-07, + "loss": 1.1086, + "mean_token_accuracy": 0.6808457374572754, + "num_tokens": 19607412.0, + "step": 758 + }, + { + "epoch": 0.08335163628376895, + "grad_norm": 2.6321828365325928, + "learning_rate": 2.774524158125915e-07, + "loss": 1.0043, + "mean_token_accuracy": 0.7096770405769348, + "num_tokens": 19630184.0, + "step": 759 + }, + { + "epoch": 0.0834614539863826, + "grad_norm": 2.6749041080474854, + "learning_rate": 2.7781844802342604e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.6822051405906677, + "num_tokens": 19655787.0, + "step": 760 + }, + { + "epoch": 0.08357127168899627, + "grad_norm": 2.3942501544952393, + "learning_rate": 2.781844802342606e-07, + "loss": 1.0519, + "mean_token_accuracy": 0.6925632953643799, + "num_tokens": 19683813.0, + "step": 761 + }, + { + "epoch": 0.08368108939160993, + "grad_norm": 2.7941484451293945, + "learning_rate": 2.7855051244509513e-07, + "loss": 1.0802, + "mean_token_accuracy": 0.6816631555557251, + "num_tokens": 19706512.0, + "step": 762 + }, + { + "epoch": 0.08379090709422359, + "grad_norm": 2.346116781234741, + "learning_rate": 2.7891654465592974e-07, + "loss": 1.1498, + "mean_token_accuracy": 0.6704909205436707, + "num_tokens": 19737097.0, + "step": 763 + }, + { + "epoch": 0.08390072479683724, + "grad_norm": 2.583561658859253, + "learning_rate": 2.792825768667643e-07, + "loss": 1.0047, + "mean_token_accuracy": 0.703596830368042, + "num_tokens": 19762815.0, + "step": 764 + }, + { + "epoch": 0.0840105424994509, + "grad_norm": 2.7466988563537598, + "learning_rate": 2.796486090775988e-07, + "loss": 1.1177, + "mean_token_accuracy": 0.6750551462173462, + "num_tokens": 19785444.0, + "step": 765 + }, + { + "epoch": 0.08412036020206458, + "grad_norm": 2.7380752563476562, + "learning_rate": 2.800146412884334e-07, + "loss": 1.1249, + "mean_token_accuracy": 0.6717573404312134, + "num_tokens": 19809678.0, + "step": 766 + }, + { + "epoch": 0.08423017790467824, + "grad_norm": 2.556196689605713, + "learning_rate": 2.8038067349926793e-07, + "loss": 1.0989, + "mean_token_accuracy": 0.6749445199966431, + "num_tokens": 19834658.0, + "step": 767 + }, + { + "epoch": 0.0843399956072919, + "grad_norm": 2.466113328933716, + "learning_rate": 2.8074670571010243e-07, + "loss": 1.1851, + "mean_token_accuracy": 0.6610159277915955, + "num_tokens": 19860076.0, + "step": 768 + }, + { + "epoch": 0.08444981330990556, + "grad_norm": 2.8001725673675537, + "learning_rate": 2.8111273792093703e-07, + "loss": 1.1394, + "mean_token_accuracy": 0.6690274477005005, + "num_tokens": 19885069.0, + "step": 769 + }, + { + "epoch": 0.08455963101251922, + "grad_norm": 2.6467783451080322, + "learning_rate": 2.814787701317716e-07, + "loss": 1.1994, + "mean_token_accuracy": 0.6611584424972534, + "num_tokens": 19909921.0, + "step": 770 + }, + { + "epoch": 0.08466944871513288, + "grad_norm": 2.4769411087036133, + "learning_rate": 2.818448023426062e-07, + "loss": 1.0551, + "mean_token_accuracy": 0.6977632641792297, + "num_tokens": 19935788.0, + "step": 771 + }, + { + "epoch": 0.08477926641774654, + "grad_norm": 2.4126033782958984, + "learning_rate": 2.822108345534407e-07, + "loss": 1.1328, + "mean_token_accuracy": 0.6675214767456055, + "num_tokens": 19965350.0, + "step": 772 + }, + { + "epoch": 0.0848890841203602, + "grad_norm": 2.714141607284546, + "learning_rate": 2.8257686676427523e-07, + "loss": 1.0055, + "mean_token_accuracy": 0.7040688395500183, + "num_tokens": 19985991.0, + "step": 773 + }, + { + "epoch": 0.08499890182297386, + "grad_norm": 2.4760661125183105, + "learning_rate": 2.8294289897510983e-07, + "loss": 1.0837, + "mean_token_accuracy": 0.6833673715591431, + "num_tokens": 20013789.0, + "step": 774 + }, + { + "epoch": 0.08510871952558753, + "grad_norm": 2.7943644523620605, + "learning_rate": 2.833089311859443e-07, + "loss": 1.0243, + "mean_token_accuracy": 0.6928861141204834, + "num_tokens": 20034001.0, + "step": 775 + }, + { + "epoch": 0.08521853722820119, + "grad_norm": 2.5749897956848145, + "learning_rate": 2.836749633967789e-07, + "loss": 1.1447, + "mean_token_accuracy": 0.6641843318939209, + "num_tokens": 20060222.0, + "step": 776 + }, + { + "epoch": 0.08532835493081485, + "grad_norm": 2.274005651473999, + "learning_rate": 2.840409956076135e-07, + "loss": 1.0789, + "mean_token_accuracy": 0.6835291981697083, + "num_tokens": 20093804.0, + "step": 777 + }, + { + "epoch": 0.08543817263342851, + "grad_norm": 2.2433972358703613, + "learning_rate": 2.8440702781844797e-07, + "loss": 1.2162, + "mean_token_accuracy": 0.6427496671676636, + "num_tokens": 20126639.0, + "step": 778 + }, + { + "epoch": 0.08554799033604217, + "grad_norm": 2.57893705368042, + "learning_rate": 2.847730600292826e-07, + "loss": 1.0142, + "mean_token_accuracy": 0.704839825630188, + "num_tokens": 20154205.0, + "step": 779 + }, + { + "epoch": 0.08565780803865583, + "grad_norm": 2.266387939453125, + "learning_rate": 2.851390922401171e-07, + "loss": 1.2064, + "mean_token_accuracy": 0.6647634506225586, + "num_tokens": 20185261.0, + "step": 780 + }, + { + "epoch": 0.08576762574126949, + "grad_norm": 2.8049938678741455, + "learning_rate": 2.8550512445095167e-07, + "loss": 1.0079, + "mean_token_accuracy": 0.69487464427948, + "num_tokens": 20205777.0, + "step": 781 + }, + { + "epoch": 0.08587744344388315, + "grad_norm": 2.2117888927459717, + "learning_rate": 2.858711566617862e-07, + "loss": 1.0528, + "mean_token_accuracy": 0.6894241571426392, + "num_tokens": 20237386.0, + "step": 782 + }, + { + "epoch": 0.08598726114649681, + "grad_norm": 2.578786849975586, + "learning_rate": 2.8623718887262077e-07, + "loss": 1.1108, + "mean_token_accuracy": 0.6784124374389648, + "num_tokens": 20261696.0, + "step": 783 + }, + { + "epoch": 0.08609707884911047, + "grad_norm": 2.4877963066101074, + "learning_rate": 2.866032210834553e-07, + "loss": 1.0787, + "mean_token_accuracy": 0.6815659999847412, + "num_tokens": 20287351.0, + "step": 784 + }, + { + "epoch": 0.08620689655172414, + "grad_norm": 2.7265686988830566, + "learning_rate": 2.8696925329428987e-07, + "loss": 0.9995, + "mean_token_accuracy": 0.7025378942489624, + "num_tokens": 20309391.0, + "step": 785 + }, + { + "epoch": 0.0863167142543378, + "grad_norm": 2.733807325363159, + "learning_rate": 2.873352855051244e-07, + "loss": 1.0592, + "mean_token_accuracy": 0.6826949119567871, + "num_tokens": 20332515.0, + "step": 786 + }, + { + "epoch": 0.08642653195695146, + "grad_norm": 2.526954412460327, + "learning_rate": 2.87701317715959e-07, + "loss": 1.0965, + "mean_token_accuracy": 0.6858435869216919, + "num_tokens": 20359379.0, + "step": 787 + }, + { + "epoch": 0.08653634965956512, + "grad_norm": 2.5793678760528564, + "learning_rate": 2.8806734992679357e-07, + "loss": 1.1262, + "mean_token_accuracy": 0.6671164035797119, + "num_tokens": 20383436.0, + "step": 788 + }, + { + "epoch": 0.08664616736217878, + "grad_norm": 2.201709270477295, + "learning_rate": 2.8843338213762807e-07, + "loss": 1.1038, + "mean_token_accuracy": 0.6767393350601196, + "num_tokens": 20414443.0, + "step": 789 + }, + { + "epoch": 0.08675598506479244, + "grad_norm": 2.752084493637085, + "learning_rate": 2.8879941434846267e-07, + "loss": 1.0645, + "mean_token_accuracy": 0.6806840896606445, + "num_tokens": 20437080.0, + "step": 790 + }, + { + "epoch": 0.0868658027674061, + "grad_norm": 2.540066957473755, + "learning_rate": 2.891654465592972e-07, + "loss": 0.9614, + "mean_token_accuracy": 0.7143040895462036, + "num_tokens": 20461294.0, + "step": 791 + }, + { + "epoch": 0.08697562047001976, + "grad_norm": 2.569772958755493, + "learning_rate": 2.895314787701317e-07, + "loss": 1.0999, + "mean_token_accuracy": 0.6737529039382935, + "num_tokens": 20486377.0, + "step": 792 + }, + { + "epoch": 0.08708543817263342, + "grad_norm": 2.7829151153564453, + "learning_rate": 2.898975109809663e-07, + "loss": 1.053, + "mean_token_accuracy": 0.690364420413971, + "num_tokens": 20508898.0, + "step": 793 + }, + { + "epoch": 0.0871952558752471, + "grad_norm": 2.3132941722869873, + "learning_rate": 2.9026354319180086e-07, + "loss": 1.1311, + "mean_token_accuracy": 0.669199526309967, + "num_tokens": 20538441.0, + "step": 794 + }, + { + "epoch": 0.08730507357786076, + "grad_norm": 2.4792027473449707, + "learning_rate": 2.9062957540263547e-07, + "loss": 1.0773, + "mean_token_accuracy": 0.6839430332183838, + "num_tokens": 20561412.0, + "step": 795 + }, + { + "epoch": 0.08741489128047442, + "grad_norm": 2.6120729446411133, + "learning_rate": 2.9099560761346996e-07, + "loss": 1.0768, + "mean_token_accuracy": 0.6825504302978516, + "num_tokens": 20586226.0, + "step": 796 + }, + { + "epoch": 0.08752470898308808, + "grad_norm": 2.256944179534912, + "learning_rate": 2.913616398243045e-07, + "loss": 1.2003, + "mean_token_accuracy": 0.6547267436981201, + "num_tokens": 20618182.0, + "step": 797 + }, + { + "epoch": 0.08763452668570174, + "grad_norm": 2.4138667583465576, + "learning_rate": 2.917276720351391e-07, + "loss": 1.1646, + "mean_token_accuracy": 0.6706714630126953, + "num_tokens": 20647672.0, + "step": 798 + }, + { + "epoch": 0.0877443443883154, + "grad_norm": 2.6284091472625732, + "learning_rate": 2.920937042459736e-07, + "loss": 1.1034, + "mean_token_accuracy": 0.6897978186607361, + "num_tokens": 20671724.0, + "step": 799 + }, + { + "epoch": 0.08785416209092906, + "grad_norm": 2.2980992794036865, + "learning_rate": 2.9245973645680816e-07, + "loss": 1.1446, + "mean_token_accuracy": 0.6825972199440002, + "num_tokens": 20700193.0, + "step": 800 + }, + { + "epoch": 0.08796397979354272, + "grad_norm": 2.564450263977051, + "learning_rate": 2.9282576866764276e-07, + "loss": 1.0705, + "mean_token_accuracy": 0.6807981729507446, + "num_tokens": 20724905.0, + "step": 801 + }, + { + "epoch": 0.08807379749615638, + "grad_norm": 2.538393259048462, + "learning_rate": 2.9319180087847726e-07, + "loss": 1.1288, + "mean_token_accuracy": 0.6703101396560669, + "num_tokens": 20749749.0, + "step": 802 + }, + { + "epoch": 0.08818361519877004, + "grad_norm": 2.2942631244659424, + "learning_rate": 2.9355783308931186e-07, + "loss": 1.0025, + "mean_token_accuracy": 0.6978905200958252, + "num_tokens": 20778139.0, + "step": 803 + }, + { + "epoch": 0.08829343290138371, + "grad_norm": 2.5246903896331787, + "learning_rate": 2.939238653001464e-07, + "loss": 1.1146, + "mean_token_accuracy": 0.6865822076797485, + "num_tokens": 20801934.0, + "step": 804 + }, + { + "epoch": 0.08840325060399737, + "grad_norm": 2.3273375034332275, + "learning_rate": 2.9428989751098096e-07, + "loss": 1.1122, + "mean_token_accuracy": 0.6770038604736328, + "num_tokens": 20832336.0, + "step": 805 + }, + { + "epoch": 0.08851306830661103, + "grad_norm": 2.4489612579345703, + "learning_rate": 2.946559297218155e-07, + "loss": 1.1165, + "mean_token_accuracy": 0.670748233795166, + "num_tokens": 20858646.0, + "step": 806 + }, + { + "epoch": 0.08862288600922469, + "grad_norm": 2.355435848236084, + "learning_rate": 2.9502196193265006e-07, + "loss": 1.141, + "mean_token_accuracy": 0.6674596071243286, + "num_tokens": 20887602.0, + "step": 807 + }, + { + "epoch": 0.08873270371183835, + "grad_norm": 2.6357712745666504, + "learning_rate": 2.953879941434846e-07, + "loss": 1.1662, + "mean_token_accuracy": 0.6594538688659668, + "num_tokens": 20912488.0, + "step": 808 + }, + { + "epoch": 0.08884252141445201, + "grad_norm": 2.3242082595825195, + "learning_rate": 2.9575402635431915e-07, + "loss": 1.1889, + "mean_token_accuracy": 0.6537724733352661, + "num_tokens": 20940624.0, + "step": 809 + }, + { + "epoch": 0.08895233911706567, + "grad_norm": 2.3227603435516357, + "learning_rate": 2.961200585651537e-07, + "loss": 1.084, + "mean_token_accuracy": 0.6831519603729248, + "num_tokens": 20970873.0, + "step": 810 + }, + { + "epoch": 0.08906215681967933, + "grad_norm": 2.6390035152435303, + "learning_rate": 2.964860907759883e-07, + "loss": 1.0231, + "mean_token_accuracy": 0.694068431854248, + "num_tokens": 20993869.0, + "step": 811 + }, + { + "epoch": 0.08917197452229299, + "grad_norm": 2.4265568256378174, + "learning_rate": 2.9685212298682285e-07, + "loss": 1.0287, + "mean_token_accuracy": 0.6931077241897583, + "num_tokens": 21019651.0, + "step": 812 + }, + { + "epoch": 0.08928179222490666, + "grad_norm": 2.2966806888580322, + "learning_rate": 2.9721815519765735e-07, + "loss": 1.0596, + "mean_token_accuracy": 0.6892102360725403, + "num_tokens": 21048736.0, + "step": 813 + }, + { + "epoch": 0.08939160992752032, + "grad_norm": 2.6838903427124023, + "learning_rate": 2.9758418740849195e-07, + "loss": 1.1898, + "mean_token_accuracy": 0.667159914970398, + "num_tokens": 21071505.0, + "step": 814 + }, + { + "epoch": 0.08950142763013398, + "grad_norm": 2.6668171882629395, + "learning_rate": 2.979502196193265e-07, + "loss": 1.0227, + "mean_token_accuracy": 0.6967074871063232, + "num_tokens": 21094803.0, + "step": 815 + }, + { + "epoch": 0.08961124533274764, + "grad_norm": 2.514035701751709, + "learning_rate": 2.98316251830161e-07, + "loss": 1.0975, + "mean_token_accuracy": 0.6808029413223267, + "num_tokens": 21120328.0, + "step": 816 + }, + { + "epoch": 0.0897210630353613, + "grad_norm": 2.4252188205718994, + "learning_rate": 2.986822840409956e-07, + "loss": 1.1126, + "mean_token_accuracy": 0.6718235015869141, + "num_tokens": 21146162.0, + "step": 817 + }, + { + "epoch": 0.08983088073797496, + "grad_norm": 2.5539159774780273, + "learning_rate": 2.9904831625183015e-07, + "loss": 1.0438, + "mean_token_accuracy": 0.6889156699180603, + "num_tokens": 21169041.0, + "step": 818 + }, + { + "epoch": 0.08994069844058862, + "grad_norm": 2.430345058441162, + "learning_rate": 2.994143484626647e-07, + "loss": 1.097, + "mean_token_accuracy": 0.6820095181465149, + "num_tokens": 21195693.0, + "step": 819 + }, + { + "epoch": 0.09005051614320228, + "grad_norm": 2.752241611480713, + "learning_rate": 2.9978038067349925e-07, + "loss": 1.0602, + "mean_token_accuracy": 0.6912758946418762, + "num_tokens": 21217125.0, + "step": 820 + }, + { + "epoch": 0.09016033384581594, + "grad_norm": 2.4454612731933594, + "learning_rate": 3.001464128843338e-07, + "loss": 1.1599, + "mean_token_accuracy": 0.6600759029388428, + "num_tokens": 21243201.0, + "step": 821 + }, + { + "epoch": 0.0902701515484296, + "grad_norm": 2.259122610092163, + "learning_rate": 3.005124450951684e-07, + "loss": 1.0628, + "mean_token_accuracy": 0.6905947327613831, + "num_tokens": 21272069.0, + "step": 822 + }, + { + "epoch": 0.09037996925104327, + "grad_norm": 2.420147180557251, + "learning_rate": 3.008784773060029e-07, + "loss": 1.1492, + "mean_token_accuracy": 0.6651281714439392, + "num_tokens": 21299739.0, + "step": 823 + }, + { + "epoch": 0.09048978695365693, + "grad_norm": 2.9544529914855957, + "learning_rate": 3.0124450951683744e-07, + "loss": 1.063, + "mean_token_accuracy": 0.6773597598075867, + "num_tokens": 21318516.0, + "step": 824 + }, + { + "epoch": 0.0905996046562706, + "grad_norm": 2.8397717475891113, + "learning_rate": 3.0161054172767204e-07, + "loss": 1.0797, + "mean_token_accuracy": 0.675229549407959, + "num_tokens": 21337893.0, + "step": 825 + }, + { + "epoch": 0.09070942235888425, + "grad_norm": 2.3552935123443604, + "learning_rate": 3.0197657393850654e-07, + "loss": 1.1949, + "mean_token_accuracy": 0.6484524011611938, + "num_tokens": 21366728.0, + "step": 826 + }, + { + "epoch": 0.09081924006149791, + "grad_norm": 2.144484281539917, + "learning_rate": 3.0234260614934114e-07, + "loss": 1.1857, + "mean_token_accuracy": 0.6515176296234131, + "num_tokens": 21398742.0, + "step": 827 + }, + { + "epoch": 0.09092905776411157, + "grad_norm": 2.6090004444122314, + "learning_rate": 3.027086383601757e-07, + "loss": 0.9928, + "mean_token_accuracy": 0.7047131061553955, + "num_tokens": 21422256.0, + "step": 828 + }, + { + "epoch": 0.09103887546672523, + "grad_norm": 2.4921953678131104, + "learning_rate": 3.0307467057101024e-07, + "loss": 1.1254, + "mean_token_accuracy": 0.6727159023284912, + "num_tokens": 21445882.0, + "step": 829 + }, + { + "epoch": 0.09114869316933889, + "grad_norm": 2.5119056701660156, + "learning_rate": 3.034407027818448e-07, + "loss": 1.1029, + "mean_token_accuracy": 0.6736388206481934, + "num_tokens": 21472863.0, + "step": 830 + }, + { + "epoch": 0.09125851087195255, + "grad_norm": 2.7427468299865723, + "learning_rate": 3.0380673499267934e-07, + "loss": 1.0402, + "mean_token_accuracy": 0.6923413276672363, + "num_tokens": 21496899.0, + "step": 831 + }, + { + "epoch": 0.09136832857456623, + "grad_norm": 2.7548635005950928, + "learning_rate": 3.041727672035139e-07, + "loss": 1.1142, + "mean_token_accuracy": 0.6761687994003296, + "num_tokens": 21519107.0, + "step": 832 + }, + { + "epoch": 0.09147814627717989, + "grad_norm": 2.5834732055664062, + "learning_rate": 3.0453879941434844e-07, + "loss": 1.1499, + "mean_token_accuracy": 0.6653053760528564, + "num_tokens": 21544059.0, + "step": 833 + }, + { + "epoch": 0.09158796397979355, + "grad_norm": 2.598339557647705, + "learning_rate": 3.04904831625183e-07, + "loss": 1.0736, + "mean_token_accuracy": 0.6863253116607666, + "num_tokens": 21568468.0, + "step": 834 + }, + { + "epoch": 0.0916977816824072, + "grad_norm": 2.513427972793579, + "learning_rate": 3.052708638360176e-07, + "loss": 1.0914, + "mean_token_accuracy": 0.6740631461143494, + "num_tokens": 21592891.0, + "step": 835 + }, + { + "epoch": 0.09180759938502087, + "grad_norm": 2.8348920345306396, + "learning_rate": 3.056368960468521e-07, + "loss": 1.0842, + "mean_token_accuracy": 0.6834783554077148, + "num_tokens": 21614569.0, + "step": 836 + }, + { + "epoch": 0.09191741708763453, + "grad_norm": 2.57419490814209, + "learning_rate": 3.0600292825768663e-07, + "loss": 1.1505, + "mean_token_accuracy": 0.6580661535263062, + "num_tokens": 21638089.0, + "step": 837 + }, + { + "epoch": 0.09202723479024819, + "grad_norm": 2.1005043983459473, + "learning_rate": 3.0636896046852124e-07, + "loss": 1.0961, + "mean_token_accuracy": 0.6880829334259033, + "num_tokens": 21668891.0, + "step": 838 + }, + { + "epoch": 0.09213705249286185, + "grad_norm": 2.371554136276245, + "learning_rate": 3.067349926793558e-07, + "loss": 1.1471, + "mean_token_accuracy": 0.6680701375007629, + "num_tokens": 21696713.0, + "step": 839 + }, + { + "epoch": 0.0922468701954755, + "grad_norm": 2.2745871543884277, + "learning_rate": 3.071010248901903e-07, + "loss": 1.0086, + "mean_token_accuracy": 0.703671395778656, + "num_tokens": 21724588.0, + "step": 840 + }, + { + "epoch": 0.09235668789808917, + "grad_norm": 2.8802106380462646, + "learning_rate": 3.074670571010249e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.6936026811599731, + "num_tokens": 21745178.0, + "step": 841 + }, + { + "epoch": 0.09246650560070284, + "grad_norm": 2.5724055767059326, + "learning_rate": 3.0783308931185943e-07, + "loss": 1.0715, + "mean_token_accuracy": 0.6887237429618835, + "num_tokens": 21768870.0, + "step": 842 + }, + { + "epoch": 0.0925763233033165, + "grad_norm": 2.3707072734832764, + "learning_rate": 3.08199121522694e-07, + "loss": 1.0569, + "mean_token_accuracy": 0.6867324113845825, + "num_tokens": 21796552.0, + "step": 843 + }, + { + "epoch": 0.09268614100593016, + "grad_norm": 2.2409825325012207, + "learning_rate": 3.0856515373352853e-07, + "loss": 1.1285, + "mean_token_accuracy": 0.6739587783813477, + "num_tokens": 21825989.0, + "step": 844 + }, + { + "epoch": 0.09279595870854382, + "grad_norm": 2.6424190998077393, + "learning_rate": 3.089311859443631e-07, + "loss": 1.0362, + "mean_token_accuracy": 0.6912240982055664, + "num_tokens": 21848075.0, + "step": 845 + }, + { + "epoch": 0.09290577641115748, + "grad_norm": 2.303002119064331, + "learning_rate": 3.092972181551977e-07, + "loss": 1.1325, + "mean_token_accuracy": 0.6737242341041565, + "num_tokens": 21877666.0, + "step": 846 + }, + { + "epoch": 0.09301559411377114, + "grad_norm": 2.2820897102355957, + "learning_rate": 3.096632503660322e-07, + "loss": 1.0816, + "mean_token_accuracy": 0.6775619387626648, + "num_tokens": 21904463.0, + "step": 847 + }, + { + "epoch": 0.0931254118163848, + "grad_norm": 2.5482921600341797, + "learning_rate": 3.1002928257686673e-07, + "loss": 1.0575, + "mean_token_accuracy": 0.6822636723518372, + "num_tokens": 21927852.0, + "step": 848 + }, + { + "epoch": 0.09323522951899846, + "grad_norm": 2.3302037715911865, + "learning_rate": 3.1039531478770133e-07, + "loss": 1.0527, + "mean_token_accuracy": 0.6946284770965576, + "num_tokens": 21956317.0, + "step": 849 + }, + { + "epoch": 0.09334504722161212, + "grad_norm": 2.3781003952026367, + "learning_rate": 3.107613469985358e-07, + "loss": 1.0535, + "mean_token_accuracy": 0.6840881705284119, + "num_tokens": 21981920.0, + "step": 850 + }, + { + "epoch": 0.09345486492422579, + "grad_norm": 2.7328438758850098, + "learning_rate": 3.1112737920937043e-07, + "loss": 1.116, + "mean_token_accuracy": 0.6715552806854248, + "num_tokens": 22006163.0, + "step": 851 + }, + { + "epoch": 0.09356468262683945, + "grad_norm": 2.5326194763183594, + "learning_rate": 3.11493411420205e-07, + "loss": 1.0867, + "mean_token_accuracy": 0.6766800880432129, + "num_tokens": 22029906.0, + "step": 852 + }, + { + "epoch": 0.09367450032945311, + "grad_norm": 2.4841151237487793, + "learning_rate": 3.1185944363103947e-07, + "loss": 1.1262, + "mean_token_accuracy": 0.6748374700546265, + "num_tokens": 22055611.0, + "step": 853 + }, + { + "epoch": 0.09378431803206677, + "grad_norm": 2.3966586589813232, + "learning_rate": 3.122254758418741e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.6751289367675781, + "num_tokens": 22081614.0, + "step": 854 + }, + { + "epoch": 0.09389413573468043, + "grad_norm": 2.5935258865356445, + "learning_rate": 3.125915080527086e-07, + "loss": 1.0579, + "mean_token_accuracy": 0.6825337409973145, + "num_tokens": 22104956.0, + "step": 855 + }, + { + "epoch": 0.09400395343729409, + "grad_norm": 2.2517824172973633, + "learning_rate": 3.1295754026354317e-07, + "loss": 1.1053, + "mean_token_accuracy": 0.6730133295059204, + "num_tokens": 22135119.0, + "step": 856 + }, + { + "epoch": 0.09411377113990775, + "grad_norm": 2.710505723953247, + "learning_rate": 3.133235724743777e-07, + "loss": 1.1203, + "mean_token_accuracy": 0.6764760613441467, + "num_tokens": 22157625.0, + "step": 857 + }, + { + "epoch": 0.09422358884252141, + "grad_norm": 2.226757049560547, + "learning_rate": 3.1368960468521227e-07, + "loss": 1.0941, + "mean_token_accuracy": 0.6813324689865112, + "num_tokens": 22188178.0, + "step": 858 + }, + { + "epoch": 0.09433340654513507, + "grad_norm": 2.316816806793213, + "learning_rate": 3.1405563689604687e-07, + "loss": 1.0966, + "mean_token_accuracy": 0.6744102239608765, + "num_tokens": 22213762.0, + "step": 859 + }, + { + "epoch": 0.09444322424774873, + "grad_norm": 2.686408519744873, + "learning_rate": 3.1442166910688137e-07, + "loss": 1.1037, + "mean_token_accuracy": 0.6773778200149536, + "num_tokens": 22235680.0, + "step": 860 + }, + { + "epoch": 0.0945530419503624, + "grad_norm": 2.702075719833374, + "learning_rate": 3.147877013177159e-07, + "loss": 1.0336, + "mean_token_accuracy": 0.6930595636367798, + "num_tokens": 22258638.0, + "step": 861 + }, + { + "epoch": 0.09466285965297606, + "grad_norm": 2.5352704524993896, + "learning_rate": 3.151537335285505e-07, + "loss": 1.0756, + "mean_token_accuracy": 0.6827784776687622, + "num_tokens": 22282129.0, + "step": 862 + }, + { + "epoch": 0.09477267735558972, + "grad_norm": 2.185467481613159, + "learning_rate": 3.1551976573938507e-07, + "loss": 0.9961, + "mean_token_accuracy": 0.7026017904281616, + "num_tokens": 22312222.0, + "step": 863 + }, + { + "epoch": 0.09488249505820338, + "grad_norm": 2.6746621131896973, + "learning_rate": 3.1588579795021956e-07, + "loss": 1.0891, + "mean_token_accuracy": 0.6739501953125, + "num_tokens": 22335476.0, + "step": 864 + }, + { + "epoch": 0.09499231276081704, + "grad_norm": 2.434049129486084, + "learning_rate": 3.1625183016105417e-07, + "loss": 1.1426, + "mean_token_accuracy": 0.6862611770629883, + "num_tokens": 22362628.0, + "step": 865 + }, + { + "epoch": 0.0951021304634307, + "grad_norm": 2.69368052482605, + "learning_rate": 3.166178623718887e-07, + "loss": 1.1433, + "mean_token_accuracy": 0.670654833316803, + "num_tokens": 22388402.0, + "step": 866 + }, + { + "epoch": 0.09521194816604436, + "grad_norm": 2.3283400535583496, + "learning_rate": 3.1698389458272327e-07, + "loss": 1.1228, + "mean_token_accuracy": 0.6768152117729187, + "num_tokens": 22417436.0, + "step": 867 + }, + { + "epoch": 0.09532176586865802, + "grad_norm": 2.7022082805633545, + "learning_rate": 3.173499267935578e-07, + "loss": 1.1509, + "mean_token_accuracy": 0.6632232666015625, + "num_tokens": 22440356.0, + "step": 868 + }, + { + "epoch": 0.09543158357127168, + "grad_norm": 2.785019874572754, + "learning_rate": 3.1771595900439236e-07, + "loss": 1.146, + "mean_token_accuracy": 0.6723323464393616, + "num_tokens": 22462445.0, + "step": 869 + }, + { + "epoch": 0.09554140127388536, + "grad_norm": 2.1603662967681885, + "learning_rate": 3.1808199121522697e-07, + "loss": 1.1417, + "mean_token_accuracy": 0.668264627456665, + "num_tokens": 22495714.0, + "step": 870 + }, + { + "epoch": 0.09565121897649902, + "grad_norm": 2.472447633743286, + "learning_rate": 3.1844802342606146e-07, + "loss": 1.1266, + "mean_token_accuracy": 0.6680711507797241, + "num_tokens": 22520890.0, + "step": 871 + }, + { + "epoch": 0.09576103667911268, + "grad_norm": 2.265918493270874, + "learning_rate": 3.18814055636896e-07, + "loss": 1.0984, + "mean_token_accuracy": 0.6844911575317383, + "num_tokens": 22548125.0, + "step": 872 + }, + { + "epoch": 0.09587085438172634, + "grad_norm": 2.5496177673339844, + "learning_rate": 3.191800878477306e-07, + "loss": 1.0436, + "mean_token_accuracy": 0.6967850923538208, + "num_tokens": 22571203.0, + "step": 873 + }, + { + "epoch": 0.09598067208434, + "grad_norm": 2.2932498455047607, + "learning_rate": 3.195461200585651e-07, + "loss": 1.0685, + "mean_token_accuracy": 0.6806175708770752, + "num_tokens": 22600078.0, + "step": 874 + }, + { + "epoch": 0.09609048978695366, + "grad_norm": 2.4985082149505615, + "learning_rate": 3.199121522693997e-07, + "loss": 1.1093, + "mean_token_accuracy": 0.6810925602912903, + "num_tokens": 22626236.0, + "step": 875 + }, + { + "epoch": 0.09620030748956732, + "grad_norm": 2.29787540435791, + "learning_rate": 3.2027818448023426e-07, + "loss": 1.0968, + "mean_token_accuracy": 0.6734969615936279, + "num_tokens": 22653233.0, + "step": 876 + }, + { + "epoch": 0.09631012519218098, + "grad_norm": 2.3649234771728516, + "learning_rate": 3.2064421669106876e-07, + "loss": 1.1141, + "mean_token_accuracy": 0.6693174839019775, + "num_tokens": 22679272.0, + "step": 877 + }, + { + "epoch": 0.09641994289479464, + "grad_norm": 2.5269227027893066, + "learning_rate": 3.2101024890190336e-07, + "loss": 1.0515, + "mean_token_accuracy": 0.6905930042266846, + "num_tokens": 22704003.0, + "step": 878 + }, + { + "epoch": 0.0965297605974083, + "grad_norm": 2.408874988555908, + "learning_rate": 3.213762811127379e-07, + "loss": 1.037, + "mean_token_accuracy": 0.6942651271820068, + "num_tokens": 22729119.0, + "step": 879 + }, + { + "epoch": 0.09663957830002197, + "grad_norm": 2.3692169189453125, + "learning_rate": 3.2174231332357246e-07, + "loss": 1.1154, + "mean_token_accuracy": 0.6741969585418701, + "num_tokens": 22755678.0, + "step": 880 + }, + { + "epoch": 0.09674939600263563, + "grad_norm": 2.1107254028320312, + "learning_rate": 3.22108345534407e-07, + "loss": 1.0923, + "mean_token_accuracy": 0.6802234649658203, + "num_tokens": 22788600.0, + "step": 881 + }, + { + "epoch": 0.09685921370524929, + "grad_norm": 2.2597007751464844, + "learning_rate": 3.2247437774524155e-07, + "loss": 1.014, + "mean_token_accuracy": 0.6975002884864807, + "num_tokens": 22814636.0, + "step": 882 + }, + { + "epoch": 0.09696903140786295, + "grad_norm": 2.572603225708008, + "learning_rate": 3.2284040995607616e-07, + "loss": 1.1278, + "mean_token_accuracy": 0.6694586873054504, + "num_tokens": 22839289.0, + "step": 883 + }, + { + "epoch": 0.09707884911047661, + "grad_norm": 2.440126895904541, + "learning_rate": 3.2320644216691065e-07, + "loss": 1.0776, + "mean_token_accuracy": 0.6781179904937744, + "num_tokens": 22864607.0, + "step": 884 + }, + { + "epoch": 0.09718866681309027, + "grad_norm": 2.5712780952453613, + "learning_rate": 3.235724743777452e-07, + "loss": 1.1277, + "mean_token_accuracy": 0.6821163296699524, + "num_tokens": 22892406.0, + "step": 885 + }, + { + "epoch": 0.09729848451570393, + "grad_norm": 2.5311830043792725, + "learning_rate": 3.239385065885798e-07, + "loss": 1.1024, + "mean_token_accuracy": 0.6720576286315918, + "num_tokens": 22917563.0, + "step": 886 + }, + { + "epoch": 0.09740830221831759, + "grad_norm": 2.3650612831115723, + "learning_rate": 3.2430453879941435e-07, + "loss": 1.0777, + "mean_token_accuracy": 0.6822086572647095, + "num_tokens": 22944690.0, + "step": 887 + }, + { + "epoch": 0.09751811992093125, + "grad_norm": 2.513615846633911, + "learning_rate": 3.2467057101024885e-07, + "loss": 1.0663, + "mean_token_accuracy": 0.6868982315063477, + "num_tokens": 22970435.0, + "step": 888 + }, + { + "epoch": 0.09762793762354492, + "grad_norm": 2.3851890563964844, + "learning_rate": 3.2503660322108345e-07, + "loss": 1.0364, + "mean_token_accuracy": 0.6964348554611206, + "num_tokens": 22996469.0, + "step": 889 + }, + { + "epoch": 0.09773775532615858, + "grad_norm": 2.593501091003418, + "learning_rate": 3.25402635431918e-07, + "loss": 1.158, + "mean_token_accuracy": 0.6695467829704285, + "num_tokens": 23019491.0, + "step": 890 + }, + { + "epoch": 0.09784757302877224, + "grad_norm": 2.251314878463745, + "learning_rate": 3.2576866764275255e-07, + "loss": 1.0757, + "mean_token_accuracy": 0.6880232095718384, + "num_tokens": 23048481.0, + "step": 891 + }, + { + "epoch": 0.0979573907313859, + "grad_norm": 2.4713165760040283, + "learning_rate": 3.261346998535871e-07, + "loss": 1.078, + "mean_token_accuracy": 0.681187629699707, + "num_tokens": 23074453.0, + "step": 892 + }, + { + "epoch": 0.09806720843399956, + "grad_norm": 2.2072255611419678, + "learning_rate": 3.2650073206442165e-07, + "loss": 0.9888, + "mean_token_accuracy": 0.7053149342536926, + "num_tokens": 23103151.0, + "step": 893 + }, + { + "epoch": 0.09817702613661322, + "grad_norm": 2.186002492904663, + "learning_rate": 3.2686676427525625e-07, + "loss": 1.099, + "mean_token_accuracy": 0.6761927604675293, + "num_tokens": 23134289.0, + "step": 894 + }, + { + "epoch": 0.09828684383922688, + "grad_norm": 2.643620729446411, + "learning_rate": 3.2723279648609075e-07, + "loss": 1.0286, + "mean_token_accuracy": 0.7006072998046875, + "num_tokens": 23156101.0, + "step": 895 + }, + { + "epoch": 0.09839666154184054, + "grad_norm": 2.5788655281066895, + "learning_rate": 3.275988286969253e-07, + "loss": 1.0022, + "mean_token_accuracy": 0.7008076906204224, + "num_tokens": 23179178.0, + "step": 896 + }, + { + "epoch": 0.0985064792444542, + "grad_norm": 2.6468536853790283, + "learning_rate": 3.279648609077599e-07, + "loss": 1.1022, + "mean_token_accuracy": 0.6782282590866089, + "num_tokens": 23200455.0, + "step": 897 + }, + { + "epoch": 0.09861629694706786, + "grad_norm": 2.4701011180877686, + "learning_rate": 3.283308931185944e-07, + "loss": 1.0955, + "mean_token_accuracy": 0.6936217546463013, + "num_tokens": 23226237.0, + "step": 898 + }, + { + "epoch": 0.09872611464968153, + "grad_norm": 2.3404996395111084, + "learning_rate": 3.28696925329429e-07, + "loss": 1.0829, + "mean_token_accuracy": 0.6877027750015259, + "num_tokens": 23251813.0, + "step": 899 + }, + { + "epoch": 0.0988359323522952, + "grad_norm": 2.354012966156006, + "learning_rate": 3.2906295754026354e-07, + "loss": 1.1456, + "mean_token_accuracy": 0.658971905708313, + "num_tokens": 23278810.0, + "step": 900 + }, + { + "epoch": 0.09894575005490885, + "grad_norm": 2.2182021141052246, + "learning_rate": 3.2942898975109804e-07, + "loss": 1.1321, + "mean_token_accuracy": 0.673058271408081, + "num_tokens": 23307162.0, + "step": 901 + }, + { + "epoch": 0.09905556775752251, + "grad_norm": 2.243695020675659, + "learning_rate": 3.2979502196193264e-07, + "loss": 1.1439, + "mean_token_accuracy": 0.6656460762023926, + "num_tokens": 23335308.0, + "step": 902 + }, + { + "epoch": 0.09916538546013617, + "grad_norm": 2.462043046951294, + "learning_rate": 3.301610541727672e-07, + "loss": 1.0444, + "mean_token_accuracy": 0.6932078003883362, + "num_tokens": 23359068.0, + "step": 903 + }, + { + "epoch": 0.09927520316274983, + "grad_norm": 2.4413647651672363, + "learning_rate": 3.3052708638360174e-07, + "loss": 1.0083, + "mean_token_accuracy": 0.7060584425926208, + "num_tokens": 23381987.0, + "step": 904 + }, + { + "epoch": 0.0993850208653635, + "grad_norm": 1.9958813190460205, + "learning_rate": 3.308931185944363e-07, + "loss": 1.1209, + "mean_token_accuracy": 0.671327531337738, + "num_tokens": 23420367.0, + "step": 905 + }, + { + "epoch": 0.09949483856797715, + "grad_norm": 2.6417324542999268, + "learning_rate": 3.3125915080527084e-07, + "loss": 1.0627, + "mean_token_accuracy": 0.688244104385376, + "num_tokens": 23442587.0, + "step": 906 + }, + { + "epoch": 0.09960465627059081, + "grad_norm": 2.3935928344726562, + "learning_rate": 3.3162518301610544e-07, + "loss": 1.1288, + "mean_token_accuracy": 0.6729621887207031, + "num_tokens": 23469268.0, + "step": 907 + }, + { + "epoch": 0.09971447397320449, + "grad_norm": 2.1781857013702393, + "learning_rate": 3.3199121522693994e-07, + "loss": 1.19, + "mean_token_accuracy": 0.6467872858047485, + "num_tokens": 23500886.0, + "step": 908 + }, + { + "epoch": 0.09982429167581815, + "grad_norm": 2.634462594985962, + "learning_rate": 3.323572474377745e-07, + "loss": 1.0924, + "mean_token_accuracy": 0.6823728680610657, + "num_tokens": 23523429.0, + "step": 909 + }, + { + "epoch": 0.0999341093784318, + "grad_norm": 2.5032365322113037, + "learning_rate": 3.327232796486091e-07, + "loss": 1.0416, + "mean_token_accuracy": 0.6977517604827881, + "num_tokens": 23545940.0, + "step": 910 + }, + { + "epoch": 0.10004392708104547, + "grad_norm": 2.5298941135406494, + "learning_rate": 3.3308931185944364e-07, + "loss": 1.1858, + "mean_token_accuracy": 0.6604012250900269, + "num_tokens": 23572000.0, + "step": 911 + }, + { + "epoch": 0.10015374478365913, + "grad_norm": 2.369685173034668, + "learning_rate": 3.3345534407027813e-07, + "loss": 1.109, + "mean_token_accuracy": 0.6785153746604919, + "num_tokens": 23600798.0, + "step": 912 + }, + { + "epoch": 0.10026356248627279, + "grad_norm": 2.49601674079895, + "learning_rate": 3.3382137628111273e-07, + "loss": 1.0236, + "mean_token_accuracy": 0.696110725402832, + "num_tokens": 23624264.0, + "step": 913 + }, + { + "epoch": 0.10037338018888645, + "grad_norm": 2.1584653854370117, + "learning_rate": 3.341874084919473e-07, + "loss": 1.0629, + "mean_token_accuracy": 0.6850155591964722, + "num_tokens": 23655171.0, + "step": 914 + }, + { + "epoch": 0.1004831978915001, + "grad_norm": 2.257697582244873, + "learning_rate": 3.3455344070278183e-07, + "loss": 1.1235, + "mean_token_accuracy": 0.672150731086731, + "num_tokens": 23683903.0, + "step": 915 + }, + { + "epoch": 0.10059301559411377, + "grad_norm": 2.5914926528930664, + "learning_rate": 3.349194729136164e-07, + "loss": 1.058, + "mean_token_accuracy": 0.6916287541389465, + "num_tokens": 23706406.0, + "step": 916 + }, + { + "epoch": 0.10070283329672743, + "grad_norm": 2.3859381675720215, + "learning_rate": 3.3528550512445093e-07, + "loss": 1.1263, + "mean_token_accuracy": 0.6669008731842041, + "num_tokens": 23732433.0, + "step": 917 + }, + { + "epoch": 0.1008126509993411, + "grad_norm": 2.2076363563537598, + "learning_rate": 3.356515373352855e-07, + "loss": 1.0313, + "mean_token_accuracy": 0.6925991177558899, + "num_tokens": 23759236.0, + "step": 918 + }, + { + "epoch": 0.10092246870195476, + "grad_norm": 2.4606637954711914, + "learning_rate": 3.3601756954612003e-07, + "loss": 1.022, + "mean_token_accuracy": 0.6974092721939087, + "num_tokens": 23784398.0, + "step": 919 + }, + { + "epoch": 0.10103228640456842, + "grad_norm": 2.3746278285980225, + "learning_rate": 3.363836017569546e-07, + "loss": 1.0055, + "mean_token_accuracy": 0.6990245580673218, + "num_tokens": 23812650.0, + "step": 920 + }, + { + "epoch": 0.10114210410718208, + "grad_norm": 2.1978931427001953, + "learning_rate": 3.367496339677892e-07, + "loss": 1.0923, + "mean_token_accuracy": 0.6814920902252197, + "num_tokens": 23841776.0, + "step": 921 + }, + { + "epoch": 0.10125192180979574, + "grad_norm": 2.543292284011841, + "learning_rate": 3.371156661786237e-07, + "loss": 1.0785, + "mean_token_accuracy": 0.6783018708229065, + "num_tokens": 23865229.0, + "step": 922 + }, + { + "epoch": 0.1013617395124094, + "grad_norm": 2.443732976913452, + "learning_rate": 3.374816983894583e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.6743537783622742, + "num_tokens": 23889843.0, + "step": 923 + }, + { + "epoch": 0.10147155721502306, + "grad_norm": 2.445971727371216, + "learning_rate": 3.3784773060029283e-07, + "loss": 1.1606, + "mean_token_accuracy": 0.6658763885498047, + "num_tokens": 23917810.0, + "step": 924 + }, + { + "epoch": 0.10158137491763672, + "grad_norm": 2.664342164993286, + "learning_rate": 3.382137628111273e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6787682175636292, + "num_tokens": 23940531.0, + "step": 925 + }, + { + "epoch": 0.10169119262025038, + "grad_norm": 2.238755941390991, + "learning_rate": 3.385797950219619e-07, + "loss": 1.074, + "mean_token_accuracy": 0.6821549534797668, + "num_tokens": 23968313.0, + "step": 926 + }, + { + "epoch": 0.10180101032286405, + "grad_norm": 2.2668044567108154, + "learning_rate": 3.389458272327965e-07, + "loss": 1.1174, + "mean_token_accuracy": 0.6762388348579407, + "num_tokens": 23999841.0, + "step": 927 + }, + { + "epoch": 0.10191082802547771, + "grad_norm": 2.2480485439300537, + "learning_rate": 3.39311859443631e-07, + "loss": 1.1391, + "mean_token_accuracy": 0.6637154817581177, + "num_tokens": 24031207.0, + "step": 928 + }, + { + "epoch": 0.10202064572809137, + "grad_norm": 2.164050340652466, + "learning_rate": 3.3967789165446557e-07, + "loss": 1.1249, + "mean_token_accuracy": 0.6729483008384705, + "num_tokens": 24060419.0, + "step": 929 + }, + { + "epoch": 0.10213046343070503, + "grad_norm": 2.404407024383545, + "learning_rate": 3.400439238653001e-07, + "loss": 1.0913, + "mean_token_accuracy": 0.685936450958252, + "num_tokens": 24086752.0, + "step": 930 + }, + { + "epoch": 0.10224028113331869, + "grad_norm": 2.490318775177002, + "learning_rate": 3.404099560761347e-07, + "loss": 1.0406, + "mean_token_accuracy": 0.6873397827148438, + "num_tokens": 24110671.0, + "step": 931 + }, + { + "epoch": 0.10235009883593235, + "grad_norm": 2.1943063735961914, + "learning_rate": 3.407759882869692e-07, + "loss": 1.0851, + "mean_token_accuracy": 0.6784844398498535, + "num_tokens": 24142021.0, + "step": 932 + }, + { + "epoch": 0.10245991653854601, + "grad_norm": 2.7286336421966553, + "learning_rate": 3.4114202049780377e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6790098547935486, + "num_tokens": 24163153.0, + "step": 933 + }, + { + "epoch": 0.10256973424115967, + "grad_norm": 2.202549457550049, + "learning_rate": 3.4150805270863837e-07, + "loss": 1.153, + "mean_token_accuracy": 0.6622832417488098, + "num_tokens": 24194395.0, + "step": 934 + }, + { + "epoch": 0.10267955194377333, + "grad_norm": 2.3439860343933105, + "learning_rate": 3.4187408491947287e-07, + "loss": 1.1366, + "mean_token_accuracy": 0.6656934022903442, + "num_tokens": 24220717.0, + "step": 935 + }, + { + "epoch": 0.10278936964638699, + "grad_norm": 2.4541404247283936, + "learning_rate": 3.422401171303074e-07, + "loss": 1.0986, + "mean_token_accuracy": 0.6816911697387695, + "num_tokens": 24245109.0, + "step": 936 + }, + { + "epoch": 0.10289918734900066, + "grad_norm": 2.857592821121216, + "learning_rate": 3.42606149341142e-07, + "loss": 1.1672, + "mean_token_accuracy": 0.668070375919342, + "num_tokens": 24264983.0, + "step": 937 + }, + { + "epoch": 0.10300900505161432, + "grad_norm": 2.5749435424804688, + "learning_rate": 3.4297218155197657e-07, + "loss": 1.0901, + "mean_token_accuracy": 0.6882440447807312, + "num_tokens": 24288101.0, + "step": 938 + }, + { + "epoch": 0.10311882275422798, + "grad_norm": 2.2654528617858887, + "learning_rate": 3.433382137628111e-07, + "loss": 1.06, + "mean_token_accuracy": 0.6860031485557556, + "num_tokens": 24314589.0, + "step": 939 + }, + { + "epoch": 0.10322864045684164, + "grad_norm": 2.63641619682312, + "learning_rate": 3.4370424597364567e-07, + "loss": 1.1415, + "mean_token_accuracy": 0.6762741804122925, + "num_tokens": 24337109.0, + "step": 940 + }, + { + "epoch": 0.1033384581594553, + "grad_norm": 2.271434783935547, + "learning_rate": 3.440702781844802e-07, + "loss": 1.1282, + "mean_token_accuracy": 0.6738343238830566, + "num_tokens": 24367221.0, + "step": 941 + }, + { + "epoch": 0.10344827586206896, + "grad_norm": 2.363109827041626, + "learning_rate": 3.4443631039531476e-07, + "loss": 1.1472, + "mean_token_accuracy": 0.6637874841690063, + "num_tokens": 24394210.0, + "step": 942 + }, + { + "epoch": 0.10355809356468262, + "grad_norm": 2.3568670749664307, + "learning_rate": 3.448023426061493e-07, + "loss": 1.0136, + "mean_token_accuracy": 0.6957489252090454, + "num_tokens": 24419005.0, + "step": 943 + }, + { + "epoch": 0.10366791126729628, + "grad_norm": 2.254913091659546, + "learning_rate": 3.4516837481698386e-07, + "loss": 1.0964, + "mean_token_accuracy": 0.6807454824447632, + "num_tokens": 24448427.0, + "step": 944 + }, + { + "epoch": 0.10377772896990994, + "grad_norm": 2.137859582901001, + "learning_rate": 3.4553440702781846e-07, + "loss": 1.2002, + "mean_token_accuracy": 0.6517117023468018, + "num_tokens": 24480209.0, + "step": 945 + }, + { + "epoch": 0.10388754667252362, + "grad_norm": 2.5321428775787354, + "learning_rate": 3.4590043923865296e-07, + "loss": 1.026, + "mean_token_accuracy": 0.6974982023239136, + "num_tokens": 24503484.0, + "step": 946 + }, + { + "epoch": 0.10399736437513728, + "grad_norm": 2.3713765144348145, + "learning_rate": 3.4626647144948756e-07, + "loss": 1.1005, + "mean_token_accuracy": 0.6733958125114441, + "num_tokens": 24528827.0, + "step": 947 + }, + { + "epoch": 0.10410718207775094, + "grad_norm": 2.738077402114868, + "learning_rate": 3.466325036603221e-07, + "loss": 1.0331, + "mean_token_accuracy": 0.6978141665458679, + "num_tokens": 24550278.0, + "step": 948 + }, + { + "epoch": 0.1042169997803646, + "grad_norm": 2.44711971282959, + "learning_rate": 3.469985358711566e-07, + "loss": 1.0062, + "mean_token_accuracy": 0.6967957019805908, + "num_tokens": 24573147.0, + "step": 949 + }, + { + "epoch": 0.10432681748297826, + "grad_norm": 2.5000250339508057, + "learning_rate": 3.473645680819912e-07, + "loss": 1.0731, + "mean_token_accuracy": 0.6765800714492798, + "num_tokens": 24597279.0, + "step": 950 + }, + { + "epoch": 0.10443663518559192, + "grad_norm": 2.489558458328247, + "learning_rate": 3.4773060029282576e-07, + "loss": 1.1228, + "mean_token_accuracy": 0.6614791750907898, + "num_tokens": 24623328.0, + "step": 951 + }, + { + "epoch": 0.10454645288820558, + "grad_norm": 2.727975606918335, + "learning_rate": 3.480966325036603e-07, + "loss": 1.0706, + "mean_token_accuracy": 0.6934745907783508, + "num_tokens": 24644301.0, + "step": 952 + }, + { + "epoch": 0.10465627059081924, + "grad_norm": 2.118152141571045, + "learning_rate": 3.4846266471449486e-07, + "loss": 1.1341, + "mean_token_accuracy": 0.6696041822433472, + "num_tokens": 24678615.0, + "step": 953 + }, + { + "epoch": 0.1047660882934329, + "grad_norm": 2.6314024925231934, + "learning_rate": 3.488286969253294e-07, + "loss": 0.9733, + "mean_token_accuracy": 0.7060809135437012, + "num_tokens": 24701336.0, + "step": 954 + }, + { + "epoch": 0.10487590599604656, + "grad_norm": 2.2896828651428223, + "learning_rate": 3.49194729136164e-07, + "loss": 1.032, + "mean_token_accuracy": 0.6937340497970581, + "num_tokens": 24727917.0, + "step": 955 + }, + { + "epoch": 0.10498572369866023, + "grad_norm": 2.7051172256469727, + "learning_rate": 3.495607613469985e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.6908214092254639, + "num_tokens": 24748147.0, + "step": 956 + }, + { + "epoch": 0.10509554140127389, + "grad_norm": 2.4300553798675537, + "learning_rate": 3.4992679355783305e-07, + "loss": 1.0651, + "mean_token_accuracy": 0.6893154382705688, + "num_tokens": 24773384.0, + "step": 957 + }, + { + "epoch": 0.10520535910388755, + "grad_norm": 2.2980308532714844, + "learning_rate": 3.5029282576866766e-07, + "loss": 1.0998, + "mean_token_accuracy": 0.6734075546264648, + "num_tokens": 24801521.0, + "step": 958 + }, + { + "epoch": 0.10531517680650121, + "grad_norm": 2.3320393562316895, + "learning_rate": 3.5065885797950215e-07, + "loss": 1.1215, + "mean_token_accuracy": 0.6747720241546631, + "num_tokens": 24830015.0, + "step": 959 + }, + { + "epoch": 0.10542499450911487, + "grad_norm": 2.5882933139801025, + "learning_rate": 3.510248901903367e-07, + "loss": 1.0835, + "mean_token_accuracy": 0.6780480742454529, + "num_tokens": 24853232.0, + "step": 960 + }, + { + "epoch": 0.10553481221172853, + "grad_norm": 2.4353129863739014, + "learning_rate": 3.513909224011713e-07, + "loss": 1.1423, + "mean_token_accuracy": 0.6680262088775635, + "num_tokens": 24877653.0, + "step": 961 + }, + { + "epoch": 0.10564462991434219, + "grad_norm": 2.483370780944824, + "learning_rate": 3.5175695461200585e-07, + "loss": 1.1235, + "mean_token_accuracy": 0.6696091890335083, + "num_tokens": 24903434.0, + "step": 962 + }, + { + "epoch": 0.10575444761695585, + "grad_norm": 2.404805898666382, + "learning_rate": 3.521229868228404e-07, + "loss": 1.1589, + "mean_token_accuracy": 0.6685867309570312, + "num_tokens": 24931347.0, + "step": 963 + }, + { + "epoch": 0.10586426531956951, + "grad_norm": 2.5934765338897705, + "learning_rate": 3.5248901903367495e-07, + "loss": 1.0593, + "mean_token_accuracy": 0.688945472240448, + "num_tokens": 24955365.0, + "step": 964 + }, + { + "epoch": 0.10597408302218318, + "grad_norm": 2.6551949977874756, + "learning_rate": 3.528550512445095e-07, + "loss": 1.1469, + "mean_token_accuracy": 0.6640695929527283, + "num_tokens": 24977249.0, + "step": 965 + }, + { + "epoch": 0.10608390072479684, + "grad_norm": 2.1972978115081787, + "learning_rate": 3.5322108345534405e-07, + "loss": 1.1581, + "mean_token_accuracy": 0.6655734777450562, + "num_tokens": 25008578.0, + "step": 966 + }, + { + "epoch": 0.1061937184274105, + "grad_norm": 2.698866128921509, + "learning_rate": 3.535871156661786e-07, + "loss": 1.02, + "mean_token_accuracy": 0.6895395517349243, + "num_tokens": 25028258.0, + "step": 967 + }, + { + "epoch": 0.10630353613002416, + "grad_norm": 2.5011146068573, + "learning_rate": 3.5395314787701315e-07, + "loss": 1.0496, + "mean_token_accuracy": 0.6889115571975708, + "num_tokens": 25052108.0, + "step": 968 + }, + { + "epoch": 0.10641335383263782, + "grad_norm": 2.0404603481292725, + "learning_rate": 3.5431918008784775e-07, + "loss": 1.1111, + "mean_token_accuracy": 0.6782435178756714, + "num_tokens": 25085379.0, + "step": 969 + }, + { + "epoch": 0.10652317153525148, + "grad_norm": 2.570936441421509, + "learning_rate": 3.5468521229868224e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6826008558273315, + "num_tokens": 25108319.0, + "step": 970 + }, + { + "epoch": 0.10663298923786514, + "grad_norm": 2.3094232082366943, + "learning_rate": 3.5505124450951685e-07, + "loss": 1.1962, + "mean_token_accuracy": 0.6581100821495056, + "num_tokens": 25139744.0, + "step": 971 + }, + { + "epoch": 0.1067428069404788, + "grad_norm": 2.3472211360931396, + "learning_rate": 3.554172767203514e-07, + "loss": 1.0785, + "mean_token_accuracy": 0.6810656189918518, + "num_tokens": 25166647.0, + "step": 972 + }, + { + "epoch": 0.10685262464309246, + "grad_norm": 2.2304515838623047, + "learning_rate": 3.557833089311859e-07, + "loss": 1.064, + "mean_token_accuracy": 0.6878283023834229, + "num_tokens": 25195503.0, + "step": 973 + }, + { + "epoch": 0.10696244234570612, + "grad_norm": 2.4598634243011475, + "learning_rate": 3.561493411420205e-07, + "loss": 1.0472, + "mean_token_accuracy": 0.6907100081443787, + "num_tokens": 25220138.0, + "step": 974 + }, + { + "epoch": 0.1070722600483198, + "grad_norm": 2.4668259620666504, + "learning_rate": 3.5651537335285504e-07, + "loss": 1.0582, + "mean_token_accuracy": 0.6847684383392334, + "num_tokens": 25244278.0, + "step": 975 + }, + { + "epoch": 0.10718207775093345, + "grad_norm": 2.1857738494873047, + "learning_rate": 3.5688140556368954e-07, + "loss": 1.0975, + "mean_token_accuracy": 0.6705430746078491, + "num_tokens": 25275205.0, + "step": 976 + }, + { + "epoch": 0.10729189545354711, + "grad_norm": 2.6014063358306885, + "learning_rate": 3.5724743777452414e-07, + "loss": 1.0333, + "mean_token_accuracy": 0.6923041343688965, + "num_tokens": 25296551.0, + "step": 977 + }, + { + "epoch": 0.10740171315616077, + "grad_norm": 2.507568359375, + "learning_rate": 3.576134699853587e-07, + "loss": 0.9429, + "mean_token_accuracy": 0.7186013460159302, + "num_tokens": 25319702.0, + "step": 978 + }, + { + "epoch": 0.10751153085877443, + "grad_norm": 2.385787010192871, + "learning_rate": 3.579795021961933e-07, + "loss": 1.1331, + "mean_token_accuracy": 0.667220413684845, + "num_tokens": 25345390.0, + "step": 979 + }, + { + "epoch": 0.1076213485613881, + "grad_norm": 2.314531087875366, + "learning_rate": 3.583455344070278e-07, + "loss": 1.0578, + "mean_token_accuracy": 0.6853652000427246, + "num_tokens": 25371403.0, + "step": 980 + }, + { + "epoch": 0.10773116626400175, + "grad_norm": 2.7458274364471436, + "learning_rate": 3.5871156661786234e-07, + "loss": 1.0099, + "mean_token_accuracy": 0.6981214880943298, + "num_tokens": 25389771.0, + "step": 981 + }, + { + "epoch": 0.10784098396661541, + "grad_norm": 2.642094135284424, + "learning_rate": 3.5907759882869694e-07, + "loss": 1.0458, + "mean_token_accuracy": 0.6871333122253418, + "num_tokens": 25410123.0, + "step": 982 + }, + { + "epoch": 0.10795080166922907, + "grad_norm": 2.616492986679077, + "learning_rate": 3.5944363103953144e-07, + "loss": 1.0564, + "mean_token_accuracy": 0.6825495958328247, + "num_tokens": 25432542.0, + "step": 983 + }, + { + "epoch": 0.10806061937184275, + "grad_norm": 2.3486251831054688, + "learning_rate": 3.59809663250366e-07, + "loss": 1.1059, + "mean_token_accuracy": 0.6708110570907593, + "num_tokens": 25458481.0, + "step": 984 + }, + { + "epoch": 0.10817043707445641, + "grad_norm": 2.403686046600342, + "learning_rate": 3.601756954612006e-07, + "loss": 1.0238, + "mean_token_accuracy": 0.7014857530593872, + "num_tokens": 25483261.0, + "step": 985 + }, + { + "epoch": 0.10828025477707007, + "grad_norm": 2.369173526763916, + "learning_rate": 3.6054172767203514e-07, + "loss": 1.0261, + "mean_token_accuracy": 0.6971492767333984, + "num_tokens": 25511391.0, + "step": 986 + }, + { + "epoch": 0.10839007247968373, + "grad_norm": 2.120908737182617, + "learning_rate": 3.609077598828697e-07, + "loss": 1.1022, + "mean_token_accuracy": 0.6756285429000854, + "num_tokens": 25541179.0, + "step": 987 + }, + { + "epoch": 0.10849989018229739, + "grad_norm": 2.568669557571411, + "learning_rate": 3.6127379209370423e-07, + "loss": 1.0124, + "mean_token_accuracy": 0.7006502151489258, + "num_tokens": 25564057.0, + "step": 988 + }, + { + "epoch": 0.10860970788491105, + "grad_norm": 2.383836030960083, + "learning_rate": 3.616398243045388e-07, + "loss": 1.0503, + "mean_token_accuracy": 0.6950105428695679, + "num_tokens": 25590617.0, + "step": 989 + }, + { + "epoch": 0.1087195255875247, + "grad_norm": 2.85473370552063, + "learning_rate": 3.6200585651537333e-07, + "loss": 1.055, + "mean_token_accuracy": 0.6903844475746155, + "num_tokens": 25610733.0, + "step": 990 + }, + { + "epoch": 0.10882934329013837, + "grad_norm": 2.420214891433716, + "learning_rate": 3.623718887262079e-07, + "loss": 1.0845, + "mean_token_accuracy": 0.6804792881011963, + "num_tokens": 25636827.0, + "step": 991 + }, + { + "epoch": 0.10893916099275203, + "grad_norm": 2.5567684173583984, + "learning_rate": 3.6273792093704243e-07, + "loss": 1.1169, + "mean_token_accuracy": 0.6770450472831726, + "num_tokens": 25659218.0, + "step": 992 + }, + { + "epoch": 0.10904897869536569, + "grad_norm": 2.5652623176574707, + "learning_rate": 3.6310395314787703e-07, + "loss": 1.0702, + "mean_token_accuracy": 0.6831117868423462, + "num_tokens": 25682816.0, + "step": 993 + }, + { + "epoch": 0.10915879639797936, + "grad_norm": 2.168358564376831, + "learning_rate": 3.6346998535871153e-07, + "loss": 1.0906, + "mean_token_accuracy": 0.6792124509811401, + "num_tokens": 25712238.0, + "step": 994 + }, + { + "epoch": 0.10926861410059302, + "grad_norm": 2.3809525966644287, + "learning_rate": 3.6383601756954613e-07, + "loss": 0.9847, + "mean_token_accuracy": 0.6975529789924622, + "num_tokens": 25737964.0, + "step": 995 + }, + { + "epoch": 0.10937843180320668, + "grad_norm": 2.2591285705566406, + "learning_rate": 3.642020497803807e-07, + "loss": 1.011, + "mean_token_accuracy": 0.6987552642822266, + "num_tokens": 25765993.0, + "step": 996 + }, + { + "epoch": 0.10948824950582034, + "grad_norm": 2.614527940750122, + "learning_rate": 3.645680819912152e-07, + "loss": 1.0556, + "mean_token_accuracy": 0.6901122331619263, + "num_tokens": 25789007.0, + "step": 997 + }, + { + "epoch": 0.109598067208434, + "grad_norm": 2.363065242767334, + "learning_rate": 3.649341142020498e-07, + "loss": 0.986, + "mean_token_accuracy": 0.7055075168609619, + "num_tokens": 25814004.0, + "step": 998 + }, + { + "epoch": 0.10970788491104766, + "grad_norm": 2.610910654067993, + "learning_rate": 3.653001464128843e-07, + "loss": 1.0602, + "mean_token_accuracy": 0.6874275207519531, + "num_tokens": 25835515.0, + "step": 999 + }, + { + "epoch": 0.10981770261366132, + "grad_norm": 2.49289870262146, + "learning_rate": 3.656661786237188e-07, + "loss": 1.1003, + "mean_token_accuracy": 0.6818499565124512, + "num_tokens": 25859300.0, + "step": 1000 + }, + { + "epoch": 0.10992752031627498, + "grad_norm": 2.343247413635254, + "learning_rate": 3.660322108345534e-07, + "loss": 1.0385, + "mean_token_accuracy": 0.6891762018203735, + "num_tokens": 25888053.0, + "step": 1001 + }, + { + "epoch": 0.11003733801888864, + "grad_norm": 2.4042763710021973, + "learning_rate": 3.66398243045388e-07, + "loss": 1.1538, + "mean_token_accuracy": 0.6606560945510864, + "num_tokens": 25913789.0, + "step": 1002 + }, + { + "epoch": 0.11014715572150231, + "grad_norm": 2.5785062313079834, + "learning_rate": 3.667642752562226e-07, + "loss": 1.163, + "mean_token_accuracy": 0.6700953245162964, + "num_tokens": 25936587.0, + "step": 1003 + }, + { + "epoch": 0.11025697342411597, + "grad_norm": 2.1614151000976562, + "learning_rate": 3.6713030746705707e-07, + "loss": 1.0475, + "mean_token_accuracy": 0.689159095287323, + "num_tokens": 25965940.0, + "step": 1004 + }, + { + "epoch": 0.11036679112672963, + "grad_norm": 2.269599676132202, + "learning_rate": 3.674963396778916e-07, + "loss": 0.9894, + "mean_token_accuracy": 0.7072023153305054, + "num_tokens": 25993188.0, + "step": 1005 + }, + { + "epoch": 0.11047660882934329, + "grad_norm": 2.4525396823883057, + "learning_rate": 3.678623718887262e-07, + "loss": 1.027, + "mean_token_accuracy": 0.6930451989173889, + "num_tokens": 26016160.0, + "step": 1006 + }, + { + "epoch": 0.11058642653195695, + "grad_norm": 2.4582695960998535, + "learning_rate": 3.682284040995607e-07, + "loss": 1.0372, + "mean_token_accuracy": 0.6891772747039795, + "num_tokens": 26039892.0, + "step": 1007 + }, + { + "epoch": 0.11069624423457061, + "grad_norm": 2.555173873901367, + "learning_rate": 3.6859443631039527e-07, + "loss": 1.0673, + "mean_token_accuracy": 0.6826508641242981, + "num_tokens": 26061680.0, + "step": 1008 + }, + { + "epoch": 0.11080606193718427, + "grad_norm": 2.3161017894744873, + "learning_rate": 3.6896046852122987e-07, + "loss": 1.1622, + "mean_token_accuracy": 0.65779709815979, + "num_tokens": 26088811.0, + "step": 1009 + }, + { + "epoch": 0.11091587963979793, + "grad_norm": 2.500910997390747, + "learning_rate": 3.693265007320644e-07, + "loss": 1.1229, + "mean_token_accuracy": 0.6703913807868958, + "num_tokens": 26113462.0, + "step": 1010 + }, + { + "epoch": 0.11102569734241159, + "grad_norm": 2.3418760299682617, + "learning_rate": 3.6969253294289897e-07, + "loss": 1.1498, + "mean_token_accuracy": 0.6637145280838013, + "num_tokens": 26140867.0, + "step": 1011 + }, + { + "epoch": 0.11113551504502525, + "grad_norm": 2.2769882678985596, + "learning_rate": 3.700585651537335e-07, + "loss": 1.0698, + "mean_token_accuracy": 0.6786097288131714, + "num_tokens": 26168121.0, + "step": 1012 + }, + { + "epoch": 0.11124533274763893, + "grad_norm": 2.7216968536376953, + "learning_rate": 3.7042459736456807e-07, + "loss": 1.0294, + "mean_token_accuracy": 0.6904330253601074, + "num_tokens": 26189626.0, + "step": 1013 + }, + { + "epoch": 0.11135515045025259, + "grad_norm": 2.575282573699951, + "learning_rate": 3.707906295754026e-07, + "loss": 1.0987, + "mean_token_accuracy": 0.675966739654541, + "num_tokens": 26213658.0, + "step": 1014 + }, + { + "epoch": 0.11146496815286625, + "grad_norm": 2.141164541244507, + "learning_rate": 3.7115666178623716e-07, + "loss": 1.0379, + "mean_token_accuracy": 0.6901103258132935, + "num_tokens": 26243731.0, + "step": 1015 + }, + { + "epoch": 0.1115747858554799, + "grad_norm": 2.616030216217041, + "learning_rate": 3.715226939970717e-07, + "loss": 1.0609, + "mean_token_accuracy": 0.6830824613571167, + "num_tokens": 26266656.0, + "step": 1016 + }, + { + "epoch": 0.11168460355809356, + "grad_norm": 2.9683542251586914, + "learning_rate": 3.7188872620790626e-07, + "loss": 1.0295, + "mean_token_accuracy": 0.6934095621109009, + "num_tokens": 26285162.0, + "step": 1017 + }, + { + "epoch": 0.11179442126070722, + "grad_norm": 2.478120803833008, + "learning_rate": 3.722547584187408e-07, + "loss": 1.0413, + "mean_token_accuracy": 0.6903218626976013, + "num_tokens": 26309229.0, + "step": 1018 + }, + { + "epoch": 0.11190423896332088, + "grad_norm": 2.4417076110839844, + "learning_rate": 3.726207906295754e-07, + "loss": 1.1016, + "mean_token_accuracy": 0.6785832047462463, + "num_tokens": 26334702.0, + "step": 1019 + }, + { + "epoch": 0.11201405666593454, + "grad_norm": 2.521195411682129, + "learning_rate": 3.7298682284040996e-07, + "loss": 1.0625, + "mean_token_accuracy": 0.6947410106658936, + "num_tokens": 26359954.0, + "step": 1020 + }, + { + "epoch": 0.1121238743685482, + "grad_norm": 2.4162261486053467, + "learning_rate": 3.7335285505124446e-07, + "loss": 1.0195, + "mean_token_accuracy": 0.6972956657409668, + "num_tokens": 26383642.0, + "step": 1021 + }, + { + "epoch": 0.11223369207116188, + "grad_norm": 2.1727874279022217, + "learning_rate": 3.7371888726207906e-07, + "loss": 1.0321, + "mean_token_accuracy": 0.6985775828361511, + "num_tokens": 26413718.0, + "step": 1022 + }, + { + "epoch": 0.11234350977377554, + "grad_norm": 2.0154671669006348, + "learning_rate": 3.740849194729136e-07, + "loss": 1.1477, + "mean_token_accuracy": 0.6662524342536926, + "num_tokens": 26448070.0, + "step": 1023 + }, + { + "epoch": 0.1124533274763892, + "grad_norm": 2.2479608058929443, + "learning_rate": 3.744509516837481e-07, + "loss": 1.1557, + "mean_token_accuracy": 0.6635657548904419, + "num_tokens": 26477876.0, + "step": 1024 + }, + { + "epoch": 0.11256314517900286, + "grad_norm": 2.489490270614624, + "learning_rate": 3.748169838945827e-07, + "loss": 1.1647, + "mean_token_accuracy": 0.6560684442520142, + "num_tokens": 26503463.0, + "step": 1025 + }, + { + "epoch": 0.11267296288161652, + "grad_norm": 2.3286702632904053, + "learning_rate": 3.7518301610541726e-07, + "loss": 1.0625, + "mean_token_accuracy": 0.6860700845718384, + "num_tokens": 26530910.0, + "step": 1026 + }, + { + "epoch": 0.11278278058423018, + "grad_norm": 2.3314971923828125, + "learning_rate": 3.7554904831625186e-07, + "loss": 1.127, + "mean_token_accuracy": 0.6651235818862915, + "num_tokens": 26556511.0, + "step": 1027 + }, + { + "epoch": 0.11289259828684384, + "grad_norm": 2.757477045059204, + "learning_rate": 3.7591508052708636e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.6737720370292664, + "num_tokens": 26577702.0, + "step": 1028 + }, + { + "epoch": 0.1130024159894575, + "grad_norm": 2.4347965717315674, + "learning_rate": 3.762811127379209e-07, + "loss": 1.1559, + "mean_token_accuracy": 0.6692007780075073, + "num_tokens": 26602465.0, + "step": 1029 + }, + { + "epoch": 0.11311223369207116, + "grad_norm": 2.759284019470215, + "learning_rate": 3.766471449487555e-07, + "loss": 1.0554, + "mean_token_accuracy": 0.6828721165657043, + "num_tokens": 26622436.0, + "step": 1030 + }, + { + "epoch": 0.11322205139468482, + "grad_norm": 2.426323175430298, + "learning_rate": 3.7701317715959e-07, + "loss": 1.0657, + "mean_token_accuracy": 0.6858416795730591, + "num_tokens": 26646386.0, + "step": 1031 + }, + { + "epoch": 0.11333186909729849, + "grad_norm": 2.3420403003692627, + "learning_rate": 3.7737920937042455e-07, + "loss": 1.1616, + "mean_token_accuracy": 0.6609072089195251, + "num_tokens": 26674495.0, + "step": 1032 + }, + { + "epoch": 0.11344168679991215, + "grad_norm": 2.2689321041107178, + "learning_rate": 3.7774524158125915e-07, + "loss": 1.039, + "mean_token_accuracy": 0.6923987865447998, + "num_tokens": 26701395.0, + "step": 1033 + }, + { + "epoch": 0.11355150450252581, + "grad_norm": 2.582639217376709, + "learning_rate": 3.781112737920937e-07, + "loss": 1.1269, + "mean_token_accuracy": 0.677029013633728, + "num_tokens": 26724537.0, + "step": 1034 + }, + { + "epoch": 0.11366132220513947, + "grad_norm": 2.4467532634735107, + "learning_rate": 3.7847730600292825e-07, + "loss": 1.1022, + "mean_token_accuracy": 0.677006721496582, + "num_tokens": 26747889.0, + "step": 1035 + }, + { + "epoch": 0.11377113990775313, + "grad_norm": 2.9735043048858643, + "learning_rate": 3.788433382137628e-07, + "loss": 1.0111, + "mean_token_accuracy": 0.6974565982818604, + "num_tokens": 26767707.0, + "step": 1036 + }, + { + "epoch": 0.11388095761036679, + "grad_norm": 2.348691701889038, + "learning_rate": 3.7920937042459735e-07, + "loss": 1.1278, + "mean_token_accuracy": 0.6818803548812866, + "num_tokens": 26793032.0, + "step": 1037 + }, + { + "epoch": 0.11399077531298045, + "grad_norm": 2.4152352809906006, + "learning_rate": 3.795754026354319e-07, + "loss": 1.0382, + "mean_token_accuracy": 0.690133273601532, + "num_tokens": 26819051.0, + "step": 1038 + }, + { + "epoch": 0.11410059301559411, + "grad_norm": 2.4473845958709717, + "learning_rate": 3.7994143484626645e-07, + "loss": 1.0428, + "mean_token_accuracy": 0.6932744383811951, + "num_tokens": 26842616.0, + "step": 1039 + }, + { + "epoch": 0.11421041071820777, + "grad_norm": 2.524994373321533, + "learning_rate": 3.80307467057101e-07, + "loss": 1.1129, + "mean_token_accuracy": 0.6829540729522705, + "num_tokens": 26865703.0, + "step": 1040 + }, + { + "epoch": 0.11432022842082144, + "grad_norm": 2.894874334335327, + "learning_rate": 3.8067349926793555e-07, + "loss": 1.1043, + "mean_token_accuracy": 0.6673097610473633, + "num_tokens": 26884984.0, + "step": 1041 + }, + { + "epoch": 0.1144300461234351, + "grad_norm": 2.346609592437744, + "learning_rate": 3.810395314787701e-07, + "loss": 1.082, + "mean_token_accuracy": 0.6838856935501099, + "num_tokens": 26912097.0, + "step": 1042 + }, + { + "epoch": 0.11453986382604876, + "grad_norm": 2.7135210037231445, + "learning_rate": 3.814055636896047e-07, + "loss": 1.0164, + "mean_token_accuracy": 0.699170708656311, + "num_tokens": 26932033.0, + "step": 1043 + }, + { + "epoch": 0.11464968152866242, + "grad_norm": 2.550924777984619, + "learning_rate": 3.8177159590043925e-07, + "loss": 1.1006, + "mean_token_accuracy": 0.6754614114761353, + "num_tokens": 26954418.0, + "step": 1044 + }, + { + "epoch": 0.11475949923127608, + "grad_norm": 2.1517701148986816, + "learning_rate": 3.8213762811127374e-07, + "loss": 1.065, + "mean_token_accuracy": 0.6868414878845215, + "num_tokens": 26984406.0, + "step": 1045 + }, + { + "epoch": 0.11486931693388974, + "grad_norm": 2.3850879669189453, + "learning_rate": 3.8250366032210835e-07, + "loss": 1.065, + "mean_token_accuracy": 0.6895577907562256, + "num_tokens": 27012181.0, + "step": 1046 + }, + { + "epoch": 0.1149791346365034, + "grad_norm": 2.2419445514678955, + "learning_rate": 3.828696925329429e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6778949499130249, + "num_tokens": 27040817.0, + "step": 1047 + }, + { + "epoch": 0.11508895233911706, + "grad_norm": 2.470062732696533, + "learning_rate": 3.832357247437774e-07, + "loss": 1.0022, + "mean_token_accuracy": 0.7108487486839294, + "num_tokens": 27063531.0, + "step": 1048 + }, + { + "epoch": 0.11519877004173072, + "grad_norm": 2.522977352142334, + "learning_rate": 3.83601756954612e-07, + "loss": 1.0395, + "mean_token_accuracy": 0.6863512992858887, + "num_tokens": 27085193.0, + "step": 1049 + }, + { + "epoch": 0.11530858774434438, + "grad_norm": 2.318726062774658, + "learning_rate": 3.8396778916544654e-07, + "loss": 1.0835, + "mean_token_accuracy": 0.6748493313789368, + "num_tokens": 27112026.0, + "step": 1050 + }, + { + "epoch": 0.11541840544695806, + "grad_norm": 2.5414342880249023, + "learning_rate": 3.8433382137628114e-07, + "loss": 1.0797, + "mean_token_accuracy": 0.6857605576515198, + "num_tokens": 27134736.0, + "step": 1051 + }, + { + "epoch": 0.11552822314957172, + "grad_norm": 2.088395833969116, + "learning_rate": 3.8469985358711564e-07, + "loss": 1.1905, + "mean_token_accuracy": 0.6508297920227051, + "num_tokens": 27167154.0, + "step": 1052 + }, + { + "epoch": 0.11563804085218538, + "grad_norm": 2.3137643337249756, + "learning_rate": 3.850658857979502e-07, + "loss": 1.0506, + "mean_token_accuracy": 0.6930873990058899, + "num_tokens": 27191779.0, + "step": 1053 + }, + { + "epoch": 0.11574785855479904, + "grad_norm": 2.6976985931396484, + "learning_rate": 3.854319180087848e-07, + "loss": 1.0669, + "mean_token_accuracy": 0.68101966381073, + "num_tokens": 27214207.0, + "step": 1054 + }, + { + "epoch": 0.1158576762574127, + "grad_norm": 2.401675224304199, + "learning_rate": 3.857979502196193e-07, + "loss": 1.1013, + "mean_token_accuracy": 0.6812283992767334, + "num_tokens": 27240519.0, + "step": 1055 + }, + { + "epoch": 0.11596749396002635, + "grad_norm": 2.3813247680664062, + "learning_rate": 3.8616398243045384e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.6972239017486572, + "num_tokens": 27265773.0, + "step": 1056 + }, + { + "epoch": 0.11607731166264001, + "grad_norm": 2.2695415019989014, + "learning_rate": 3.8653001464128844e-07, + "loss": 1.1263, + "mean_token_accuracy": 0.6668758392333984, + "num_tokens": 27294062.0, + "step": 1057 + }, + { + "epoch": 0.11618712936525367, + "grad_norm": 2.5665860176086426, + "learning_rate": 3.8689604685212293e-07, + "loss": 1.0597, + "mean_token_accuracy": 0.6867109537124634, + "num_tokens": 27317200.0, + "step": 1058 + }, + { + "epoch": 0.11629694706786733, + "grad_norm": 2.244823455810547, + "learning_rate": 3.8726207906295754e-07, + "loss": 1.1156, + "mean_token_accuracy": 0.672724723815918, + "num_tokens": 27345129.0, + "step": 1059 + }, + { + "epoch": 0.11640676477048101, + "grad_norm": 2.1726157665252686, + "learning_rate": 3.876281112737921e-07, + "loss": 1.1299, + "mean_token_accuracy": 0.6692795157432556, + "num_tokens": 27376500.0, + "step": 1060 + }, + { + "epoch": 0.11651658247309467, + "grad_norm": 2.424267053604126, + "learning_rate": 3.8799414348462663e-07, + "loss": 1.1113, + "mean_token_accuracy": 0.6677703857421875, + "num_tokens": 27405136.0, + "step": 1061 + }, + { + "epoch": 0.11662640017570833, + "grad_norm": 2.391723155975342, + "learning_rate": 3.883601756954612e-07, + "loss": 1.067, + "mean_token_accuracy": 0.6794434785842896, + "num_tokens": 27430922.0, + "step": 1062 + }, + { + "epoch": 0.11673621787832199, + "grad_norm": 2.681997299194336, + "learning_rate": 3.8872620790629573e-07, + "loss": 1.0342, + "mean_token_accuracy": 0.6932459473609924, + "num_tokens": 27452160.0, + "step": 1063 + }, + { + "epoch": 0.11684603558093565, + "grad_norm": 2.386410713195801, + "learning_rate": 3.890922401171303e-07, + "loss": 1.0959, + "mean_token_accuracy": 0.679915189743042, + "num_tokens": 27477265.0, + "step": 1064 + }, + { + "epoch": 0.11695585328354931, + "grad_norm": 2.4279673099517822, + "learning_rate": 3.8945827232796483e-07, + "loss": 1.1279, + "mean_token_accuracy": 0.6687041521072388, + "num_tokens": 27503052.0, + "step": 1065 + }, + { + "epoch": 0.11706567098616297, + "grad_norm": 2.5268311500549316, + "learning_rate": 3.898243045387994e-07, + "loss": 1.0518, + "mean_token_accuracy": 0.6933548450469971, + "num_tokens": 27526173.0, + "step": 1066 + }, + { + "epoch": 0.11717548868877663, + "grad_norm": 2.152935028076172, + "learning_rate": 3.90190336749634e-07, + "loss": 1.0965, + "mean_token_accuracy": 0.686783492565155, + "num_tokens": 27556183.0, + "step": 1067 + }, + { + "epoch": 0.11728530639139029, + "grad_norm": 2.2652602195739746, + "learning_rate": 3.9055636896046853e-07, + "loss": 1.0697, + "mean_token_accuracy": 0.6846076250076294, + "num_tokens": 27583306.0, + "step": 1068 + }, + { + "epoch": 0.11739512409400395, + "grad_norm": 2.4040510654449463, + "learning_rate": 3.9092240117130303e-07, + "loss": 1.0284, + "mean_token_accuracy": 0.6929308176040649, + "num_tokens": 27607080.0, + "step": 1069 + }, + { + "epoch": 0.11750494179661762, + "grad_norm": 2.072540760040283, + "learning_rate": 3.9128843338213763e-07, + "loss": 1.1304, + "mean_token_accuracy": 0.6685150861740112, + "num_tokens": 27639131.0, + "step": 1070 + }, + { + "epoch": 0.11761475949923128, + "grad_norm": 2.5412678718566895, + "learning_rate": 3.916544655929722e-07, + "loss": 1.0565, + "mean_token_accuracy": 0.6821625232696533, + "num_tokens": 27660915.0, + "step": 1071 + }, + { + "epoch": 0.11772457720184494, + "grad_norm": 2.1467039585113525, + "learning_rate": 3.920204978038067e-07, + "loss": 1.1065, + "mean_token_accuracy": 0.6735667586326599, + "num_tokens": 27693707.0, + "step": 1072 + }, + { + "epoch": 0.1178343949044586, + "grad_norm": 2.3307487964630127, + "learning_rate": 3.923865300146413e-07, + "loss": 1.1863, + "mean_token_accuracy": 0.6569669246673584, + "num_tokens": 27721535.0, + "step": 1073 + }, + { + "epoch": 0.11794421260707226, + "grad_norm": 2.5332982540130615, + "learning_rate": 3.927525622254758e-07, + "loss": 0.987, + "mean_token_accuracy": 0.7012380957603455, + "num_tokens": 27744277.0, + "step": 1074 + }, + { + "epoch": 0.11805403030968592, + "grad_norm": 2.2359185218811035, + "learning_rate": 3.9311859443631043e-07, + "loss": 1.1023, + "mean_token_accuracy": 0.6751954555511475, + "num_tokens": 27773010.0, + "step": 1075 + }, + { + "epoch": 0.11816384801229958, + "grad_norm": 2.742396831512451, + "learning_rate": 3.934846266471449e-07, + "loss": 1.1289, + "mean_token_accuracy": 0.6809896230697632, + "num_tokens": 27794162.0, + "step": 1076 + }, + { + "epoch": 0.11827366571491324, + "grad_norm": 2.4598991870880127, + "learning_rate": 3.9385065885797947e-07, + "loss": 1.0845, + "mean_token_accuracy": 0.6776337623596191, + "num_tokens": 27817334.0, + "step": 1077 + }, + { + "epoch": 0.1183834834175269, + "grad_norm": 2.5049564838409424, + "learning_rate": 3.942166910688141e-07, + "loss": 1.1483, + "mean_token_accuracy": 0.6589092016220093, + "num_tokens": 27841083.0, + "step": 1078 + }, + { + "epoch": 0.11849330112014057, + "grad_norm": 2.683666467666626, + "learning_rate": 3.9458272327964857e-07, + "loss": 1.0132, + "mean_token_accuracy": 0.6927131414413452, + "num_tokens": 27861503.0, + "step": 1079 + }, + { + "epoch": 0.11860311882275423, + "grad_norm": 2.3276216983795166, + "learning_rate": 3.949487554904831e-07, + "loss": 1.0557, + "mean_token_accuracy": 0.6953656077384949, + "num_tokens": 27888130.0, + "step": 1080 + }, + { + "epoch": 0.1187129365253679, + "grad_norm": 2.3859846591949463, + "learning_rate": 3.953147877013177e-07, + "loss": 1.0638, + "mean_token_accuracy": 0.6813187599182129, + "num_tokens": 27913796.0, + "step": 1081 + }, + { + "epoch": 0.11882275422798155, + "grad_norm": 2.445350408554077, + "learning_rate": 3.956808199121522e-07, + "loss": 1.0904, + "mean_token_accuracy": 0.6803663969039917, + "num_tokens": 27938790.0, + "step": 1082 + }, + { + "epoch": 0.11893257193059521, + "grad_norm": 2.2792532444000244, + "learning_rate": 3.960468521229868e-07, + "loss": 1.0602, + "mean_token_accuracy": 0.6842944622039795, + "num_tokens": 27964042.0, + "step": 1083 + }, + { + "epoch": 0.11904238963320887, + "grad_norm": 2.307920217514038, + "learning_rate": 3.9641288433382137e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6872172355651855, + "num_tokens": 27990871.0, + "step": 1084 + }, + { + "epoch": 0.11915220733582253, + "grad_norm": 2.3162283897399902, + "learning_rate": 3.967789165446559e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6730228662490845, + "num_tokens": 28019249.0, + "step": 1085 + }, + { + "epoch": 0.11926202503843619, + "grad_norm": 2.2915797233581543, + "learning_rate": 3.9714494875549047e-07, + "loss": 1.0797, + "mean_token_accuracy": 0.6741214990615845, + "num_tokens": 28047209.0, + "step": 1086 + }, + { + "epoch": 0.11937184274104985, + "grad_norm": 2.1999635696411133, + "learning_rate": 3.97510980966325e-07, + "loss": 1.1573, + "mean_token_accuracy": 0.6718043088912964, + "num_tokens": 28076909.0, + "step": 1087 + }, + { + "epoch": 0.11948166044366351, + "grad_norm": 2.2774100303649902, + "learning_rate": 3.9787701317715957e-07, + "loss": 1.1595, + "mean_token_accuracy": 0.660725474357605, + "num_tokens": 28105066.0, + "step": 1088 + }, + { + "epoch": 0.11959147814627719, + "grad_norm": 2.7392237186431885, + "learning_rate": 3.982430453879941e-07, + "loss": 1.0631, + "mean_token_accuracy": 0.6773344874382019, + "num_tokens": 28126388.0, + "step": 1089 + }, + { + "epoch": 0.11970129584889085, + "grad_norm": 2.3948051929473877, + "learning_rate": 3.9860907759882866e-07, + "loss": 1.0194, + "mean_token_accuracy": 0.6925560235977173, + "num_tokens": 28152525.0, + "step": 1090 + }, + { + "epoch": 0.1198111135515045, + "grad_norm": 2.411407232284546, + "learning_rate": 3.9897510980966327e-07, + "loss": 0.9926, + "mean_token_accuracy": 0.7058228850364685, + "num_tokens": 28176649.0, + "step": 1091 + }, + { + "epoch": 0.11992093125411817, + "grad_norm": 2.819448709487915, + "learning_rate": 3.993411420204978e-07, + "loss": 0.9766, + "mean_token_accuracy": 0.7147974967956543, + "num_tokens": 28194986.0, + "step": 1092 + }, + { + "epoch": 0.12003074895673183, + "grad_norm": 2.205357789993286, + "learning_rate": 3.997071742313323e-07, + "loss": 1.0701, + "mean_token_accuracy": 0.6849117279052734, + "num_tokens": 28225155.0, + "step": 1093 + }, + { + "epoch": 0.12014056665934549, + "grad_norm": 2.2319211959838867, + "learning_rate": 4.000732064421669e-07, + "loss": 1.0624, + "mean_token_accuracy": 0.6816300749778748, + "num_tokens": 28254599.0, + "step": 1094 + }, + { + "epoch": 0.12025038436195915, + "grad_norm": 2.5574417114257812, + "learning_rate": 4.0043923865300146e-07, + "loss": 1.0067, + "mean_token_accuracy": 0.6955536007881165, + "num_tokens": 28277178.0, + "step": 1095 + }, + { + "epoch": 0.1203602020645728, + "grad_norm": 2.4359447956085205, + "learning_rate": 4.0080527086383596e-07, + "loss": 1.0737, + "mean_token_accuracy": 0.6850723028182983, + "num_tokens": 28304301.0, + "step": 1096 + }, + { + "epoch": 0.12047001976718646, + "grad_norm": 2.338148355484009, + "learning_rate": 4.0117130307467056e-07, + "loss": 0.9603, + "mean_token_accuracy": 0.7116266489028931, + "num_tokens": 28328572.0, + "step": 1097 + }, + { + "epoch": 0.12057983746980014, + "grad_norm": 2.541919231414795, + "learning_rate": 4.015373352855051e-07, + "loss": 1.0518, + "mean_token_accuracy": 0.6869342923164368, + "num_tokens": 28352705.0, + "step": 1098 + }, + { + "epoch": 0.1206896551724138, + "grad_norm": 2.38617205619812, + "learning_rate": 4.0190336749633966e-07, + "loss": 1.1039, + "mean_token_accuracy": 0.6785970330238342, + "num_tokens": 28379284.0, + "step": 1099 + }, + { + "epoch": 0.12079947287502746, + "grad_norm": 1.9845595359802246, + "learning_rate": 4.022693997071742e-07, + "loss": 1.0981, + "mean_token_accuracy": 0.6787306666374207, + "num_tokens": 28412910.0, + "step": 1100 + }, + { + "epoch": 0.12090929057764112, + "grad_norm": 2.200389862060547, + "learning_rate": 4.0263543191800876e-07, + "loss": 1.0162, + "mean_token_accuracy": 0.7035322189331055, + "num_tokens": 28439611.0, + "step": 1101 + }, + { + "epoch": 0.12101910828025478, + "grad_norm": 2.4285762310028076, + "learning_rate": 4.0300146412884336e-07, + "loss": 1.1241, + "mean_token_accuracy": 0.673492431640625, + "num_tokens": 28464781.0, + "step": 1102 + }, + { + "epoch": 0.12112892598286844, + "grad_norm": 2.6050593852996826, + "learning_rate": 4.0336749633967785e-07, + "loss": 1.0294, + "mean_token_accuracy": 0.6915048360824585, + "num_tokens": 28486031.0, + "step": 1103 + }, + { + "epoch": 0.1212387436854821, + "grad_norm": 2.568925380706787, + "learning_rate": 4.037335285505124e-07, + "loss": 0.9751, + "mean_token_accuracy": 0.7058770656585693, + "num_tokens": 28507282.0, + "step": 1104 + }, + { + "epoch": 0.12134856138809576, + "grad_norm": 2.463566303253174, + "learning_rate": 4.04099560761347e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.6820849776268005, + "num_tokens": 28532491.0, + "step": 1105 + }, + { + "epoch": 0.12145837909070942, + "grad_norm": 2.20703125, + "learning_rate": 4.044655929721815e-07, + "loss": 1.0158, + "mean_token_accuracy": 0.7062997817993164, + "num_tokens": 28559439.0, + "step": 1106 + }, + { + "epoch": 0.12156819679332308, + "grad_norm": 2.3386011123657227, + "learning_rate": 4.048316251830161e-07, + "loss": 1.1146, + "mean_token_accuracy": 0.6719095706939697, + "num_tokens": 28584914.0, + "step": 1107 + }, + { + "epoch": 0.12167801449593675, + "grad_norm": 2.2280519008636475, + "learning_rate": 4.0519765739385065e-07, + "loss": 1.078, + "mean_token_accuracy": 0.6847671270370483, + "num_tokens": 28613358.0, + "step": 1108 + }, + { + "epoch": 0.12178783219855041, + "grad_norm": 2.3997409343719482, + "learning_rate": 4.055636896046852e-07, + "loss": 1.1091, + "mean_token_accuracy": 0.6854047179222107, + "num_tokens": 28637827.0, + "step": 1109 + }, + { + "epoch": 0.12189764990116407, + "grad_norm": 2.3254239559173584, + "learning_rate": 4.0592972181551975e-07, + "loss": 1.1621, + "mean_token_accuracy": 0.6633415222167969, + "num_tokens": 28663784.0, + "step": 1110 + }, + { + "epoch": 0.12200746760377773, + "grad_norm": 2.4499258995056152, + "learning_rate": 4.062957540263543e-07, + "loss": 1.1241, + "mean_token_accuracy": 0.6692578792572021, + "num_tokens": 28686562.0, + "step": 1111 + }, + { + "epoch": 0.12211728530639139, + "grad_norm": 2.4304323196411133, + "learning_rate": 4.0666178623718885e-07, + "loss": 1.1196, + "mean_token_accuracy": 0.6788274049758911, + "num_tokens": 28713361.0, + "step": 1112 + }, + { + "epoch": 0.12222710300900505, + "grad_norm": 2.5004193782806396, + "learning_rate": 4.070278184480234e-07, + "loss": 1.0399, + "mean_token_accuracy": 0.6903141736984253, + "num_tokens": 28736221.0, + "step": 1113 + }, + { + "epoch": 0.12233692071161871, + "grad_norm": 2.068646192550659, + "learning_rate": 4.0739385065885795e-07, + "loss": 1.02, + "mean_token_accuracy": 0.6939507722854614, + "num_tokens": 28767480.0, + "step": 1114 + }, + { + "epoch": 0.12244673841423237, + "grad_norm": 2.6152374744415283, + "learning_rate": 4.0775988286969255e-07, + "loss": 1.0386, + "mean_token_accuracy": 0.6956140995025635, + "num_tokens": 28790622.0, + "step": 1115 + }, + { + "epoch": 0.12255655611684603, + "grad_norm": 2.2809624671936035, + "learning_rate": 4.081259150805271e-07, + "loss": 1.0794, + "mean_token_accuracy": 0.6899782419204712, + "num_tokens": 28816316.0, + "step": 1116 + }, + { + "epoch": 0.1226663738194597, + "grad_norm": 2.4637773036956787, + "learning_rate": 4.084919472913616e-07, + "loss": 1.0497, + "mean_token_accuracy": 0.687053918838501, + "num_tokens": 28838421.0, + "step": 1117 + }, + { + "epoch": 0.12277619152207336, + "grad_norm": 2.5834529399871826, + "learning_rate": 4.088579795021962e-07, + "loss": 1.1655, + "mean_token_accuracy": 0.6630122065544128, + "num_tokens": 28863527.0, + "step": 1118 + }, + { + "epoch": 0.12288600922468702, + "grad_norm": 2.724616765975952, + "learning_rate": 4.0922401171303075e-07, + "loss": 0.9151, + "mean_token_accuracy": 0.7244236469268799, + "num_tokens": 28882731.0, + "step": 1119 + }, + { + "epoch": 0.12299582692730068, + "grad_norm": 2.5454390048980713, + "learning_rate": 4.0959004392386524e-07, + "loss": 1.0837, + "mean_token_accuracy": 0.6774715781211853, + "num_tokens": 28906336.0, + "step": 1120 + }, + { + "epoch": 0.12310564462991434, + "grad_norm": 2.493520736694336, + "learning_rate": 4.0995607613469984e-07, + "loss": 0.9895, + "mean_token_accuracy": 0.7078804969787598, + "num_tokens": 28927718.0, + "step": 1121 + }, + { + "epoch": 0.123215462332528, + "grad_norm": 2.324289560317993, + "learning_rate": 4.103221083455344e-07, + "loss": 1.1504, + "mean_token_accuracy": 0.6605507731437683, + "num_tokens": 28956062.0, + "step": 1122 + }, + { + "epoch": 0.12332528003514166, + "grad_norm": 2.3967690467834473, + "learning_rate": 4.1068814055636894e-07, + "loss": 1.0908, + "mean_token_accuracy": 0.6821309924125671, + "num_tokens": 28980636.0, + "step": 1123 + }, + { + "epoch": 0.12343509773775532, + "grad_norm": 2.415834665298462, + "learning_rate": 4.110541727672035e-07, + "loss": 1.1203, + "mean_token_accuracy": 0.6646761298179626, + "num_tokens": 29005838.0, + "step": 1124 + }, + { + "epoch": 0.12354491544036898, + "grad_norm": 2.309438705444336, + "learning_rate": 4.1142020497803804e-07, + "loss": 1.1355, + "mean_token_accuracy": 0.6631471514701843, + "num_tokens": 29032584.0, + "step": 1125 + }, + { + "epoch": 0.12365473314298264, + "grad_norm": 2.5408501625061035, + "learning_rate": 4.1178623718887264e-07, + "loss": 1.07, + "mean_token_accuracy": 0.6834555268287659, + "num_tokens": 29054249.0, + "step": 1126 + }, + { + "epoch": 0.12376455084559632, + "grad_norm": 2.1550068855285645, + "learning_rate": 4.1215226939970714e-07, + "loss": 1.0872, + "mean_token_accuracy": 0.6781456470489502, + "num_tokens": 29083825.0, + "step": 1127 + }, + { + "epoch": 0.12387436854820998, + "grad_norm": 2.1699886322021484, + "learning_rate": 4.125183016105417e-07, + "loss": 1.0942, + "mean_token_accuracy": 0.6861288547515869, + "num_tokens": 29113619.0, + "step": 1128 + }, + { + "epoch": 0.12398418625082364, + "grad_norm": 2.2742395401000977, + "learning_rate": 4.128843338213763e-07, + "loss": 1.1341, + "mean_token_accuracy": 0.6719455718994141, + "num_tokens": 29141354.0, + "step": 1129 + }, + { + "epoch": 0.1240940039534373, + "grad_norm": 2.537862777709961, + "learning_rate": 4.132503660322108e-07, + "loss": 1.0741, + "mean_token_accuracy": 0.677521824836731, + "num_tokens": 29163113.0, + "step": 1130 + }, + { + "epoch": 0.12420382165605096, + "grad_norm": 2.228372573852539, + "learning_rate": 4.136163982430454e-07, + "loss": 1.0301, + "mean_token_accuracy": 0.6932919025421143, + "num_tokens": 29190986.0, + "step": 1131 + }, + { + "epoch": 0.12431363935866462, + "grad_norm": 2.151348352432251, + "learning_rate": 4.1398243045387994e-07, + "loss": 0.9624, + "mean_token_accuracy": 0.7112988233566284, + "num_tokens": 29218126.0, + "step": 1132 + }, + { + "epoch": 0.12442345706127828, + "grad_norm": 2.277275562286377, + "learning_rate": 4.143484626647145e-07, + "loss": 1.0816, + "mean_token_accuracy": 0.6765413284301758, + "num_tokens": 29246177.0, + "step": 1133 + }, + { + "epoch": 0.12453327476389194, + "grad_norm": 2.421872138977051, + "learning_rate": 4.1471449487554904e-07, + "loss": 1.0468, + "mean_token_accuracy": 0.6971226930618286, + "num_tokens": 29270132.0, + "step": 1134 + }, + { + "epoch": 0.1246430924665056, + "grad_norm": 2.2113382816314697, + "learning_rate": 4.150805270863836e-07, + "loss": 1.1168, + "mean_token_accuracy": 0.6691528558731079, + "num_tokens": 29298680.0, + "step": 1135 + }, + { + "epoch": 0.12475291016911927, + "grad_norm": 2.59818434715271, + "learning_rate": 4.1544655929721813e-07, + "loss": 1.0205, + "mean_token_accuracy": 0.6977518796920776, + "num_tokens": 29319341.0, + "step": 1136 + }, + { + "epoch": 0.12486272787173293, + "grad_norm": 2.3300628662109375, + "learning_rate": 4.158125915080527e-07, + "loss": 1.0645, + "mean_token_accuracy": 0.6979856491088867, + "num_tokens": 29344645.0, + "step": 1137 + }, + { + "epoch": 0.12497254557434659, + "grad_norm": 2.2435569763183594, + "learning_rate": 4.1617862371888723e-07, + "loss": 1.0362, + "mean_token_accuracy": 0.691257119178772, + "num_tokens": 29370995.0, + "step": 1138 + }, + { + "epoch": 0.12508236327696023, + "grad_norm": 2.408529758453369, + "learning_rate": 4.1654465592972183e-07, + "loss": 1.1478, + "mean_token_accuracy": 0.6668035984039307, + "num_tokens": 29396692.0, + "step": 1139 + }, + { + "epoch": 0.1251921809795739, + "grad_norm": 2.187314033508301, + "learning_rate": 4.1691068814055633e-07, + "loss": 1.096, + "mean_token_accuracy": 0.6723572015762329, + "num_tokens": 29426669.0, + "step": 1140 + }, + { + "epoch": 0.12530199868218758, + "grad_norm": 2.3342816829681396, + "learning_rate": 4.172767203513909e-07, + "loss": 1.0041, + "mean_token_accuracy": 0.6992186903953552, + "num_tokens": 29451681.0, + "step": 1141 + }, + { + "epoch": 0.12541181638480123, + "grad_norm": 2.027444839477539, + "learning_rate": 4.176427525622255e-07, + "loss": 1.1303, + "mean_token_accuracy": 0.6722565293312073, + "num_tokens": 29484702.0, + "step": 1142 + }, + { + "epoch": 0.1255216340874149, + "grad_norm": 2.6326253414154053, + "learning_rate": 4.1800878477306003e-07, + "loss": 0.9795, + "mean_token_accuracy": 0.704829216003418, + "num_tokens": 29504341.0, + "step": 1143 + }, + { + "epoch": 0.12563145179002855, + "grad_norm": 2.2914247512817383, + "learning_rate": 4.183748169838945e-07, + "loss": 1.0092, + "mean_token_accuracy": 0.6917015314102173, + "num_tokens": 29530130.0, + "step": 1144 + }, + { + "epoch": 0.12574126949264222, + "grad_norm": 2.175969362258911, + "learning_rate": 4.1874084919472913e-07, + "loss": 1.1085, + "mean_token_accuracy": 0.673133134841919, + "num_tokens": 29559283.0, + "step": 1145 + }, + { + "epoch": 0.12585108719525587, + "grad_norm": 2.4005212783813477, + "learning_rate": 4.191068814055637e-07, + "loss": 1.1016, + "mean_token_accuracy": 0.6762024760246277, + "num_tokens": 29584911.0, + "step": 1146 + }, + { + "epoch": 0.12596090489786954, + "grad_norm": 2.2948641777038574, + "learning_rate": 4.194729136163982e-07, + "loss": 0.987, + "mean_token_accuracy": 0.7013576030731201, + "num_tokens": 29610399.0, + "step": 1147 + }, + { + "epoch": 0.1260707226004832, + "grad_norm": 2.4764888286590576, + "learning_rate": 4.198389458272328e-07, + "loss": 0.9572, + "mean_token_accuracy": 0.7082735896110535, + "num_tokens": 29633032.0, + "step": 1148 + }, + { + "epoch": 0.12618054030309686, + "grad_norm": 2.3719425201416016, + "learning_rate": 4.202049780380673e-07, + "loss": 1.067, + "mean_token_accuracy": 0.6808634400367737, + "num_tokens": 29657214.0, + "step": 1149 + }, + { + "epoch": 0.1262903580057105, + "grad_norm": 2.0913712978363037, + "learning_rate": 4.205710102489019e-07, + "loss": 1.0744, + "mean_token_accuracy": 0.6827908158302307, + "num_tokens": 29688013.0, + "step": 1150 + }, + { + "epoch": 0.12640017570832418, + "grad_norm": 2.5474486351013184, + "learning_rate": 4.209370424597364e-07, + "loss": 1.0124, + "mean_token_accuracy": 0.6961080431938171, + "num_tokens": 29710451.0, + "step": 1151 + }, + { + "epoch": 0.12650999341093785, + "grad_norm": 2.3921022415161133, + "learning_rate": 4.2130307467057097e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.6775725483894348, + "num_tokens": 29735654.0, + "step": 1152 + }, + { + "epoch": 0.1266198111135515, + "grad_norm": 2.3729937076568604, + "learning_rate": 4.2166910688140557e-07, + "loss": 1.1352, + "mean_token_accuracy": 0.6623997688293457, + "num_tokens": 29764641.0, + "step": 1153 + }, + { + "epoch": 0.12672962881616517, + "grad_norm": 2.4832677841186523, + "learning_rate": 4.2203513909224007e-07, + "loss": 1.1451, + "mean_token_accuracy": 0.6703879237174988, + "num_tokens": 29790458.0, + "step": 1154 + }, + { + "epoch": 0.12683944651877882, + "grad_norm": 2.3887436389923096, + "learning_rate": 4.2240117130307467e-07, + "loss": 1.0512, + "mean_token_accuracy": 0.6879705190658569, + "num_tokens": 29816208.0, + "step": 1155 + }, + { + "epoch": 0.1269492642213925, + "grad_norm": 2.409111499786377, + "learning_rate": 4.227672035139092e-07, + "loss": 1.0827, + "mean_token_accuracy": 0.6888706088066101, + "num_tokens": 29840722.0, + "step": 1156 + }, + { + "epoch": 0.12705908192400614, + "grad_norm": 2.2663586139678955, + "learning_rate": 4.231332357247437e-07, + "loss": 1.1734, + "mean_token_accuracy": 0.6577234268188477, + "num_tokens": 29869138.0, + "step": 1157 + }, + { + "epoch": 0.1271688996266198, + "grad_norm": 2.532405138015747, + "learning_rate": 4.234992679355783e-07, + "loss": 1.0455, + "mean_token_accuracy": 0.6859573125839233, + "num_tokens": 29891804.0, + "step": 1158 + }, + { + "epoch": 0.12727871732923346, + "grad_norm": 2.2894558906555176, + "learning_rate": 4.2386530014641287e-07, + "loss": 1.1324, + "mean_token_accuracy": 0.6735965609550476, + "num_tokens": 29918368.0, + "step": 1159 + }, + { + "epoch": 0.12738853503184713, + "grad_norm": 2.2631757259368896, + "learning_rate": 4.242313323572474e-07, + "loss": 1.0932, + "mean_token_accuracy": 0.6827664375305176, + "num_tokens": 29947243.0, + "step": 1160 + }, + { + "epoch": 0.1274983527344608, + "grad_norm": 2.201087236404419, + "learning_rate": 4.2459736456808197e-07, + "loss": 1.1245, + "mean_token_accuracy": 0.6715007424354553, + "num_tokens": 29976185.0, + "step": 1161 + }, + { + "epoch": 0.12760817043707445, + "grad_norm": 2.2350008487701416, + "learning_rate": 4.249633967789165e-07, + "loss": 1.0911, + "mean_token_accuracy": 0.6748554706573486, + "num_tokens": 30004893.0, + "step": 1162 + }, + { + "epoch": 0.12771798813968813, + "grad_norm": 2.1546993255615234, + "learning_rate": 4.253294289897511e-07, + "loss": 1.1819, + "mean_token_accuracy": 0.657321572303772, + "num_tokens": 30033645.0, + "step": 1163 + }, + { + "epoch": 0.12782780584230177, + "grad_norm": 2.5030007362365723, + "learning_rate": 4.256954612005856e-07, + "loss": 1.0834, + "mean_token_accuracy": 0.6855176687240601, + "num_tokens": 30055734.0, + "step": 1164 + }, + { + "epoch": 0.12793762354491545, + "grad_norm": 2.5374064445495605, + "learning_rate": 4.2606149341142016e-07, + "loss": 1.0043, + "mean_token_accuracy": 0.6945376396179199, + "num_tokens": 30076609.0, + "step": 1165 + }, + { + "epoch": 0.1280474412475291, + "grad_norm": 2.530912399291992, + "learning_rate": 4.2642752562225476e-07, + "loss": 1.1211, + "mean_token_accuracy": 0.6634238958358765, + "num_tokens": 30101695.0, + "step": 1166 + }, + { + "epoch": 0.12815725895014277, + "grad_norm": 2.3894031047821045, + "learning_rate": 4.267935578330893e-07, + "loss": 1.1276, + "mean_token_accuracy": 0.6732683777809143, + "num_tokens": 30128385.0, + "step": 1167 + }, + { + "epoch": 0.1282670766527564, + "grad_norm": 2.57753586769104, + "learning_rate": 4.271595900439238e-07, + "loss": 0.909, + "mean_token_accuracy": 0.7287189960479736, + "num_tokens": 30149434.0, + "step": 1168 + }, + { + "epoch": 0.12837689435537009, + "grad_norm": 2.2794666290283203, + "learning_rate": 4.275256222547584e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.6967371106147766, + "num_tokens": 30175072.0, + "step": 1169 + }, + { + "epoch": 0.12848671205798376, + "grad_norm": 2.171645402908325, + "learning_rate": 4.2789165446559296e-07, + "loss": 1.0784, + "mean_token_accuracy": 0.6835165023803711, + "num_tokens": 30204221.0, + "step": 1170 + }, + { + "epoch": 0.1285965297605974, + "grad_norm": 2.198729991912842, + "learning_rate": 4.282576866764275e-07, + "loss": 1.0994, + "mean_token_accuracy": 0.6769664287567139, + "num_tokens": 30232624.0, + "step": 1171 + }, + { + "epoch": 0.12870634746321108, + "grad_norm": 2.251528739929199, + "learning_rate": 4.2862371888726206e-07, + "loss": 1.0834, + "mean_token_accuracy": 0.6846597194671631, + "num_tokens": 30259309.0, + "step": 1172 + }, + { + "epoch": 0.12881616516582473, + "grad_norm": 2.283550500869751, + "learning_rate": 4.289897510980966e-07, + "loss": 1.0784, + "mean_token_accuracy": 0.6862390041351318, + "num_tokens": 30286533.0, + "step": 1173 + }, + { + "epoch": 0.1289259828684384, + "grad_norm": 2.455364942550659, + "learning_rate": 4.293557833089312e-07, + "loss": 1.0399, + "mean_token_accuracy": 0.6889116168022156, + "num_tokens": 30311150.0, + "step": 1174 + }, + { + "epoch": 0.12903580057105205, + "grad_norm": 2.4577584266662598, + "learning_rate": 4.297218155197657e-07, + "loss": 1.0061, + "mean_token_accuracy": 0.6963770389556885, + "num_tokens": 30334858.0, + "step": 1175 + }, + { + "epoch": 0.12914561827366572, + "grad_norm": 2.5663156509399414, + "learning_rate": 4.3008784773060026e-07, + "loss": 0.9823, + "mean_token_accuracy": 0.7098066806793213, + "num_tokens": 30358502.0, + "step": 1176 + }, + { + "epoch": 0.12925543597627936, + "grad_norm": 2.334804058074951, + "learning_rate": 4.3045387994143486e-07, + "loss": 1.1137, + "mean_token_accuracy": 0.6701538562774658, + "num_tokens": 30384783.0, + "step": 1177 + }, + { + "epoch": 0.12936525367889304, + "grad_norm": 2.440227508544922, + "learning_rate": 4.3081991215226935e-07, + "loss": 1.0846, + "mean_token_accuracy": 0.6905848979949951, + "num_tokens": 30408913.0, + "step": 1178 + }, + { + "epoch": 0.1294750713815067, + "grad_norm": 2.391160011291504, + "learning_rate": 4.3118594436310396e-07, + "loss": 1.0382, + "mean_token_accuracy": 0.6947424411773682, + "num_tokens": 30433900.0, + "step": 1179 + }, + { + "epoch": 0.12958488908412036, + "grad_norm": 2.492603302001953, + "learning_rate": 4.315519765739385e-07, + "loss": 1.0016, + "mean_token_accuracy": 0.7035067081451416, + "num_tokens": 30461190.0, + "step": 1180 + }, + { + "epoch": 0.12969470678673403, + "grad_norm": 2.395242691040039, + "learning_rate": 4.31918008784773e-07, + "loss": 1.1008, + "mean_token_accuracy": 0.6753692626953125, + "num_tokens": 30489651.0, + "step": 1181 + }, + { + "epoch": 0.12980452448934768, + "grad_norm": 2.3375329971313477, + "learning_rate": 4.322840409956076e-07, + "loss": 1.0771, + "mean_token_accuracy": 0.6942634582519531, + "num_tokens": 30519052.0, + "step": 1182 + }, + { + "epoch": 0.12991434219196135, + "grad_norm": 2.338141918182373, + "learning_rate": 4.3265007320644215e-07, + "loss": 0.9852, + "mean_token_accuracy": 0.7044277787208557, + "num_tokens": 30543441.0, + "step": 1183 + }, + { + "epoch": 0.130024159894575, + "grad_norm": 2.4641261100769043, + "learning_rate": 4.330161054172767e-07, + "loss": 1.133, + "mean_token_accuracy": 0.6700233221054077, + "num_tokens": 30567401.0, + "step": 1184 + }, + { + "epoch": 0.13013397759718867, + "grad_norm": 2.345489501953125, + "learning_rate": 4.3338213762811125e-07, + "loss": 1.1315, + "mean_token_accuracy": 0.6676216125488281, + "num_tokens": 30594937.0, + "step": 1185 + }, + { + "epoch": 0.13024379529980232, + "grad_norm": 2.1916072368621826, + "learning_rate": 4.337481698389458e-07, + "loss": 0.9842, + "mean_token_accuracy": 0.7078807353973389, + "num_tokens": 30623647.0, + "step": 1186 + }, + { + "epoch": 0.130353613002416, + "grad_norm": 2.302164077758789, + "learning_rate": 4.341142020497804e-07, + "loss": 1.093, + "mean_token_accuracy": 0.683877170085907, + "num_tokens": 30650386.0, + "step": 1187 + }, + { + "epoch": 0.13046343070502964, + "grad_norm": 2.586156129837036, + "learning_rate": 4.344802342606149e-07, + "loss": 1.0675, + "mean_token_accuracy": 0.6813502907752991, + "num_tokens": 30673298.0, + "step": 1188 + }, + { + "epoch": 0.1305732484076433, + "grad_norm": 2.182748317718506, + "learning_rate": 4.3484626647144945e-07, + "loss": 1.0917, + "mean_token_accuracy": 0.6849731206893921, + "num_tokens": 30702162.0, + "step": 1189 + }, + { + "epoch": 0.13068306611025698, + "grad_norm": 2.37206768989563, + "learning_rate": 4.3521229868228405e-07, + "loss": 1.1211, + "mean_token_accuracy": 0.6758624315261841, + "num_tokens": 30728125.0, + "step": 1190 + }, + { + "epoch": 0.13079288381287063, + "grad_norm": 2.368927478790283, + "learning_rate": 4.355783308931186e-07, + "loss": 1.0019, + "mean_token_accuracy": 0.7012432813644409, + "num_tokens": 30753050.0, + "step": 1191 + }, + { + "epoch": 0.1309027015154843, + "grad_norm": 2.397585153579712, + "learning_rate": 4.359443631039531e-07, + "loss": 1.0523, + "mean_token_accuracy": 0.6876215934753418, + "num_tokens": 30778703.0, + "step": 1192 + }, + { + "epoch": 0.13101251921809795, + "grad_norm": 2.0767369270324707, + "learning_rate": 4.363103953147877e-07, + "loss": 1.0472, + "mean_token_accuracy": 0.6873078346252441, + "num_tokens": 30810814.0, + "step": 1193 + }, + { + "epoch": 0.13112233692071162, + "grad_norm": 2.2827279567718506, + "learning_rate": 4.3667642752562224e-07, + "loss": 1.0576, + "mean_token_accuracy": 0.6990712881088257, + "num_tokens": 30836427.0, + "step": 1194 + }, + { + "epoch": 0.13123215462332527, + "grad_norm": 2.1867270469665527, + "learning_rate": 4.370424597364568e-07, + "loss": 1.123, + "mean_token_accuracy": 0.6679629683494568, + "num_tokens": 30865532.0, + "step": 1195 + }, + { + "epoch": 0.13134197232593894, + "grad_norm": 2.6115574836730957, + "learning_rate": 4.3740849194729134e-07, + "loss": 1.0462, + "mean_token_accuracy": 0.6863243579864502, + "num_tokens": 30885416.0, + "step": 1196 + }, + { + "epoch": 0.1314517900285526, + "grad_norm": 2.4007792472839355, + "learning_rate": 4.377745241581259e-07, + "loss": 1.0197, + "mean_token_accuracy": 0.6917524337768555, + "num_tokens": 30909752.0, + "step": 1197 + }, + { + "epoch": 0.13156160773116626, + "grad_norm": 2.7789487838745117, + "learning_rate": 4.381405563689605e-07, + "loss": 1.0022, + "mean_token_accuracy": 0.7014362215995789, + "num_tokens": 30928463.0, + "step": 1198 + }, + { + "epoch": 0.13167142543377994, + "grad_norm": 2.6030101776123047, + "learning_rate": 4.38506588579795e-07, + "loss": 1.0433, + "mean_token_accuracy": 0.6901962757110596, + "num_tokens": 30950612.0, + "step": 1199 + }, + { + "epoch": 0.13178124313639358, + "grad_norm": 2.2986321449279785, + "learning_rate": 4.3887262079062954e-07, + "loss": 1.0407, + "mean_token_accuracy": 0.698881983757019, + "num_tokens": 30975408.0, + "step": 1200 + }, + { + "epoch": 0.13189106083900726, + "grad_norm": 2.4589626789093018, + "learning_rate": 4.3923865300146414e-07, + "loss": 0.9495, + "mean_token_accuracy": 0.712681770324707, + "num_tokens": 30998382.0, + "step": 1201 + }, + { + "epoch": 0.1320008785416209, + "grad_norm": 2.20348858833313, + "learning_rate": 4.3960468521229864e-07, + "loss": 1.003, + "mean_token_accuracy": 0.702782928943634, + "num_tokens": 31024645.0, + "step": 1202 + }, + { + "epoch": 0.13211069624423458, + "grad_norm": 2.4291844367980957, + "learning_rate": 4.3997071742313324e-07, + "loss": 1.0605, + "mean_token_accuracy": 0.6936054825782776, + "num_tokens": 31050392.0, + "step": 1203 + }, + { + "epoch": 0.13222051394684822, + "grad_norm": 2.4460294246673584, + "learning_rate": 4.403367496339678e-07, + "loss": 0.995, + "mean_token_accuracy": 0.702476978302002, + "num_tokens": 31072099.0, + "step": 1204 + }, + { + "epoch": 0.1323303316494619, + "grad_norm": 2.2189552783966064, + "learning_rate": 4.407027818448023e-07, + "loss": 1.0305, + "mean_token_accuracy": 0.6898212432861328, + "num_tokens": 31099348.0, + "step": 1205 + }, + { + "epoch": 0.13244014935207554, + "grad_norm": 2.2610080242156982, + "learning_rate": 4.410688140556369e-07, + "loss": 1.03, + "mean_token_accuracy": 0.6909153461456299, + "num_tokens": 31127674.0, + "step": 1206 + }, + { + "epoch": 0.13254996705468922, + "grad_norm": 2.6599183082580566, + "learning_rate": 4.4143484626647144e-07, + "loss": 1.0226, + "mean_token_accuracy": 0.702622652053833, + "num_tokens": 31147826.0, + "step": 1207 + }, + { + "epoch": 0.1326597847573029, + "grad_norm": 2.669041872024536, + "learning_rate": 4.41800878477306e-07, + "loss": 0.9702, + "mean_token_accuracy": 0.7058910131454468, + "num_tokens": 31168192.0, + "step": 1208 + }, + { + "epoch": 0.13276960245991654, + "grad_norm": 2.763950824737549, + "learning_rate": 4.4216691068814053e-07, + "loss": 0.864, + "mean_token_accuracy": 0.7248986959457397, + "num_tokens": 31186404.0, + "step": 1209 + }, + { + "epoch": 0.1328794201625302, + "grad_norm": 2.16217303276062, + "learning_rate": 4.425329428989751e-07, + "loss": 1.118, + "mean_token_accuracy": 0.6657146215438843, + "num_tokens": 31216377.0, + "step": 1210 + }, + { + "epoch": 0.13298923786514386, + "grad_norm": 2.465715169906616, + "learning_rate": 4.428989751098097e-07, + "loss": 1.0706, + "mean_token_accuracy": 0.6827074885368347, + "num_tokens": 31241083.0, + "step": 1211 + }, + { + "epoch": 0.13309905556775753, + "grad_norm": 2.3043594360351562, + "learning_rate": 4.432650073206442e-07, + "loss": 1.1045, + "mean_token_accuracy": 0.6767438650131226, + "num_tokens": 31269362.0, + "step": 1212 + }, + { + "epoch": 0.13320887327037118, + "grad_norm": 1.9868229627609253, + "learning_rate": 4.4363103953147873e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.6968153715133667, + "num_tokens": 31301121.0, + "step": 1213 + }, + { + "epoch": 0.13331869097298485, + "grad_norm": 2.6146652698516846, + "learning_rate": 4.4399707174231333e-07, + "loss": 1.1048, + "mean_token_accuracy": 0.6678427457809448, + "num_tokens": 31323850.0, + "step": 1214 + }, + { + "epoch": 0.1334285086755985, + "grad_norm": 2.3330156803131104, + "learning_rate": 4.443631039531479e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6848459839820862, + "num_tokens": 31349231.0, + "step": 1215 + }, + { + "epoch": 0.13353832637821217, + "grad_norm": 2.7242321968078613, + "learning_rate": 4.447291361639824e-07, + "loss": 1.1015, + "mean_token_accuracy": 0.6783826351165771, + "num_tokens": 31368491.0, + "step": 1216 + }, + { + "epoch": 0.13364814408082584, + "grad_norm": 2.3878121376037598, + "learning_rate": 4.45095168374817e-07, + "loss": 0.968, + "mean_token_accuracy": 0.7041847705841064, + "num_tokens": 31392465.0, + "step": 1217 + }, + { + "epoch": 0.1337579617834395, + "grad_norm": 2.456578493118286, + "learning_rate": 4.4546120058565153e-07, + "loss": 1.0731, + "mean_token_accuracy": 0.6773847937583923, + "num_tokens": 31417587.0, + "step": 1218 + }, + { + "epoch": 0.13386777948605316, + "grad_norm": 2.106383800506592, + "learning_rate": 4.458272327964861e-07, + "loss": 1.0265, + "mean_token_accuracy": 0.6955444812774658, + "num_tokens": 31448313.0, + "step": 1219 + }, + { + "epoch": 0.1339775971886668, + "grad_norm": 2.491579532623291, + "learning_rate": 4.4619326500732063e-07, + "loss": 1.0301, + "mean_token_accuracy": 0.6896505355834961, + "num_tokens": 31472179.0, + "step": 1220 + }, + { + "epoch": 0.13408741489128048, + "grad_norm": 2.1729423999786377, + "learning_rate": 4.465592972181552e-07, + "loss": 1.09, + "mean_token_accuracy": 0.685196042060852, + "num_tokens": 31501837.0, + "step": 1221 + }, + { + "epoch": 0.13419723259389413, + "grad_norm": 2.47400164604187, + "learning_rate": 4.469253294289897e-07, + "loss": 1.011, + "mean_token_accuracy": 0.698586106300354, + "num_tokens": 31523276.0, + "step": 1222 + }, + { + "epoch": 0.1343070502965078, + "grad_norm": 2.501985549926758, + "learning_rate": 4.472913616398243e-07, + "loss": 1.1594, + "mean_token_accuracy": 0.663594126701355, + "num_tokens": 31547514.0, + "step": 1223 + }, + { + "epoch": 0.13441686799912145, + "grad_norm": 2.1671621799468994, + "learning_rate": 4.476573938506588e-07, + "loss": 0.9929, + "mean_token_accuracy": 0.7006608247756958, + "num_tokens": 31575007.0, + "step": 1224 + }, + { + "epoch": 0.13452668570173512, + "grad_norm": 2.380038022994995, + "learning_rate": 4.480234260614934e-07, + "loss": 1.037, + "mean_token_accuracy": 0.6890693306922913, + "num_tokens": 31600483.0, + "step": 1225 + }, + { + "epoch": 0.13463650340434877, + "grad_norm": 2.497714042663574, + "learning_rate": 4.483894582723279e-07, + "loss": 1.0251, + "mean_token_accuracy": 0.6991919875144958, + "num_tokens": 31622639.0, + "step": 1226 + }, + { + "epoch": 0.13474632110696244, + "grad_norm": 2.4689528942108154, + "learning_rate": 4.487554904831625e-07, + "loss": 1.046, + "mean_token_accuracy": 0.7004215121269226, + "num_tokens": 31647351.0, + "step": 1227 + }, + { + "epoch": 0.13485613880957611, + "grad_norm": 2.7528581619262695, + "learning_rate": 4.4912152269399707e-07, + "loss": 1.1051, + "mean_token_accuracy": 0.6756384372711182, + "num_tokens": 31667220.0, + "step": 1228 + }, + { + "epoch": 0.13496595651218976, + "grad_norm": 2.54508113861084, + "learning_rate": 4.4948755490483157e-07, + "loss": 1.1164, + "mean_token_accuracy": 0.6793631315231323, + "num_tokens": 31689044.0, + "step": 1229 + }, + { + "epoch": 0.13507577421480343, + "grad_norm": 2.37357759475708, + "learning_rate": 4.4985358711566617e-07, + "loss": 1.2474, + "mean_token_accuracy": 0.6459565758705139, + "num_tokens": 31717794.0, + "step": 1230 + }, + { + "epoch": 0.13518559191741708, + "grad_norm": 2.3444550037384033, + "learning_rate": 4.502196193265007e-07, + "loss": 1.1121, + "mean_token_accuracy": 0.6728112697601318, + "num_tokens": 31743349.0, + "step": 1231 + }, + { + "epoch": 0.13529540962003075, + "grad_norm": 2.818902015686035, + "learning_rate": 4.5058565153733527e-07, + "loss": 0.9946, + "mean_token_accuracy": 0.6965161561965942, + "num_tokens": 31761698.0, + "step": 1232 + }, + { + "epoch": 0.1354052273226444, + "grad_norm": 2.881321668624878, + "learning_rate": 4.509516837481698e-07, + "loss": 0.9896, + "mean_token_accuracy": 0.7067734003067017, + "num_tokens": 31780060.0, + "step": 1233 + }, + { + "epoch": 0.13551504502525807, + "grad_norm": 2.11897873878479, + "learning_rate": 4.5131771595900437e-07, + "loss": 1.0049, + "mean_token_accuracy": 0.7013705968856812, + "num_tokens": 31810081.0, + "step": 1234 + }, + { + "epoch": 0.13562486272787172, + "grad_norm": 2.2233707904815674, + "learning_rate": 4.5168374816983897e-07, + "loss": 1.0723, + "mean_token_accuracy": 0.6817072629928589, + "num_tokens": 31838983.0, + "step": 1235 + }, + { + "epoch": 0.1357346804304854, + "grad_norm": 2.2973594665527344, + "learning_rate": 4.5204978038067347e-07, + "loss": 1.0802, + "mean_token_accuracy": 0.6870145797729492, + "num_tokens": 31864058.0, + "step": 1236 + }, + { + "epoch": 0.13584449813309907, + "grad_norm": 2.3988091945648193, + "learning_rate": 4.52415812591508e-07, + "loss": 1.0584, + "mean_token_accuracy": 0.6845042109489441, + "num_tokens": 31889614.0, + "step": 1237 + }, + { + "epoch": 0.1359543158357127, + "grad_norm": 2.3572990894317627, + "learning_rate": 4.527818448023426e-07, + "loss": 1.1617, + "mean_token_accuracy": 0.6715127229690552, + "num_tokens": 31915897.0, + "step": 1238 + }, + { + "epoch": 0.1360641335383264, + "grad_norm": 2.1999454498291016, + "learning_rate": 4.531478770131771e-07, + "loss": 1.0151, + "mean_token_accuracy": 0.6922159790992737, + "num_tokens": 31943588.0, + "step": 1239 + }, + { + "epoch": 0.13617395124094003, + "grad_norm": 2.300368309020996, + "learning_rate": 4.5351390922401166e-07, + "loss": 1.077, + "mean_token_accuracy": 0.6833131313323975, + "num_tokens": 31970532.0, + "step": 1240 + }, + { + "epoch": 0.1362837689435537, + "grad_norm": 2.3178136348724365, + "learning_rate": 4.5387994143484626e-07, + "loss": 1.0423, + "mean_token_accuracy": 0.6962330341339111, + "num_tokens": 31996697.0, + "step": 1241 + }, + { + "epoch": 0.13639358664616735, + "grad_norm": 2.6201584339141846, + "learning_rate": 4.542459736456808e-07, + "loss": 0.9952, + "mean_token_accuracy": 0.7055785655975342, + "num_tokens": 32018303.0, + "step": 1242 + }, + { + "epoch": 0.13650340434878103, + "grad_norm": 2.1510610580444336, + "learning_rate": 4.5461200585651536e-07, + "loss": 1.1181, + "mean_token_accuracy": 0.6669231653213501, + "num_tokens": 32046571.0, + "step": 1243 + }, + { + "epoch": 0.13661322205139467, + "grad_norm": 2.44258713722229, + "learning_rate": 4.549780380673499e-07, + "loss": 1.1415, + "mean_token_accuracy": 0.6703335046768188, + "num_tokens": 32072281.0, + "step": 1244 + }, + { + "epoch": 0.13672303975400835, + "grad_norm": 2.9422483444213867, + "learning_rate": 4.5534407027818446e-07, + "loss": 1.1779, + "mean_token_accuracy": 0.6611398458480835, + "num_tokens": 32099451.0, + "step": 1245 + }, + { + "epoch": 0.13683285745662202, + "grad_norm": 2.35931658744812, + "learning_rate": 4.55710102489019e-07, + "loss": 1.0433, + "mean_token_accuracy": 0.6928471922874451, + "num_tokens": 32124080.0, + "step": 1246 + }, + { + "epoch": 0.13694267515923567, + "grad_norm": 2.230501890182495, + "learning_rate": 4.5607613469985356e-07, + "loss": 1.1452, + "mean_token_accuracy": 0.6641011238098145, + "num_tokens": 32153434.0, + "step": 1247 + }, + { + "epoch": 0.13705249286184934, + "grad_norm": 2.5080502033233643, + "learning_rate": 4.564421669106881e-07, + "loss": 1.0209, + "mean_token_accuracy": 0.6963638067245483, + "num_tokens": 32176510.0, + "step": 1248 + }, + { + "epoch": 0.13716231056446299, + "grad_norm": 2.0775582790374756, + "learning_rate": 4.568081991215227e-07, + "loss": 1.0739, + "mean_token_accuracy": 0.678318977355957, + "num_tokens": 32206030.0, + "step": 1249 + }, + { + "epoch": 0.13727212826707666, + "grad_norm": 2.2532033920288086, + "learning_rate": 4.571742313323572e-07, + "loss": 1.0509, + "mean_token_accuracy": 0.6916120648384094, + "num_tokens": 32233255.0, + "step": 1250 + }, + { + "epoch": 0.1373819459696903, + "grad_norm": 2.341150999069214, + "learning_rate": 4.575402635431918e-07, + "loss": 1.0406, + "mean_token_accuracy": 0.6932820081710815, + "num_tokens": 32257818.0, + "step": 1251 + }, + { + "epoch": 0.13749176367230398, + "grad_norm": 2.2033114433288574, + "learning_rate": 4.5790629575402636e-07, + "loss": 1.0454, + "mean_token_accuracy": 0.6891587972640991, + "num_tokens": 32286826.0, + "step": 1252 + }, + { + "epoch": 0.13760158137491763, + "grad_norm": 2.459940195083618, + "learning_rate": 4.5827232796486085e-07, + "loss": 1.0527, + "mean_token_accuracy": 0.6863068342208862, + "num_tokens": 32310447.0, + "step": 1253 + }, + { + "epoch": 0.1377113990775313, + "grad_norm": 2.372678756713867, + "learning_rate": 4.5863836017569545e-07, + "loss": 1.0959, + "mean_token_accuracy": 0.6911530494689941, + "num_tokens": 32335713.0, + "step": 1254 + }, + { + "epoch": 0.13782121678014497, + "grad_norm": 2.2568814754486084, + "learning_rate": 4.5900439238653e-07, + "loss": 1.0631, + "mean_token_accuracy": 0.6891027092933655, + "num_tokens": 32361259.0, + "step": 1255 + }, + { + "epoch": 0.13793103448275862, + "grad_norm": 2.610930919647217, + "learning_rate": 4.593704245973645e-07, + "loss": 1.0484, + "mean_token_accuracy": 0.6879401803016663, + "num_tokens": 32383138.0, + "step": 1256 + }, + { + "epoch": 0.1380408521853723, + "grad_norm": 2.259490489959717, + "learning_rate": 4.597364568081991e-07, + "loss": 1.0488, + "mean_token_accuracy": 0.6859354972839355, + "num_tokens": 32409879.0, + "step": 1257 + }, + { + "epoch": 0.13815066988798594, + "grad_norm": 2.6916215419769287, + "learning_rate": 4.6010248901903365e-07, + "loss": 1.0289, + "mean_token_accuracy": 0.6905195713043213, + "num_tokens": 32430039.0, + "step": 1258 + }, + { + "epoch": 0.1382604875905996, + "grad_norm": 2.321918249130249, + "learning_rate": 4.6046852122986825e-07, + "loss": 1.0326, + "mean_token_accuracy": 0.6910154223442078, + "num_tokens": 32456043.0, + "step": 1259 + }, + { + "epoch": 0.13837030529321326, + "grad_norm": 2.0398852825164795, + "learning_rate": 4.6083455344070275e-07, + "loss": 1.1406, + "mean_token_accuracy": 0.6650717854499817, + "num_tokens": 32489703.0, + "step": 1260 + }, + { + "epoch": 0.13848012299582693, + "grad_norm": 2.003500461578369, + "learning_rate": 4.612005856515373e-07, + "loss": 1.0943, + "mean_token_accuracy": 0.6753607392311096, + "num_tokens": 32523481.0, + "step": 1261 + }, + { + "epoch": 0.13858994069844058, + "grad_norm": 2.668422222137451, + "learning_rate": 4.615666178623719e-07, + "loss": 1.0311, + "mean_token_accuracy": 0.6948420405387878, + "num_tokens": 32544230.0, + "step": 1262 + }, + { + "epoch": 0.13869975840105425, + "grad_norm": 2.672374963760376, + "learning_rate": 4.619326500732064e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.6971681118011475, + "num_tokens": 32563592.0, + "step": 1263 + }, + { + "epoch": 0.1388095761036679, + "grad_norm": 2.304234266281128, + "learning_rate": 4.6229868228404095e-07, + "loss": 1.1587, + "mean_token_accuracy": 0.6727135181427002, + "num_tokens": 32592096.0, + "step": 1264 + }, + { + "epoch": 0.13891939380628157, + "grad_norm": 2.4287734031677246, + "learning_rate": 4.6266471449487555e-07, + "loss": 1.0273, + "mean_token_accuracy": 0.6912821531295776, + "num_tokens": 32615477.0, + "step": 1265 + }, + { + "epoch": 0.13902921150889525, + "grad_norm": 2.5482099056243896, + "learning_rate": 4.630307467057101e-07, + "loss": 1.0233, + "mean_token_accuracy": 0.6925524473190308, + "num_tokens": 32637654.0, + "step": 1266 + }, + { + "epoch": 0.1391390292115089, + "grad_norm": 2.310539484024048, + "learning_rate": 4.6339677891654465e-07, + "loss": 0.9994, + "mean_token_accuracy": 0.6993199586868286, + "num_tokens": 32661199.0, + "step": 1267 + }, + { + "epoch": 0.13924884691412256, + "grad_norm": 2.5369014739990234, + "learning_rate": 4.637628111273792e-07, + "loss": 1.0945, + "mean_token_accuracy": 0.6752843856811523, + "num_tokens": 32684302.0, + "step": 1268 + }, + { + "epoch": 0.1393586646167362, + "grad_norm": 2.161068916320801, + "learning_rate": 4.6412884333821374e-07, + "loss": 1.13, + "mean_token_accuracy": 0.6669653654098511, + "num_tokens": 32711546.0, + "step": 1269 + }, + { + "epoch": 0.13946848231934988, + "grad_norm": 2.148710250854492, + "learning_rate": 4.644948755490483e-07, + "loss": 1.1443, + "mean_token_accuracy": 0.6638570427894592, + "num_tokens": 32741165.0, + "step": 1270 + }, + { + "epoch": 0.13957830002196353, + "grad_norm": 2.0794811248779297, + "learning_rate": 4.6486090775988284e-07, + "loss": 1.0579, + "mean_token_accuracy": 0.6870812177658081, + "num_tokens": 32774701.0, + "step": 1271 + }, + { + "epoch": 0.1396881177245772, + "grad_norm": 2.4168546199798584, + "learning_rate": 4.652269399707174e-07, + "loss": 1.1436, + "mean_token_accuracy": 0.6658343076705933, + "num_tokens": 32798129.0, + "step": 1272 + }, + { + "epoch": 0.13979793542719085, + "grad_norm": 2.0742275714874268, + "learning_rate": 4.65592972181552e-07, + "loss": 1.0609, + "mean_token_accuracy": 0.689683198928833, + "num_tokens": 32829323.0, + "step": 1273 + }, + { + "epoch": 0.13990775312980452, + "grad_norm": 2.2599472999572754, + "learning_rate": 4.659590043923865e-07, + "loss": 0.9853, + "mean_token_accuracy": 0.705630898475647, + "num_tokens": 32856609.0, + "step": 1274 + }, + { + "epoch": 0.1400175708324182, + "grad_norm": 2.4082419872283936, + "learning_rate": 4.663250366032211e-07, + "loss": 1.0386, + "mean_token_accuracy": 0.686806321144104, + "num_tokens": 32881060.0, + "step": 1275 + }, + { + "epoch": 0.14012738853503184, + "grad_norm": 2.3449547290802, + "learning_rate": 4.6669106881405564e-07, + "loss": 1.0043, + "mean_token_accuracy": 0.711695671081543, + "num_tokens": 32904678.0, + "step": 1276 + }, + { + "epoch": 0.14023720623764552, + "grad_norm": 2.473179340362549, + "learning_rate": 4.6705710102489014e-07, + "loss": 1.0142, + "mean_token_accuracy": 0.6905529499053955, + "num_tokens": 32928285.0, + "step": 1277 + }, + { + "epoch": 0.14034702394025916, + "grad_norm": 2.4101366996765137, + "learning_rate": 4.6742313323572474e-07, + "loss": 1.0457, + "mean_token_accuracy": 0.6909178495407104, + "num_tokens": 32953717.0, + "step": 1278 + }, + { + "epoch": 0.14045684164287284, + "grad_norm": 2.007829427719116, + "learning_rate": 4.677891654465593e-07, + "loss": 1.0231, + "mean_token_accuracy": 0.6956433057785034, + "num_tokens": 32985152.0, + "step": 1279 + }, + { + "epoch": 0.14056665934548648, + "grad_norm": 2.438462495803833, + "learning_rate": 4.681551976573938e-07, + "loss": 1.0383, + "mean_token_accuracy": 0.6908630132675171, + "num_tokens": 33009420.0, + "step": 1280 + }, + { + "epoch": 0.14067647704810016, + "grad_norm": 2.314420223236084, + "learning_rate": 4.685212298682284e-07, + "loss": 1.1668, + "mean_token_accuracy": 0.6547598242759705, + "num_tokens": 33042302.0, + "step": 1281 + }, + { + "epoch": 0.1407862947507138, + "grad_norm": 2.073517084121704, + "learning_rate": 4.6888726207906293e-07, + "loss": 1.0595, + "mean_token_accuracy": 0.6832067966461182, + "num_tokens": 33072431.0, + "step": 1282 + }, + { + "epoch": 0.14089611245332748, + "grad_norm": 2.2796177864074707, + "learning_rate": 4.6925329428989754e-07, + "loss": 1.104, + "mean_token_accuracy": 0.6695857048034668, + "num_tokens": 33098031.0, + "step": 1283 + }, + { + "epoch": 0.14100593015594115, + "grad_norm": 2.3960278034210205, + "learning_rate": 4.6961932650073203e-07, + "loss": 1.0883, + "mean_token_accuracy": 0.684101939201355, + "num_tokens": 33121574.0, + "step": 1284 + }, + { + "epoch": 0.1411157478585548, + "grad_norm": 2.550199031829834, + "learning_rate": 4.699853587115666e-07, + "loss": 1.101, + "mean_token_accuracy": 0.6823867559432983, + "num_tokens": 33145414.0, + "step": 1285 + }, + { + "epoch": 0.14122556556116847, + "grad_norm": 2.521268606185913, + "learning_rate": 4.703513909224012e-07, + "loss": 0.999, + "mean_token_accuracy": 0.6995866894721985, + "num_tokens": 33166555.0, + "step": 1286 + }, + { + "epoch": 0.14133538326378212, + "grad_norm": 2.002866744995117, + "learning_rate": 4.707174231332357e-07, + "loss": 1.1242, + "mean_token_accuracy": 0.6701062321662903, + "num_tokens": 33200074.0, + "step": 1287 + }, + { + "epoch": 0.1414452009663958, + "grad_norm": 2.6430797576904297, + "learning_rate": 4.7108345534407023e-07, + "loss": 0.8995, + "mean_token_accuracy": 0.7293510437011719, + "num_tokens": 33220090.0, + "step": 1288 + }, + { + "epoch": 0.14155501866900944, + "grad_norm": 2.217536687850952, + "learning_rate": 4.7144948755490483e-07, + "loss": 1.0174, + "mean_token_accuracy": 0.6898154020309448, + "num_tokens": 33247080.0, + "step": 1289 + }, + { + "epoch": 0.1416648363716231, + "grad_norm": 2.0777506828308105, + "learning_rate": 4.718155197657394e-07, + "loss": 1.1153, + "mean_token_accuracy": 0.6778391599655151, + "num_tokens": 33276588.0, + "step": 1290 + }, + { + "epoch": 0.14177465407423676, + "grad_norm": 2.6367013454437256, + "learning_rate": 4.7218155197657393e-07, + "loss": 0.9709, + "mean_token_accuracy": 0.7068687677383423, + "num_tokens": 33296291.0, + "step": 1291 + }, + { + "epoch": 0.14188447177685043, + "grad_norm": 2.18192195892334, + "learning_rate": 4.725475841874085e-07, + "loss": 1.1214, + "mean_token_accuracy": 0.6680413484573364, + "num_tokens": 33327544.0, + "step": 1292 + }, + { + "epoch": 0.1419942894794641, + "grad_norm": 2.008930206298828, + "learning_rate": 4.7291361639824303e-07, + "loss": 1.0464, + "mean_token_accuracy": 0.6921355724334717, + "num_tokens": 33360733.0, + "step": 1293 + }, + { + "epoch": 0.14210410718207775, + "grad_norm": 2.2605888843536377, + "learning_rate": 4.732796486090776e-07, + "loss": 1.0169, + "mean_token_accuracy": 0.6982976198196411, + "num_tokens": 33386974.0, + "step": 1294 + }, + { + "epoch": 0.14221392488469142, + "grad_norm": 2.4529366493225098, + "learning_rate": 4.736456808199121e-07, + "loss": 1.045, + "mean_token_accuracy": 0.6995375752449036, + "num_tokens": 33410360.0, + "step": 1295 + }, + { + "epoch": 0.14232374258730507, + "grad_norm": 2.3046205043792725, + "learning_rate": 4.740117130307467e-07, + "loss": 1.0447, + "mean_token_accuracy": 0.6885802745819092, + "num_tokens": 33435322.0, + "step": 1296 + }, + { + "epoch": 0.14243356028991874, + "grad_norm": 2.269519805908203, + "learning_rate": 4.743777452415813e-07, + "loss": 1.2058, + "mean_token_accuracy": 0.6538702249526978, + "num_tokens": 33464229.0, + "step": 1297 + }, + { + "epoch": 0.1425433779925324, + "grad_norm": 2.5316686630249023, + "learning_rate": 4.7474377745241577e-07, + "loss": 0.9523, + "mean_token_accuracy": 0.7188043594360352, + "num_tokens": 33485248.0, + "step": 1298 + }, + { + "epoch": 0.14265319569514606, + "grad_norm": 2.2501931190490723, + "learning_rate": 4.751098096632504e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6867610216140747, + "num_tokens": 33516065.0, + "step": 1299 + }, + { + "epoch": 0.1427630133977597, + "grad_norm": 1.9970179796218872, + "learning_rate": 4.754758418740849e-07, + "loss": 0.9967, + "mean_token_accuracy": 0.7054612636566162, + "num_tokens": 33548790.0, + "step": 1300 + }, + { + "epoch": 0.14287283110037338, + "grad_norm": 2.174708127975464, + "learning_rate": 4.758418740849194e-07, + "loss": 1.0663, + "mean_token_accuracy": 0.6875282526016235, + "num_tokens": 33577167.0, + "step": 1301 + }, + { + "epoch": 0.14298264880298703, + "grad_norm": 2.4325053691864014, + "learning_rate": 4.76207906295754e-07, + "loss": 1.0376, + "mean_token_accuracy": 0.6971954703330994, + "num_tokens": 33599933.0, + "step": 1302 + }, + { + "epoch": 0.1430924665056007, + "grad_norm": 2.580379009246826, + "learning_rate": 4.7657393850658857e-07, + "loss": 0.9514, + "mean_token_accuracy": 0.7094119787216187, + "num_tokens": 33618648.0, + "step": 1303 + }, + { + "epoch": 0.14320228420821438, + "grad_norm": 2.2738876342773438, + "learning_rate": 4.769399707174231e-07, + "loss": 1.0231, + "mean_token_accuracy": 0.693604588508606, + "num_tokens": 33643320.0, + "step": 1304 + }, + { + "epoch": 0.14331210191082802, + "grad_norm": 2.384453058242798, + "learning_rate": 4.773060029282577e-07, + "loss": 1.1088, + "mean_token_accuracy": 0.6692377924919128, + "num_tokens": 33667215.0, + "step": 1305 + }, + { + "epoch": 0.1434219196134417, + "grad_norm": 2.441540002822876, + "learning_rate": 4.776720351390922e-07, + "loss": 1.1184, + "mean_token_accuracy": 0.6679764986038208, + "num_tokens": 33691314.0, + "step": 1306 + }, + { + "epoch": 0.14353173731605534, + "grad_norm": 2.429004192352295, + "learning_rate": 4.780380673499268e-07, + "loss": 1.0898, + "mean_token_accuracy": 0.6824609041213989, + "num_tokens": 33716231.0, + "step": 1307 + }, + { + "epoch": 0.14364155501866901, + "grad_norm": 1.9500312805175781, + "learning_rate": 4.784040995607613e-07, + "loss": 1.0659, + "mean_token_accuracy": 0.6824396252632141, + "num_tokens": 33750223.0, + "step": 1308 + }, + { + "epoch": 0.14375137272128266, + "grad_norm": 2.478240966796875, + "learning_rate": 4.787701317715958e-07, + "loss": 0.9963, + "mean_token_accuracy": 0.7058166861534119, + "num_tokens": 33772320.0, + "step": 1309 + }, + { + "epoch": 0.14386119042389633, + "grad_norm": 2.5757625102996826, + "learning_rate": 4.791361639824304e-07, + "loss": 1.0449, + "mean_token_accuracy": 0.6905869245529175, + "num_tokens": 33793342.0, + "step": 1310 + }, + { + "epoch": 0.14397100812650998, + "grad_norm": 2.6064982414245605, + "learning_rate": 4.79502196193265e-07, + "loss": 0.9614, + "mean_token_accuracy": 0.7059409022331238, + "num_tokens": 33813186.0, + "step": 1311 + }, + { + "epoch": 0.14408082582912365, + "grad_norm": 2.343437910079956, + "learning_rate": 4.798682284040995e-07, + "loss": 1.01, + "mean_token_accuracy": 0.6915148496627808, + "num_tokens": 33840214.0, + "step": 1312 + }, + { + "epoch": 0.14419064353173733, + "grad_norm": 2.589820623397827, + "learning_rate": 4.802342606149341e-07, + "loss": 1.0286, + "mean_token_accuracy": 0.6895705461502075, + "num_tokens": 33863647.0, + "step": 1313 + }, + { + "epoch": 0.14430046123435097, + "grad_norm": 2.0441927909851074, + "learning_rate": 4.806002928257686e-07, + "loss": 1.1011, + "mean_token_accuracy": 0.6788641214370728, + "num_tokens": 33896827.0, + "step": 1314 + }, + { + "epoch": 0.14441027893696465, + "grad_norm": 2.423750400543213, + "learning_rate": 4.809663250366032e-07, + "loss": 1.0616, + "mean_token_accuracy": 0.68110191822052, + "num_tokens": 33920697.0, + "step": 1315 + }, + { + "epoch": 0.1445200966395783, + "grad_norm": 2.2622878551483154, + "learning_rate": 4.813323572474377e-07, + "loss": 1.0182, + "mean_token_accuracy": 0.6986842155456543, + "num_tokens": 33945998.0, + "step": 1316 + }, + { + "epoch": 0.14462991434219197, + "grad_norm": 2.451852798461914, + "learning_rate": 4.816983894582723e-07, + "loss": 1.0379, + "mean_token_accuracy": 0.6966204643249512, + "num_tokens": 33969471.0, + "step": 1317 + }, + { + "epoch": 0.1447397320448056, + "grad_norm": 2.295973777770996, + "learning_rate": 4.820644216691069e-07, + "loss": 1.0712, + "mean_token_accuracy": 0.678676426410675, + "num_tokens": 33995373.0, + "step": 1318 + }, + { + "epoch": 0.1448495497474193, + "grad_norm": 2.583833694458008, + "learning_rate": 4.824304538799414e-07, + "loss": 0.9728, + "mean_token_accuracy": 0.7082421183586121, + "num_tokens": 34019470.0, + "step": 1319 + }, + { + "epoch": 0.14495936745003293, + "grad_norm": 2.3040919303894043, + "learning_rate": 4.827964860907759e-07, + "loss": 1.0927, + "mean_token_accuracy": 0.6784384846687317, + "num_tokens": 34046792.0, + "step": 1320 + }, + { + "epoch": 0.1450691851526466, + "grad_norm": 2.280946731567383, + "learning_rate": 4.831625183016105e-07, + "loss": 1.0709, + "mean_token_accuracy": 0.6794667840003967, + "num_tokens": 34073908.0, + "step": 1321 + }, + { + "epoch": 0.14517900285526028, + "grad_norm": 2.16202712059021, + "learning_rate": 4.835285505124451e-07, + "loss": 0.9322, + "mean_token_accuracy": 0.7153509855270386, + "num_tokens": 34101665.0, + "step": 1322 + }, + { + "epoch": 0.14528882055787393, + "grad_norm": 2.5492653846740723, + "learning_rate": 4.838945827232796e-07, + "loss": 0.9585, + "mean_token_accuracy": 0.710796594619751, + "num_tokens": 34123530.0, + "step": 1323 + }, + { + "epoch": 0.1453986382604876, + "grad_norm": 2.3433048725128174, + "learning_rate": 4.842606149341142e-07, + "loss": 1.0505, + "mean_token_accuracy": 0.6774759888648987, + "num_tokens": 34148038.0, + "step": 1324 + }, + { + "epoch": 0.14550845596310125, + "grad_norm": 2.389906644821167, + "learning_rate": 4.846266471449487e-07, + "loss": 1.0551, + "mean_token_accuracy": 0.6824579238891602, + "num_tokens": 34171039.0, + "step": 1325 + }, + { + "epoch": 0.14561827366571492, + "grad_norm": 2.2938427925109863, + "learning_rate": 4.849926793557833e-07, + "loss": 1.0584, + "mean_token_accuracy": 0.682651162147522, + "num_tokens": 34197557.0, + "step": 1326 + }, + { + "epoch": 0.14572809136832857, + "grad_norm": 2.4550840854644775, + "learning_rate": 4.853587115666178e-07, + "loss": 0.9975, + "mean_token_accuracy": 0.6975124478340149, + "num_tokens": 34219659.0, + "step": 1327 + }, + { + "epoch": 0.14583790907094224, + "grad_norm": 2.5745298862457275, + "learning_rate": 4.857247437774524e-07, + "loss": 1.0617, + "mean_token_accuracy": 0.6812248826026917, + "num_tokens": 34241179.0, + "step": 1328 + }, + { + "epoch": 0.14594772677355589, + "grad_norm": 2.546809434890747, + "learning_rate": 4.86090775988287e-07, + "loss": 0.9421, + "mean_token_accuracy": 0.7164311408996582, + "num_tokens": 34261000.0, + "step": 1329 + }, + { + "epoch": 0.14605754447616956, + "grad_norm": 2.4541985988616943, + "learning_rate": 4.864568081991215e-07, + "loss": 1.0492, + "mean_token_accuracy": 0.6860387325286865, + "num_tokens": 34285817.0, + "step": 1330 + }, + { + "epoch": 0.14616736217878323, + "grad_norm": 2.1805577278137207, + "learning_rate": 4.868228404099561e-07, + "loss": 1.1327, + "mean_token_accuracy": 0.6656460165977478, + "num_tokens": 34315537.0, + "step": 1331 + }, + { + "epoch": 0.14627717988139688, + "grad_norm": 2.3619070053100586, + "learning_rate": 4.871888726207906e-07, + "loss": 1.0561, + "mean_token_accuracy": 0.6855508685112, + "num_tokens": 34338741.0, + "step": 1332 + }, + { + "epoch": 0.14638699758401055, + "grad_norm": 2.4076473712921143, + "learning_rate": 4.875549048316251e-07, + "loss": 1.0359, + "mean_token_accuracy": 0.6919925808906555, + "num_tokens": 34362546.0, + "step": 1333 + }, + { + "epoch": 0.1464968152866242, + "grad_norm": 1.9941058158874512, + "learning_rate": 4.879209370424597e-07, + "loss": 1.1286, + "mean_token_accuracy": 0.6656336784362793, + "num_tokens": 34394885.0, + "step": 1334 + }, + { + "epoch": 0.14660663298923787, + "grad_norm": 2.0383541584014893, + "learning_rate": 4.882869692532943e-07, + "loss": 1.0778, + "mean_token_accuracy": 0.6740219593048096, + "num_tokens": 34425299.0, + "step": 1335 + }, + { + "epoch": 0.14671645069185152, + "grad_norm": 2.3934431076049805, + "learning_rate": 4.886530014641288e-07, + "loss": 1.0355, + "mean_token_accuracy": 0.6928893327713013, + "num_tokens": 34448680.0, + "step": 1336 + }, + { + "epoch": 0.1468262683944652, + "grad_norm": 2.28312349319458, + "learning_rate": 4.890190336749634e-07, + "loss": 1.0852, + "mean_token_accuracy": 0.6812373399734497, + "num_tokens": 34474831.0, + "step": 1337 + }, + { + "epoch": 0.14693608609707884, + "grad_norm": 2.3955397605895996, + "learning_rate": 4.893850658857979e-07, + "loss": 0.9188, + "mean_token_accuracy": 0.7198169231414795, + "num_tokens": 34496472.0, + "step": 1338 + }, + { + "epoch": 0.1470459037996925, + "grad_norm": 2.4556260108947754, + "learning_rate": 4.897510980966325e-07, + "loss": 0.9867, + "mean_token_accuracy": 0.713093638420105, + "num_tokens": 34519372.0, + "step": 1339 + }, + { + "epoch": 0.14715572150230616, + "grad_norm": 2.312079429626465, + "learning_rate": 4.90117130307467e-07, + "loss": 1.08, + "mean_token_accuracy": 0.6930245161056519, + "num_tokens": 34545661.0, + "step": 1340 + }, + { + "epoch": 0.14726553920491983, + "grad_norm": 2.178640842437744, + "learning_rate": 4.904831625183016e-07, + "loss": 1.1633, + "mean_token_accuracy": 0.6641075611114502, + "num_tokens": 34576126.0, + "step": 1341 + }, + { + "epoch": 0.1473753569075335, + "grad_norm": 2.394110918045044, + "learning_rate": 4.908491947291362e-07, + "loss": 1.0842, + "mean_token_accuracy": 0.680875301361084, + "num_tokens": 34600086.0, + "step": 1342 + }, + { + "epoch": 0.14748517461014715, + "grad_norm": 2.2466061115264893, + "learning_rate": 4.912152269399707e-07, + "loss": 0.9958, + "mean_token_accuracy": 0.7044306993484497, + "num_tokens": 34626401.0, + "step": 1343 + }, + { + "epoch": 0.14759499231276083, + "grad_norm": 2.2814738750457764, + "learning_rate": 4.915812591508052e-07, + "loss": 1.0924, + "mean_token_accuracy": 0.6760477423667908, + "num_tokens": 34653201.0, + "step": 1344 + }, + { + "epoch": 0.14770481001537447, + "grad_norm": 2.3438069820404053, + "learning_rate": 4.919472913616398e-07, + "loss": 1.0746, + "mean_token_accuracy": 0.678378701210022, + "num_tokens": 34680307.0, + "step": 1345 + }, + { + "epoch": 0.14781462771798815, + "grad_norm": 2.4827253818511963, + "learning_rate": 4.923133235724744e-07, + "loss": 1.0312, + "mean_token_accuracy": 0.6953426599502563, + "num_tokens": 34702531.0, + "step": 1346 + }, + { + "epoch": 0.1479244454206018, + "grad_norm": 2.0433456897735596, + "learning_rate": 4.926793557833089e-07, + "loss": 1.0597, + "mean_token_accuracy": 0.6841834783554077, + "num_tokens": 34733708.0, + "step": 1347 + }, + { + "epoch": 0.14803426312321546, + "grad_norm": 2.429661273956299, + "learning_rate": 4.930453879941435e-07, + "loss": 1.0902, + "mean_token_accuracy": 0.6764568090438843, + "num_tokens": 34758625.0, + "step": 1348 + }, + { + "epoch": 0.1481440808258291, + "grad_norm": 2.2103769779205322, + "learning_rate": 4.93411420204978e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6953988075256348, + "num_tokens": 34786952.0, + "step": 1349 + }, + { + "epoch": 0.14825389852844278, + "grad_norm": 2.2437589168548584, + "learning_rate": 4.937774524158126e-07, + "loss": 1.1154, + "mean_token_accuracy": 0.6717184782028198, + "num_tokens": 34813181.0, + "step": 1350 + }, + { + "epoch": 0.14836371623105646, + "grad_norm": 2.1126108169555664, + "learning_rate": 4.941434846266471e-07, + "loss": 1.0095, + "mean_token_accuracy": 0.6952868103981018, + "num_tokens": 34845742.0, + "step": 1351 + }, + { + "epoch": 0.1484735339336701, + "grad_norm": 2.2218828201293945, + "learning_rate": 4.945095168374817e-07, + "loss": 1.0548, + "mean_token_accuracy": 0.6854334473609924, + "num_tokens": 34872515.0, + "step": 1352 + }, + { + "epoch": 0.14858335163628378, + "grad_norm": 2.3680293560028076, + "learning_rate": 4.948755490483163e-07, + "loss": 1.1108, + "mean_token_accuracy": 0.6681824922561646, + "num_tokens": 34899778.0, + "step": 1353 + }, + { + "epoch": 0.14869316933889742, + "grad_norm": 2.2915892601013184, + "learning_rate": 4.952415812591508e-07, + "loss": 0.9949, + "mean_token_accuracy": 0.6946365237236023, + "num_tokens": 34926103.0, + "step": 1354 + }, + { + "epoch": 0.1488029870415111, + "grad_norm": 2.189629316329956, + "learning_rate": 4.956076134699854e-07, + "loss": 0.965, + "mean_token_accuracy": 0.7073271870613098, + "num_tokens": 34952471.0, + "step": 1355 + }, + { + "epoch": 0.14891280474412474, + "grad_norm": 2.0904574394226074, + "learning_rate": 4.959736456808199e-07, + "loss": 1.0939, + "mean_token_accuracy": 0.67230623960495, + "num_tokens": 34981653.0, + "step": 1356 + }, + { + "epoch": 0.14902262244673842, + "grad_norm": 2.469757556915283, + "learning_rate": 4.963396778916544e-07, + "loss": 0.9795, + "mean_token_accuracy": 0.7096619606018066, + "num_tokens": 35004160.0, + "step": 1357 + }, + { + "epoch": 0.14913244014935206, + "grad_norm": 2.1677916049957275, + "learning_rate": 4.96705710102489e-07, + "loss": 1.0763, + "mean_token_accuracy": 0.6847352981567383, + "num_tokens": 35033695.0, + "step": 1358 + }, + { + "epoch": 0.14924225785196574, + "grad_norm": 2.266615629196167, + "learning_rate": 4.970717423133236e-07, + "loss": 1.0072, + "mean_token_accuracy": 0.6904715895652771, + "num_tokens": 35058960.0, + "step": 1359 + }, + { + "epoch": 0.1493520755545794, + "grad_norm": 2.290813446044922, + "learning_rate": 4.974377745241581e-07, + "loss": 1.0286, + "mean_token_accuracy": 0.6932991147041321, + "num_tokens": 35082458.0, + "step": 1360 + }, + { + "epoch": 0.14946189325719306, + "grad_norm": 2.15773344039917, + "learning_rate": 4.978038067349927e-07, + "loss": 1.0704, + "mean_token_accuracy": 0.6909747123718262, + "num_tokens": 35111320.0, + "step": 1361 + }, + { + "epoch": 0.14957171095980673, + "grad_norm": 2.486654758453369, + "learning_rate": 4.981698389458272e-07, + "loss": 1.0977, + "mean_token_accuracy": 0.6685775518417358, + "num_tokens": 35136034.0, + "step": 1362 + }, + { + "epoch": 0.14968152866242038, + "grad_norm": 2.4016880989074707, + "learning_rate": 4.985358711566618e-07, + "loss": 1.0409, + "mean_token_accuracy": 0.6914806365966797, + "num_tokens": 35160498.0, + "step": 1363 + }, + { + "epoch": 0.14979134636503405, + "grad_norm": 2.061107635498047, + "learning_rate": 4.989019033674963e-07, + "loss": 1.0782, + "mean_token_accuracy": 0.684070885181427, + "num_tokens": 35191074.0, + "step": 1364 + }, + { + "epoch": 0.1499011640676477, + "grad_norm": 2.3396177291870117, + "learning_rate": 4.992679355783309e-07, + "loss": 1.0492, + "mean_token_accuracy": 0.6916173100471497, + "num_tokens": 35216079.0, + "step": 1365 + }, + { + "epoch": 0.15001098177026137, + "grad_norm": 2.4558651447296143, + "learning_rate": 4.996339677891655e-07, + "loss": 1.0525, + "mean_token_accuracy": 0.6945352554321289, + "num_tokens": 35239149.0, + "step": 1366 + }, + { + "epoch": 0.15012079947287502, + "grad_norm": 2.2121670246124268, + "learning_rate": 5e-07, + "loss": 1.1018, + "mean_token_accuracy": 0.6782844662666321, + "num_tokens": 35267266.0, + "step": 1367 + }, + { + "epoch": 0.1502306171754887, + "grad_norm": 2.200930118560791, + "learning_rate": 5.003660322108345e-07, + "loss": 1.1085, + "mean_token_accuracy": 0.6758867502212524, + "num_tokens": 35295774.0, + "step": 1368 + }, + { + "epoch": 0.15034043487810236, + "grad_norm": 2.6652278900146484, + "learning_rate": 5.007320644216691e-07, + "loss": 1.0368, + "mean_token_accuracy": 0.6875271797180176, + "num_tokens": 35315866.0, + "step": 1369 + }, + { + "epoch": 0.150450252580716, + "grad_norm": 2.3804166316986084, + "learning_rate": 5.010980966325037e-07, + "loss": 1.0711, + "mean_token_accuracy": 0.6876465082168579, + "num_tokens": 35340461.0, + "step": 1370 + }, + { + "epoch": 0.15056007028332968, + "grad_norm": 2.1092634201049805, + "learning_rate": 5.014641288433382e-07, + "loss": 0.9819, + "mean_token_accuracy": 0.703368067741394, + "num_tokens": 35367551.0, + "step": 1371 + }, + { + "epoch": 0.15066988798594333, + "grad_norm": 2.1279263496398926, + "learning_rate": 5.018301610541728e-07, + "loss": 1.1114, + "mean_token_accuracy": 0.6663691401481628, + "num_tokens": 35397737.0, + "step": 1372 + }, + { + "epoch": 0.150779705688557, + "grad_norm": 2.433434009552002, + "learning_rate": 5.021961932650073e-07, + "loss": 1.0272, + "mean_token_accuracy": 0.6939836740493774, + "num_tokens": 35422047.0, + "step": 1373 + }, + { + "epoch": 0.15088952339117065, + "grad_norm": 2.154697895050049, + "learning_rate": 5.025622254758418e-07, + "loss": 1.0491, + "mean_token_accuracy": 0.6884704828262329, + "num_tokens": 35450974.0, + "step": 1374 + }, + { + "epoch": 0.15099934109378432, + "grad_norm": 2.1773035526275635, + "learning_rate": 5.029282576866764e-07, + "loss": 0.9847, + "mean_token_accuracy": 0.705015242099762, + "num_tokens": 35478196.0, + "step": 1375 + }, + { + "epoch": 0.15110915879639797, + "grad_norm": 2.0776009559631348, + "learning_rate": 5.03294289897511e-07, + "loss": 1.1131, + "mean_token_accuracy": 0.6687760353088379, + "num_tokens": 35510077.0, + "step": 1376 + }, + { + "epoch": 0.15121897649901164, + "grad_norm": 2.276146411895752, + "learning_rate": 5.036603221083456e-07, + "loss": 1.0609, + "mean_token_accuracy": 0.6841700077056885, + "num_tokens": 35538238.0, + "step": 1377 + }, + { + "epoch": 0.1513287942016253, + "grad_norm": 2.135308265686035, + "learning_rate": 5.040263543191801e-07, + "loss": 1.1086, + "mean_token_accuracy": 0.6685284376144409, + "num_tokens": 35569282.0, + "step": 1378 + }, + { + "epoch": 0.15143861190423896, + "grad_norm": 2.342716932296753, + "learning_rate": 5.043923865300146e-07, + "loss": 1.1513, + "mean_token_accuracy": 0.6719959378242493, + "num_tokens": 35595162.0, + "step": 1379 + }, + { + "epoch": 0.15154842960685264, + "grad_norm": 2.3281090259552, + "learning_rate": 5.047584187408492e-07, + "loss": 0.9932, + "mean_token_accuracy": 0.6980867385864258, + "num_tokens": 35620173.0, + "step": 1380 + }, + { + "epoch": 0.15165824730946628, + "grad_norm": 2.5368168354034424, + "learning_rate": 5.051244509516838e-07, + "loss": 0.984, + "mean_token_accuracy": 0.7048782110214233, + "num_tokens": 35641699.0, + "step": 1381 + }, + { + "epoch": 0.15176806501207996, + "grad_norm": 2.1876626014709473, + "learning_rate": 5.054904831625183e-07, + "loss": 1.0929, + "mean_token_accuracy": 0.6737250685691833, + "num_tokens": 35672048.0, + "step": 1382 + }, + { + "epoch": 0.1518778827146936, + "grad_norm": 2.321106195449829, + "learning_rate": 5.058565153733529e-07, + "loss": 1.088, + "mean_token_accuracy": 0.6787188053131104, + "num_tokens": 35697483.0, + "step": 1383 + }, + { + "epoch": 0.15198770041730728, + "grad_norm": 2.280531406402588, + "learning_rate": 5.062225475841874e-07, + "loss": 1.0583, + "mean_token_accuracy": 0.6966581344604492, + "num_tokens": 35723408.0, + "step": 1384 + }, + { + "epoch": 0.15209751811992092, + "grad_norm": 2.322547674179077, + "learning_rate": 5.065885797950219e-07, + "loss": 1.0285, + "mean_token_accuracy": 0.691029965877533, + "num_tokens": 35747189.0, + "step": 1385 + }, + { + "epoch": 0.1522073358225346, + "grad_norm": 2.102731466293335, + "learning_rate": 5.069546120058566e-07, + "loss": 1.0575, + "mean_token_accuracy": 0.6825284957885742, + "num_tokens": 35777084.0, + "step": 1386 + }, + { + "epoch": 0.15231715352514824, + "grad_norm": 2.334054946899414, + "learning_rate": 5.073206442166911e-07, + "loss": 0.9714, + "mean_token_accuracy": 0.7096595764160156, + "num_tokens": 35801318.0, + "step": 1387 + }, + { + "epoch": 0.15242697122776191, + "grad_norm": 2.772400379180908, + "learning_rate": 5.076866764275256e-07, + "loss": 0.9904, + "mean_token_accuracy": 0.7036679983139038, + "num_tokens": 35820852.0, + "step": 1388 + }, + { + "epoch": 0.1525367889303756, + "grad_norm": 2.323340654373169, + "learning_rate": 5.080527086383602e-07, + "loss": 1.0808, + "mean_token_accuracy": 0.6761853694915771, + "num_tokens": 35846653.0, + "step": 1389 + }, + { + "epoch": 0.15264660663298923, + "grad_norm": 2.0723321437835693, + "learning_rate": 5.084187408491947e-07, + "loss": 1.0867, + "mean_token_accuracy": 0.6757655143737793, + "num_tokens": 35877674.0, + "step": 1390 + }, + { + "epoch": 0.1527564243356029, + "grad_norm": 2.0999021530151367, + "learning_rate": 5.087847730600292e-07, + "loss": 1.0799, + "mean_token_accuracy": 0.6806398630142212, + "num_tokens": 35908450.0, + "step": 1391 + }, + { + "epoch": 0.15286624203821655, + "grad_norm": 2.267784357070923, + "learning_rate": 5.091508052708639e-07, + "loss": 1.0435, + "mean_token_accuracy": 0.6933692097663879, + "num_tokens": 35934858.0, + "step": 1392 + }, + { + "epoch": 0.15297605974083023, + "grad_norm": 2.2357428073883057, + "learning_rate": 5.095168374816984e-07, + "loss": 1.0469, + "mean_token_accuracy": 0.686190664768219, + "num_tokens": 35962276.0, + "step": 1393 + }, + { + "epoch": 0.15308587744344387, + "grad_norm": 2.541769027709961, + "learning_rate": 5.09882869692533e-07, + "loss": 1.1718, + "mean_token_accuracy": 0.6627906560897827, + "num_tokens": 35987873.0, + "step": 1394 + }, + { + "epoch": 0.15319569514605755, + "grad_norm": 2.395216703414917, + "learning_rate": 5.102489019033675e-07, + "loss": 1.0452, + "mean_token_accuracy": 0.685847282409668, + "num_tokens": 36013354.0, + "step": 1395 + }, + { + "epoch": 0.1533055128486712, + "grad_norm": 2.2029035091400146, + "learning_rate": 5.10614934114202e-07, + "loss": 1.0621, + "mean_token_accuracy": 0.6817206740379333, + "num_tokens": 36040211.0, + "step": 1396 + }, + { + "epoch": 0.15341533055128487, + "grad_norm": 2.240814208984375, + "learning_rate": 5.109809663250367e-07, + "loss": 1.09, + "mean_token_accuracy": 0.6770620346069336, + "num_tokens": 36066374.0, + "step": 1397 + }, + { + "epoch": 0.15352514825389854, + "grad_norm": 2.302957057952881, + "learning_rate": 5.113469985358712e-07, + "loss": 0.9885, + "mean_token_accuracy": 0.7000230550765991, + "num_tokens": 36092114.0, + "step": 1398 + }, + { + "epoch": 0.1536349659565122, + "grad_norm": 2.463291883468628, + "learning_rate": 5.117130307467057e-07, + "loss": 1.0511, + "mean_token_accuracy": 0.6894669532775879, + "num_tokens": 36117150.0, + "step": 1399 + }, + { + "epoch": 0.15374478365912586, + "grad_norm": 2.1149630546569824, + "learning_rate": 5.120790629575403e-07, + "loss": 1.1441, + "mean_token_accuracy": 0.6675069332122803, + "num_tokens": 36146676.0, + "step": 1400 + }, + { + "epoch": 0.1538546013617395, + "grad_norm": 2.3044769763946533, + "learning_rate": 5.124450951683748e-07, + "loss": 1.0989, + "mean_token_accuracy": 0.6727938652038574, + "num_tokens": 36173244.0, + "step": 1401 + }, + { + "epoch": 0.15396441906435318, + "grad_norm": 2.152919292449951, + "learning_rate": 5.128111273792094e-07, + "loss": 1.1006, + "mean_token_accuracy": 0.6777250170707703, + "num_tokens": 36203197.0, + "step": 1402 + }, + { + "epoch": 0.15407423676696683, + "grad_norm": 2.2176904678344727, + "learning_rate": 5.13177159590044e-07, + "loss": 1.1016, + "mean_token_accuracy": 0.6745572686195374, + "num_tokens": 36231276.0, + "step": 1403 + }, + { + "epoch": 0.1541840544695805, + "grad_norm": 2.0810370445251465, + "learning_rate": 5.135431918008785e-07, + "loss": 1.0665, + "mean_token_accuracy": 0.6908162832260132, + "num_tokens": 36261923.0, + "step": 1404 + }, + { + "epoch": 0.15429387217219415, + "grad_norm": 2.4359166622161865, + "learning_rate": 5.13909224011713e-07, + "loss": 0.9809, + "mean_token_accuracy": 0.7098432779312134, + "num_tokens": 36285200.0, + "step": 1405 + }, + { + "epoch": 0.15440368987480782, + "grad_norm": 2.2750539779663086, + "learning_rate": 5.142752562225476e-07, + "loss": 1.0682, + "mean_token_accuracy": 0.6800129413604736, + "num_tokens": 36310732.0, + "step": 1406 + }, + { + "epoch": 0.1545135075774215, + "grad_norm": 2.422652006149292, + "learning_rate": 5.14641288433382e-07, + "loss": 0.9953, + "mean_token_accuracy": 0.7009868621826172, + "num_tokens": 36333833.0, + "step": 1407 + }, + { + "epoch": 0.15462332528003514, + "grad_norm": 2.3580679893493652, + "learning_rate": 5.150073206442168e-07, + "loss": 1.0059, + "mean_token_accuracy": 0.70323646068573, + "num_tokens": 36357548.0, + "step": 1408 + }, + { + "epoch": 0.1547331429826488, + "grad_norm": 2.1177783012390137, + "learning_rate": 5.153733528550513e-07, + "loss": 0.9635, + "mean_token_accuracy": 0.7077805995941162, + "num_tokens": 36386854.0, + "step": 1409 + }, + { + "epoch": 0.15484296068526246, + "grad_norm": 2.0544211864471436, + "learning_rate": 5.157393850658857e-07, + "loss": 1.1586, + "mean_token_accuracy": 0.6616991758346558, + "num_tokens": 36425311.0, + "step": 1410 + }, + { + "epoch": 0.15495277838787613, + "grad_norm": 2.070923089981079, + "learning_rate": 5.161054172767203e-07, + "loss": 0.9747, + "mean_token_accuracy": 0.7018684148788452, + "num_tokens": 36453543.0, + "step": 1411 + }, + { + "epoch": 0.15506259609048978, + "grad_norm": 2.4090495109558105, + "learning_rate": 5.164714494875548e-07, + "loss": 1.0839, + "mean_token_accuracy": 0.6881023049354553, + "num_tokens": 36479897.0, + "step": 1412 + }, + { + "epoch": 0.15517241379310345, + "grad_norm": 2.338798761367798, + "learning_rate": 5.168374816983894e-07, + "loss": 1.0779, + "mean_token_accuracy": 0.682207465171814, + "num_tokens": 36508596.0, + "step": 1413 + }, + { + "epoch": 0.1552822314957171, + "grad_norm": 2.5255794525146484, + "learning_rate": 5.17203513909224e-07, + "loss": 1.0728, + "mean_token_accuracy": 0.6795178651809692, + "num_tokens": 36532995.0, + "step": 1414 + }, + { + "epoch": 0.15539204919833077, + "grad_norm": 2.1325502395629883, + "learning_rate": 5.175695461200585e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.6795332431793213, + "num_tokens": 36565082.0, + "step": 1415 + }, + { + "epoch": 0.15550186690094442, + "grad_norm": 2.2242419719696045, + "learning_rate": 5.17935578330893e-07, + "loss": 1.1603, + "mean_token_accuracy": 0.6564387083053589, + "num_tokens": 36595531.0, + "step": 1416 + }, + { + "epoch": 0.1556116846035581, + "grad_norm": 2.4247865676879883, + "learning_rate": 5.183016105417276e-07, + "loss": 1.0151, + "mean_token_accuracy": 0.6994357705116272, + "num_tokens": 36618337.0, + "step": 1417 + }, + { + "epoch": 0.15572150230617177, + "grad_norm": 2.559694528579712, + "learning_rate": 5.186676427525622e-07, + "loss": 1.1121, + "mean_token_accuracy": 0.669903039932251, + "num_tokens": 36641123.0, + "step": 1418 + }, + { + "epoch": 0.1558313200087854, + "grad_norm": 2.340179204940796, + "learning_rate": 5.190336749633967e-07, + "loss": 1.0594, + "mean_token_accuracy": 0.6912086009979248, + "num_tokens": 36667870.0, + "step": 1419 + }, + { + "epoch": 0.15594113771139909, + "grad_norm": 2.206639528274536, + "learning_rate": 5.193997071742313e-07, + "loss": 1.0466, + "mean_token_accuracy": 0.6896668672561646, + "num_tokens": 36697458.0, + "step": 1420 + }, + { + "epoch": 0.15605095541401273, + "grad_norm": 2.1337101459503174, + "learning_rate": 5.197657393850658e-07, + "loss": 1.1004, + "mean_token_accuracy": 0.6780178546905518, + "num_tokens": 36728999.0, + "step": 1421 + }, + { + "epoch": 0.1561607731166264, + "grad_norm": 2.7840282917022705, + "learning_rate": 5.201317715959003e-07, + "loss": 1.1376, + "mean_token_accuracy": 0.6671460866928101, + "num_tokens": 36751360.0, + "step": 1422 + }, + { + "epoch": 0.15627059081924005, + "grad_norm": 2.2902486324310303, + "learning_rate": 5.204978038067349e-07, + "loss": 1.095, + "mean_token_accuracy": 0.6812284588813782, + "num_tokens": 36778609.0, + "step": 1423 + }, + { + "epoch": 0.15638040852185373, + "grad_norm": 2.372972011566162, + "learning_rate": 5.208638360175695e-07, + "loss": 1.0165, + "mean_token_accuracy": 0.6935238838195801, + "num_tokens": 36801124.0, + "step": 1424 + }, + { + "epoch": 0.15649022622446737, + "grad_norm": 2.331251382827759, + "learning_rate": 5.212298682284041e-07, + "loss": 1.0964, + "mean_token_accuracy": 0.6727505326271057, + "num_tokens": 36826030.0, + "step": 1425 + }, + { + "epoch": 0.15660004392708105, + "grad_norm": 2.197559118270874, + "learning_rate": 5.215959004392386e-07, + "loss": 1.1204, + "mean_token_accuracy": 0.669613242149353, + "num_tokens": 36857373.0, + "step": 1426 + }, + { + "epoch": 0.15670986162969472, + "grad_norm": 2.354064702987671, + "learning_rate": 5.219619326500731e-07, + "loss": 1.0214, + "mean_token_accuracy": 0.6968653798103333, + "num_tokens": 36882743.0, + "step": 1427 + }, + { + "epoch": 0.15681967933230836, + "grad_norm": 2.2345306873321533, + "learning_rate": 5.223279648609077e-07, + "loss": 1.0352, + "mean_token_accuracy": 0.6908993721008301, + "num_tokens": 36908317.0, + "step": 1428 + }, + { + "epoch": 0.15692949703492204, + "grad_norm": 2.3579368591308594, + "learning_rate": 5.226939970717423e-07, + "loss": 1.068, + "mean_token_accuracy": 0.6793192625045776, + "num_tokens": 36933934.0, + "step": 1429 + }, + { + "epoch": 0.15703931473753568, + "grad_norm": 2.196413993835449, + "learning_rate": 5.230600292825768e-07, + "loss": 1.0059, + "mean_token_accuracy": 0.6966009736061096, + "num_tokens": 36961658.0, + "step": 1430 + }, + { + "epoch": 0.15714913244014936, + "grad_norm": 2.2704312801361084, + "learning_rate": 5.234260614934114e-07, + "loss": 1.0349, + "mean_token_accuracy": 0.6856536865234375, + "num_tokens": 36989890.0, + "step": 1431 + }, + { + "epoch": 0.157258950142763, + "grad_norm": 2.4830708503723145, + "learning_rate": 5.237920937042459e-07, + "loss": 1.1271, + "mean_token_accuracy": 0.6708431243896484, + "num_tokens": 37014359.0, + "step": 1432 + }, + { + "epoch": 0.15736876784537668, + "grad_norm": 2.45174241065979, + "learning_rate": 5.241581259150804e-07, + "loss": 1.0332, + "mean_token_accuracy": 0.6931481957435608, + "num_tokens": 37036466.0, + "step": 1433 + }, + { + "epoch": 0.15747858554799032, + "grad_norm": 2.610792636871338, + "learning_rate": 5.245241581259151e-07, + "loss": 0.981, + "mean_token_accuracy": 0.705224871635437, + "num_tokens": 37056557.0, + "step": 1434 + }, + { + "epoch": 0.157588403250604, + "grad_norm": 2.4107344150543213, + "learning_rate": 5.248901903367496e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.673519492149353, + "num_tokens": 37079546.0, + "step": 1435 + }, + { + "epoch": 0.15769822095321767, + "grad_norm": 2.4994101524353027, + "learning_rate": 5.252562225475841e-07, + "loss": 1.1209, + "mean_token_accuracy": 0.6704230308532715, + "num_tokens": 37103143.0, + "step": 1436 + }, + { + "epoch": 0.15780803865583132, + "grad_norm": 2.305529832839966, + "learning_rate": 5.256222547584187e-07, + "loss": 0.9869, + "mean_token_accuracy": 0.7200730443000793, + "num_tokens": 37128958.0, + "step": 1437 + }, + { + "epoch": 0.157917856358445, + "grad_norm": 2.61940860748291, + "learning_rate": 5.259882869692532e-07, + "loss": 0.981, + "mean_token_accuracy": 0.7068555951118469, + "num_tokens": 37148862.0, + "step": 1438 + }, + { + "epoch": 0.15802767406105864, + "grad_norm": 2.6053128242492676, + "learning_rate": 5.263543191800877e-07, + "loss": 0.9409, + "mean_token_accuracy": 0.7133756875991821, + "num_tokens": 37168500.0, + "step": 1439 + }, + { + "epoch": 0.1581374917636723, + "grad_norm": 2.505356550216675, + "learning_rate": 5.267203513909224e-07, + "loss": 1.0647, + "mean_token_accuracy": 0.6812302470207214, + "num_tokens": 37191202.0, + "step": 1440 + }, + { + "epoch": 0.15824730946628596, + "grad_norm": 2.459718704223633, + "learning_rate": 5.270863836017569e-07, + "loss": 1.121, + "mean_token_accuracy": 0.6636503338813782, + "num_tokens": 37215939.0, + "step": 1441 + }, + { + "epoch": 0.15835712716889963, + "grad_norm": 2.200208902359009, + "learning_rate": 5.274524158125915e-07, + "loss": 1.2373, + "mean_token_accuracy": 0.6614638566970825, + "num_tokens": 37246502.0, + "step": 1442 + }, + { + "epoch": 0.15846694487151328, + "grad_norm": 2.4319093227386475, + "learning_rate": 5.27818448023426e-07, + "loss": 0.943, + "mean_token_accuracy": 0.714350700378418, + "num_tokens": 37269519.0, + "step": 1443 + }, + { + "epoch": 0.15857676257412695, + "grad_norm": 2.1731479167938232, + "learning_rate": 5.281844802342605e-07, + "loss": 1.0123, + "mean_token_accuracy": 0.6990000009536743, + "num_tokens": 37295472.0, + "step": 1444 + }, + { + "epoch": 0.15868658027674062, + "grad_norm": 2.250302791595459, + "learning_rate": 5.285505124450952e-07, + "loss": 1.0923, + "mean_token_accuracy": 0.6789708137512207, + "num_tokens": 37323089.0, + "step": 1445 + }, + { + "epoch": 0.15879639797935427, + "grad_norm": 2.395758867263794, + "learning_rate": 5.289165446559297e-07, + "loss": 1.0988, + "mean_token_accuracy": 0.6698101758956909, + "num_tokens": 37349083.0, + "step": 1446 + }, + { + "epoch": 0.15890621568196794, + "grad_norm": 2.5284740924835205, + "learning_rate": 5.292825768667642e-07, + "loss": 0.9144, + "mean_token_accuracy": 0.7187700271606445, + "num_tokens": 37370736.0, + "step": 1447 + }, + { + "epoch": 0.1590160333845816, + "grad_norm": 2.3916666507720947, + "learning_rate": 5.296486090775988e-07, + "loss": 1.0749, + "mean_token_accuracy": 0.681726336479187, + "num_tokens": 37395554.0, + "step": 1448 + }, + { + "epoch": 0.15912585108719526, + "grad_norm": 2.3729259967803955, + "learning_rate": 5.300146412884333e-07, + "loss": 1.0416, + "mean_token_accuracy": 0.687462568283081, + "num_tokens": 37418863.0, + "step": 1449 + }, + { + "epoch": 0.1592356687898089, + "grad_norm": 2.9141392707824707, + "learning_rate": 5.303806734992679e-07, + "loss": 1.0321, + "mean_token_accuracy": 0.6832531094551086, + "num_tokens": 37435803.0, + "step": 1450 + }, + { + "epoch": 0.15934548649242258, + "grad_norm": 2.3098511695861816, + "learning_rate": 5.307467057101025e-07, + "loss": 1.1499, + "mean_token_accuracy": 0.6573442816734314, + "num_tokens": 37465254.0, + "step": 1451 + }, + { + "epoch": 0.15945530419503623, + "grad_norm": 2.292492151260376, + "learning_rate": 5.31112737920937e-07, + "loss": 0.9883, + "mean_token_accuracy": 0.7074750661849976, + "num_tokens": 37490392.0, + "step": 1452 + }, + { + "epoch": 0.1595651218976499, + "grad_norm": 2.3009941577911377, + "learning_rate": 5.314787701317715e-07, + "loss": 1.0447, + "mean_token_accuracy": 0.6876119375228882, + "num_tokens": 37515868.0, + "step": 1453 + }, + { + "epoch": 0.15967493960026355, + "grad_norm": 2.3609728813171387, + "learning_rate": 5.318448023426061e-07, + "loss": 0.9897, + "mean_token_accuracy": 0.7043418884277344, + "num_tokens": 37539607.0, + "step": 1454 + }, + { + "epoch": 0.15978475730287722, + "grad_norm": 2.8560781478881836, + "learning_rate": 5.322108345534406e-07, + "loss": 0.9217, + "mean_token_accuracy": 0.7161433696746826, + "num_tokens": 37557033.0, + "step": 1455 + }, + { + "epoch": 0.1598945750054909, + "grad_norm": 2.3579628467559814, + "learning_rate": 5.325768667642752e-07, + "loss": 0.9806, + "mean_token_accuracy": 0.7003841996192932, + "num_tokens": 37581118.0, + "step": 1456 + }, + { + "epoch": 0.16000439270810454, + "grad_norm": 2.339226722717285, + "learning_rate": 5.329428989751098e-07, + "loss": 1.1314, + "mean_token_accuracy": 0.667583167552948, + "num_tokens": 37609075.0, + "step": 1457 + }, + { + "epoch": 0.16011421041071822, + "grad_norm": 2.1235742568969727, + "learning_rate": 5.333089311859443e-07, + "loss": 1.0401, + "mean_token_accuracy": 0.6891235113143921, + "num_tokens": 37636878.0, + "step": 1458 + }, + { + "epoch": 0.16022402811333186, + "grad_norm": 2.1337475776672363, + "learning_rate": 5.336749633967789e-07, + "loss": 1.0991, + "mean_token_accuracy": 0.6683434844017029, + "num_tokens": 37667267.0, + "step": 1459 + }, + { + "epoch": 0.16033384581594554, + "grad_norm": 2.626617193222046, + "learning_rate": 5.340409956076134e-07, + "loss": 1.0072, + "mean_token_accuracy": 0.7000125646591187, + "num_tokens": 37688240.0, + "step": 1460 + }, + { + "epoch": 0.16044366351855918, + "grad_norm": 2.6160144805908203, + "learning_rate": 5.34407027818448e-07, + "loss": 0.928, + "mean_token_accuracy": 0.7142786979675293, + "num_tokens": 37707416.0, + "step": 1461 + }, + { + "epoch": 0.16055348122117286, + "grad_norm": 2.0190417766571045, + "learning_rate": 5.347730600292826e-07, + "loss": 1.0773, + "mean_token_accuracy": 0.6832205653190613, + "num_tokens": 37740039.0, + "step": 1462 + }, + { + "epoch": 0.1606632989237865, + "grad_norm": 2.0718328952789307, + "learning_rate": 5.351390922401171e-07, + "loss": 1.1347, + "mean_token_accuracy": 0.6699882745742798, + "num_tokens": 37773476.0, + "step": 1463 + }, + { + "epoch": 0.16077311662640018, + "grad_norm": 2.198026418685913, + "learning_rate": 5.355051244509516e-07, + "loss": 1.0574, + "mean_token_accuracy": 0.6902638077735901, + "num_tokens": 37802625.0, + "step": 1464 + }, + { + "epoch": 0.16088293432901385, + "grad_norm": 2.740053415298462, + "learning_rate": 5.358711566617862e-07, + "loss": 0.9618, + "mean_token_accuracy": 0.7088761925697327, + "num_tokens": 37822336.0, + "step": 1465 + }, + { + "epoch": 0.1609927520316275, + "grad_norm": 2.295747756958008, + "learning_rate": 5.362371888726208e-07, + "loss": 0.9968, + "mean_token_accuracy": 0.6987466812133789, + "num_tokens": 37845801.0, + "step": 1466 + }, + { + "epoch": 0.16110256973424117, + "grad_norm": 2.3483920097351074, + "learning_rate": 5.366032210834553e-07, + "loss": 1.1074, + "mean_token_accuracy": 0.6780867576599121, + "num_tokens": 37872498.0, + "step": 1467 + }, + { + "epoch": 0.16121238743685481, + "grad_norm": 1.9937318563461304, + "learning_rate": 5.369692532942899e-07, + "loss": 1.0483, + "mean_token_accuracy": 0.687330961227417, + "num_tokens": 37904152.0, + "step": 1468 + }, + { + "epoch": 0.1613222051394685, + "grad_norm": 2.1053884029388428, + "learning_rate": 5.373352855051244e-07, + "loss": 0.9741, + "mean_token_accuracy": 0.7106304168701172, + "num_tokens": 37931150.0, + "step": 1469 + }, + { + "epoch": 0.16143202284208213, + "grad_norm": 2.3302557468414307, + "learning_rate": 5.377013177159589e-07, + "loss": 1.0478, + "mean_token_accuracy": 0.6850367784500122, + "num_tokens": 37955011.0, + "step": 1470 + }, + { + "epoch": 0.1615418405446958, + "grad_norm": 2.399446964263916, + "learning_rate": 5.380673499267935e-07, + "loss": 1.1136, + "mean_token_accuracy": 0.6716781854629517, + "num_tokens": 37981518.0, + "step": 1471 + }, + { + "epoch": 0.16165165824730945, + "grad_norm": 2.246284008026123, + "learning_rate": 5.384333821376281e-07, + "loss": 1.0812, + "mean_token_accuracy": 0.6814975142478943, + "num_tokens": 38008938.0, + "step": 1472 + }, + { + "epoch": 0.16176147594992313, + "grad_norm": 2.296086072921753, + "learning_rate": 5.387994143484626e-07, + "loss": 1.0388, + "mean_token_accuracy": 0.6866783499717712, + "num_tokens": 38036626.0, + "step": 1473 + }, + { + "epoch": 0.1618712936525368, + "grad_norm": 1.9944976568222046, + "learning_rate": 5.391654465592972e-07, + "loss": 1.0965, + "mean_token_accuracy": 0.6742771863937378, + "num_tokens": 38070141.0, + "step": 1474 + }, + { + "epoch": 0.16198111135515045, + "grad_norm": 2.803293228149414, + "learning_rate": 5.395314787701317e-07, + "loss": 1.0145, + "mean_token_accuracy": 0.6937214136123657, + "num_tokens": 38092134.0, + "step": 1475 + }, + { + "epoch": 0.16209092905776412, + "grad_norm": 2.648883104324341, + "learning_rate": 5.398975109809663e-07, + "loss": 0.9776, + "mean_token_accuracy": 0.7018635272979736, + "num_tokens": 38111775.0, + "step": 1476 + }, + { + "epoch": 0.16220074676037777, + "grad_norm": 2.5859546661376953, + "learning_rate": 5.402635431918009e-07, + "loss": 1.0879, + "mean_token_accuracy": 0.6767747402191162, + "num_tokens": 38134366.0, + "step": 1477 + }, + { + "epoch": 0.16231056446299144, + "grad_norm": 2.2919487953186035, + "learning_rate": 5.406295754026354e-07, + "loss": 1.0766, + "mean_token_accuracy": 0.6785234212875366, + "num_tokens": 38159731.0, + "step": 1478 + }, + { + "epoch": 0.1624203821656051, + "grad_norm": 2.2435402870178223, + "learning_rate": 5.4099560761347e-07, + "loss": 1.0551, + "mean_token_accuracy": 0.680763304233551, + "num_tokens": 38189414.0, + "step": 1479 + }, + { + "epoch": 0.16253019986821876, + "grad_norm": 2.3856542110443115, + "learning_rate": 5.413616398243045e-07, + "loss": 1.0225, + "mean_token_accuracy": 0.6911889314651489, + "num_tokens": 38213059.0, + "step": 1480 + }, + { + "epoch": 0.1626400175708324, + "grad_norm": 2.1769397258758545, + "learning_rate": 5.41727672035139e-07, + "loss": 1.1124, + "mean_token_accuracy": 0.6683104038238525, + "num_tokens": 38242367.0, + "step": 1481 + }, + { + "epoch": 0.16274983527344608, + "grad_norm": 2.252507209777832, + "learning_rate": 5.420937042459737e-07, + "loss": 1.1282, + "mean_token_accuracy": 0.6625720262527466, + "num_tokens": 38269270.0, + "step": 1482 + }, + { + "epoch": 0.16285965297605975, + "grad_norm": 2.2109594345092773, + "learning_rate": 5.424597364568082e-07, + "loss": 1.0912, + "mean_token_accuracy": 0.6875765919685364, + "num_tokens": 38297171.0, + "step": 1483 + }, + { + "epoch": 0.1629694706786734, + "grad_norm": 2.6146740913391113, + "learning_rate": 5.428257686676427e-07, + "loss": 1.0918, + "mean_token_accuracy": 0.6741479635238647, + "num_tokens": 38319751.0, + "step": 1484 + }, + { + "epoch": 0.16307928838128707, + "grad_norm": 2.3918726444244385, + "learning_rate": 5.431918008784773e-07, + "loss": 1.0225, + "mean_token_accuracy": 0.6925700902938843, + "num_tokens": 38344590.0, + "step": 1485 + }, + { + "epoch": 0.16318910608390072, + "grad_norm": 2.325105905532837, + "learning_rate": 5.435578330893118e-07, + "loss": 1.0018, + "mean_token_accuracy": 0.7029340267181396, + "num_tokens": 38369984.0, + "step": 1486 + }, + { + "epoch": 0.1632989237865144, + "grad_norm": 2.609196662902832, + "learning_rate": 5.439238653001463e-07, + "loss": 0.9989, + "mean_token_accuracy": 0.7004256248474121, + "num_tokens": 38392266.0, + "step": 1487 + }, + { + "epoch": 0.16340874148912804, + "grad_norm": 2.000401735305786, + "learning_rate": 5.44289897510981e-07, + "loss": 1.1691, + "mean_token_accuracy": 0.6599587202072144, + "num_tokens": 38426150.0, + "step": 1488 + }, + { + "epoch": 0.1635185591917417, + "grad_norm": 2.033644914627075, + "learning_rate": 5.446559297218155e-07, + "loss": 1.0087, + "mean_token_accuracy": 0.7029205560684204, + "num_tokens": 38458153.0, + "step": 1489 + }, + { + "epoch": 0.16362837689435536, + "grad_norm": 2.614518642425537, + "learning_rate": 5.4502196193265e-07, + "loss": 1.0419, + "mean_token_accuracy": 0.695292592048645, + "num_tokens": 38478408.0, + "step": 1490 + }, + { + "epoch": 0.16373819459696903, + "grad_norm": 2.1933908462524414, + "learning_rate": 5.453879941434846e-07, + "loss": 0.961, + "mean_token_accuracy": 0.7045871019363403, + "num_tokens": 38507099.0, + "step": 1491 + }, + { + "epoch": 0.16384801229958268, + "grad_norm": 2.2437546253204346, + "learning_rate": 5.457540263543191e-07, + "loss": 1.0311, + "mean_token_accuracy": 0.6924747228622437, + "num_tokens": 38532923.0, + "step": 1492 + }, + { + "epoch": 0.16395783000219635, + "grad_norm": 2.369605302810669, + "learning_rate": 5.461200585651538e-07, + "loss": 1.1048, + "mean_token_accuracy": 0.6748452186584473, + "num_tokens": 38558137.0, + "step": 1493 + }, + { + "epoch": 0.16406764770481003, + "grad_norm": 2.2160139083862305, + "learning_rate": 5.464860907759883e-07, + "loss": 1.0443, + "mean_token_accuracy": 0.6890190839767456, + "num_tokens": 38584890.0, + "step": 1494 + }, + { + "epoch": 0.16417746540742367, + "grad_norm": 2.2313756942749023, + "learning_rate": 5.468521229868228e-07, + "loss": 1.0479, + "mean_token_accuracy": 0.6868633031845093, + "num_tokens": 38611310.0, + "step": 1495 + }, + { + "epoch": 0.16428728311003735, + "grad_norm": 2.2412474155426025, + "learning_rate": 5.472181551976574e-07, + "loss": 1.0962, + "mean_token_accuracy": 0.6775031089782715, + "num_tokens": 38638118.0, + "step": 1496 + }, + { + "epoch": 0.164397100812651, + "grad_norm": 2.4821090698242188, + "learning_rate": 5.475841874084919e-07, + "loss": 0.9592, + "mean_token_accuracy": 0.7092792987823486, + "num_tokens": 38660015.0, + "step": 1497 + }, + { + "epoch": 0.16450691851526467, + "grad_norm": 2.2274677753448486, + "learning_rate": 5.479502196193265e-07, + "loss": 1.0693, + "mean_token_accuracy": 0.6808262467384338, + "num_tokens": 38688580.0, + "step": 1498 + }, + { + "epoch": 0.1646167362178783, + "grad_norm": 2.359994411468506, + "learning_rate": 5.483162518301611e-07, + "loss": 1.0808, + "mean_token_accuracy": 0.6790705323219299, + "num_tokens": 38714193.0, + "step": 1499 + }, + { + "epoch": 0.16472655392049199, + "grad_norm": 2.2185144424438477, + "learning_rate": 5.486822840409956e-07, + "loss": 1.0706, + "mean_token_accuracy": 0.6791196465492249, + "num_tokens": 38742448.0, + "step": 1500 + }, + { + "epoch": 0.16483637162310563, + "grad_norm": 2.439730644226074, + "learning_rate": 5.490483162518301e-07, + "loss": 0.9775, + "mean_token_accuracy": 0.7085849046707153, + "num_tokens": 38765620.0, + "step": 1501 + }, + { + "epoch": 0.1649461893257193, + "grad_norm": 2.418790102005005, + "learning_rate": 5.494143484626647e-07, + "loss": 0.9519, + "mean_token_accuracy": 0.7130105495452881, + "num_tokens": 38788077.0, + "step": 1502 + }, + { + "epoch": 0.16505600702833298, + "grad_norm": 2.380723237991333, + "learning_rate": 5.497803806734992e-07, + "loss": 1.0137, + "mean_token_accuracy": 0.6993364095687866, + "num_tokens": 38811822.0, + "step": 1503 + }, + { + "epoch": 0.16516582473094663, + "grad_norm": 2.3779258728027344, + "learning_rate": 5.501464128843338e-07, + "loss": 1.0329, + "mean_token_accuracy": 0.6901930570602417, + "num_tokens": 38835722.0, + "step": 1504 + }, + { + "epoch": 0.1652756424335603, + "grad_norm": 2.2435708045959473, + "learning_rate": 5.505124450951684e-07, + "loss": 0.9875, + "mean_token_accuracy": 0.7012511491775513, + "num_tokens": 38859889.0, + "step": 1505 + }, + { + "epoch": 0.16538546013617395, + "grad_norm": 2.0821926593780518, + "learning_rate": 5.508784773060029e-07, + "loss": 0.9877, + "mean_token_accuracy": 0.7153289318084717, + "num_tokens": 38890641.0, + "step": 1506 + }, + { + "epoch": 0.16549527783878762, + "grad_norm": 2.3635668754577637, + "learning_rate": 5.512445095168374e-07, + "loss": 1.0942, + "mean_token_accuracy": 0.6806544065475464, + "num_tokens": 38914684.0, + "step": 1507 + }, + { + "epoch": 0.16560509554140126, + "grad_norm": 2.5526351928710938, + "learning_rate": 5.51610541727672e-07, + "loss": 0.997, + "mean_token_accuracy": 0.7021938562393188, + "num_tokens": 38935499.0, + "step": 1508 + }, + { + "epoch": 0.16571491324401494, + "grad_norm": 2.3887040615081787, + "learning_rate": 5.519765739385066e-07, + "loss": 1.0363, + "mean_token_accuracy": 0.6865290403366089, + "num_tokens": 38960393.0, + "step": 1509 + }, + { + "epoch": 0.16582473094662858, + "grad_norm": 2.518939256668091, + "learning_rate": 5.523426061493412e-07, + "loss": 1.0648, + "mean_token_accuracy": 0.6906571388244629, + "num_tokens": 38983330.0, + "step": 1510 + }, + { + "epoch": 0.16593454864924226, + "grad_norm": 2.1938388347625732, + "learning_rate": 5.527086383601757e-07, + "loss": 1.1263, + "mean_token_accuracy": 0.6688031554222107, + "num_tokens": 39011686.0, + "step": 1511 + }, + { + "epoch": 0.16604436635185593, + "grad_norm": 2.344820499420166, + "learning_rate": 5.530746705710102e-07, + "loss": 1.0536, + "mean_token_accuracy": 0.6878054738044739, + "num_tokens": 39037527.0, + "step": 1512 + }, + { + "epoch": 0.16615418405446958, + "grad_norm": 2.502321720123291, + "learning_rate": 5.534407027818448e-07, + "loss": 1.0975, + "mean_token_accuracy": 0.6732074022293091, + "num_tokens": 39061584.0, + "step": 1513 + }, + { + "epoch": 0.16626400175708325, + "grad_norm": 2.380232334136963, + "learning_rate": 5.538067349926794e-07, + "loss": 1.013, + "mean_token_accuracy": 0.6950703859329224, + "num_tokens": 39085730.0, + "step": 1514 + }, + { + "epoch": 0.1663738194596969, + "grad_norm": 2.8414146900177, + "learning_rate": 5.541727672035139e-07, + "loss": 1.0243, + "mean_token_accuracy": 0.6991514563560486, + "num_tokens": 39103747.0, + "step": 1515 + }, + { + "epoch": 0.16648363716231057, + "grad_norm": 2.4908716678619385, + "learning_rate": 5.545387994143485e-07, + "loss": 0.9792, + "mean_token_accuracy": 0.7013140320777893, + "num_tokens": 39124300.0, + "step": 1516 + }, + { + "epoch": 0.16659345486492422, + "grad_norm": 2.4015631675720215, + "learning_rate": 5.54904831625183e-07, + "loss": 0.9461, + "mean_token_accuracy": 0.7095775604248047, + "num_tokens": 39148700.0, + "step": 1517 + }, + { + "epoch": 0.1667032725675379, + "grad_norm": 2.605893611907959, + "learning_rate": 5.552708638360175e-07, + "loss": 1.1076, + "mean_token_accuracy": 0.6687907576560974, + "num_tokens": 39172508.0, + "step": 1518 + }, + { + "epoch": 0.16681309027015154, + "grad_norm": 2.262875556945801, + "learning_rate": 5.556368960468521e-07, + "loss": 0.9538, + "mean_token_accuracy": 0.7097836136817932, + "num_tokens": 39197782.0, + "step": 1519 + }, + { + "epoch": 0.1669229079727652, + "grad_norm": 2.2503514289855957, + "learning_rate": 5.560029282576867e-07, + "loss": 1.0048, + "mean_token_accuracy": 0.6978820562362671, + "num_tokens": 39224616.0, + "step": 1520 + }, + { + "epoch": 0.16703272567537888, + "grad_norm": 2.288672924041748, + "learning_rate": 5.563689604685212e-07, + "loss": 0.9938, + "mean_token_accuracy": 0.7001767158508301, + "num_tokens": 39249771.0, + "step": 1521 + }, + { + "epoch": 0.16714254337799253, + "grad_norm": 2.40399432182312, + "learning_rate": 5.567349926793558e-07, + "loss": 0.9242, + "mean_token_accuracy": 0.7152349352836609, + "num_tokens": 39271524.0, + "step": 1522 + }, + { + "epoch": 0.1672523610806062, + "grad_norm": 2.1903858184814453, + "learning_rate": 5.571010248901903e-07, + "loss": 0.9941, + "mean_token_accuracy": 0.6999549865722656, + "num_tokens": 39299086.0, + "step": 1523 + }, + { + "epoch": 0.16736217878321985, + "grad_norm": 2.025735855102539, + "learning_rate": 5.574670571010248e-07, + "loss": 1.006, + "mean_token_accuracy": 0.7046303749084473, + "num_tokens": 39331423.0, + "step": 1524 + }, + { + "epoch": 0.16747199648583352, + "grad_norm": 2.541682481765747, + "learning_rate": 5.578330893118595e-07, + "loss": 1.06, + "mean_token_accuracy": 0.6830800771713257, + "num_tokens": 39352747.0, + "step": 1525 + }, + { + "epoch": 0.16758181418844717, + "grad_norm": 2.3848280906677246, + "learning_rate": 5.58199121522694e-07, + "loss": 1.0737, + "mean_token_accuracy": 0.6828184723854065, + "num_tokens": 39377552.0, + "step": 1526 + }, + { + "epoch": 0.16769163189106084, + "grad_norm": 2.4345285892486572, + "learning_rate": 5.585651537335286e-07, + "loss": 1.0495, + "mean_token_accuracy": 0.6855342388153076, + "num_tokens": 39401174.0, + "step": 1527 + }, + { + "epoch": 0.1678014495936745, + "grad_norm": 2.1818559169769287, + "learning_rate": 5.589311859443631e-07, + "loss": 0.952, + "mean_token_accuracy": 0.7116169929504395, + "num_tokens": 39427179.0, + "step": 1528 + }, + { + "epoch": 0.16791126729628816, + "grad_norm": 2.504570960998535, + "learning_rate": 5.592972181551976e-07, + "loss": 0.9454, + "mean_token_accuracy": 0.7087352275848389, + "num_tokens": 39446853.0, + "step": 1529 + }, + { + "epoch": 0.1680210849989018, + "grad_norm": 2.4998908042907715, + "learning_rate": 5.596632503660323e-07, + "loss": 0.9571, + "mean_token_accuracy": 0.7127382755279541, + "num_tokens": 39467826.0, + "step": 1530 + }, + { + "epoch": 0.16813090270151548, + "grad_norm": 2.310150384902954, + "learning_rate": 5.600292825768668e-07, + "loss": 0.9148, + "mean_token_accuracy": 0.718069314956665, + "num_tokens": 39492608.0, + "step": 1531 + }, + { + "epoch": 0.16824072040412916, + "grad_norm": 2.5591328144073486, + "learning_rate": 5.603953147877013e-07, + "loss": 1.0576, + "mean_token_accuracy": 0.690179169178009, + "num_tokens": 39513984.0, + "step": 1532 + }, + { + "epoch": 0.1683505381067428, + "grad_norm": 2.159175157546997, + "learning_rate": 5.607613469985359e-07, + "loss": 0.9762, + "mean_token_accuracy": 0.703987717628479, + "num_tokens": 39541191.0, + "step": 1533 + }, + { + "epoch": 0.16846035580935648, + "grad_norm": 2.338866949081421, + "learning_rate": 5.611273792093704e-07, + "loss": 0.9846, + "mean_token_accuracy": 0.7160253524780273, + "num_tokens": 39565937.0, + "step": 1534 + }, + { + "epoch": 0.16857017351197012, + "grad_norm": 2.69417667388916, + "learning_rate": 5.614934114202049e-07, + "loss": 1.0264, + "mean_token_accuracy": 0.6963381767272949, + "num_tokens": 39585586.0, + "step": 1535 + }, + { + "epoch": 0.1686799912145838, + "grad_norm": 1.9973052740097046, + "learning_rate": 5.618594436310396e-07, + "loss": 1.1407, + "mean_token_accuracy": 0.6650521755218506, + "num_tokens": 39621116.0, + "step": 1536 + }, + { + "epoch": 0.16878980891719744, + "grad_norm": 2.5043787956237793, + "learning_rate": 5.622254758418741e-07, + "loss": 1.0733, + "mean_token_accuracy": 0.6844034194946289, + "num_tokens": 39645545.0, + "step": 1537 + }, + { + "epoch": 0.16889962661981112, + "grad_norm": 2.308725595474243, + "learning_rate": 5.625915080527086e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.6878054141998291, + "num_tokens": 39671881.0, + "step": 1538 + }, + { + "epoch": 0.16900944432242476, + "grad_norm": 2.444974422454834, + "learning_rate": 5.629575402635432e-07, + "loss": 1.0155, + "mean_token_accuracy": 0.6961317658424377, + "num_tokens": 39695290.0, + "step": 1539 + }, + { + "epoch": 0.16911926202503844, + "grad_norm": 2.210258960723877, + "learning_rate": 5.633235724743777e-07, + "loss": 1.0211, + "mean_token_accuracy": 0.7010210752487183, + "num_tokens": 39722135.0, + "step": 1540 + }, + { + "epoch": 0.1692290797276521, + "grad_norm": 2.270078420639038, + "learning_rate": 5.636896046852124e-07, + "loss": 1.15, + "mean_token_accuracy": 0.6617167592048645, + "num_tokens": 39752087.0, + "step": 1541 + }, + { + "epoch": 0.16933889743026576, + "grad_norm": 2.109419345855713, + "learning_rate": 5.640556368960469e-07, + "loss": 1.0254, + "mean_token_accuracy": 0.6934658288955688, + "num_tokens": 39782108.0, + "step": 1542 + }, + { + "epoch": 0.16944871513287943, + "grad_norm": 2.222062110900879, + "learning_rate": 5.644216691068814e-07, + "loss": 1.0731, + "mean_token_accuracy": 0.675371527671814, + "num_tokens": 39810707.0, + "step": 1543 + }, + { + "epoch": 0.16955853283549308, + "grad_norm": 2.283102512359619, + "learning_rate": 5.64787701317716e-07, + "loss": 0.9884, + "mean_token_accuracy": 0.6998631358146667, + "num_tokens": 39836838.0, + "step": 1544 + }, + { + "epoch": 0.16966835053810675, + "grad_norm": 2.4577810764312744, + "learning_rate": 5.651537335285505e-07, + "loss": 0.9686, + "mean_token_accuracy": 0.7108678817749023, + "num_tokens": 39860798.0, + "step": 1545 + }, + { + "epoch": 0.1697781682407204, + "grad_norm": 2.3531835079193115, + "learning_rate": 5.655197657393851e-07, + "loss": 1.1198, + "mean_token_accuracy": 0.6720527410507202, + "num_tokens": 39886558.0, + "step": 1546 + }, + { + "epoch": 0.16988798594333407, + "grad_norm": 2.575234889984131, + "learning_rate": 5.658857979502197e-07, + "loss": 0.9474, + "mean_token_accuracy": 0.7070711255073547, + "num_tokens": 39907445.0, + "step": 1547 + }, + { + "epoch": 0.16999780364594771, + "grad_norm": 2.1753861904144287, + "learning_rate": 5.662518301610542e-07, + "loss": 0.9461, + "mean_token_accuracy": 0.7186204791069031, + "num_tokens": 39937347.0, + "step": 1548 + }, + { + "epoch": 0.1701076213485614, + "grad_norm": 2.15962553024292, + "learning_rate": 5.666178623718887e-07, + "loss": 0.9842, + "mean_token_accuracy": 0.7055522203445435, + "num_tokens": 39966512.0, + "step": 1549 + }, + { + "epoch": 0.17021743905117506, + "grad_norm": 2.381989002227783, + "learning_rate": 5.669838945827233e-07, + "loss": 1.0685, + "mean_token_accuracy": 0.6814039945602417, + "num_tokens": 39993941.0, + "step": 1550 + }, + { + "epoch": 0.1703272567537887, + "grad_norm": 2.992835760116577, + "learning_rate": 5.673499267935577e-07, + "loss": 0.8887, + "mean_token_accuracy": 0.7310321927070618, + "num_tokens": 40012552.0, + "step": 1551 + }, + { + "epoch": 0.17043707445640238, + "grad_norm": 2.157390832901001, + "learning_rate": 5.677159590043924e-07, + "loss": 1.0587, + "mean_token_accuracy": 0.6873670816421509, + "num_tokens": 40043398.0, + "step": 1552 + }, + { + "epoch": 0.17054689215901603, + "grad_norm": 2.396151065826416, + "learning_rate": 5.68081991215227e-07, + "loss": 1.0925, + "mean_token_accuracy": 0.6792099475860596, + "num_tokens": 40070386.0, + "step": 1553 + }, + { + "epoch": 0.1706567098616297, + "grad_norm": 2.4042937755584717, + "learning_rate": 5.684480234260615e-07, + "loss": 1.0212, + "mean_token_accuracy": 0.6950595378875732, + "num_tokens": 40094339.0, + "step": 1554 + }, + { + "epoch": 0.17076652756424335, + "grad_norm": 2.2818830013275146, + "learning_rate": 5.688140556368959e-07, + "loss": 0.9351, + "mean_token_accuracy": 0.7210057377815247, + "num_tokens": 40118938.0, + "step": 1555 + }, + { + "epoch": 0.17087634526685702, + "grad_norm": 2.19637131690979, + "learning_rate": 5.691800878477305e-07, + "loss": 1.1112, + "mean_token_accuracy": 0.6672136187553406, + "num_tokens": 40148530.0, + "step": 1556 + }, + { + "epoch": 0.17098616296947067, + "grad_norm": 2.2481067180633545, + "learning_rate": 5.695461200585652e-07, + "loss": 1.066, + "mean_token_accuracy": 0.6876349449157715, + "num_tokens": 40175048.0, + "step": 1557 + }, + { + "epoch": 0.17109598067208434, + "grad_norm": 2.442040205001831, + "learning_rate": 5.699121522693998e-07, + "loss": 1.0229, + "mean_token_accuracy": 0.6940107345581055, + "num_tokens": 40198815.0, + "step": 1558 + }, + { + "epoch": 0.17120579837469801, + "grad_norm": 2.427168846130371, + "learning_rate": 5.702781844802342e-07, + "loss": 0.995, + "mean_token_accuracy": 0.7072909474372864, + "num_tokens": 40221652.0, + "step": 1559 + }, + { + "epoch": 0.17131561607731166, + "grad_norm": 2.3035390377044678, + "learning_rate": 5.706442166910687e-07, + "loss": 1.0114, + "mean_token_accuracy": 0.6961264610290527, + "num_tokens": 40248573.0, + "step": 1560 + }, + { + "epoch": 0.17142543377992533, + "grad_norm": 2.4548332691192627, + "learning_rate": 5.710102489019033e-07, + "loss": 1.0852, + "mean_token_accuracy": 0.680486798286438, + "num_tokens": 40273242.0, + "step": 1561 + }, + { + "epoch": 0.17153525148253898, + "grad_norm": 2.431441307067871, + "learning_rate": 5.71376281112738e-07, + "loss": 1.0104, + "mean_token_accuracy": 0.6970536708831787, + "num_tokens": 40295379.0, + "step": 1562 + }, + { + "epoch": 0.17164506918515265, + "grad_norm": 2.8023900985717773, + "learning_rate": 5.717423133235724e-07, + "loss": 1.0149, + "mean_token_accuracy": 0.6945093274116516, + "num_tokens": 40313918.0, + "step": 1563 + }, + { + "epoch": 0.1717548868877663, + "grad_norm": 2.35904860496521, + "learning_rate": 5.72108345534407e-07, + "loss": 1.032, + "mean_token_accuracy": 0.6931995749473572, + "num_tokens": 40338811.0, + "step": 1564 + }, + { + "epoch": 0.17186470459037997, + "grad_norm": 2.4317057132720947, + "learning_rate": 5.724743777452415e-07, + "loss": 1.0387, + "mean_token_accuracy": 0.6898938417434692, + "num_tokens": 40361823.0, + "step": 1565 + }, + { + "epoch": 0.17197452229299362, + "grad_norm": 2.361219882965088, + "learning_rate": 5.72840409956076e-07, + "loss": 1.0719, + "mean_token_accuracy": 0.6901206374168396, + "num_tokens": 40386526.0, + "step": 1566 + }, + { + "epoch": 0.1720843399956073, + "grad_norm": 2.2780404090881348, + "learning_rate": 5.732064421669106e-07, + "loss": 0.9979, + "mean_token_accuracy": 0.7013404369354248, + "num_tokens": 40412670.0, + "step": 1567 + }, + { + "epoch": 0.17219415769822094, + "grad_norm": 2.2931671142578125, + "learning_rate": 5.735724743777452e-07, + "loss": 1.0184, + "mean_token_accuracy": 0.6967191696166992, + "num_tokens": 40439359.0, + "step": 1568 + }, + { + "epoch": 0.1723039754008346, + "grad_norm": 2.1950488090515137, + "learning_rate": 5.739385065885797e-07, + "loss": 1.0124, + "mean_token_accuracy": 0.6997883319854736, + "num_tokens": 40466494.0, + "step": 1569 + }, + { + "epoch": 0.1724137931034483, + "grad_norm": 2.5362982749938965, + "learning_rate": 5.743045387994143e-07, + "loss": 0.971, + "mean_token_accuracy": 0.7037607431411743, + "num_tokens": 40488412.0, + "step": 1570 + }, + { + "epoch": 0.17252361080606193, + "grad_norm": 2.340975046157837, + "learning_rate": 5.746705710102488e-07, + "loss": 1.0655, + "mean_token_accuracy": 0.6821228861808777, + "num_tokens": 40511460.0, + "step": 1571 + }, + { + "epoch": 0.1726334285086756, + "grad_norm": 2.232487916946411, + "learning_rate": 5.750366032210833e-07, + "loss": 0.9839, + "mean_token_accuracy": 0.7063220739364624, + "num_tokens": 40537397.0, + "step": 1572 + }, + { + "epoch": 0.17274324621128925, + "grad_norm": 2.693403720855713, + "learning_rate": 5.75402635431918e-07, + "loss": 0.9924, + "mean_token_accuracy": 0.7010452151298523, + "num_tokens": 40556838.0, + "step": 1573 + }, + { + "epoch": 0.17285306391390293, + "grad_norm": 2.786227226257324, + "learning_rate": 5.757686676427525e-07, + "loss": 1.0021, + "mean_token_accuracy": 0.6938157081604004, + "num_tokens": 40575543.0, + "step": 1574 + }, + { + "epoch": 0.17296288161651657, + "grad_norm": 2.1698646545410156, + "learning_rate": 5.761346998535871e-07, + "loss": 1.056, + "mean_token_accuracy": 0.688494861125946, + "num_tokens": 40604152.0, + "step": 1575 + }, + { + "epoch": 0.17307269931913025, + "grad_norm": 2.370185136795044, + "learning_rate": 5.765007320644216e-07, + "loss": 1.0503, + "mean_token_accuracy": 0.6865882873535156, + "num_tokens": 40627574.0, + "step": 1576 + }, + { + "epoch": 0.1731825170217439, + "grad_norm": 2.220339059829712, + "learning_rate": 5.768667642752561e-07, + "loss": 0.9903, + "mean_token_accuracy": 0.6995095610618591, + "num_tokens": 40651057.0, + "step": 1577 + }, + { + "epoch": 0.17329233472435757, + "grad_norm": 2.221599578857422, + "learning_rate": 5.772327964860908e-07, + "loss": 1.0257, + "mean_token_accuracy": 0.6890242099761963, + "num_tokens": 40678662.0, + "step": 1578 + }, + { + "epoch": 0.17340215242697124, + "grad_norm": 2.307269811630249, + "learning_rate": 5.775988286969253e-07, + "loss": 1.0893, + "mean_token_accuracy": 0.6838871836662292, + "num_tokens": 40704905.0, + "step": 1579 + }, + { + "epoch": 0.17351197012958489, + "grad_norm": 2.5684382915496826, + "learning_rate": 5.779648609077598e-07, + "loss": 0.9529, + "mean_token_accuracy": 0.7125042676925659, + "num_tokens": 40725916.0, + "step": 1580 + }, + { + "epoch": 0.17362178783219856, + "grad_norm": 2.55507493019104, + "learning_rate": 5.783308931185944e-07, + "loss": 1.0518, + "mean_token_accuracy": 0.6844428777694702, + "num_tokens": 40746748.0, + "step": 1581 + }, + { + "epoch": 0.1737316055348122, + "grad_norm": 2.1756691932678223, + "learning_rate": 5.786969253294289e-07, + "loss": 1.1228, + "mean_token_accuracy": 0.6852937340736389, + "num_tokens": 40779091.0, + "step": 1582 + }, + { + "epoch": 0.17384142323742588, + "grad_norm": 2.0630128383636475, + "learning_rate": 5.790629575402634e-07, + "loss": 1.1014, + "mean_token_accuracy": 0.6709854006767273, + "num_tokens": 40809761.0, + "step": 1583 + }, + { + "epoch": 0.17395124094003953, + "grad_norm": 2.133960485458374, + "learning_rate": 5.794289897510981e-07, + "loss": 1.0674, + "mean_token_accuracy": 0.6854323148727417, + "num_tokens": 40841486.0, + "step": 1584 + }, + { + "epoch": 0.1740610586426532, + "grad_norm": 2.1109817028045654, + "learning_rate": 5.797950219619326e-07, + "loss": 1.0835, + "mean_token_accuracy": 0.6833840012550354, + "num_tokens": 40870417.0, + "step": 1585 + }, + { + "epoch": 0.17417087634526685, + "grad_norm": 2.1986098289489746, + "learning_rate": 5.801610541727671e-07, + "loss": 1.0117, + "mean_token_accuracy": 0.7028540968894958, + "num_tokens": 40898422.0, + "step": 1586 + }, + { + "epoch": 0.17428069404788052, + "grad_norm": 2.497751474380493, + "learning_rate": 5.805270863836017e-07, + "loss": 1.1255, + "mean_token_accuracy": 0.6697672605514526, + "num_tokens": 40921246.0, + "step": 1587 + }, + { + "epoch": 0.1743905117504942, + "grad_norm": 2.2115750312805176, + "learning_rate": 5.808931185944362e-07, + "loss": 1.014, + "mean_token_accuracy": 0.6895596981048584, + "num_tokens": 40947697.0, + "step": 1588 + }, + { + "epoch": 0.17450032945310784, + "grad_norm": 2.281057596206665, + "learning_rate": 5.812591508052709e-07, + "loss": 0.9398, + "mean_token_accuracy": 0.7175296545028687, + "num_tokens": 40971954.0, + "step": 1589 + }, + { + "epoch": 0.1746101471557215, + "grad_norm": 2.115609645843506, + "learning_rate": 5.816251830161054e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.680611789226532, + "num_tokens": 41003026.0, + "step": 1590 + }, + { + "epoch": 0.17471996485833516, + "grad_norm": 2.3088767528533936, + "learning_rate": 5.819912152269399e-07, + "loss": 1.0634, + "mean_token_accuracy": 0.6794067025184631, + "num_tokens": 41027889.0, + "step": 1591 + }, + { + "epoch": 0.17482978256094883, + "grad_norm": 2.6810755729675293, + "learning_rate": 5.823572474377745e-07, + "loss": 1.0324, + "mean_token_accuracy": 0.6891739368438721, + "num_tokens": 41048983.0, + "step": 1592 + }, + { + "epoch": 0.17493960026356248, + "grad_norm": 2.3382914066314697, + "learning_rate": 5.82723279648609e-07, + "loss": 1.0843, + "mean_token_accuracy": 0.6726385354995728, + "num_tokens": 41074025.0, + "step": 1593 + }, + { + "epoch": 0.17504941796617615, + "grad_norm": 2.1966946125030518, + "learning_rate": 5.830893118594436e-07, + "loss": 1.0838, + "mean_token_accuracy": 0.6772397756576538, + "num_tokens": 41104142.0, + "step": 1594 + }, + { + "epoch": 0.1751592356687898, + "grad_norm": 2.3270938396453857, + "learning_rate": 5.834553440702782e-07, + "loss": 1.0502, + "mean_token_accuracy": 0.687483012676239, + "num_tokens": 41128610.0, + "step": 1595 + }, + { + "epoch": 0.17526905337140347, + "grad_norm": 2.2088489532470703, + "learning_rate": 5.838213762811127e-07, + "loss": 0.9908, + "mean_token_accuracy": 0.702506959438324, + "num_tokens": 41154215.0, + "step": 1596 + }, + { + "epoch": 0.17537887107401715, + "grad_norm": 2.133497476577759, + "learning_rate": 5.841874084919472e-07, + "loss": 1.0807, + "mean_token_accuracy": 0.6737655997276306, + "num_tokens": 41186683.0, + "step": 1597 + }, + { + "epoch": 0.1754886887766308, + "grad_norm": 2.3089048862457275, + "learning_rate": 5.845534407027818e-07, + "loss": 0.999, + "mean_token_accuracy": 0.694904625415802, + "num_tokens": 41211138.0, + "step": 1598 + }, + { + "epoch": 0.17559850647924446, + "grad_norm": 2.8164477348327637, + "learning_rate": 5.849194729136163e-07, + "loss": 0.9268, + "mean_token_accuracy": 0.7182216644287109, + "num_tokens": 41229133.0, + "step": 1599 + }, + { + "epoch": 0.1757083241818581, + "grad_norm": 2.4320783615112305, + "learning_rate": 5.852855051244509e-07, + "loss": 1.0598, + "mean_token_accuracy": 0.6882476806640625, + "num_tokens": 41254358.0, + "step": 1600 + }, + { + "epoch": 0.17581814188447178, + "grad_norm": 2.203343152999878, + "learning_rate": 5.856515373352855e-07, + "loss": 1.0871, + "mean_token_accuracy": 0.6846399307250977, + "num_tokens": 41282152.0, + "step": 1601 + }, + { + "epoch": 0.17592795958708543, + "grad_norm": 2.33988881111145, + "learning_rate": 5.8601756954612e-07, + "loss": 0.9628, + "mean_token_accuracy": 0.714012622833252, + "num_tokens": 41307760.0, + "step": 1602 + }, + { + "epoch": 0.1760377772896991, + "grad_norm": 2.2687363624572754, + "learning_rate": 5.863836017569545e-07, + "loss": 0.9823, + "mean_token_accuracy": 0.7026407122612, + "num_tokens": 41333522.0, + "step": 1603 + }, + { + "epoch": 0.17614759499231275, + "grad_norm": 2.3251068592071533, + "learning_rate": 5.867496339677891e-07, + "loss": 1.0177, + "mean_token_accuracy": 0.6967591047286987, + "num_tokens": 41357614.0, + "step": 1604 + }, + { + "epoch": 0.17625741269492642, + "grad_norm": 2.2193868160247803, + "learning_rate": 5.871156661786237e-07, + "loss": 0.9702, + "mean_token_accuracy": 0.709210991859436, + "num_tokens": 41383591.0, + "step": 1605 + }, + { + "epoch": 0.17636723039754007, + "grad_norm": 2.271904706954956, + "learning_rate": 5.874816983894583e-07, + "loss": 1.0968, + "mean_token_accuracy": 0.6729092597961426, + "num_tokens": 41410821.0, + "step": 1606 + }, + { + "epoch": 0.17647704810015374, + "grad_norm": 2.537780523300171, + "learning_rate": 5.878477306002928e-07, + "loss": 1.0689, + "mean_token_accuracy": 0.6859707832336426, + "num_tokens": 41433212.0, + "step": 1607 + }, + { + "epoch": 0.17658686580276742, + "grad_norm": 2.2113513946533203, + "learning_rate": 5.882137628111273e-07, + "loss": 1.1436, + "mean_token_accuracy": 0.6594667434692383, + "num_tokens": 41461618.0, + "step": 1608 + }, + { + "epoch": 0.17669668350538106, + "grad_norm": 2.181825876235962, + "learning_rate": 5.885797950219619e-07, + "loss": 1.0558, + "mean_token_accuracy": 0.6812527179718018, + "num_tokens": 41490369.0, + "step": 1609 + }, + { + "epoch": 0.17680650120799474, + "grad_norm": 2.2864201068878174, + "learning_rate": 5.889458272327965e-07, + "loss": 1.2098, + "mean_token_accuracy": 0.6474982500076294, + "num_tokens": 41518493.0, + "step": 1610 + }, + { + "epoch": 0.17691631891060838, + "grad_norm": 2.264575719833374, + "learning_rate": 5.89311859443631e-07, + "loss": 1.1193, + "mean_token_accuracy": 0.6745367050170898, + "num_tokens": 41545826.0, + "step": 1611 + }, + { + "epoch": 0.17702613661322206, + "grad_norm": 2.493701696395874, + "learning_rate": 5.896778916544656e-07, + "loss": 1.0139, + "mean_token_accuracy": 0.6907610297203064, + "num_tokens": 41568877.0, + "step": 1612 + }, + { + "epoch": 0.1771359543158357, + "grad_norm": 2.347412586212158, + "learning_rate": 5.900439238653001e-07, + "loss": 0.9824, + "mean_token_accuracy": 0.7049407958984375, + "num_tokens": 41594228.0, + "step": 1613 + }, + { + "epoch": 0.17724577201844938, + "grad_norm": 2.3610496520996094, + "learning_rate": 5.904099560761346e-07, + "loss": 1.0899, + "mean_token_accuracy": 0.6799055337905884, + "num_tokens": 41620861.0, + "step": 1614 + }, + { + "epoch": 0.17735558972106302, + "grad_norm": 2.374225378036499, + "learning_rate": 5.907759882869692e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6814360618591309, + "num_tokens": 41646859.0, + "step": 1615 + }, + { + "epoch": 0.1774654074236767, + "grad_norm": 2.4235634803771973, + "learning_rate": 5.911420204978038e-07, + "loss": 1.0501, + "mean_token_accuracy": 0.6870713233947754, + "num_tokens": 41670100.0, + "step": 1616 + }, + { + "epoch": 0.17757522512629037, + "grad_norm": 2.5501596927642822, + "learning_rate": 5.915080527086383e-07, + "loss": 0.9896, + "mean_token_accuracy": 0.7075978517532349, + "num_tokens": 41690417.0, + "step": 1617 + }, + { + "epoch": 0.17768504282890402, + "grad_norm": 2.2797343730926514, + "learning_rate": 5.918740849194729e-07, + "loss": 0.9551, + "mean_token_accuracy": 0.7203304767608643, + "num_tokens": 41723702.0, + "step": 1618 + }, + { + "epoch": 0.1777948605315177, + "grad_norm": 2.313589334487915, + "learning_rate": 5.922401171303074e-07, + "loss": 0.9711, + "mean_token_accuracy": 0.7047269344329834, + "num_tokens": 41747292.0, + "step": 1619 + }, + { + "epoch": 0.17790467823413134, + "grad_norm": 2.5482373237609863, + "learning_rate": 5.926061493411419e-07, + "loss": 1.0745, + "mean_token_accuracy": 0.6812134385108948, + "num_tokens": 41771664.0, + "step": 1620 + }, + { + "epoch": 0.178014495936745, + "grad_norm": 2.109189510345459, + "learning_rate": 5.929721815519766e-07, + "loss": 1.0842, + "mean_token_accuracy": 0.6779646873474121, + "num_tokens": 41802272.0, + "step": 1621 + }, + { + "epoch": 0.17812431363935866, + "grad_norm": 2.2538633346557617, + "learning_rate": 5.933382137628111e-07, + "loss": 1.0255, + "mean_token_accuracy": 0.6898040771484375, + "num_tokens": 41826323.0, + "step": 1622 + }, + { + "epoch": 0.17823413134197233, + "grad_norm": 2.044185161590576, + "learning_rate": 5.937042459736457e-07, + "loss": 1.1344, + "mean_token_accuracy": 0.6660110354423523, + "num_tokens": 41859043.0, + "step": 1623 + }, + { + "epoch": 0.17834394904458598, + "grad_norm": 2.486645221710205, + "learning_rate": 5.940702781844802e-07, + "loss": 0.9994, + "mean_token_accuracy": 0.6992120146751404, + "num_tokens": 41880055.0, + "step": 1624 + }, + { + "epoch": 0.17845376674719965, + "grad_norm": 2.3337864875793457, + "learning_rate": 5.944363103953147e-07, + "loss": 1.0649, + "mean_token_accuracy": 0.6777001619338989, + "num_tokens": 41906987.0, + "step": 1625 + }, + { + "epoch": 0.17856358444981332, + "grad_norm": 2.385885238647461, + "learning_rate": 5.948023426061494e-07, + "loss": 1.0397, + "mean_token_accuracy": 0.7006902098655701, + "num_tokens": 41929196.0, + "step": 1626 + }, + { + "epoch": 0.17867340215242697, + "grad_norm": 2.3095695972442627, + "learning_rate": 5.951683748169839e-07, + "loss": 1.1469, + "mean_token_accuracy": 0.6660809516906738, + "num_tokens": 41955354.0, + "step": 1627 + }, + { + "epoch": 0.17878321985504064, + "grad_norm": 2.255755662918091, + "learning_rate": 5.955344070278184e-07, + "loss": 1.146, + "mean_token_accuracy": 0.6582125425338745, + "num_tokens": 41982518.0, + "step": 1628 + }, + { + "epoch": 0.1788930375576543, + "grad_norm": 2.3237462043762207, + "learning_rate": 5.95900439238653e-07, + "loss": 0.9966, + "mean_token_accuracy": 0.6991581320762634, + "num_tokens": 42006184.0, + "step": 1629 + }, + { + "epoch": 0.17900285526026796, + "grad_norm": 2.7677290439605713, + "learning_rate": 5.962664714494875e-07, + "loss": 0.9922, + "mean_token_accuracy": 0.7010945081710815, + "num_tokens": 42025392.0, + "step": 1630 + }, + { + "epoch": 0.1791126729628816, + "grad_norm": 2.162986993789673, + "learning_rate": 5.96632503660322e-07, + "loss": 1.1743, + "mean_token_accuracy": 0.6614450216293335, + "num_tokens": 42055514.0, + "step": 1631 + }, + { + "epoch": 0.17922249066549528, + "grad_norm": 2.3502159118652344, + "learning_rate": 5.969985358711567e-07, + "loss": 1.0703, + "mean_token_accuracy": 0.6826711893081665, + "num_tokens": 42081033.0, + "step": 1632 + }, + { + "epoch": 0.17933230836810893, + "grad_norm": 2.401615619659424, + "learning_rate": 5.973645680819912e-07, + "loss": 0.9756, + "mean_token_accuracy": 0.7106883525848389, + "num_tokens": 42104399.0, + "step": 1633 + }, + { + "epoch": 0.1794421260707226, + "grad_norm": 2.334669351577759, + "learning_rate": 5.977306002928257e-07, + "loss": 1.0543, + "mean_token_accuracy": 0.6839743852615356, + "num_tokens": 42128912.0, + "step": 1634 + }, + { + "epoch": 0.17955194377333628, + "grad_norm": 2.6207871437072754, + "learning_rate": 5.980966325036603e-07, + "loss": 1.09, + "mean_token_accuracy": 0.6762047410011292, + "num_tokens": 42150892.0, + "step": 1635 + }, + { + "epoch": 0.17966176147594992, + "grad_norm": 2.4845902919769287, + "learning_rate": 5.984626647144948e-07, + "loss": 0.9753, + "mean_token_accuracy": 0.711798369884491, + "num_tokens": 42172510.0, + "step": 1636 + }, + { + "epoch": 0.1797715791785636, + "grad_norm": 2.208911657333374, + "learning_rate": 5.988286969253294e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.6843385696411133, + "num_tokens": 42199310.0, + "step": 1637 + }, + { + "epoch": 0.17988139688117724, + "grad_norm": 2.1074721813201904, + "learning_rate": 5.99194729136164e-07, + "loss": 1.0863, + "mean_token_accuracy": 0.6746960878372192, + "num_tokens": 42229990.0, + "step": 1638 + }, + { + "epoch": 0.17999121458379091, + "grad_norm": 2.5708911418914795, + "learning_rate": 5.995607613469985e-07, + "loss": 0.9576, + "mean_token_accuracy": 0.7134044170379639, + "num_tokens": 42249858.0, + "step": 1639 + }, + { + "epoch": 0.18010103228640456, + "grad_norm": 2.5591132640838623, + "learning_rate": 5.999267935578331e-07, + "loss": 0.953, + "mean_token_accuracy": 0.7093793153762817, + "num_tokens": 42271355.0, + "step": 1640 + }, + { + "epoch": 0.18021084998901823, + "grad_norm": 2.2761688232421875, + "learning_rate": 6.002928257686676e-07, + "loss": 1.0222, + "mean_token_accuracy": 0.6988310813903809, + "num_tokens": 42297624.0, + "step": 1641 + }, + { + "epoch": 0.18032066769163188, + "grad_norm": 2.2623238563537598, + "learning_rate": 6.006588579795022e-07, + "loss": 0.9624, + "mean_token_accuracy": 0.7065621018409729, + "num_tokens": 42322395.0, + "step": 1642 + }, + { + "epoch": 0.18043048539424555, + "grad_norm": 2.6426568031311035, + "learning_rate": 6.010248901903368e-07, + "loss": 1.016, + "mean_token_accuracy": 0.7084029912948608, + "num_tokens": 42344634.0, + "step": 1643 + }, + { + "epoch": 0.1805403030968592, + "grad_norm": 2.359342575073242, + "learning_rate": 6.013909224011713e-07, + "loss": 1.0977, + "mean_token_accuracy": 0.6773092150688171, + "num_tokens": 42370501.0, + "step": 1644 + }, + { + "epoch": 0.18065012079947287, + "grad_norm": 2.2395992279052734, + "learning_rate": 6.017569546120058e-07, + "loss": 1.0466, + "mean_token_accuracy": 0.6916664838790894, + "num_tokens": 42396966.0, + "step": 1645 + }, + { + "epoch": 0.18075993850208655, + "grad_norm": 2.3016529083251953, + "learning_rate": 6.021229868228404e-07, + "loss": 1.0938, + "mean_token_accuracy": 0.6852397918701172, + "num_tokens": 42423033.0, + "step": 1646 + }, + { + "epoch": 0.1808697562047002, + "grad_norm": 1.9946544170379639, + "learning_rate": 6.024890190336749e-07, + "loss": 1.042, + "mean_token_accuracy": 0.6939613223075867, + "num_tokens": 42456449.0, + "step": 1647 + }, + { + "epoch": 0.18097957390731387, + "grad_norm": 2.370236396789551, + "learning_rate": 6.028550512445095e-07, + "loss": 1.0074, + "mean_token_accuracy": 0.6997400522232056, + "num_tokens": 42480035.0, + "step": 1648 + }, + { + "epoch": 0.1810893916099275, + "grad_norm": 2.420858383178711, + "learning_rate": 6.032210834553441e-07, + "loss": 1.0167, + "mean_token_accuracy": 0.7006616592407227, + "num_tokens": 42503308.0, + "step": 1649 + }, + { + "epoch": 0.1811992093125412, + "grad_norm": 2.215991735458374, + "learning_rate": 6.035871156661786e-07, + "loss": 1.0827, + "mean_token_accuracy": 0.6786342859268188, + "num_tokens": 42532075.0, + "step": 1650 + }, + { + "epoch": 0.18130902701515483, + "grad_norm": 2.359473466873169, + "learning_rate": 6.039531478770131e-07, + "loss": 0.9938, + "mean_token_accuracy": 0.7037600874900818, + "num_tokens": 42556184.0, + "step": 1651 + }, + { + "epoch": 0.1814188447177685, + "grad_norm": 2.560459613800049, + "learning_rate": 6.043191800878477e-07, + "loss": 1.0939, + "mean_token_accuracy": 0.6809897422790527, + "num_tokens": 42578435.0, + "step": 1652 + }, + { + "epoch": 0.18152866242038215, + "grad_norm": 2.3542864322662354, + "learning_rate": 6.046852122986823e-07, + "loss": 1.0488, + "mean_token_accuracy": 0.6872497200965881, + "num_tokens": 42604295.0, + "step": 1653 + }, + { + "epoch": 0.18163848012299583, + "grad_norm": 2.278775215148926, + "learning_rate": 6.050512445095168e-07, + "loss": 1.12, + "mean_token_accuracy": 0.6799773573875427, + "num_tokens": 42632476.0, + "step": 1654 + }, + { + "epoch": 0.1817482978256095, + "grad_norm": 2.4182212352752686, + "learning_rate": 6.054172767203514e-07, + "loss": 1.0985, + "mean_token_accuracy": 0.6751473546028137, + "num_tokens": 42657921.0, + "step": 1655 + }, + { + "epoch": 0.18185811552822315, + "grad_norm": 2.508443593978882, + "learning_rate": 6.057833089311859e-07, + "loss": 1.1408, + "mean_token_accuracy": 0.664941132068634, + "num_tokens": 42683798.0, + "step": 1656 + }, + { + "epoch": 0.18196793323083682, + "grad_norm": 2.369713306427002, + "learning_rate": 6.061493411420205e-07, + "loss": 0.994, + "mean_token_accuracy": 0.702407717704773, + "num_tokens": 42707333.0, + "step": 1657 + }, + { + "epoch": 0.18207775093345047, + "grad_norm": 2.2431929111480713, + "learning_rate": 6.065153733528551e-07, + "loss": 1.0451, + "mean_token_accuracy": 0.6947814226150513, + "num_tokens": 42733776.0, + "step": 1658 + }, + { + "epoch": 0.18218756863606414, + "grad_norm": 1.9379651546478271, + "learning_rate": 6.068814055636896e-07, + "loss": 1.0863, + "mean_token_accuracy": 0.6782952547073364, + "num_tokens": 42769503.0, + "step": 1659 + }, + { + "epoch": 0.18229738633867779, + "grad_norm": 2.567296266555786, + "learning_rate": 6.072474377745242e-07, + "loss": 1.0042, + "mean_token_accuracy": 0.7043793201446533, + "num_tokens": 42791718.0, + "step": 1660 + }, + { + "epoch": 0.18240720404129146, + "grad_norm": 2.379189968109131, + "learning_rate": 6.076134699853587e-07, + "loss": 1.0297, + "mean_token_accuracy": 0.6931684613227844, + "num_tokens": 42816722.0, + "step": 1661 + }, + { + "epoch": 0.1825170217439051, + "grad_norm": 2.1454010009765625, + "learning_rate": 6.079795021961932e-07, + "loss": 0.9813, + "mean_token_accuracy": 0.699904203414917, + "num_tokens": 42844961.0, + "step": 1662 + }, + { + "epoch": 0.18262683944651878, + "grad_norm": 2.202930212020874, + "learning_rate": 6.083455344070278e-07, + "loss": 0.9386, + "mean_token_accuracy": 0.7107822299003601, + "num_tokens": 42870511.0, + "step": 1663 + }, + { + "epoch": 0.18273665714913245, + "grad_norm": 2.1795542240142822, + "learning_rate": 6.087115666178624e-07, + "loss": 0.9663, + "mean_token_accuracy": 0.7067681550979614, + "num_tokens": 42897649.0, + "step": 1664 + }, + { + "epoch": 0.1828464748517461, + "grad_norm": 2.09212064743042, + "learning_rate": 6.090775988286969e-07, + "loss": 0.9759, + "mean_token_accuracy": 0.705199658870697, + "num_tokens": 42926306.0, + "step": 1665 + }, + { + "epoch": 0.18295629255435977, + "grad_norm": 2.2892611026763916, + "learning_rate": 6.094436310395315e-07, + "loss": 1.0154, + "mean_token_accuracy": 0.6929370760917664, + "num_tokens": 42951568.0, + "step": 1666 + }, + { + "epoch": 0.18306611025697342, + "grad_norm": 2.40297532081604, + "learning_rate": 6.09809663250366e-07, + "loss": 1.0953, + "mean_token_accuracy": 0.6875419616699219, + "num_tokens": 42975783.0, + "step": 1667 + }, + { + "epoch": 0.1831759279595871, + "grad_norm": 2.1667110919952393, + "learning_rate": 6.101756954612005e-07, + "loss": 1.1179, + "mean_token_accuracy": 0.6715439558029175, + "num_tokens": 43004209.0, + "step": 1668 + }, + { + "epoch": 0.18328574566220074, + "grad_norm": 2.384671211242676, + "learning_rate": 6.105417276720352e-07, + "loss": 1.033, + "mean_token_accuracy": 0.6870214939117432, + "num_tokens": 43028852.0, + "step": 1669 + }, + { + "epoch": 0.1833955633648144, + "grad_norm": 2.377577543258667, + "learning_rate": 6.109077598828697e-07, + "loss": 1.0037, + "mean_token_accuracy": 0.6943919658660889, + "num_tokens": 43051679.0, + "step": 1670 + }, + { + "epoch": 0.18350538106742806, + "grad_norm": 2.2308382987976074, + "learning_rate": 6.112737920937042e-07, + "loss": 1.0924, + "mean_token_accuracy": 0.6774795651435852, + "num_tokens": 43079437.0, + "step": 1671 + }, + { + "epoch": 0.18361519877004173, + "grad_norm": 1.9992890357971191, + "learning_rate": 6.116398243045388e-07, + "loss": 1.0635, + "mean_token_accuracy": 0.6817002296447754, + "num_tokens": 43111882.0, + "step": 1672 + }, + { + "epoch": 0.1837250164726554, + "grad_norm": 2.3606908321380615, + "learning_rate": 6.120058565153733e-07, + "loss": 1.0685, + "mean_token_accuracy": 0.6916301250457764, + "num_tokens": 43138332.0, + "step": 1673 + }, + { + "epoch": 0.18383483417526905, + "grad_norm": 2.3212106227874756, + "learning_rate": 6.12371888726208e-07, + "loss": 1.0653, + "mean_token_accuracy": 0.6864703893661499, + "num_tokens": 43163166.0, + "step": 1674 + }, + { + "epoch": 0.18394465187788273, + "grad_norm": 2.1007606983184814, + "learning_rate": 6.127379209370425e-07, + "loss": 1.0184, + "mean_token_accuracy": 0.6976349353790283, + "num_tokens": 43192052.0, + "step": 1675 + }, + { + "epoch": 0.18405446958049637, + "grad_norm": 2.1975107192993164, + "learning_rate": 6.13103953147877e-07, + "loss": 0.9762, + "mean_token_accuracy": 0.7027170658111572, + "num_tokens": 43218645.0, + "step": 1676 + }, + { + "epoch": 0.18416428728311005, + "grad_norm": 2.459773063659668, + "learning_rate": 6.134699853587116e-07, + "loss": 1.0743, + "mean_token_accuracy": 0.6752883791923523, + "num_tokens": 43241509.0, + "step": 1677 + }, + { + "epoch": 0.1842741049857237, + "grad_norm": 2.493513822555542, + "learning_rate": 6.138360175695461e-07, + "loss": 1.0153, + "mean_token_accuracy": 0.6947500109672546, + "num_tokens": 43264698.0, + "step": 1678 + }, + { + "epoch": 0.18438392268833736, + "grad_norm": 2.4801440238952637, + "learning_rate": 6.142020497803806e-07, + "loss": 0.9871, + "mean_token_accuracy": 0.7065014839172363, + "num_tokens": 43287337.0, + "step": 1679 + }, + { + "epoch": 0.184493740390951, + "grad_norm": 2.4525187015533447, + "learning_rate": 6.145680819912153e-07, + "loss": 1.0732, + "mean_token_accuracy": 0.6848723292350769, + "num_tokens": 43313430.0, + "step": 1680 + }, + { + "epoch": 0.18460355809356468, + "grad_norm": 2.3379693031311035, + "learning_rate": 6.149341142020498e-07, + "loss": 1.0464, + "mean_token_accuracy": 0.6885470151901245, + "num_tokens": 43337787.0, + "step": 1681 + }, + { + "epoch": 0.18471337579617833, + "grad_norm": 1.9952267408370972, + "learning_rate": 6.153001464128843e-07, + "loss": 1.098, + "mean_token_accuracy": 0.674980103969574, + "num_tokens": 43373378.0, + "step": 1682 + }, + { + "epoch": 0.184823193498792, + "grad_norm": 1.9986106157302856, + "learning_rate": 6.156661786237189e-07, + "loss": 0.956, + "mean_token_accuracy": 0.7098100185394287, + "num_tokens": 43404353.0, + "step": 1683 + }, + { + "epoch": 0.18493301120140568, + "grad_norm": 2.1110901832580566, + "learning_rate": 6.160322108345534e-07, + "loss": 1.0883, + "mean_token_accuracy": 0.6806181073188782, + "num_tokens": 43433934.0, + "step": 1684 + }, + { + "epoch": 0.18504282890401932, + "grad_norm": 1.9634453058242798, + "learning_rate": 6.16398243045388e-07, + "loss": 0.985, + "mean_token_accuracy": 0.6991365551948547, + "num_tokens": 43466980.0, + "step": 1685 + }, + { + "epoch": 0.185152646606633, + "grad_norm": 2.1980926990509033, + "learning_rate": 6.167642752562226e-07, + "loss": 1.016, + "mean_token_accuracy": 0.7036987543106079, + "num_tokens": 43494439.0, + "step": 1686 + }, + { + "epoch": 0.18526246430924664, + "grad_norm": 2.234251022338867, + "learning_rate": 6.171303074670571e-07, + "loss": 1.1315, + "mean_token_accuracy": 0.6717097759246826, + "num_tokens": 43521827.0, + "step": 1687 + }, + { + "epoch": 0.18537228201186032, + "grad_norm": 2.452692747116089, + "learning_rate": 6.174963396778916e-07, + "loss": 1.0784, + "mean_token_accuracy": 0.6746580600738525, + "num_tokens": 43546844.0, + "step": 1688 + }, + { + "epoch": 0.18548209971447396, + "grad_norm": 2.1677474975585938, + "learning_rate": 6.178623718887262e-07, + "loss": 1.1069, + "mean_token_accuracy": 0.6681105494499207, + "num_tokens": 43579098.0, + "step": 1689 + }, + { + "epoch": 0.18559191741708764, + "grad_norm": 2.6092991828918457, + "learning_rate": 6.182284040995608e-07, + "loss": 0.9045, + "mean_token_accuracy": 0.7256506681442261, + "num_tokens": 43599155.0, + "step": 1690 + }, + { + "epoch": 0.18570173511970128, + "grad_norm": 2.443331003189087, + "learning_rate": 6.185944363103954e-07, + "loss": 1.0092, + "mean_token_accuracy": 0.6992776989936829, + "num_tokens": 43622669.0, + "step": 1691 + }, + { + "epoch": 0.18581155282231496, + "grad_norm": 2.379070997238159, + "learning_rate": 6.189604685212299e-07, + "loss": 0.9985, + "mean_token_accuracy": 0.6989597678184509, + "num_tokens": 43646422.0, + "step": 1692 + }, + { + "epoch": 0.18592137052492863, + "grad_norm": 2.0670082569122314, + "learning_rate": 6.193265007320644e-07, + "loss": 1.0005, + "mean_token_accuracy": 0.69953453540802, + "num_tokens": 43676400.0, + "step": 1693 + }, + { + "epoch": 0.18603118822754228, + "grad_norm": 2.6176586151123047, + "learning_rate": 6.19692532942899e-07, + "loss": 1.0079, + "mean_token_accuracy": 0.6988248825073242, + "num_tokens": 43697193.0, + "step": 1694 + }, + { + "epoch": 0.18614100593015595, + "grad_norm": 2.434765577316284, + "learning_rate": 6.200585651537335e-07, + "loss": 1.047, + "mean_token_accuracy": 0.6882709860801697, + "num_tokens": 43721079.0, + "step": 1695 + }, + { + "epoch": 0.1862508236327696, + "grad_norm": 2.3654539585113525, + "learning_rate": 6.204245973645681e-07, + "loss": 1.0395, + "mean_token_accuracy": 0.6948114037513733, + "num_tokens": 43747631.0, + "step": 1696 + }, + { + "epoch": 0.18636064133538327, + "grad_norm": 2.153515338897705, + "learning_rate": 6.207906295754027e-07, + "loss": 1.0096, + "mean_token_accuracy": 0.6963958740234375, + "num_tokens": 43778341.0, + "step": 1697 + }, + { + "epoch": 0.18647045903799692, + "grad_norm": 2.3957231044769287, + "learning_rate": 6.211566617862372e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.6979448795318604, + "num_tokens": 43802076.0, + "step": 1698 + }, + { + "epoch": 0.1865802767406106, + "grad_norm": 2.2743844985961914, + "learning_rate": 6.215226939970716e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6784523725509644, + "num_tokens": 43829357.0, + "step": 1699 + }, + { + "epoch": 0.18669009444322424, + "grad_norm": 2.4533119201660156, + "learning_rate": 6.218887262079063e-07, + "loss": 0.9577, + "mean_token_accuracy": 0.7044068574905396, + "num_tokens": 43852687.0, + "step": 1700 + }, + { + "epoch": 0.1867999121458379, + "grad_norm": 2.3536524772644043, + "learning_rate": 6.222547584187409e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.691655158996582, + "num_tokens": 43877297.0, + "step": 1701 + }, + { + "epoch": 0.18690972984845158, + "grad_norm": 2.4567558765411377, + "learning_rate": 6.226207906295753e-07, + "loss": 0.9951, + "mean_token_accuracy": 0.6958112120628357, + "num_tokens": 43899048.0, + "step": 1702 + }, + { + "epoch": 0.18701954755106523, + "grad_norm": 2.4092702865600586, + "learning_rate": 6.2298682284041e-07, + "loss": 0.9276, + "mean_token_accuracy": 0.716286301612854, + "num_tokens": 43921618.0, + "step": 1703 + }, + { + "epoch": 0.1871293652536789, + "grad_norm": 1.993667721748352, + "learning_rate": 6.233528550512444e-07, + "loss": 1.0798, + "mean_token_accuracy": 0.6892043352127075, + "num_tokens": 43952213.0, + "step": 1704 + }, + { + "epoch": 0.18723918295629255, + "grad_norm": 2.1537108421325684, + "learning_rate": 6.237188872620789e-07, + "loss": 1.0086, + "mean_token_accuracy": 0.7020379900932312, + "num_tokens": 43979710.0, + "step": 1705 + }, + { + "epoch": 0.18734900065890622, + "grad_norm": 2.225947618484497, + "learning_rate": 6.240849194729137e-07, + "loss": 1.0102, + "mean_token_accuracy": 0.6960881352424622, + "num_tokens": 44005736.0, + "step": 1706 + }, + { + "epoch": 0.18745881836151987, + "grad_norm": 2.007962703704834, + "learning_rate": 6.244509516837481e-07, + "loss": 1.1087, + "mean_token_accuracy": 0.6790338158607483, + "num_tokens": 44036907.0, + "step": 1707 + }, + { + "epoch": 0.18756863606413354, + "grad_norm": 2.0031909942626953, + "learning_rate": 6.248169838945828e-07, + "loss": 1.0655, + "mean_token_accuracy": 0.6807241439819336, + "num_tokens": 44067574.0, + "step": 1708 + }, + { + "epoch": 0.1876784537667472, + "grad_norm": 2.2977967262268066, + "learning_rate": 6.251830161054172e-07, + "loss": 1.0982, + "mean_token_accuracy": 0.674607515335083, + "num_tokens": 44092129.0, + "step": 1709 + }, + { + "epoch": 0.18778827146936086, + "grad_norm": 2.067491292953491, + "learning_rate": 6.255490483162517e-07, + "loss": 1.1195, + "mean_token_accuracy": 0.6730945110321045, + "num_tokens": 44124520.0, + "step": 1710 + }, + { + "epoch": 0.18789808917197454, + "grad_norm": 2.3470051288604736, + "learning_rate": 6.259150805270863e-07, + "loss": 1.0701, + "mean_token_accuracy": 0.6896623373031616, + "num_tokens": 44150189.0, + "step": 1711 + }, + { + "epoch": 0.18800790687458818, + "grad_norm": 2.548161268234253, + "learning_rate": 6.262811127379209e-07, + "loss": 0.9825, + "mean_token_accuracy": 0.7056487798690796, + "num_tokens": 44171622.0, + "step": 1712 + }, + { + "epoch": 0.18811772457720186, + "grad_norm": 2.3294613361358643, + "learning_rate": 6.266471449487554e-07, + "loss": 1.0736, + "mean_token_accuracy": 0.691084623336792, + "num_tokens": 44196722.0, + "step": 1713 + }, + { + "epoch": 0.1882275422798155, + "grad_norm": 2.0390713214874268, + "learning_rate": 6.2701317715959e-07, + "loss": 1.1113, + "mean_token_accuracy": 0.6692349314689636, + "num_tokens": 44228788.0, + "step": 1714 + }, + { + "epoch": 0.18833735998242918, + "grad_norm": 2.488115072250366, + "learning_rate": 6.273792093704245e-07, + "loss": 1.0581, + "mean_token_accuracy": 0.6841093301773071, + "num_tokens": 44253946.0, + "step": 1715 + }, + { + "epoch": 0.18844717768504282, + "grad_norm": 2.3999335765838623, + "learning_rate": 6.27745241581259e-07, + "loss": 1.0617, + "mean_token_accuracy": 0.6827631592750549, + "num_tokens": 44277666.0, + "step": 1716 + }, + { + "epoch": 0.1885569953876565, + "grad_norm": 2.0870437622070312, + "learning_rate": 6.281112737920937e-07, + "loss": 0.9944, + "mean_token_accuracy": 0.7014222145080566, + "num_tokens": 44306333.0, + "step": 1717 + }, + { + "epoch": 0.18866681309027014, + "grad_norm": 2.2046267986297607, + "learning_rate": 6.284773060029282e-07, + "loss": 1.0783, + "mean_token_accuracy": 0.6763390302658081, + "num_tokens": 44333933.0, + "step": 1718 + }, + { + "epoch": 0.18877663079288381, + "grad_norm": 2.3510398864746094, + "learning_rate": 6.288433382137627e-07, + "loss": 0.9919, + "mean_token_accuracy": 0.6988526582717896, + "num_tokens": 44358642.0, + "step": 1719 + }, + { + "epoch": 0.18888644849549746, + "grad_norm": 2.482846975326538, + "learning_rate": 6.292093704245973e-07, + "loss": 1.0245, + "mean_token_accuracy": 0.6877254843711853, + "num_tokens": 44381083.0, + "step": 1720 + }, + { + "epoch": 0.18899626619811113, + "grad_norm": 2.2712464332580566, + "learning_rate": 6.295754026354318e-07, + "loss": 1.0757, + "mean_token_accuracy": 0.6846674680709839, + "num_tokens": 44406847.0, + "step": 1721 + }, + { + "epoch": 0.1891060839007248, + "grad_norm": 2.2200143337249756, + "learning_rate": 6.299414348462665e-07, + "loss": 1.0622, + "mean_token_accuracy": 0.6776686906814575, + "num_tokens": 44433809.0, + "step": 1722 + }, + { + "epoch": 0.18921590160333845, + "grad_norm": 2.4284136295318604, + "learning_rate": 6.30307467057101e-07, + "loss": 1.0526, + "mean_token_accuracy": 0.6838750839233398, + "num_tokens": 44458042.0, + "step": 1723 + }, + { + "epoch": 0.18932571930595213, + "grad_norm": 2.6901509761810303, + "learning_rate": 6.306734992679355e-07, + "loss": 1.0051, + "mean_token_accuracy": 0.6970158219337463, + "num_tokens": 44478035.0, + "step": 1724 + }, + { + "epoch": 0.18943553700856577, + "grad_norm": 1.9727234840393066, + "learning_rate": 6.310395314787701e-07, + "loss": 1.1102, + "mean_token_accuracy": 0.6674429178237915, + "num_tokens": 44513865.0, + "step": 1725 + }, + { + "epoch": 0.18954535471117945, + "grad_norm": 2.5994062423706055, + "learning_rate": 6.314055636896046e-07, + "loss": 1.0287, + "mean_token_accuracy": 0.6993454098701477, + "num_tokens": 44534733.0, + "step": 1726 + }, + { + "epoch": 0.1896551724137931, + "grad_norm": 2.2455666065216064, + "learning_rate": 6.317715959004391e-07, + "loss": 0.9803, + "mean_token_accuracy": 0.7028515934944153, + "num_tokens": 44560374.0, + "step": 1727 + }, + { + "epoch": 0.18976499011640677, + "grad_norm": 2.749329090118408, + "learning_rate": 6.321376281112738e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.7092772126197815, + "num_tokens": 44579550.0, + "step": 1728 + }, + { + "epoch": 0.1898748078190204, + "grad_norm": 2.195249557495117, + "learning_rate": 6.325036603221083e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.6815152764320374, + "num_tokens": 44605935.0, + "step": 1729 + }, + { + "epoch": 0.1899846255216341, + "grad_norm": 2.4660189151763916, + "learning_rate": 6.328696925329428e-07, + "loss": 1.0229, + "mean_token_accuracy": 0.6950715780258179, + "num_tokens": 44628919.0, + "step": 1730 + }, + { + "epoch": 0.19009444322424776, + "grad_norm": 2.223597764968872, + "learning_rate": 6.332357247437774e-07, + "loss": 1.0518, + "mean_token_accuracy": 0.681980311870575, + "num_tokens": 44656105.0, + "step": 1731 + }, + { + "epoch": 0.1902042609268614, + "grad_norm": 2.1221764087677, + "learning_rate": 6.336017569546119e-07, + "loss": 1.0601, + "mean_token_accuracy": 0.6785813570022583, + "num_tokens": 44686900.0, + "step": 1732 + }, + { + "epoch": 0.19031407862947508, + "grad_norm": 2.546833038330078, + "learning_rate": 6.339677891654465e-07, + "loss": 1.0048, + "mean_token_accuracy": 0.6979748010635376, + "num_tokens": 44710054.0, + "step": 1733 + }, + { + "epoch": 0.19042389633208873, + "grad_norm": 2.677579402923584, + "learning_rate": 6.343338213762811e-07, + "loss": 0.9874, + "mean_token_accuracy": 0.7057231068611145, + "num_tokens": 44730033.0, + "step": 1734 + }, + { + "epoch": 0.1905337140347024, + "grad_norm": 2.290006160736084, + "learning_rate": 6.346998535871156e-07, + "loss": 0.9539, + "mean_token_accuracy": 0.707557201385498, + "num_tokens": 44756831.0, + "step": 1735 + }, + { + "epoch": 0.19064353173731605, + "grad_norm": 2.265625, + "learning_rate": 6.350658857979501e-07, + "loss": 1.0322, + "mean_token_accuracy": 0.6889469623565674, + "num_tokens": 44782982.0, + "step": 1736 + }, + { + "epoch": 0.19075334943992972, + "grad_norm": 2.257507801055908, + "learning_rate": 6.354319180087847e-07, + "loss": 1.1553, + "mean_token_accuracy": 0.6659512519836426, + "num_tokens": 44811212.0, + "step": 1737 + }, + { + "epoch": 0.19086316714254337, + "grad_norm": 2.4844253063201904, + "learning_rate": 6.357979502196193e-07, + "loss": 1.0084, + "mean_token_accuracy": 0.6997947096824646, + "num_tokens": 44833233.0, + "step": 1738 + }, + { + "epoch": 0.19097298484515704, + "grad_norm": 2.0847887992858887, + "learning_rate": 6.361639824304539e-07, + "loss": 1.0554, + "mean_token_accuracy": 0.68880695104599, + "num_tokens": 44861753.0, + "step": 1739 + }, + { + "epoch": 0.1910828025477707, + "grad_norm": 2.215519666671753, + "learning_rate": 6.365300146412884e-07, + "loss": 1.0383, + "mean_token_accuracy": 0.6984747648239136, + "num_tokens": 44887824.0, + "step": 1740 + }, + { + "epoch": 0.19119262025038436, + "grad_norm": 2.2528514862060547, + "learning_rate": 6.368960468521229e-07, + "loss": 1.0239, + "mean_token_accuracy": 0.6860471963882446, + "num_tokens": 44914371.0, + "step": 1741 + }, + { + "epoch": 0.19130243795299803, + "grad_norm": 2.0718657970428467, + "learning_rate": 6.372620790629575e-07, + "loss": 1.0909, + "mean_token_accuracy": 0.6755315661430359, + "num_tokens": 44943379.0, + "step": 1742 + }, + { + "epoch": 0.19141225565561168, + "grad_norm": 2.4291305541992188, + "learning_rate": 6.37628111273792e-07, + "loss": 1.108, + "mean_token_accuracy": 0.6802940368652344, + "num_tokens": 44967204.0, + "step": 1743 + }, + { + "epoch": 0.19152207335822535, + "grad_norm": 2.184821367263794, + "learning_rate": 6.379941434846266e-07, + "loss": 0.9926, + "mean_token_accuracy": 0.6988977193832397, + "num_tokens": 44993218.0, + "step": 1744 + }, + { + "epoch": 0.191631891060839, + "grad_norm": 2.403404951095581, + "learning_rate": 6.383601756954612e-07, + "loss": 1.0142, + "mean_token_accuracy": 0.6934701204299927, + "num_tokens": 45016771.0, + "step": 1745 + }, + { + "epoch": 0.19174170876345267, + "grad_norm": 2.450913429260254, + "learning_rate": 6.387262079062957e-07, + "loss": 1.0566, + "mean_token_accuracy": 0.6854035258293152, + "num_tokens": 45041696.0, + "step": 1746 + }, + { + "epoch": 0.19185152646606632, + "grad_norm": 2.2651784420013428, + "learning_rate": 6.390922401171302e-07, + "loss": 1.0682, + "mean_token_accuracy": 0.6840733885765076, + "num_tokens": 45067615.0, + "step": 1747 + }, + { + "epoch": 0.19196134416868, + "grad_norm": 2.44230580329895, + "learning_rate": 6.394582723279648e-07, + "loss": 1.0681, + "mean_token_accuracy": 0.6919176578521729, + "num_tokens": 45092065.0, + "step": 1748 + }, + { + "epoch": 0.19207116187129367, + "grad_norm": 2.3880364894866943, + "learning_rate": 6.398243045387994e-07, + "loss": 1.085, + "mean_token_accuracy": 0.6823176145553589, + "num_tokens": 45116555.0, + "step": 1749 + }, + { + "epoch": 0.1921809795739073, + "grad_norm": 2.222748279571533, + "learning_rate": 6.401903367496339e-07, + "loss": 1.0937, + "mean_token_accuracy": 0.679986298084259, + "num_tokens": 45142968.0, + "step": 1750 + }, + { + "epoch": 0.19229079727652099, + "grad_norm": 2.594987154006958, + "learning_rate": 6.405563689604685e-07, + "loss": 0.9798, + "mean_token_accuracy": 0.7017030715942383, + "num_tokens": 45162421.0, + "step": 1751 + }, + { + "epoch": 0.19240061497913463, + "grad_norm": 2.240358829498291, + "learning_rate": 6.40922401171303e-07, + "loss": 0.9503, + "mean_token_accuracy": 0.7111507654190063, + "num_tokens": 45188853.0, + "step": 1752 + }, + { + "epoch": 0.1925104326817483, + "grad_norm": 2.355940818786621, + "learning_rate": 6.412884333821375e-07, + "loss": 1.0898, + "mean_token_accuracy": 0.6958510279655457, + "num_tokens": 45211908.0, + "step": 1753 + }, + { + "epoch": 0.19262025038436195, + "grad_norm": 2.2686946392059326, + "learning_rate": 6.416544655929722e-07, + "loss": 0.9508, + "mean_token_accuracy": 0.7123785018920898, + "num_tokens": 45235897.0, + "step": 1754 + }, + { + "epoch": 0.19273006808697563, + "grad_norm": 2.1491119861602783, + "learning_rate": 6.420204978038067e-07, + "loss": 1.1879, + "mean_token_accuracy": 0.6540006399154663, + "num_tokens": 45266668.0, + "step": 1755 + }, + { + "epoch": 0.19283988578958927, + "grad_norm": 2.3308839797973633, + "learning_rate": 6.423865300146413e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.6735117435455322, + "num_tokens": 45291961.0, + "step": 1756 + }, + { + "epoch": 0.19294970349220295, + "grad_norm": 2.1147358417510986, + "learning_rate": 6.427525622254758e-07, + "loss": 1.0498, + "mean_token_accuracy": 0.6888605356216431, + "num_tokens": 45321224.0, + "step": 1757 + }, + { + "epoch": 0.1930595211948166, + "grad_norm": 2.2034647464752197, + "learning_rate": 6.431185944363103e-07, + "loss": 1.0402, + "mean_token_accuracy": 0.69422447681427, + "num_tokens": 45349120.0, + "step": 1758 + }, + { + "epoch": 0.19316933889743026, + "grad_norm": 2.44146466255188, + "learning_rate": 6.434846266471449e-07, + "loss": 0.9591, + "mean_token_accuracy": 0.709479033946991, + "num_tokens": 45370609.0, + "step": 1759 + }, + { + "epoch": 0.19327915660004394, + "grad_norm": 2.0355072021484375, + "learning_rate": 6.438506588579795e-07, + "loss": 1.0207, + "mean_token_accuracy": 0.6950353384017944, + "num_tokens": 45402004.0, + "step": 1760 + }, + { + "epoch": 0.19338897430265758, + "grad_norm": 2.559037923812866, + "learning_rate": 6.44216691068814e-07, + "loss": 0.9701, + "mean_token_accuracy": 0.7072434425354004, + "num_tokens": 45423421.0, + "step": 1761 + }, + { + "epoch": 0.19349879200527126, + "grad_norm": 2.140207290649414, + "learning_rate": 6.445827232796486e-07, + "loss": 0.9751, + "mean_token_accuracy": 0.7090973854064941, + "num_tokens": 45449405.0, + "step": 1762 + }, + { + "epoch": 0.1936086097078849, + "grad_norm": 2.496297597885132, + "learning_rate": 6.449487554904831e-07, + "loss": 0.9842, + "mean_token_accuracy": 0.6943784952163696, + "num_tokens": 45470264.0, + "step": 1763 + }, + { + "epoch": 0.19371842741049858, + "grad_norm": 2.033134698867798, + "learning_rate": 6.453147877013176e-07, + "loss": 0.9828, + "mean_token_accuracy": 0.7095232605934143, + "num_tokens": 45501935.0, + "step": 1764 + }, + { + "epoch": 0.19382824511311222, + "grad_norm": 2.2631540298461914, + "learning_rate": 6.456808199121523e-07, + "loss": 1.0792, + "mean_token_accuracy": 0.6811703443527222, + "num_tokens": 45528933.0, + "step": 1765 + }, + { + "epoch": 0.1939380628157259, + "grad_norm": 2.3987276554107666, + "learning_rate": 6.460468521229868e-07, + "loss": 0.9392, + "mean_token_accuracy": 0.7133880257606506, + "num_tokens": 45551417.0, + "step": 1766 + }, + { + "epoch": 0.19404788051833954, + "grad_norm": 2.0230932235717773, + "learning_rate": 6.464128843338213e-07, + "loss": 1.0318, + "mean_token_accuracy": 0.692611038684845, + "num_tokens": 45582175.0, + "step": 1767 + }, + { + "epoch": 0.19415769822095322, + "grad_norm": 2.4033596515655518, + "learning_rate": 6.467789165446559e-07, + "loss": 1.1004, + "mean_token_accuracy": 0.6719163656234741, + "num_tokens": 45606661.0, + "step": 1768 + }, + { + "epoch": 0.1942675159235669, + "grad_norm": 2.3769874572753906, + "learning_rate": 6.471449487554904e-07, + "loss": 1.0598, + "mean_token_accuracy": 0.6818756461143494, + "num_tokens": 45632245.0, + "step": 1769 + }, + { + "epoch": 0.19437733362618054, + "grad_norm": 1.9385191202163696, + "learning_rate": 6.475109809663251e-07, + "loss": 1.0773, + "mean_token_accuracy": 0.6831281185150146, + "num_tokens": 45666175.0, + "step": 1770 + }, + { + "epoch": 0.1944871513287942, + "grad_norm": 2.318735122680664, + "learning_rate": 6.478770131771596e-07, + "loss": 1.0407, + "mean_token_accuracy": 0.6982566118240356, + "num_tokens": 45692819.0, + "step": 1771 + }, + { + "epoch": 0.19459696903140786, + "grad_norm": 2.092839002609253, + "learning_rate": 6.482430453879941e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.6932083368301392, + "num_tokens": 45721537.0, + "step": 1772 + }, + { + "epoch": 0.19470678673402153, + "grad_norm": 2.4718873500823975, + "learning_rate": 6.486090775988287e-07, + "loss": 1.0791, + "mean_token_accuracy": 0.687586784362793, + "num_tokens": 45745690.0, + "step": 1773 + }, + { + "epoch": 0.19481660443663518, + "grad_norm": 2.3850326538085938, + "learning_rate": 6.489751098096632e-07, + "loss": 1.1064, + "mean_token_accuracy": 0.6743960976600647, + "num_tokens": 45771981.0, + "step": 1774 + }, + { + "epoch": 0.19492642213924885, + "grad_norm": 2.4331257343292236, + "learning_rate": 6.493411420204977e-07, + "loss": 1.0363, + "mean_token_accuracy": 0.6946683526039124, + "num_tokens": 45796644.0, + "step": 1775 + }, + { + "epoch": 0.1950362398418625, + "grad_norm": 2.780794620513916, + "learning_rate": 6.497071742313324e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.6781989932060242, + "num_tokens": 45815886.0, + "step": 1776 + }, + { + "epoch": 0.19514605754447617, + "grad_norm": 2.2134711742401123, + "learning_rate": 6.500732064421669e-07, + "loss": 1.0482, + "mean_token_accuracy": 0.6879653930664062, + "num_tokens": 45843574.0, + "step": 1777 + }, + { + "epoch": 0.19525587524708984, + "grad_norm": 2.194843292236328, + "learning_rate": 6.504392386530014e-07, + "loss": 1.0154, + "mean_token_accuracy": 0.6985284686088562, + "num_tokens": 45870615.0, + "step": 1778 + }, + { + "epoch": 0.1953656929497035, + "grad_norm": 2.451476573944092, + "learning_rate": 6.50805270863836e-07, + "loss": 1.0233, + "mean_token_accuracy": 0.6969709396362305, + "num_tokens": 45892574.0, + "step": 1779 + }, + { + "epoch": 0.19547551065231716, + "grad_norm": 2.3938252925872803, + "learning_rate": 6.511713030746705e-07, + "loss": 1.0927, + "mean_token_accuracy": 0.6762371063232422, + "num_tokens": 45917428.0, + "step": 1780 + }, + { + "epoch": 0.1955853283549308, + "grad_norm": 2.067688226699829, + "learning_rate": 6.515373352855051e-07, + "loss": 1.0182, + "mean_token_accuracy": 0.691169023513794, + "num_tokens": 45947191.0, + "step": 1781 + }, + { + "epoch": 0.19569514605754448, + "grad_norm": 2.228858709335327, + "learning_rate": 6.519033674963397e-07, + "loss": 0.9953, + "mean_token_accuracy": 0.700087308883667, + "num_tokens": 45973562.0, + "step": 1782 + }, + { + "epoch": 0.19580496376015813, + "grad_norm": 2.2940962314605713, + "learning_rate": 6.522693997071742e-07, + "loss": 1.0022, + "mean_token_accuracy": 0.7025551199913025, + "num_tokens": 46000779.0, + "step": 1783 + }, + { + "epoch": 0.1959147814627718, + "grad_norm": 2.284043550491333, + "learning_rate": 6.526354319180087e-07, + "loss": 1.0347, + "mean_token_accuracy": 0.6932264566421509, + "num_tokens": 46025924.0, + "step": 1784 + }, + { + "epoch": 0.19602459916538545, + "grad_norm": 2.2395691871643066, + "learning_rate": 6.530014641288433e-07, + "loss": 1.0455, + "mean_token_accuracy": 0.686177134513855, + "num_tokens": 46052335.0, + "step": 1785 + }, + { + "epoch": 0.19613441686799912, + "grad_norm": 2.32196307182312, + "learning_rate": 6.533674963396779e-07, + "loss": 0.9707, + "mean_token_accuracy": 0.7078368663787842, + "num_tokens": 46075495.0, + "step": 1786 + }, + { + "epoch": 0.1962442345706128, + "grad_norm": 2.6468000411987305, + "learning_rate": 6.537335285505125e-07, + "loss": 0.9912, + "mean_token_accuracy": 0.697124719619751, + "num_tokens": 46094772.0, + "step": 1787 + }, + { + "epoch": 0.19635405227322644, + "grad_norm": 2.292501211166382, + "learning_rate": 6.54099560761347e-07, + "loss": 1.0985, + "mean_token_accuracy": 0.6734908819198608, + "num_tokens": 46123044.0, + "step": 1788 + }, + { + "epoch": 0.19646386997584012, + "grad_norm": 2.3284080028533936, + "learning_rate": 6.544655929721815e-07, + "loss": 1.0327, + "mean_token_accuracy": 0.6921846866607666, + "num_tokens": 46146887.0, + "step": 1789 + }, + { + "epoch": 0.19657368767845376, + "grad_norm": 2.489703416824341, + "learning_rate": 6.548316251830161e-07, + "loss": 1.0255, + "mean_token_accuracy": 0.7002179026603699, + "num_tokens": 46168260.0, + "step": 1790 + }, + { + "epoch": 0.19668350538106744, + "grad_norm": 2.4890615940093994, + "learning_rate": 6.551976573938506e-07, + "loss": 1.0835, + "mean_token_accuracy": 0.6855080723762512, + "num_tokens": 46190266.0, + "step": 1791 + }, + { + "epoch": 0.19679332308368108, + "grad_norm": 2.347811222076416, + "learning_rate": 6.555636896046852e-07, + "loss": 1.0511, + "mean_token_accuracy": 0.6873109936714172, + "num_tokens": 46214139.0, + "step": 1792 + }, + { + "epoch": 0.19690314078629476, + "grad_norm": 2.6516175270080566, + "learning_rate": 6.559297218155198e-07, + "loss": 1.0538, + "mean_token_accuracy": 0.6916270852088928, + "num_tokens": 46235396.0, + "step": 1793 + }, + { + "epoch": 0.1970129584889084, + "grad_norm": 2.2366256713867188, + "learning_rate": 6.562957540263543e-07, + "loss": 1.0156, + "mean_token_accuracy": 0.6940175294876099, + "num_tokens": 46261539.0, + "step": 1794 + }, + { + "epoch": 0.19712277619152208, + "grad_norm": 2.291290760040283, + "learning_rate": 6.566617862371888e-07, + "loss": 1.0195, + "mean_token_accuracy": 0.6957566738128662, + "num_tokens": 46286833.0, + "step": 1795 + }, + { + "epoch": 0.19723259389413572, + "grad_norm": 2.2405126094818115, + "learning_rate": 6.570278184480234e-07, + "loss": 1.0648, + "mean_token_accuracy": 0.6826828718185425, + "num_tokens": 46313733.0, + "step": 1796 + }, + { + "epoch": 0.1973424115967494, + "grad_norm": 2.314321994781494, + "learning_rate": 6.57393850658858e-07, + "loss": 1.067, + "mean_token_accuracy": 0.684144139289856, + "num_tokens": 46337087.0, + "step": 1797 + }, + { + "epoch": 0.19745222929936307, + "grad_norm": 2.7761497497558594, + "learning_rate": 6.577598828696925e-07, + "loss": 0.9647, + "mean_token_accuracy": 0.7025664448738098, + "num_tokens": 46356648.0, + "step": 1798 + }, + { + "epoch": 0.19756204700197671, + "grad_norm": 2.351515769958496, + "learning_rate": 6.581259150805271e-07, + "loss": 0.935, + "mean_token_accuracy": 0.7124844789505005, + "num_tokens": 46378939.0, + "step": 1799 + }, + { + "epoch": 0.1976718647045904, + "grad_norm": 2.13641619682312, + "learning_rate": 6.584919472913616e-07, + "loss": 1.0369, + "mean_token_accuracy": 0.6860105395317078, + "num_tokens": 46408171.0, + "step": 1800 + }, + { + "epoch": 0.19778168240720403, + "grad_norm": 2.432846784591675, + "learning_rate": 6.588579795021961e-07, + "loss": 0.9937, + "mean_token_accuracy": 0.7022329568862915, + "num_tokens": 46431563.0, + "step": 1801 + }, + { + "epoch": 0.1978915001098177, + "grad_norm": 2.407905101776123, + "learning_rate": 6.592240117130308e-07, + "loss": 1.0238, + "mean_token_accuracy": 0.700808048248291, + "num_tokens": 46455897.0, + "step": 1802 + }, + { + "epoch": 0.19800131781243135, + "grad_norm": 2.142205238342285, + "learning_rate": 6.595900439238653e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.6754710078239441, + "num_tokens": 46485421.0, + "step": 1803 + }, + { + "epoch": 0.19811113551504503, + "grad_norm": 2.734729528427124, + "learning_rate": 6.599560761346999e-07, + "loss": 0.9979, + "mean_token_accuracy": 0.6995112895965576, + "num_tokens": 46504392.0, + "step": 1804 + }, + { + "epoch": 0.19822095321765867, + "grad_norm": 2.4543752670288086, + "learning_rate": 6.603221083455344e-07, + "loss": 0.9468, + "mean_token_accuracy": 0.7147172689437866, + "num_tokens": 46526987.0, + "step": 1805 + }, + { + "epoch": 0.19833077092027235, + "grad_norm": 2.33254075050354, + "learning_rate": 6.606881405563689e-07, + "loss": 1.0102, + "mean_token_accuracy": 0.6973426938056946, + "num_tokens": 46552167.0, + "step": 1806 + }, + { + "epoch": 0.19844058862288602, + "grad_norm": 2.331977128982544, + "learning_rate": 6.610541727672035e-07, + "loss": 1.0309, + "mean_token_accuracy": 0.6869221925735474, + "num_tokens": 46578431.0, + "step": 1807 + }, + { + "epoch": 0.19855040632549967, + "grad_norm": 2.448321580886841, + "learning_rate": 6.614202049780381e-07, + "loss": 0.9137, + "mean_token_accuracy": 0.7183874249458313, + "num_tokens": 46597882.0, + "step": 1808 + }, + { + "epoch": 0.19866022402811334, + "grad_norm": 2.1407055854797363, + "learning_rate": 6.617862371888726e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.6915563344955444, + "num_tokens": 46627173.0, + "step": 1809 + }, + { + "epoch": 0.198770041730727, + "grad_norm": 2.3323943614959717, + "learning_rate": 6.621522693997072e-07, + "loss": 1.043, + "mean_token_accuracy": 0.695023238658905, + "num_tokens": 46652241.0, + "step": 1810 + }, + { + "epoch": 0.19887985943334066, + "grad_norm": 2.211742877960205, + "learning_rate": 6.625183016105417e-07, + "loss": 1.0851, + "mean_token_accuracy": 0.6759032011032104, + "num_tokens": 46681172.0, + "step": 1811 + }, + { + "epoch": 0.1989896771359543, + "grad_norm": 2.054898500442505, + "learning_rate": 6.628843338213762e-07, + "loss": 1.0673, + "mean_token_accuracy": 0.6833339929580688, + "num_tokens": 46711892.0, + "step": 1812 + }, + { + "epoch": 0.19909949483856798, + "grad_norm": 1.995814323425293, + "learning_rate": 6.632503660322109e-07, + "loss": 1.0685, + "mean_token_accuracy": 0.6794617176055908, + "num_tokens": 46744839.0, + "step": 1813 + }, + { + "epoch": 0.19920931254118163, + "grad_norm": 2.3209259510040283, + "learning_rate": 6.636163982430454e-07, + "loss": 0.934, + "mean_token_accuracy": 0.7063193321228027, + "num_tokens": 46768706.0, + "step": 1814 + }, + { + "epoch": 0.1993191302437953, + "grad_norm": 2.645171880722046, + "learning_rate": 6.639824304538799e-07, + "loss": 0.9876, + "mean_token_accuracy": 0.7019503116607666, + "num_tokens": 46789029.0, + "step": 1815 + }, + { + "epoch": 0.19942894794640897, + "grad_norm": 2.153186798095703, + "learning_rate": 6.643484626647145e-07, + "loss": 1.0691, + "mean_token_accuracy": 0.6821576356887817, + "num_tokens": 46819385.0, + "step": 1816 + }, + { + "epoch": 0.19953876564902262, + "grad_norm": 2.2973926067352295, + "learning_rate": 6.64714494875549e-07, + "loss": 1.051, + "mean_token_accuracy": 0.6854262351989746, + "num_tokens": 46845617.0, + "step": 1817 + }, + { + "epoch": 0.1996485833516363, + "grad_norm": 2.4532225131988525, + "learning_rate": 6.650805270863836e-07, + "loss": 0.977, + "mean_token_accuracy": 0.7029972076416016, + "num_tokens": 46867168.0, + "step": 1818 + }, + { + "epoch": 0.19975840105424994, + "grad_norm": 2.439206838607788, + "learning_rate": 6.654465592972182e-07, + "loss": 0.9782, + "mean_token_accuracy": 0.7053508758544922, + "num_tokens": 46893199.0, + "step": 1819 + }, + { + "epoch": 0.1998682187568636, + "grad_norm": 2.944164991378784, + "learning_rate": 6.658125915080527e-07, + "loss": 1.0176, + "mean_token_accuracy": 0.6952248811721802, + "num_tokens": 46912484.0, + "step": 1820 + }, + { + "epoch": 0.19997803645947726, + "grad_norm": 1.9319146871566772, + "learning_rate": 6.661786237188873e-07, + "loss": 1.0997, + "mean_token_accuracy": 0.684721052646637, + "num_tokens": 46949071.0, + "step": 1821 + }, + { + "epoch": 0.20008785416209093, + "grad_norm": 2.2125422954559326, + "learning_rate": 6.665446559297218e-07, + "loss": 0.9886, + "mean_token_accuracy": 0.6973790526390076, + "num_tokens": 46976660.0, + "step": 1822 + }, + { + "epoch": 0.20019767186470458, + "grad_norm": 2.2563679218292236, + "learning_rate": 6.669106881405563e-07, + "loss": 0.9125, + "mean_token_accuracy": 0.7270331978797913, + "num_tokens": 47003264.0, + "step": 1823 + }, + { + "epoch": 0.20030748956731825, + "grad_norm": 1.9676249027252197, + "learning_rate": 6.67276720351391e-07, + "loss": 1.0331, + "mean_token_accuracy": 0.6872894763946533, + "num_tokens": 47036317.0, + "step": 1824 + }, + { + "epoch": 0.2004173072699319, + "grad_norm": 2.4576072692871094, + "learning_rate": 6.676427525622255e-07, + "loss": 1.0601, + "mean_token_accuracy": 0.6853722333908081, + "num_tokens": 47059959.0, + "step": 1825 + }, + { + "epoch": 0.20052712497254557, + "grad_norm": 2.3880462646484375, + "learning_rate": 6.6800878477306e-07, + "loss": 1.009, + "mean_token_accuracy": 0.6969733238220215, + "num_tokens": 47083996.0, + "step": 1826 + }, + { + "epoch": 0.20063694267515925, + "grad_norm": 2.147470712661743, + "learning_rate": 6.683748169838946e-07, + "loss": 1.0131, + "mean_token_accuracy": 0.7167603969573975, + "num_tokens": 47111746.0, + "step": 1827 + }, + { + "epoch": 0.2007467603777729, + "grad_norm": 2.2656302452087402, + "learning_rate": 6.687408491947291e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.685370683670044, + "num_tokens": 47137621.0, + "step": 1828 + }, + { + "epoch": 0.20085657808038657, + "grad_norm": 2.288611650466919, + "learning_rate": 6.691068814055637e-07, + "loss": 0.9869, + "mean_token_accuracy": 0.6992028951644897, + "num_tokens": 47162400.0, + "step": 1829 + }, + { + "epoch": 0.2009663957830002, + "grad_norm": 2.9208292961120605, + "learning_rate": 6.694729136163983e-07, + "loss": 0.9843, + "mean_token_accuracy": 0.7028524875640869, + "num_tokens": 47186479.0, + "step": 1830 + }, + { + "epoch": 0.20107621348561389, + "grad_norm": 2.0497429370880127, + "learning_rate": 6.698389458272328e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6883175373077393, + "num_tokens": 47217322.0, + "step": 1831 + }, + { + "epoch": 0.20118603118822753, + "grad_norm": 2.0169527530670166, + "learning_rate": 6.702049780380673e-07, + "loss": 1.0208, + "mean_token_accuracy": 0.6874312162399292, + "num_tokens": 47247815.0, + "step": 1832 + }, + { + "epoch": 0.2012958488908412, + "grad_norm": 2.284884452819824, + "learning_rate": 6.705710102489019e-07, + "loss": 1.0127, + "mean_token_accuracy": 0.6969946026802063, + "num_tokens": 47274328.0, + "step": 1833 + }, + { + "epoch": 0.20140566659345485, + "grad_norm": 2.0593788623809814, + "learning_rate": 6.709370424597365e-07, + "loss": 1.0297, + "mean_token_accuracy": 0.6875343322753906, + "num_tokens": 47303590.0, + "step": 1834 + }, + { + "epoch": 0.20151548429606853, + "grad_norm": 2.1432273387908936, + "learning_rate": 6.71303074670571e-07, + "loss": 0.8966, + "mean_token_accuracy": 0.730068564414978, + "num_tokens": 47331291.0, + "step": 1835 + }, + { + "epoch": 0.2016253019986822, + "grad_norm": 2.3063323497772217, + "learning_rate": 6.716691068814056e-07, + "loss": 1.0076, + "mean_token_accuracy": 0.696763813495636, + "num_tokens": 47356358.0, + "step": 1836 + }, + { + "epoch": 0.20173511970129585, + "grad_norm": 2.19769287109375, + "learning_rate": 6.720351390922401e-07, + "loss": 0.9963, + "mean_token_accuracy": 0.7008731365203857, + "num_tokens": 47383864.0, + "step": 1837 + }, + { + "epoch": 0.20184493740390952, + "grad_norm": 2.3283703327178955, + "learning_rate": 6.724011713030747e-07, + "loss": 1.0275, + "mean_token_accuracy": 0.6918555498123169, + "num_tokens": 47408405.0, + "step": 1838 + }, + { + "epoch": 0.20195475510652316, + "grad_norm": 2.3600118160247803, + "learning_rate": 6.727672035139092e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.7028148174285889, + "num_tokens": 47432899.0, + "step": 1839 + }, + { + "epoch": 0.20206457280913684, + "grad_norm": 2.791607618331909, + "learning_rate": 6.731332357247438e-07, + "loss": 1.1234, + "mean_token_accuracy": 0.6719594597816467, + "num_tokens": 47459200.0, + "step": 1840 + }, + { + "epoch": 0.20217439051175048, + "grad_norm": 2.1105220317840576, + "learning_rate": 6.734992679355784e-07, + "loss": 1.1149, + "mean_token_accuracy": 0.6735992431640625, + "num_tokens": 47489755.0, + "step": 1841 + }, + { + "epoch": 0.20228420821436416, + "grad_norm": 1.9810128211975098, + "learning_rate": 6.738653001464129e-07, + "loss": 1.038, + "mean_token_accuracy": 0.6890114545822144, + "num_tokens": 47521733.0, + "step": 1842 + }, + { + "epoch": 0.2023940259169778, + "grad_norm": 2.578113079071045, + "learning_rate": 6.742313323572474e-07, + "loss": 0.953, + "mean_token_accuracy": 0.7129106521606445, + "num_tokens": 47541998.0, + "step": 1843 + }, + { + "epoch": 0.20250384361959148, + "grad_norm": 2.149749994277954, + "learning_rate": 6.74597364568082e-07, + "loss": 1.0293, + "mean_token_accuracy": 0.6877401471138, + "num_tokens": 47569865.0, + "step": 1844 + }, + { + "epoch": 0.20261366132220515, + "grad_norm": 2.270406723022461, + "learning_rate": 6.749633967789166e-07, + "loss": 1.116, + "mean_token_accuracy": 0.6743999719619751, + "num_tokens": 47598795.0, + "step": 1845 + }, + { + "epoch": 0.2027234790248188, + "grad_norm": 2.175609827041626, + "learning_rate": 6.753294289897511e-07, + "loss": 1.1149, + "mean_token_accuracy": 0.671876072883606, + "num_tokens": 47627078.0, + "step": 1846 + }, + { + "epoch": 0.20283329672743247, + "grad_norm": 1.954152226448059, + "learning_rate": 6.756954612005857e-07, + "loss": 1.0229, + "mean_token_accuracy": 0.696233868598938, + "num_tokens": 47660542.0, + "step": 1847 + }, + { + "epoch": 0.20294311443004612, + "grad_norm": 2.3024513721466064, + "learning_rate": 6.760614934114202e-07, + "loss": 1.0809, + "mean_token_accuracy": 0.6808634996414185, + "num_tokens": 47687250.0, + "step": 1848 + }, + { + "epoch": 0.2030529321326598, + "grad_norm": 2.520329713821411, + "learning_rate": 6.764275256222546e-07, + "loss": 0.9768, + "mean_token_accuracy": 0.6965718269348145, + "num_tokens": 47708672.0, + "step": 1849 + }, + { + "epoch": 0.20316274983527344, + "grad_norm": 2.572859525680542, + "learning_rate": 6.767935578330894e-07, + "loss": 0.9582, + "mean_token_accuracy": 0.7136181592941284, + "num_tokens": 47730052.0, + "step": 1850 + }, + { + "epoch": 0.2032725675378871, + "grad_norm": 2.320467948913574, + "learning_rate": 6.771595900439239e-07, + "loss": 1.0945, + "mean_token_accuracy": 0.6738573312759399, + "num_tokens": 47757372.0, + "step": 1851 + }, + { + "epoch": 0.20338238524050076, + "grad_norm": 2.280911684036255, + "learning_rate": 6.775256222547583e-07, + "loss": 1.0529, + "mean_token_accuracy": 0.6779686212539673, + "num_tokens": 47784543.0, + "step": 1852 + }, + { + "epoch": 0.20349220294311443, + "grad_norm": 2.1112728118896484, + "learning_rate": 6.77891654465593e-07, + "loss": 1.0454, + "mean_token_accuracy": 0.6860618591308594, + "num_tokens": 47813373.0, + "step": 1853 + }, + { + "epoch": 0.2036020206457281, + "grad_norm": 2.301743507385254, + "learning_rate": 6.782576866764274e-07, + "loss": 1.0625, + "mean_token_accuracy": 0.6861646175384521, + "num_tokens": 47837686.0, + "step": 1854 + }, + { + "epoch": 0.20371183834834175, + "grad_norm": 2.0763187408447266, + "learning_rate": 6.78623718887262e-07, + "loss": 1.0676, + "mean_token_accuracy": 0.6910367012023926, + "num_tokens": 47870281.0, + "step": 1855 + }, + { + "epoch": 0.20382165605095542, + "grad_norm": 2.1470444202423096, + "learning_rate": 6.789897510980966e-07, + "loss": 1.0871, + "mean_token_accuracy": 0.6738348007202148, + "num_tokens": 47899080.0, + "step": 1856 + }, + { + "epoch": 0.20393147375356907, + "grad_norm": 2.274440050125122, + "learning_rate": 6.793557833089311e-07, + "loss": 1.0636, + "mean_token_accuracy": 0.6856366395950317, + "num_tokens": 47923965.0, + "step": 1857 + }, + { + "epoch": 0.20404129145618274, + "grad_norm": 2.5599632263183594, + "learning_rate": 6.797218155197657e-07, + "loss": 0.9927, + "mean_token_accuracy": 0.7008475065231323, + "num_tokens": 47945835.0, + "step": 1858 + }, + { + "epoch": 0.2041511091587964, + "grad_norm": 2.3816936016082764, + "learning_rate": 6.800878477306002e-07, + "loss": 1.0317, + "mean_token_accuracy": 0.6888065338134766, + "num_tokens": 47971101.0, + "step": 1859 + }, + { + "epoch": 0.20426092686141006, + "grad_norm": 2.3951754570007324, + "learning_rate": 6.804538799414347e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6921821236610413, + "num_tokens": 47995297.0, + "step": 1860 + }, + { + "epoch": 0.2043707445640237, + "grad_norm": 2.0242958068847656, + "learning_rate": 6.808199121522694e-07, + "loss": 1.0755, + "mean_token_accuracy": 0.6779261231422424, + "num_tokens": 48028033.0, + "step": 1861 + }, + { + "epoch": 0.20448056226663738, + "grad_norm": 2.3945469856262207, + "learning_rate": 6.811859443631039e-07, + "loss": 1.0419, + "mean_token_accuracy": 0.6865475177764893, + "num_tokens": 48050471.0, + "step": 1862 + }, + { + "epoch": 0.20459037996925103, + "grad_norm": 2.440898895263672, + "learning_rate": 6.815519765739384e-07, + "loss": 1.0342, + "mean_token_accuracy": 0.6905707716941833, + "num_tokens": 48073222.0, + "step": 1863 + }, + { + "epoch": 0.2047001976718647, + "grad_norm": 2.320216655731201, + "learning_rate": 6.81918008784773e-07, + "loss": 1.025, + "mean_token_accuracy": 0.6930677890777588, + "num_tokens": 48097610.0, + "step": 1864 + }, + { + "epoch": 0.20481001537447838, + "grad_norm": 2.5732524394989014, + "learning_rate": 6.822840409956075e-07, + "loss": 0.9991, + "mean_token_accuracy": 0.7013818621635437, + "num_tokens": 48119076.0, + "step": 1865 + }, + { + "epoch": 0.20491983307709202, + "grad_norm": 2.2138617038726807, + "learning_rate": 6.826500732064421e-07, + "loss": 0.9882, + "mean_token_accuracy": 0.6995000243186951, + "num_tokens": 48145067.0, + "step": 1866 + }, + { + "epoch": 0.2050296507797057, + "grad_norm": 2.3254952430725098, + "learning_rate": 6.830161054172767e-07, + "loss": 1.0526, + "mean_token_accuracy": 0.677850067615509, + "num_tokens": 48169280.0, + "step": 1867 + }, + { + "epoch": 0.20513946848231934, + "grad_norm": 2.032834529876709, + "learning_rate": 6.833821376281112e-07, + "loss": 0.9635, + "mean_token_accuracy": 0.7047697305679321, + "num_tokens": 48201385.0, + "step": 1868 + }, + { + "epoch": 0.20524928618493302, + "grad_norm": 2.3835976123809814, + "learning_rate": 6.837481698389457e-07, + "loss": 0.9982, + "mean_token_accuracy": 0.6959778666496277, + "num_tokens": 48227893.0, + "step": 1869 + }, + { + "epoch": 0.20535910388754666, + "grad_norm": 2.284686326980591, + "learning_rate": 6.841142020497803e-07, + "loss": 1.0594, + "mean_token_accuracy": 0.6857846975326538, + "num_tokens": 48255573.0, + "step": 1870 + }, + { + "epoch": 0.20546892159016034, + "grad_norm": 2.424947738647461, + "learning_rate": 6.844802342606148e-07, + "loss": 0.9403, + "mean_token_accuracy": 0.7209906578063965, + "num_tokens": 48276990.0, + "step": 1871 + }, + { + "epoch": 0.20557873929277398, + "grad_norm": 2.4623725414276123, + "learning_rate": 6.848462664714495e-07, + "loss": 1.0298, + "mean_token_accuracy": 0.6942669153213501, + "num_tokens": 48298559.0, + "step": 1872 + }, + { + "epoch": 0.20568855699538766, + "grad_norm": 2.128357410430908, + "learning_rate": 6.85212298682284e-07, + "loss": 1.0742, + "mean_token_accuracy": 0.675024151802063, + "num_tokens": 48325901.0, + "step": 1873 + }, + { + "epoch": 0.20579837469800133, + "grad_norm": 2.584459066390991, + "learning_rate": 6.855783308931185e-07, + "loss": 1.084, + "mean_token_accuracy": 0.6853216886520386, + "num_tokens": 48347441.0, + "step": 1874 + }, + { + "epoch": 0.20590819240061498, + "grad_norm": 3.0888898372650146, + "learning_rate": 6.859443631039531e-07, + "loss": 1.0294, + "mean_token_accuracy": 0.6946820020675659, + "num_tokens": 48365276.0, + "step": 1875 + }, + { + "epoch": 0.20601801010322865, + "grad_norm": 2.65405535697937, + "learning_rate": 6.863103953147876e-07, + "loss": 0.9371, + "mean_token_accuracy": 0.715431272983551, + "num_tokens": 48385953.0, + "step": 1876 + }, + { + "epoch": 0.2061278278058423, + "grad_norm": 2.3683204650878906, + "learning_rate": 6.866764275256222e-07, + "loss": 0.9871, + "mean_token_accuracy": 0.7022296190261841, + "num_tokens": 48410901.0, + "step": 1877 + }, + { + "epoch": 0.20623764550845597, + "grad_norm": 2.288756847381592, + "learning_rate": 6.870424597364568e-07, + "loss": 1.0809, + "mean_token_accuracy": 0.6808438897132874, + "num_tokens": 48438408.0, + "step": 1878 + }, + { + "epoch": 0.20634746321106961, + "grad_norm": 2.6784512996673584, + "learning_rate": 6.874084919472913e-07, + "loss": 0.9983, + "mean_token_accuracy": 0.7022943496704102, + "num_tokens": 48457518.0, + "step": 1879 + }, + { + "epoch": 0.2064572809136833, + "grad_norm": 2.156243324279785, + "learning_rate": 6.877745241581258e-07, + "loss": 0.985, + "mean_token_accuracy": 0.70916748046875, + "num_tokens": 48486392.0, + "step": 1880 + }, + { + "epoch": 0.20656709861629693, + "grad_norm": 2.1161575317382812, + "learning_rate": 6.881405563689604e-07, + "loss": 1.1211, + "mean_token_accuracy": 0.6770950555801392, + "num_tokens": 48516058.0, + "step": 1881 + }, + { + "epoch": 0.2066769163189106, + "grad_norm": 2.108499526977539, + "learning_rate": 6.88506588579795e-07, + "loss": 1.0728, + "mean_token_accuracy": 0.6787242293357849, + "num_tokens": 48546658.0, + "step": 1882 + }, + { + "epoch": 0.20678673402152428, + "grad_norm": 2.488863945007324, + "learning_rate": 6.888726207906295e-07, + "loss": 1.0565, + "mean_token_accuracy": 0.6811732053756714, + "num_tokens": 48569941.0, + "step": 1883 + }, + { + "epoch": 0.20689655172413793, + "grad_norm": 2.697824239730835, + "learning_rate": 6.892386530014641e-07, + "loss": 0.998, + "mean_token_accuracy": 0.6987698078155518, + "num_tokens": 48590774.0, + "step": 1884 + }, + { + "epoch": 0.2070063694267516, + "grad_norm": 2.1664836406707764, + "learning_rate": 6.896046852122986e-07, + "loss": 1.059, + "mean_token_accuracy": 0.6813820600509644, + "num_tokens": 48617652.0, + "step": 1885 + }, + { + "epoch": 0.20711618712936525, + "grad_norm": 2.2649338245391846, + "learning_rate": 6.899707174231331e-07, + "loss": 1.0054, + "mean_token_accuracy": 0.691862165927887, + "num_tokens": 48642381.0, + "step": 1886 + }, + { + "epoch": 0.20722600483197892, + "grad_norm": 2.0936062335968018, + "learning_rate": 6.903367496339677e-07, + "loss": 0.9996, + "mean_token_accuracy": 0.7022188305854797, + "num_tokens": 48671585.0, + "step": 1887 + }, + { + "epoch": 0.20733582253459257, + "grad_norm": 2.058084726333618, + "learning_rate": 6.907027818448023e-07, + "loss": 1.0678, + "mean_token_accuracy": 0.6789608001708984, + "num_tokens": 48703218.0, + "step": 1888 + }, + { + "epoch": 0.20744564023720624, + "grad_norm": 2.1921892166137695, + "learning_rate": 6.910688140556369e-07, + "loss": 1.0691, + "mean_token_accuracy": 0.6802053451538086, + "num_tokens": 48733166.0, + "step": 1889 + }, + { + "epoch": 0.2075554579398199, + "grad_norm": 2.394618272781372, + "learning_rate": 6.914348462664714e-07, + "loss": 1.025, + "mean_token_accuracy": 0.6952054500579834, + "num_tokens": 48756462.0, + "step": 1890 + }, + { + "epoch": 0.20766527564243356, + "grad_norm": 2.1528985500335693, + "learning_rate": 6.918008784773059e-07, + "loss": 1.08, + "mean_token_accuracy": 0.6903865337371826, + "num_tokens": 48785002.0, + "step": 1891 + }, + { + "epoch": 0.20777509334504723, + "grad_norm": 2.2278661727905273, + "learning_rate": 6.921669106881405e-07, + "loss": 0.9717, + "mean_token_accuracy": 0.7088109254837036, + "num_tokens": 48810763.0, + "step": 1892 + }, + { + "epoch": 0.20788491104766088, + "grad_norm": 2.3760058879852295, + "learning_rate": 6.925329428989751e-07, + "loss": 1.0002, + "mean_token_accuracy": 0.696535587310791, + "num_tokens": 48835009.0, + "step": 1893 + }, + { + "epoch": 0.20799472875027455, + "grad_norm": 2.4106640815734863, + "learning_rate": 6.928989751098096e-07, + "loss": 0.9973, + "mean_token_accuracy": 0.6982080936431885, + "num_tokens": 48858156.0, + "step": 1894 + }, + { + "epoch": 0.2081045464528882, + "grad_norm": 2.5095019340515137, + "learning_rate": 6.932650073206442e-07, + "loss": 1.0635, + "mean_token_accuracy": 0.6820849180221558, + "num_tokens": 48882263.0, + "step": 1895 + }, + { + "epoch": 0.20821436415550187, + "grad_norm": 2.501992702484131, + "learning_rate": 6.936310395314787e-07, + "loss": 1.0564, + "mean_token_accuracy": 0.6875781416893005, + "num_tokens": 48905461.0, + "step": 1896 + }, + { + "epoch": 0.20832418185811552, + "grad_norm": 2.207475185394287, + "learning_rate": 6.939970717423132e-07, + "loss": 1.0284, + "mean_token_accuracy": 0.6915205717086792, + "num_tokens": 48931619.0, + "step": 1897 + }, + { + "epoch": 0.2084339995607292, + "grad_norm": 2.2975118160247803, + "learning_rate": 6.943631039531479e-07, + "loss": 1.0652, + "mean_token_accuracy": 0.6823943853378296, + "num_tokens": 48957444.0, + "step": 1898 + }, + { + "epoch": 0.20854381726334284, + "grad_norm": 2.14607572555542, + "learning_rate": 6.947291361639824e-07, + "loss": 0.9665, + "mean_token_accuracy": 0.7037900686264038, + "num_tokens": 48986415.0, + "step": 1899 + }, + { + "epoch": 0.2086536349659565, + "grad_norm": 2.4283745288848877, + "learning_rate": 6.950951683748169e-07, + "loss": 0.9964, + "mean_token_accuracy": 0.7000342607498169, + "num_tokens": 49010061.0, + "step": 1900 + }, + { + "epoch": 0.20876345266857016, + "grad_norm": 2.5206809043884277, + "learning_rate": 6.954612005856515e-07, + "loss": 0.9533, + "mean_token_accuracy": 0.7137002348899841, + "num_tokens": 49033376.0, + "step": 1901 + }, + { + "epoch": 0.20887327037118383, + "grad_norm": 2.3628451824188232, + "learning_rate": 6.95827232796486e-07, + "loss": 1.019, + "mean_token_accuracy": 0.6946548819541931, + "num_tokens": 49057563.0, + "step": 1902 + }, + { + "epoch": 0.2089830880737975, + "grad_norm": 2.2446513175964355, + "learning_rate": 6.961932650073206e-07, + "loss": 1.0991, + "mean_token_accuracy": 0.6697374582290649, + "num_tokens": 49085030.0, + "step": 1903 + }, + { + "epoch": 0.20909290577641115, + "grad_norm": 2.508599281311035, + "learning_rate": 6.965592972181552e-07, + "loss": 0.9363, + "mean_token_accuracy": 0.7102611064910889, + "num_tokens": 49104353.0, + "step": 1904 + }, + { + "epoch": 0.20920272347902483, + "grad_norm": 2.2086589336395264, + "learning_rate": 6.969253294289897e-07, + "loss": 1.0406, + "mean_token_accuracy": 0.6955440044403076, + "num_tokens": 49134413.0, + "step": 1905 + }, + { + "epoch": 0.20931254118163847, + "grad_norm": 2.1951041221618652, + "learning_rate": 6.972913616398243e-07, + "loss": 1.0398, + "mean_token_accuracy": 0.6922212839126587, + "num_tokens": 49164137.0, + "step": 1906 + }, + { + "epoch": 0.20942235888425215, + "grad_norm": 2.2549240589141846, + "learning_rate": 6.976573938506588e-07, + "loss": 1.0235, + "mean_token_accuracy": 0.687610924243927, + "num_tokens": 49192401.0, + "step": 1907 + }, + { + "epoch": 0.2095321765868658, + "grad_norm": 2.2979748249053955, + "learning_rate": 6.980234260614933e-07, + "loss": 1.0247, + "mean_token_accuracy": 0.6914914846420288, + "num_tokens": 49218341.0, + "step": 1908 + }, + { + "epoch": 0.20964199428947947, + "grad_norm": 2.589830160140991, + "learning_rate": 6.98389458272328e-07, + "loss": 1.01, + "mean_token_accuracy": 0.6996660828590393, + "num_tokens": 49239163.0, + "step": 1909 + }, + { + "epoch": 0.2097518119920931, + "grad_norm": 2.1426243782043457, + "learning_rate": 6.987554904831625e-07, + "loss": 1.0336, + "mean_token_accuracy": 0.6928219199180603, + "num_tokens": 49268705.0, + "step": 1910 + }, + { + "epoch": 0.20986162969470679, + "grad_norm": 2.321084976196289, + "learning_rate": 6.99121522693997e-07, + "loss": 0.961, + "mean_token_accuracy": 0.7094758749008179, + "num_tokens": 49293437.0, + "step": 1911 + }, + { + "epoch": 0.20997144739732046, + "grad_norm": 2.3541371822357178, + "learning_rate": 6.994875549048316e-07, + "loss": 1.0944, + "mean_token_accuracy": 0.6694499850273132, + "num_tokens": 49317640.0, + "step": 1912 + }, + { + "epoch": 0.2100812650999341, + "grad_norm": 2.0695688724517822, + "learning_rate": 6.998535871156661e-07, + "loss": 1.0365, + "mean_token_accuracy": 0.6912627220153809, + "num_tokens": 49348204.0, + "step": 1913 + }, + { + "epoch": 0.21019108280254778, + "grad_norm": 2.382939338684082, + "learning_rate": 7.002196193265007e-07, + "loss": 1.0223, + "mean_token_accuracy": 0.6989309191703796, + "num_tokens": 49376892.0, + "step": 1914 + }, + { + "epoch": 0.21030090050516143, + "grad_norm": 2.139897346496582, + "learning_rate": 7.005856515373353e-07, + "loss": 0.9946, + "mean_token_accuracy": 0.6909146308898926, + "num_tokens": 49405788.0, + "step": 1915 + }, + { + "epoch": 0.2104107182077751, + "grad_norm": 2.398928165435791, + "learning_rate": 7.009516837481698e-07, + "loss": 1.0526, + "mean_token_accuracy": 0.685448169708252, + "num_tokens": 49428328.0, + "step": 1916 + }, + { + "epoch": 0.21052053591038875, + "grad_norm": 2.1320383548736572, + "learning_rate": 7.013177159590043e-07, + "loss": 1.0722, + "mean_token_accuracy": 0.6905522346496582, + "num_tokens": 49456756.0, + "step": 1917 + }, + { + "epoch": 0.21063035361300242, + "grad_norm": 2.384110689163208, + "learning_rate": 7.016837481698389e-07, + "loss": 1.0235, + "mean_token_accuracy": 0.6936984658241272, + "num_tokens": 49481855.0, + "step": 1918 + }, + { + "epoch": 0.21074017131561606, + "grad_norm": 2.3409223556518555, + "learning_rate": 7.020497803806734e-07, + "loss": 1.0532, + "mean_token_accuracy": 0.6866140365600586, + "num_tokens": 49507527.0, + "step": 1919 + }, + { + "epoch": 0.21084998901822974, + "grad_norm": 2.170593023300171, + "learning_rate": 7.024158125915081e-07, + "loss": 1.0615, + "mean_token_accuracy": 0.678674578666687, + "num_tokens": 49536120.0, + "step": 1920 + }, + { + "epoch": 0.2109598067208434, + "grad_norm": 2.5317811965942383, + "learning_rate": 7.027818448023426e-07, + "loss": 0.9945, + "mean_token_accuracy": 0.6985429525375366, + "num_tokens": 49557633.0, + "step": 1921 + }, + { + "epoch": 0.21106962442345706, + "grad_norm": 2.4047605991363525, + "learning_rate": 7.031478770131771e-07, + "loss": 0.9893, + "mean_token_accuracy": 0.7052098512649536, + "num_tokens": 49581184.0, + "step": 1922 + }, + { + "epoch": 0.21117944212607073, + "grad_norm": 2.1796038150787354, + "learning_rate": 7.035139092240117e-07, + "loss": 1.054, + "mean_token_accuracy": 0.6863576173782349, + "num_tokens": 49612774.0, + "step": 1923 + }, + { + "epoch": 0.21128925982868438, + "grad_norm": 2.708116292953491, + "learning_rate": 7.038799414348462e-07, + "loss": 0.9898, + "mean_token_accuracy": 0.6989525556564331, + "num_tokens": 49632537.0, + "step": 1924 + }, + { + "epoch": 0.21139907753129805, + "grad_norm": 2.1046597957611084, + "learning_rate": 7.042459736456808e-07, + "loss": 0.9418, + "mean_token_accuracy": 0.7214810848236084, + "num_tokens": 49661533.0, + "step": 1925 + }, + { + "epoch": 0.2115088952339117, + "grad_norm": 2.0700557231903076, + "learning_rate": 7.046120058565154e-07, + "loss": 0.9916, + "mean_token_accuracy": 0.6992970705032349, + "num_tokens": 49691168.0, + "step": 1926 + }, + { + "epoch": 0.21161871293652537, + "grad_norm": 2.2037241458892822, + "learning_rate": 7.049780380673499e-07, + "loss": 1.125, + "mean_token_accuracy": 0.6712850332260132, + "num_tokens": 49718846.0, + "step": 1927 + }, + { + "epoch": 0.21172853063913902, + "grad_norm": 2.666888952255249, + "learning_rate": 7.053440702781844e-07, + "loss": 1.0223, + "mean_token_accuracy": 0.6961337327957153, + "num_tokens": 49738425.0, + "step": 1928 + }, + { + "epoch": 0.2118383483417527, + "grad_norm": 2.229339122772217, + "learning_rate": 7.05710102489019e-07, + "loss": 0.9684, + "mean_token_accuracy": 0.7082154154777527, + "num_tokens": 49762843.0, + "step": 1929 + }, + { + "epoch": 0.21194816604436637, + "grad_norm": 2.17680287361145, + "learning_rate": 7.060761346998536e-07, + "loss": 0.9163, + "mean_token_accuracy": 0.725746750831604, + "num_tokens": 49787535.0, + "step": 1930 + }, + { + "epoch": 0.21205798374698, + "grad_norm": 2.5144143104553223, + "learning_rate": 7.064421669106881e-07, + "loss": 0.9492, + "mean_token_accuracy": 0.7175010442733765, + "num_tokens": 49808954.0, + "step": 1931 + }, + { + "epoch": 0.21216780144959368, + "grad_norm": 2.240332841873169, + "learning_rate": 7.068081991215227e-07, + "loss": 0.9588, + "mean_token_accuracy": 0.7133026719093323, + "num_tokens": 49833932.0, + "step": 1932 + }, + { + "epoch": 0.21227761915220733, + "grad_norm": 2.231567859649658, + "learning_rate": 7.071742313323572e-07, + "loss": 1.0453, + "mean_token_accuracy": 0.689091682434082, + "num_tokens": 49861910.0, + "step": 1933 + }, + { + "epoch": 0.212387436854821, + "grad_norm": 2.4241931438446045, + "learning_rate": 7.075402635431917e-07, + "loss": 1.0503, + "mean_token_accuracy": 0.6878501772880554, + "num_tokens": 49885140.0, + "step": 1934 + }, + { + "epoch": 0.21249725455743465, + "grad_norm": 2.1956026554107666, + "learning_rate": 7.079062957540263e-07, + "loss": 0.9443, + "mean_token_accuracy": 0.7132340669631958, + "num_tokens": 49909653.0, + "step": 1935 + }, + { + "epoch": 0.21260707226004832, + "grad_norm": 2.252800226211548, + "learning_rate": 7.082723279648609e-07, + "loss": 1.0561, + "mean_token_accuracy": 0.6871456503868103, + "num_tokens": 49936584.0, + "step": 1936 + }, + { + "epoch": 0.21271688996266197, + "grad_norm": 2.5784380435943604, + "learning_rate": 7.086383601756955e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.6895796656608582, + "num_tokens": 49956702.0, + "step": 1937 + }, + { + "epoch": 0.21282670766527564, + "grad_norm": 2.4814884662628174, + "learning_rate": 7.0900439238653e-07, + "loss": 0.9725, + "mean_token_accuracy": 0.6956737041473389, + "num_tokens": 49977896.0, + "step": 1938 + }, + { + "epoch": 0.2129365253678893, + "grad_norm": 2.268409252166748, + "learning_rate": 7.093704245973645e-07, + "loss": 0.9603, + "mean_token_accuracy": 0.7117072343826294, + "num_tokens": 50003208.0, + "step": 1939 + }, + { + "epoch": 0.21304634307050296, + "grad_norm": 2.479567527770996, + "learning_rate": 7.097364568081991e-07, + "loss": 1.0251, + "mean_token_accuracy": 0.6914820671081543, + "num_tokens": 50024940.0, + "step": 1940 + }, + { + "epoch": 0.21315616077311664, + "grad_norm": 2.1009790897369385, + "learning_rate": 7.101024890190337e-07, + "loss": 0.9148, + "mean_token_accuracy": 0.7220964431762695, + "num_tokens": 50051034.0, + "step": 1941 + }, + { + "epoch": 0.21326597847573028, + "grad_norm": 2.2631478309631348, + "learning_rate": 7.104685212298682e-07, + "loss": 1.0462, + "mean_token_accuracy": 0.6869019269943237, + "num_tokens": 50077175.0, + "step": 1942 + }, + { + "epoch": 0.21337579617834396, + "grad_norm": 1.922659993171692, + "learning_rate": 7.108345534407028e-07, + "loss": 1.1601, + "mean_token_accuracy": 0.6550042033195496, + "num_tokens": 50115357.0, + "step": 1943 + }, + { + "epoch": 0.2134856138809576, + "grad_norm": 2.1279544830322266, + "learning_rate": 7.112005856515373e-07, + "loss": 0.9506, + "mean_token_accuracy": 0.7123106718063354, + "num_tokens": 50143631.0, + "step": 1944 + }, + { + "epoch": 0.21359543158357128, + "grad_norm": 2.4208643436431885, + "learning_rate": 7.115666178623718e-07, + "loss": 0.9845, + "mean_token_accuracy": 0.7067749500274658, + "num_tokens": 50167265.0, + "step": 1945 + }, + { + "epoch": 0.21370524928618492, + "grad_norm": 2.4677035808563232, + "learning_rate": 7.119326500732065e-07, + "loss": 1.116, + "mean_token_accuracy": 0.667765736579895, + "num_tokens": 50191169.0, + "step": 1946 + }, + { + "epoch": 0.2138150669887986, + "grad_norm": 2.2107768058776855, + "learning_rate": 7.12298682284041e-07, + "loss": 1.0336, + "mean_token_accuracy": 0.6890326738357544, + "num_tokens": 50218920.0, + "step": 1947 + }, + { + "epoch": 0.21392488469141224, + "grad_norm": 2.3896472454071045, + "learning_rate": 7.126647144948755e-07, + "loss": 1.1289, + "mean_token_accuracy": 0.6650978922843933, + "num_tokens": 50244790.0, + "step": 1948 + }, + { + "epoch": 0.21403470239402592, + "grad_norm": 2.262329339981079, + "learning_rate": 7.130307467057101e-07, + "loss": 0.9947, + "mean_token_accuracy": 0.700920045375824, + "num_tokens": 50270316.0, + "step": 1949 + }, + { + "epoch": 0.2141445200966396, + "grad_norm": 2.2332074642181396, + "learning_rate": 7.133967789165446e-07, + "loss": 1.0073, + "mean_token_accuracy": 0.698441743850708, + "num_tokens": 50296368.0, + "step": 1950 + }, + { + "epoch": 0.21425433779925324, + "grad_norm": 2.5316338539123535, + "learning_rate": 7.137628111273791e-07, + "loss": 0.9883, + "mean_token_accuracy": 0.6964044570922852, + "num_tokens": 50317194.0, + "step": 1951 + }, + { + "epoch": 0.2143641555018669, + "grad_norm": 2.407346487045288, + "learning_rate": 7.141288433382138e-07, + "loss": 0.9551, + "mean_token_accuracy": 0.7107572555541992, + "num_tokens": 50338475.0, + "step": 1952 + }, + { + "epoch": 0.21447397320448056, + "grad_norm": 2.3037467002868652, + "learning_rate": 7.144948755490483e-07, + "loss": 1.068, + "mean_token_accuracy": 0.6765286326408386, + "num_tokens": 50363753.0, + "step": 1953 + }, + { + "epoch": 0.21458379090709423, + "grad_norm": 2.3373160362243652, + "learning_rate": 7.148609077598829e-07, + "loss": 0.9665, + "mean_token_accuracy": 0.7049686908721924, + "num_tokens": 50387871.0, + "step": 1954 + }, + { + "epoch": 0.21469360860970788, + "grad_norm": 2.504004955291748, + "learning_rate": 7.152269399707174e-07, + "loss": 1.0109, + "mean_token_accuracy": 0.6964324712753296, + "num_tokens": 50408307.0, + "step": 1955 + }, + { + "epoch": 0.21480342631232155, + "grad_norm": 2.297776222229004, + "learning_rate": 7.155929721815519e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6890924572944641, + "num_tokens": 50434392.0, + "step": 1956 + }, + { + "epoch": 0.2149132440149352, + "grad_norm": 2.0723752975463867, + "learning_rate": 7.159590043923866e-07, + "loss": 1.0351, + "mean_token_accuracy": 0.6879292726516724, + "num_tokens": 50464382.0, + "step": 1957 + }, + { + "epoch": 0.21502306171754887, + "grad_norm": 2.6634931564331055, + "learning_rate": 7.163250366032211e-07, + "loss": 1.1375, + "mean_token_accuracy": 0.6584306359291077, + "num_tokens": 50486233.0, + "step": 1958 + }, + { + "epoch": 0.21513287942016254, + "grad_norm": 2.1714982986450195, + "learning_rate": 7.166910688140556e-07, + "loss": 1.0803, + "mean_token_accuracy": 0.6728745698928833, + "num_tokens": 50516822.0, + "step": 1959 + }, + { + "epoch": 0.2152426971227762, + "grad_norm": 2.442145586013794, + "learning_rate": 7.170571010248902e-07, + "loss": 1.1084, + "mean_token_accuracy": 0.6738582849502563, + "num_tokens": 50541270.0, + "step": 1960 + }, + { + "epoch": 0.21535251482538986, + "grad_norm": 2.2165205478668213, + "learning_rate": 7.174231332357247e-07, + "loss": 0.9795, + "mean_token_accuracy": 0.7011014819145203, + "num_tokens": 50566736.0, + "step": 1961 + }, + { + "epoch": 0.2154623325280035, + "grad_norm": 2.260403633117676, + "learning_rate": 7.177891654465593e-07, + "loss": 1.0178, + "mean_token_accuracy": 0.693117618560791, + "num_tokens": 50591212.0, + "step": 1962 + }, + { + "epoch": 0.21557215023061718, + "grad_norm": 2.535071611404419, + "learning_rate": 7.181551976573939e-07, + "loss": 0.9425, + "mean_token_accuracy": 0.7131264209747314, + "num_tokens": 50613334.0, + "step": 1963 + }, + { + "epoch": 0.21568196793323083, + "grad_norm": 2.038257122039795, + "learning_rate": 7.185212298682284e-07, + "loss": 0.9495, + "mean_token_accuracy": 0.7108492851257324, + "num_tokens": 50642793.0, + "step": 1964 + }, + { + "epoch": 0.2157917856358445, + "grad_norm": 2.261021614074707, + "learning_rate": 7.188872620790629e-07, + "loss": 1.1178, + "mean_token_accuracy": 0.6794954538345337, + "num_tokens": 50670184.0, + "step": 1965 + }, + { + "epoch": 0.21590160333845815, + "grad_norm": 2.830693483352661, + "learning_rate": 7.192532942898975e-07, + "loss": 0.9415, + "mean_token_accuracy": 0.7071571946144104, + "num_tokens": 50688140.0, + "step": 1966 + }, + { + "epoch": 0.21601142104107182, + "grad_norm": 2.3315558433532715, + "learning_rate": 7.19619326500732e-07, + "loss": 1.044, + "mean_token_accuracy": 0.6849932670593262, + "num_tokens": 50712549.0, + "step": 1967 + }, + { + "epoch": 0.2161212387436855, + "grad_norm": 2.3891143798828125, + "learning_rate": 7.199853587115667e-07, + "loss": 1.1308, + "mean_token_accuracy": 0.665006160736084, + "num_tokens": 50738520.0, + "step": 1968 + }, + { + "epoch": 0.21623105644629914, + "grad_norm": 2.232496738433838, + "learning_rate": 7.203513909224012e-07, + "loss": 0.9683, + "mean_token_accuracy": 0.7150580883026123, + "num_tokens": 50763757.0, + "step": 1969 + }, + { + "epoch": 0.21634087414891282, + "grad_norm": 2.0082035064697266, + "learning_rate": 7.207174231332357e-07, + "loss": 1.0265, + "mean_token_accuracy": 0.7022790908813477, + "num_tokens": 50793832.0, + "step": 1970 + }, + { + "epoch": 0.21645069185152646, + "grad_norm": 2.5842738151550293, + "learning_rate": 7.210834553440703e-07, + "loss": 0.939, + "mean_token_accuracy": 0.710091233253479, + "num_tokens": 50814416.0, + "step": 1971 + }, + { + "epoch": 0.21656050955414013, + "grad_norm": 2.670804023742676, + "learning_rate": 7.214494875549048e-07, + "loss": 0.9442, + "mean_token_accuracy": 0.7061242461204529, + "num_tokens": 50833389.0, + "step": 1972 + }, + { + "epoch": 0.21667032725675378, + "grad_norm": 2.183370351791382, + "learning_rate": 7.218155197657394e-07, + "loss": 1.1334, + "mean_token_accuracy": 0.6714473366737366, + "num_tokens": 50863979.0, + "step": 1973 + }, + { + "epoch": 0.21678014495936745, + "grad_norm": 2.427790880203247, + "learning_rate": 7.22181551976574e-07, + "loss": 0.9577, + "mean_token_accuracy": 0.70528244972229, + "num_tokens": 50886208.0, + "step": 1974 + }, + { + "epoch": 0.2168899626619811, + "grad_norm": 2.4670779705047607, + "learning_rate": 7.225475841874085e-07, + "loss": 1.0316, + "mean_token_accuracy": 0.6860615611076355, + "num_tokens": 50908463.0, + "step": 1975 + }, + { + "epoch": 0.21699978036459477, + "grad_norm": 2.4932541847229004, + "learning_rate": 7.22913616398243e-07, + "loss": 0.9832, + "mean_token_accuracy": 0.710071861743927, + "num_tokens": 50931728.0, + "step": 1976 + }, + { + "epoch": 0.21710959806720842, + "grad_norm": 2.2130892276763916, + "learning_rate": 7.232796486090776e-07, + "loss": 1.0595, + "mean_token_accuracy": 0.682213544845581, + "num_tokens": 50956836.0, + "step": 1977 + }, + { + "epoch": 0.2172194157698221, + "grad_norm": 2.122641086578369, + "learning_rate": 7.236456808199122e-07, + "loss": 1.14, + "mean_token_accuracy": 0.6617356538772583, + "num_tokens": 50988874.0, + "step": 1978 + }, + { + "epoch": 0.21732923347243577, + "grad_norm": 2.2577147483825684, + "learning_rate": 7.240117130307467e-07, + "loss": 1.11, + "mean_token_accuracy": 0.6838604807853699, + "num_tokens": 51017825.0, + "step": 1979 + }, + { + "epoch": 0.2174390511750494, + "grad_norm": 2.1366186141967773, + "learning_rate": 7.243777452415813e-07, + "loss": 0.9931, + "mean_token_accuracy": 0.7004177570343018, + "num_tokens": 51045955.0, + "step": 1980 + }, + { + "epoch": 0.2175488688776631, + "grad_norm": 2.437851905822754, + "learning_rate": 7.247437774524158e-07, + "loss": 1.0378, + "mean_token_accuracy": 0.6808009147644043, + "num_tokens": 51069330.0, + "step": 1981 + }, + { + "epoch": 0.21765868658027673, + "grad_norm": 2.6718225479125977, + "learning_rate": 7.251098096632503e-07, + "loss": 1.0684, + "mean_token_accuracy": 0.6927223801612854, + "num_tokens": 51092178.0, + "step": 1982 + }, + { + "epoch": 0.2177685042828904, + "grad_norm": 2.3236894607543945, + "learning_rate": 7.254758418740849e-07, + "loss": 0.9918, + "mean_token_accuracy": 0.6931989789009094, + "num_tokens": 51115554.0, + "step": 1983 + }, + { + "epoch": 0.21787832198550405, + "grad_norm": 2.4995334148406982, + "learning_rate": 7.258418740849195e-07, + "loss": 0.9793, + "mean_token_accuracy": 0.7041566371917725, + "num_tokens": 51135508.0, + "step": 1984 + }, + { + "epoch": 0.21798813968811773, + "grad_norm": 2.2184793949127197, + "learning_rate": 7.262079062957541e-07, + "loss": 0.9787, + "mean_token_accuracy": 0.7067208290100098, + "num_tokens": 51159938.0, + "step": 1985 + }, + { + "epoch": 0.21809795739073137, + "grad_norm": 2.11283540725708, + "learning_rate": 7.265739385065886e-07, + "loss": 1.1236, + "mean_token_accuracy": 0.6678785085678101, + "num_tokens": 51189162.0, + "step": 1986 + }, + { + "epoch": 0.21820777509334505, + "grad_norm": 2.355032205581665, + "learning_rate": 7.269399707174231e-07, + "loss": 1.0142, + "mean_token_accuracy": 0.6926445960998535, + "num_tokens": 51213141.0, + "step": 1987 + }, + { + "epoch": 0.21831759279595872, + "grad_norm": 2.1187477111816406, + "learning_rate": 7.273060029282577e-07, + "loss": 1.0364, + "mean_token_accuracy": 0.6887520551681519, + "num_tokens": 51243740.0, + "step": 1988 + }, + { + "epoch": 0.21842741049857237, + "grad_norm": 2.319401264190674, + "learning_rate": 7.276720351390923e-07, + "loss": 1.0287, + "mean_token_accuracy": 0.6989991068840027, + "num_tokens": 51267656.0, + "step": 1989 + }, + { + "epoch": 0.21853722820118604, + "grad_norm": 2.313368558883667, + "learning_rate": 7.280380673499268e-07, + "loss": 0.9436, + "mean_token_accuracy": 0.7130433917045593, + "num_tokens": 51291884.0, + "step": 1990 + }, + { + "epoch": 0.21864704590379969, + "grad_norm": 2.468311309814453, + "learning_rate": 7.284040995607614e-07, + "loss": 1.0478, + "mean_token_accuracy": 0.6849757432937622, + "num_tokens": 51313957.0, + "step": 1991 + }, + { + "epoch": 0.21875686360641336, + "grad_norm": 2.717376947402954, + "learning_rate": 7.287701317715959e-07, + "loss": 0.9863, + "mean_token_accuracy": 0.7007172107696533, + "num_tokens": 51333146.0, + "step": 1992 + }, + { + "epoch": 0.218866681309027, + "grad_norm": 2.562932252883911, + "learning_rate": 7.291361639824304e-07, + "loss": 1.0722, + "mean_token_accuracy": 0.6819937229156494, + "num_tokens": 51354676.0, + "step": 1993 + }, + { + "epoch": 0.21897649901164068, + "grad_norm": 2.3918731212615967, + "learning_rate": 7.295021961932651e-07, + "loss": 0.999, + "mean_token_accuracy": 0.6998291015625, + "num_tokens": 51377314.0, + "step": 1994 + }, + { + "epoch": 0.21908631671425433, + "grad_norm": 2.3220391273498535, + "learning_rate": 7.298682284040996e-07, + "loss": 1.053, + "mean_token_accuracy": 0.6867360472679138, + "num_tokens": 51404483.0, + "step": 1995 + }, + { + "epoch": 0.219196134416868, + "grad_norm": 2.3883166313171387, + "learning_rate": 7.30234260614934e-07, + "loss": 1.0036, + "mean_token_accuracy": 0.6956725120544434, + "num_tokens": 51427270.0, + "step": 1996 + }, + { + "epoch": 0.21930595211948167, + "grad_norm": 2.481931209564209, + "learning_rate": 7.306002928257687e-07, + "loss": 1.0681, + "mean_token_accuracy": 0.6861374378204346, + "num_tokens": 51451033.0, + "step": 1997 + }, + { + "epoch": 0.21941576982209532, + "grad_norm": 2.342400312423706, + "learning_rate": 7.309663250366031e-07, + "loss": 0.9917, + "mean_token_accuracy": 0.7020227909088135, + "num_tokens": 51477625.0, + "step": 1998 + }, + { + "epoch": 0.219525587524709, + "grad_norm": 2.5730998516082764, + "learning_rate": 7.313323572474376e-07, + "loss": 1.0429, + "mean_token_accuracy": 0.6878342032432556, + "num_tokens": 51499313.0, + "step": 1999 + }, + { + "epoch": 0.21963540522732264, + "grad_norm": 2.4309537410736084, + "learning_rate": 7.316983894582724e-07, + "loss": 1.0071, + "mean_token_accuracy": 0.7052612900733948, + "num_tokens": 51523584.0, + "step": 2000 + }, + { + "epoch": 0.2197452229299363, + "grad_norm": 2.52296781539917, + "learning_rate": 7.320644216691068e-07, + "loss": 1.0674, + "mean_token_accuracy": 0.6820781826972961, + "num_tokens": 51544804.0, + "step": 2001 + }, + { + "epoch": 0.21985504063254996, + "grad_norm": 2.3623082637786865, + "learning_rate": 7.324304538799415e-07, + "loss": 0.9746, + "mean_token_accuracy": 0.7038019895553589, + "num_tokens": 51568871.0, + "step": 2002 + }, + { + "epoch": 0.21996485833516363, + "grad_norm": 2.2098336219787598, + "learning_rate": 7.32796486090776e-07, + "loss": 1.0166, + "mean_token_accuracy": 0.7053807377815247, + "num_tokens": 51596345.0, + "step": 2003 + }, + { + "epoch": 0.22007467603777728, + "grad_norm": 2.0929698944091797, + "learning_rate": 7.331625183016104e-07, + "loss": 1.0614, + "mean_token_accuracy": 0.6866485476493835, + "num_tokens": 51624644.0, + "step": 2004 + }, + { + "epoch": 0.22018449374039095, + "grad_norm": 2.325866460800171, + "learning_rate": 7.335285505124452e-07, + "loss": 0.9297, + "mean_token_accuracy": 0.7156045436859131, + "num_tokens": 51647338.0, + "step": 2005 + }, + { + "epoch": 0.22029431144300463, + "grad_norm": 2.4305639266967773, + "learning_rate": 7.338945827232796e-07, + "loss": 1.0394, + "mean_token_accuracy": 0.6901103258132935, + "num_tokens": 51670588.0, + "step": 2006 + }, + { + "epoch": 0.22040412914561827, + "grad_norm": 2.2447688579559326, + "learning_rate": 7.342606149341141e-07, + "loss": 0.9998, + "mean_token_accuracy": 0.7098885178565979, + "num_tokens": 51694915.0, + "step": 2007 + }, + { + "epoch": 0.22051394684823195, + "grad_norm": 2.171945810317993, + "learning_rate": 7.346266471449487e-07, + "loss": 1.1016, + "mean_token_accuracy": 0.6705615520477295, + "num_tokens": 51722461.0, + "step": 2008 + }, + { + "epoch": 0.2206237645508456, + "grad_norm": 2.2320213317871094, + "learning_rate": 7.349926793557832e-07, + "loss": 0.9922, + "mean_token_accuracy": 0.6977515816688538, + "num_tokens": 51750114.0, + "step": 2009 + }, + { + "epoch": 0.22073358225345927, + "grad_norm": 2.506735324859619, + "learning_rate": 7.353587115666178e-07, + "loss": 0.9723, + "mean_token_accuracy": 0.7048401832580566, + "num_tokens": 51769519.0, + "step": 2010 + }, + { + "epoch": 0.2208433999560729, + "grad_norm": 2.491044759750366, + "learning_rate": 7.357247437774524e-07, + "loss": 0.9557, + "mean_token_accuracy": 0.7144492864608765, + "num_tokens": 51790952.0, + "step": 2011 + }, + { + "epoch": 0.22095321765868658, + "grad_norm": 2.0535547733306885, + "learning_rate": 7.360907759882869e-07, + "loss": 1.0138, + "mean_token_accuracy": 0.7002086639404297, + "num_tokens": 51820766.0, + "step": 2012 + }, + { + "epoch": 0.22106303536130023, + "grad_norm": 2.1879920959472656, + "learning_rate": 7.364568081991214e-07, + "loss": 1.078, + "mean_token_accuracy": 0.681513786315918, + "num_tokens": 51848779.0, + "step": 2013 + }, + { + "epoch": 0.2211728530639139, + "grad_norm": 2.475994348526001, + "learning_rate": 7.36822840409956e-07, + "loss": 1.0507, + "mean_token_accuracy": 0.6879735589027405, + "num_tokens": 51872683.0, + "step": 2014 + }, + { + "epoch": 0.22128267076652755, + "grad_norm": 2.548386812210083, + "learning_rate": 7.371888726207905e-07, + "loss": 0.9815, + "mean_token_accuracy": 0.7041374444961548, + "num_tokens": 51892998.0, + "step": 2015 + }, + { + "epoch": 0.22139248846914122, + "grad_norm": 2.3798112869262695, + "learning_rate": 7.375549048316251e-07, + "loss": 0.9959, + "mean_token_accuracy": 0.696942925453186, + "num_tokens": 51916800.0, + "step": 2016 + }, + { + "epoch": 0.2215023061717549, + "grad_norm": 2.4643945693969727, + "learning_rate": 7.379209370424597e-07, + "loss": 0.9713, + "mean_token_accuracy": 0.7071805596351624, + "num_tokens": 51938004.0, + "step": 2017 + }, + { + "epoch": 0.22161212387436854, + "grad_norm": 2.5109074115753174, + "learning_rate": 7.382869692532942e-07, + "loss": 0.9874, + "mean_token_accuracy": 0.6992460489273071, + "num_tokens": 51957847.0, + "step": 2018 + }, + { + "epoch": 0.22172194157698222, + "grad_norm": 2.4320485591888428, + "learning_rate": 7.386530014641288e-07, + "loss": 0.9772, + "mean_token_accuracy": 0.7053232192993164, + "num_tokens": 51980724.0, + "step": 2019 + }, + { + "epoch": 0.22183175927959586, + "grad_norm": 2.2778379917144775, + "learning_rate": 7.390190336749633e-07, + "loss": 0.9545, + "mean_token_accuracy": 0.7049460411071777, + "num_tokens": 52004838.0, + "step": 2020 + }, + { + "epoch": 0.22194157698220954, + "grad_norm": 2.1693217754364014, + "learning_rate": 7.393850658857979e-07, + "loss": 1.0909, + "mean_token_accuracy": 0.6875752806663513, + "num_tokens": 52031951.0, + "step": 2021 + }, + { + "epoch": 0.22205139468482318, + "grad_norm": 2.5173490047454834, + "learning_rate": 7.397510980966325e-07, + "loss": 1.0384, + "mean_token_accuracy": 0.6845933198928833, + "num_tokens": 52055321.0, + "step": 2022 + }, + { + "epoch": 0.22216121238743686, + "grad_norm": 2.2764432430267334, + "learning_rate": 7.40117130307467e-07, + "loss": 1.0903, + "mean_token_accuracy": 0.6732887029647827, + "num_tokens": 52081626.0, + "step": 2023 + }, + { + "epoch": 0.2222710300900505, + "grad_norm": 1.8944998979568481, + "learning_rate": 7.404831625183015e-07, + "loss": 1.0604, + "mean_token_accuracy": 0.685036301612854, + "num_tokens": 52116812.0, + "step": 2024 + }, + { + "epoch": 0.22238084779266418, + "grad_norm": 2.2430226802825928, + "learning_rate": 7.408491947291361e-07, + "loss": 1.0626, + "mean_token_accuracy": 0.6926454305648804, + "num_tokens": 52146142.0, + "step": 2025 + }, + { + "epoch": 0.22249066549527785, + "grad_norm": 2.2831475734710693, + "learning_rate": 7.412152269399707e-07, + "loss": 1.0481, + "mean_token_accuracy": 0.6863961815834045, + "num_tokens": 52171070.0, + "step": 2026 + }, + { + "epoch": 0.2226004831978915, + "grad_norm": 2.1776673793792725, + "learning_rate": 7.415812591508052e-07, + "loss": 0.9421, + "mean_token_accuracy": 0.7103443741798401, + "num_tokens": 52198273.0, + "step": 2027 + }, + { + "epoch": 0.22271030090050517, + "grad_norm": 2.4715688228607178, + "learning_rate": 7.419472913616398e-07, + "loss": 1.017, + "mean_token_accuracy": 0.6956961750984192, + "num_tokens": 52220792.0, + "step": 2028 + }, + { + "epoch": 0.22282011860311882, + "grad_norm": 2.212251901626587, + "learning_rate": 7.423133235724743e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.7070741653442383, + "num_tokens": 52248973.0, + "step": 2029 + }, + { + "epoch": 0.2229299363057325, + "grad_norm": 2.4496986865997314, + "learning_rate": 7.426793557833088e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6784892082214355, + "num_tokens": 52271488.0, + "step": 2030 + }, + { + "epoch": 0.22303975400834614, + "grad_norm": 2.2226548194885254, + "learning_rate": 7.430453879941434e-07, + "loss": 1.1306, + "mean_token_accuracy": 0.663291871547699, + "num_tokens": 52299754.0, + "step": 2031 + }, + { + "epoch": 0.2231495717109598, + "grad_norm": 2.3967549800872803, + "learning_rate": 7.43411420204978e-07, + "loss": 1.0902, + "mean_token_accuracy": 0.6785504221916199, + "num_tokens": 52324906.0, + "step": 2032 + }, + { + "epoch": 0.22325938941357346, + "grad_norm": 2.3253419399261475, + "learning_rate": 7.437774524158125e-07, + "loss": 0.9792, + "mean_token_accuracy": 0.7022407054901123, + "num_tokens": 52348354.0, + "step": 2033 + }, + { + "epoch": 0.22336920711618713, + "grad_norm": 2.5824711322784424, + "learning_rate": 7.441434846266471e-07, + "loss": 1.0284, + "mean_token_accuracy": 0.689609944820404, + "num_tokens": 52369396.0, + "step": 2034 + }, + { + "epoch": 0.2234790248188008, + "grad_norm": 2.5266077518463135, + "learning_rate": 7.445095168374816e-07, + "loss": 0.8775, + "mean_token_accuracy": 0.7209360003471375, + "num_tokens": 52391080.0, + "step": 2035 + }, + { + "epoch": 0.22358884252141445, + "grad_norm": 2.6077499389648438, + "learning_rate": 7.448755490483162e-07, + "loss": 0.9615, + "mean_token_accuracy": 0.7084342241287231, + "num_tokens": 52410968.0, + "step": 2036 + }, + { + "epoch": 0.22369866022402812, + "grad_norm": 2.0432586669921875, + "learning_rate": 7.452415812591508e-07, + "loss": 1.0979, + "mean_token_accuracy": 0.6689953207969666, + "num_tokens": 52441502.0, + "step": 2037 + }, + { + "epoch": 0.22380847792664177, + "grad_norm": 2.4233388900756836, + "learning_rate": 7.456076134699853e-07, + "loss": 1.0374, + "mean_token_accuracy": 0.6895556449890137, + "num_tokens": 52464356.0, + "step": 2038 + }, + { + "epoch": 0.22391829562925544, + "grad_norm": 2.2414798736572266, + "learning_rate": 7.459736456808199e-07, + "loss": 1.0277, + "mean_token_accuracy": 0.6907755136489868, + "num_tokens": 52491196.0, + "step": 2039 + }, + { + "epoch": 0.2240281133318691, + "grad_norm": 2.6968443393707275, + "learning_rate": 7.463396778916544e-07, + "loss": 1.0236, + "mean_token_accuracy": 0.6921577453613281, + "num_tokens": 52510079.0, + "step": 2040 + }, + { + "epoch": 0.22413793103448276, + "grad_norm": 2.2624754905700684, + "learning_rate": 7.467057101024889e-07, + "loss": 1.1051, + "mean_token_accuracy": 0.677519679069519, + "num_tokens": 52536899.0, + "step": 2041 + }, + { + "epoch": 0.2242477487370964, + "grad_norm": 2.257392644882202, + "learning_rate": 7.470717423133236e-07, + "loss": 1.1124, + "mean_token_accuracy": 0.6731043457984924, + "num_tokens": 52565125.0, + "step": 2042 + }, + { + "epoch": 0.22435756643971008, + "grad_norm": 2.305565595626831, + "learning_rate": 7.474377745241581e-07, + "loss": 0.9959, + "mean_token_accuracy": 0.7008006572723389, + "num_tokens": 52589724.0, + "step": 2043 + }, + { + "epoch": 0.22446738414232376, + "grad_norm": 2.1508946418762207, + "learning_rate": 7.478038067349926e-07, + "loss": 1.127, + "mean_token_accuracy": 0.6667200326919556, + "num_tokens": 52620905.0, + "step": 2044 + }, + { + "epoch": 0.2245772018449374, + "grad_norm": 2.46272611618042, + "learning_rate": 7.481698389458272e-07, + "loss": 0.9888, + "mean_token_accuracy": 0.7043903470039368, + "num_tokens": 52647382.0, + "step": 2045 + }, + { + "epoch": 0.22468701954755108, + "grad_norm": 2.426569700241089, + "learning_rate": 7.485358711566617e-07, + "loss": 1.0325, + "mean_token_accuracy": 0.6876767873764038, + "num_tokens": 52670035.0, + "step": 2046 + }, + { + "epoch": 0.22479683725016472, + "grad_norm": 1.9609650373458862, + "learning_rate": 7.489019033674962e-07, + "loss": 1.0074, + "mean_token_accuracy": 0.6941758990287781, + "num_tokens": 52702091.0, + "step": 2047 + }, + { + "epoch": 0.2249066549527784, + "grad_norm": 2.397120475769043, + "learning_rate": 7.492679355783309e-07, + "loss": 0.9778, + "mean_token_accuracy": 0.7022594809532166, + "num_tokens": 52723476.0, + "step": 2048 + }, + { + "epoch": 0.22501647265539204, + "grad_norm": 2.1357247829437256, + "learning_rate": 7.496339677891654e-07, + "loss": 0.9282, + "mean_token_accuracy": 0.7155802249908447, + "num_tokens": 52750250.0, + "step": 2049 + }, + { + "epoch": 0.22512629035800572, + "grad_norm": 2.289659261703491, + "learning_rate": 7.5e-07, + "loss": 1.0069, + "mean_token_accuracy": 0.6943583488464355, + "num_tokens": 52774913.0, + "step": 2050 + }, + { + "epoch": 0.22523610806061936, + "grad_norm": 2.3182365894317627, + "learning_rate": 7.503660322108345e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.681769847869873, + "num_tokens": 52799061.0, + "step": 2051 + }, + { + "epoch": 0.22534592576323303, + "grad_norm": 2.1646316051483154, + "learning_rate": 7.50732064421669e-07, + "loss": 1.0307, + "mean_token_accuracy": 0.6885135173797607, + "num_tokens": 52827682.0, + "step": 2052 + }, + { + "epoch": 0.22545574346584668, + "grad_norm": 2.25484037399292, + "learning_rate": 7.510980966325037e-07, + "loss": 1.0789, + "mean_token_accuracy": 0.6769209504127502, + "num_tokens": 52854004.0, + "step": 2053 + }, + { + "epoch": 0.22556556116846035, + "grad_norm": 2.5970892906188965, + "learning_rate": 7.514641288433382e-07, + "loss": 1.0487, + "mean_token_accuracy": 0.6961839199066162, + "num_tokens": 52875334.0, + "step": 2054 + }, + { + "epoch": 0.22567537887107403, + "grad_norm": 2.2892730236053467, + "learning_rate": 7.518301610541727e-07, + "loss": 1.0385, + "mean_token_accuracy": 0.6937482953071594, + "num_tokens": 52901072.0, + "step": 2055 + }, + { + "epoch": 0.22578519657368767, + "grad_norm": 2.1666336059570312, + "learning_rate": 7.521961932650073e-07, + "loss": 1.0381, + "mean_token_accuracy": 0.6934636235237122, + "num_tokens": 52929140.0, + "step": 2056 + }, + { + "epoch": 0.22589501427630135, + "grad_norm": 2.356468915939331, + "learning_rate": 7.525622254758418e-07, + "loss": 0.9772, + "mean_token_accuracy": 0.7026914358139038, + "num_tokens": 52953279.0, + "step": 2057 + }, + { + "epoch": 0.226004831978915, + "grad_norm": 2.06756591796875, + "learning_rate": 7.529282576866763e-07, + "loss": 1.0718, + "mean_token_accuracy": 0.6767777800559998, + "num_tokens": 52985489.0, + "step": 2058 + }, + { + "epoch": 0.22611464968152867, + "grad_norm": 2.5145835876464844, + "learning_rate": 7.53294289897511e-07, + "loss": 0.9943, + "mean_token_accuracy": 0.6972439289093018, + "num_tokens": 53008432.0, + "step": 2059 + }, + { + "epoch": 0.2262244673841423, + "grad_norm": 2.573150873184204, + "learning_rate": 7.536603221083455e-07, + "loss": 1.0053, + "mean_token_accuracy": 0.7067480087280273, + "num_tokens": 53030196.0, + "step": 2060 + }, + { + "epoch": 0.226334285086756, + "grad_norm": 2.2325141429901123, + "learning_rate": 7.5402635431918e-07, + "loss": 1.0035, + "mean_token_accuracy": 0.7034478783607483, + "num_tokens": 53059442.0, + "step": 2061 + }, + { + "epoch": 0.22644410278936963, + "grad_norm": 2.5817525386810303, + "learning_rate": 7.543923865300146e-07, + "loss": 0.9891, + "mean_token_accuracy": 0.7017431855201721, + "num_tokens": 53081527.0, + "step": 2062 + }, + { + "epoch": 0.2265539204919833, + "grad_norm": 2.156114339828491, + "learning_rate": 7.547584187408491e-07, + "loss": 1.043, + "mean_token_accuracy": 0.6903235912322998, + "num_tokens": 53109281.0, + "step": 2063 + }, + { + "epoch": 0.22666373819459698, + "grad_norm": 2.2441983222961426, + "learning_rate": 7.551244509516837e-07, + "loss": 1.0, + "mean_token_accuracy": 0.7062033414840698, + "num_tokens": 53134575.0, + "step": 2064 + }, + { + "epoch": 0.22677355589721063, + "grad_norm": 2.2235400676727295, + "learning_rate": 7.554904831625183e-07, + "loss": 1.0006, + "mean_token_accuracy": 0.6993112564086914, + "num_tokens": 53163685.0, + "step": 2065 + }, + { + "epoch": 0.2268833735998243, + "grad_norm": 2.113633394241333, + "learning_rate": 7.558565153733528e-07, + "loss": 1.0318, + "mean_token_accuracy": 0.6936506032943726, + "num_tokens": 53193254.0, + "step": 2066 + }, + { + "epoch": 0.22699319130243795, + "grad_norm": 2.4434814453125, + "learning_rate": 7.562225475841874e-07, + "loss": 1.0247, + "mean_token_accuracy": 0.6926084756851196, + "num_tokens": 53216268.0, + "step": 2067 + }, + { + "epoch": 0.22710300900505162, + "grad_norm": 2.364332437515259, + "learning_rate": 7.565885797950219e-07, + "loss": 0.9271, + "mean_token_accuracy": 0.714521050453186, + "num_tokens": 53239331.0, + "step": 2068 + }, + { + "epoch": 0.22721282670766527, + "grad_norm": 2.205355644226074, + "learning_rate": 7.569546120058565e-07, + "loss": 1.061, + "mean_token_accuracy": 0.691194474697113, + "num_tokens": 53269366.0, + "step": 2069 + }, + { + "epoch": 0.22732264441027894, + "grad_norm": 2.4165337085723877, + "learning_rate": 7.573206442166911e-07, + "loss": 1.0193, + "mean_token_accuracy": 0.6937565803527832, + "num_tokens": 53293551.0, + "step": 2070 + }, + { + "epoch": 0.22743246211289259, + "grad_norm": 2.384812593460083, + "learning_rate": 7.576866764275256e-07, + "loss": 0.9669, + "mean_token_accuracy": 0.7016454339027405, + "num_tokens": 53317407.0, + "step": 2071 + }, + { + "epoch": 0.22754227981550626, + "grad_norm": 2.087482452392578, + "learning_rate": 7.580527086383601e-07, + "loss": 1.1033, + "mean_token_accuracy": 0.6812877655029297, + "num_tokens": 53349129.0, + "step": 2072 + }, + { + "epoch": 0.22765209751811993, + "grad_norm": 2.396867275238037, + "learning_rate": 7.584187408491947e-07, + "loss": 0.993, + "mean_token_accuracy": 0.6980288028717041, + "num_tokens": 53373807.0, + "step": 2073 + }, + { + "epoch": 0.22776191522073358, + "grad_norm": 2.421626329421997, + "learning_rate": 7.587847730600292e-07, + "loss": 1.1019, + "mean_token_accuracy": 0.673821747303009, + "num_tokens": 53398993.0, + "step": 2074 + }, + { + "epoch": 0.22787173292334725, + "grad_norm": 2.0990445613861084, + "learning_rate": 7.591508052708638e-07, + "loss": 1.1134, + "mean_token_accuracy": 0.673633873462677, + "num_tokens": 53430064.0, + "step": 2075 + }, + { + "epoch": 0.2279815506259609, + "grad_norm": 2.133960008621216, + "learning_rate": 7.595168374816984e-07, + "loss": 0.9883, + "mean_token_accuracy": 0.6946349143981934, + "num_tokens": 53459080.0, + "step": 2076 + }, + { + "epoch": 0.22809136832857457, + "grad_norm": 2.2094826698303223, + "learning_rate": 7.598828696925329e-07, + "loss": 1.1084, + "mean_token_accuracy": 0.6734322309494019, + "num_tokens": 53488249.0, + "step": 2077 + }, + { + "epoch": 0.22820118603118822, + "grad_norm": 2.294854164123535, + "learning_rate": 7.602489019033674e-07, + "loss": 0.892, + "mean_token_accuracy": 0.725636899471283, + "num_tokens": 53512224.0, + "step": 2078 + }, + { + "epoch": 0.2283110037338019, + "grad_norm": 2.2616734504699707, + "learning_rate": 7.60614934114202e-07, + "loss": 1.0439, + "mean_token_accuracy": 0.6847789287567139, + "num_tokens": 53539216.0, + "step": 2079 + }, + { + "epoch": 0.22842082143641554, + "grad_norm": 2.6028268337249756, + "learning_rate": 7.609809663250366e-07, + "loss": 0.9556, + "mean_token_accuracy": 0.7082473635673523, + "num_tokens": 53558834.0, + "step": 2080 + }, + { + "epoch": 0.2285306391390292, + "grad_norm": 2.1063663959503174, + "learning_rate": 7.613469985358711e-07, + "loss": 1.0275, + "mean_token_accuracy": 0.6910150051116943, + "num_tokens": 53587249.0, + "step": 2081 + }, + { + "epoch": 0.2286404568416429, + "grad_norm": 2.4738662242889404, + "learning_rate": 7.617130307467057e-07, + "loss": 0.9877, + "mean_token_accuracy": 0.701708197593689, + "num_tokens": 53610048.0, + "step": 2082 + }, + { + "epoch": 0.22875027454425653, + "grad_norm": 2.1175410747528076, + "learning_rate": 7.620790629575402e-07, + "loss": 1.01, + "mean_token_accuracy": 0.6980440020561218, + "num_tokens": 53639322.0, + "step": 2083 + }, + { + "epoch": 0.2288600922468702, + "grad_norm": 2.413588523864746, + "learning_rate": 7.624450951683748e-07, + "loss": 0.9583, + "mean_token_accuracy": 0.710809588432312, + "num_tokens": 53661912.0, + "step": 2084 + }, + { + "epoch": 0.22896990994948385, + "grad_norm": 2.192713499069214, + "learning_rate": 7.628111273792094e-07, + "loss": 1.0006, + "mean_token_accuracy": 0.7032063603401184, + "num_tokens": 53692100.0, + "step": 2085 + }, + { + "epoch": 0.22907972765209753, + "grad_norm": 2.076669454574585, + "learning_rate": 7.631771595900439e-07, + "loss": 1.0617, + "mean_token_accuracy": 0.6820303201675415, + "num_tokens": 53721317.0, + "step": 2086 + }, + { + "epoch": 0.22918954535471117, + "grad_norm": 2.4772441387176514, + "learning_rate": 7.635431918008785e-07, + "loss": 1.0697, + "mean_token_accuracy": 0.6885683536529541, + "num_tokens": 53745152.0, + "step": 2087 + }, + { + "epoch": 0.22929936305732485, + "grad_norm": 2.643831253051758, + "learning_rate": 7.63909224011713e-07, + "loss": 0.97, + "mean_token_accuracy": 0.7113335132598877, + "num_tokens": 53766720.0, + "step": 2088 + }, + { + "epoch": 0.2294091807599385, + "grad_norm": 2.126378297805786, + "learning_rate": 7.642752562225475e-07, + "loss": 0.992, + "mean_token_accuracy": 0.6953829526901245, + "num_tokens": 53794504.0, + "step": 2089 + }, + { + "epoch": 0.22951899846255217, + "grad_norm": 2.269981622695923, + "learning_rate": 7.646412884333821e-07, + "loss": 1.0227, + "mean_token_accuracy": 0.6967278718948364, + "num_tokens": 53820018.0, + "step": 2090 + }, + { + "epoch": 0.2296288161651658, + "grad_norm": 2.0699174404144287, + "learning_rate": 7.650073206442167e-07, + "loss": 1.0508, + "mean_token_accuracy": 0.6872116327285767, + "num_tokens": 53851435.0, + "step": 2091 + }, + { + "epoch": 0.22973863386777948, + "grad_norm": 2.5429940223693848, + "learning_rate": 7.653733528550512e-07, + "loss": 0.9804, + "mean_token_accuracy": 0.7032345533370972, + "num_tokens": 53873241.0, + "step": 2092 + }, + { + "epoch": 0.22984845157039316, + "grad_norm": 2.1135525703430176, + "learning_rate": 7.657393850658858e-07, + "loss": 1.0525, + "mean_token_accuracy": 0.6880972385406494, + "num_tokens": 53903030.0, + "step": 2093 + }, + { + "epoch": 0.2299582692730068, + "grad_norm": 2.555469036102295, + "learning_rate": 7.661054172767203e-07, + "loss": 1.0197, + "mean_token_accuracy": 0.6952723860740662, + "num_tokens": 53925218.0, + "step": 2094 + }, + { + "epoch": 0.23006808697562048, + "grad_norm": 2.441556453704834, + "learning_rate": 7.664714494875548e-07, + "loss": 1.0771, + "mean_token_accuracy": 0.6847646236419678, + "num_tokens": 53948843.0, + "step": 2095 + }, + { + "epoch": 0.23017790467823412, + "grad_norm": 2.058182716369629, + "learning_rate": 7.668374816983895e-07, + "loss": 0.9743, + "mean_token_accuracy": 0.7035567760467529, + "num_tokens": 53977112.0, + "step": 2096 + }, + { + "epoch": 0.2302877223808478, + "grad_norm": 2.337064027786255, + "learning_rate": 7.67203513909224e-07, + "loss": 1.0052, + "mean_token_accuracy": 0.7025979161262512, + "num_tokens": 54002810.0, + "step": 2097 + }, + { + "epoch": 0.23039754008346144, + "grad_norm": 2.290099620819092, + "learning_rate": 7.675695461200585e-07, + "loss": 1.0018, + "mean_token_accuracy": 0.7052657604217529, + "num_tokens": 54029780.0, + "step": 2098 + }, + { + "epoch": 0.23050735778607512, + "grad_norm": 2.255666971206665, + "learning_rate": 7.679355783308931e-07, + "loss": 1.0767, + "mean_token_accuracy": 0.6857205629348755, + "num_tokens": 54054899.0, + "step": 2099 + }, + { + "epoch": 0.23061717548868876, + "grad_norm": 2.1574947834014893, + "learning_rate": 7.683016105417276e-07, + "loss": 1.1001, + "mean_token_accuracy": 0.6815775632858276, + "num_tokens": 54084137.0, + "step": 2100 + }, + { + "epoch": 0.23072699319130244, + "grad_norm": 2.3367972373962402, + "learning_rate": 7.686676427525623e-07, + "loss": 0.9588, + "mean_token_accuracy": 0.7157549858093262, + "num_tokens": 54108823.0, + "step": 2101 + }, + { + "epoch": 0.2308368108939161, + "grad_norm": 2.4002621173858643, + "learning_rate": 7.690336749633968e-07, + "loss": 0.9792, + "mean_token_accuracy": 0.7019229531288147, + "num_tokens": 54131262.0, + "step": 2102 + }, + { + "epoch": 0.23094662859652976, + "grad_norm": 2.3693184852600098, + "learning_rate": 7.693997071742313e-07, + "loss": 1.0131, + "mean_token_accuracy": 0.6927939653396606, + "num_tokens": 54154640.0, + "step": 2103 + }, + { + "epoch": 0.23105644629914343, + "grad_norm": 2.0028223991394043, + "learning_rate": 7.697657393850659e-07, + "loss": 0.9929, + "mean_token_accuracy": 0.7008158564567566, + "num_tokens": 54185330.0, + "step": 2104 + }, + { + "epoch": 0.23116626400175708, + "grad_norm": 2.5483295917510986, + "learning_rate": 7.701317715959004e-07, + "loss": 0.9628, + "mean_token_accuracy": 0.7085906267166138, + "num_tokens": 54206266.0, + "step": 2105 + }, + { + "epoch": 0.23127608170437075, + "grad_norm": 2.1748673915863037, + "learning_rate": 7.704978038067349e-07, + "loss": 1.074, + "mean_token_accuracy": 0.6766347885131836, + "num_tokens": 54234830.0, + "step": 2106 + }, + { + "epoch": 0.2313858994069844, + "grad_norm": 2.3464601039886475, + "learning_rate": 7.708638360175696e-07, + "loss": 0.8863, + "mean_token_accuracy": 0.723684549331665, + "num_tokens": 54258518.0, + "step": 2107 + }, + { + "epoch": 0.23149571710959807, + "grad_norm": 2.184946060180664, + "learning_rate": 7.712298682284041e-07, + "loss": 1.0783, + "mean_token_accuracy": 0.6801700592041016, + "num_tokens": 54285139.0, + "step": 2108 + }, + { + "epoch": 0.23160553481221172, + "grad_norm": 2.382205009460449, + "learning_rate": 7.715959004392386e-07, + "loss": 0.9837, + "mean_token_accuracy": 0.7018136382102966, + "num_tokens": 54308341.0, + "step": 2109 + }, + { + "epoch": 0.2317153525148254, + "grad_norm": 2.195338726043701, + "learning_rate": 7.719619326500732e-07, + "loss": 1.1336, + "mean_token_accuracy": 0.6649538278579712, + "num_tokens": 54338009.0, + "step": 2110 + }, + { + "epoch": 0.23182517021743906, + "grad_norm": 2.4697184562683105, + "learning_rate": 7.723279648609077e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6807138323783875, + "num_tokens": 54361156.0, + "step": 2111 + }, + { + "epoch": 0.2319349879200527, + "grad_norm": 1.9449551105499268, + "learning_rate": 7.726939970717423e-07, + "loss": 0.8923, + "mean_token_accuracy": 0.7273458242416382, + "num_tokens": 54390645.0, + "step": 2112 + }, + { + "epoch": 0.23204480562266638, + "grad_norm": 2.4477128982543945, + "learning_rate": 7.730600292825769e-07, + "loss": 0.9941, + "mean_token_accuracy": 0.6947113871574402, + "num_tokens": 54412841.0, + "step": 2113 + }, + { + "epoch": 0.23215462332528003, + "grad_norm": 2.1313741207122803, + "learning_rate": 7.734260614934114e-07, + "loss": 0.9695, + "mean_token_accuracy": 0.7042170763015747, + "num_tokens": 54440731.0, + "step": 2114 + }, + { + "epoch": 0.2322644410278937, + "grad_norm": 2.7091541290283203, + "learning_rate": 7.737920937042459e-07, + "loss": 1.0107, + "mean_token_accuracy": 0.6936259269714355, + "num_tokens": 54460803.0, + "step": 2115 + }, + { + "epoch": 0.23237425873050735, + "grad_norm": 2.4670135974884033, + "learning_rate": 7.741581259150805e-07, + "loss": 0.986, + "mean_token_accuracy": 0.7080355882644653, + "num_tokens": 54483265.0, + "step": 2116 + }, + { + "epoch": 0.23248407643312102, + "grad_norm": 2.155243396759033, + "learning_rate": 7.745241581259151e-07, + "loss": 0.9864, + "mean_token_accuracy": 0.701980710029602, + "num_tokens": 54509995.0, + "step": 2117 + }, + { + "epoch": 0.23259389413573467, + "grad_norm": 2.391850471496582, + "learning_rate": 7.748901903367497e-07, + "loss": 0.9831, + "mean_token_accuracy": 0.7003142833709717, + "num_tokens": 54533911.0, + "step": 2118 + }, + { + "epoch": 0.23270371183834834, + "grad_norm": 2.5354902744293213, + "learning_rate": 7.752562225475842e-07, + "loss": 0.9894, + "mean_token_accuracy": 0.6997088193893433, + "num_tokens": 54555628.0, + "step": 2119 + }, + { + "epoch": 0.23281352954096202, + "grad_norm": 2.1267096996307373, + "learning_rate": 7.756222547584187e-07, + "loss": 1.0419, + "mean_token_accuracy": 0.6918469667434692, + "num_tokens": 54584663.0, + "step": 2120 + }, + { + "epoch": 0.23292334724357566, + "grad_norm": 2.6290102005004883, + "learning_rate": 7.759882869692533e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.7018557786941528, + "num_tokens": 54606332.0, + "step": 2121 + }, + { + "epoch": 0.23303316494618934, + "grad_norm": 2.0485787391662598, + "learning_rate": 7.763543191800878e-07, + "loss": 1.1006, + "mean_token_accuracy": 0.6765363216400146, + "num_tokens": 54639008.0, + "step": 2122 + }, + { + "epoch": 0.23314298264880298, + "grad_norm": 2.1211531162261963, + "learning_rate": 7.767203513909224e-07, + "loss": 1.041, + "mean_token_accuracy": 0.6855889558792114, + "num_tokens": 54668941.0, + "step": 2123 + }, + { + "epoch": 0.23325280035141666, + "grad_norm": 2.4807679653167725, + "learning_rate": 7.77086383601757e-07, + "loss": 0.9899, + "mean_token_accuracy": 0.6996324062347412, + "num_tokens": 54691348.0, + "step": 2124 + }, + { + "epoch": 0.2333626180540303, + "grad_norm": 2.2534890174865723, + "learning_rate": 7.774524158125915e-07, + "loss": 0.9811, + "mean_token_accuracy": 0.7058374881744385, + "num_tokens": 54715840.0, + "step": 2125 + }, + { + "epoch": 0.23347243575664398, + "grad_norm": 2.278053045272827, + "learning_rate": 7.77818448023426e-07, + "loss": 1.108, + "mean_token_accuracy": 0.6681499481201172, + "num_tokens": 54742261.0, + "step": 2126 + }, + { + "epoch": 0.23358225345925762, + "grad_norm": 1.9679501056671143, + "learning_rate": 7.781844802342606e-07, + "loss": 1.0918, + "mean_token_accuracy": 0.6729052066802979, + "num_tokens": 54775001.0, + "step": 2127 + }, + { + "epoch": 0.2336920711618713, + "grad_norm": 2.2787063121795654, + "learning_rate": 7.785505124450952e-07, + "loss": 1.066, + "mean_token_accuracy": 0.6824033260345459, + "num_tokens": 54800287.0, + "step": 2128 + }, + { + "epoch": 0.23380188886448494, + "grad_norm": 2.2067699432373047, + "learning_rate": 7.789165446559297e-07, + "loss": 1.0869, + "mean_token_accuracy": 0.6817241311073303, + "num_tokens": 54827212.0, + "step": 2129 + }, + { + "epoch": 0.23391170656709862, + "grad_norm": 2.2311723232269287, + "learning_rate": 7.792825768667643e-07, + "loss": 1.1597, + "mean_token_accuracy": 0.6586935520172119, + "num_tokens": 54855265.0, + "step": 2130 + }, + { + "epoch": 0.2340215242697123, + "grad_norm": 2.4303815364837646, + "learning_rate": 7.796486090775988e-07, + "loss": 1.0707, + "mean_token_accuracy": 0.6774979829788208, + "num_tokens": 54880558.0, + "step": 2131 + }, + { + "epoch": 0.23413134197232593, + "grad_norm": 2.201418876647949, + "learning_rate": 7.800146412884333e-07, + "loss": 1.1037, + "mean_token_accuracy": 0.6745655536651611, + "num_tokens": 54908760.0, + "step": 2132 + }, + { + "epoch": 0.2342411596749396, + "grad_norm": 2.4350662231445312, + "learning_rate": 7.80380673499268e-07, + "loss": 0.92, + "mean_token_accuracy": 0.7149373292922974, + "num_tokens": 54927989.0, + "step": 2133 + }, + { + "epoch": 0.23435097737755325, + "grad_norm": 2.1735923290252686, + "learning_rate": 7.807467057101025e-07, + "loss": 1.0473, + "mean_token_accuracy": 0.6868649125099182, + "num_tokens": 54955289.0, + "step": 2134 + }, + { + "epoch": 0.23446079508016693, + "grad_norm": 2.2488698959350586, + "learning_rate": 7.811127379209371e-07, + "loss": 0.9778, + "mean_token_accuracy": 0.699059009552002, + "num_tokens": 54981124.0, + "step": 2135 + }, + { + "epoch": 0.23457061278278057, + "grad_norm": 2.117642879486084, + "learning_rate": 7.814787701317716e-07, + "loss": 1.0691, + "mean_token_accuracy": 0.683396577835083, + "num_tokens": 55009205.0, + "step": 2136 + }, + { + "epoch": 0.23468043048539425, + "grad_norm": 2.239741802215576, + "learning_rate": 7.818448023426061e-07, + "loss": 1.1208, + "mean_token_accuracy": 0.6680766344070435, + "num_tokens": 55036118.0, + "step": 2137 + }, + { + "epoch": 0.2347902481880079, + "grad_norm": 2.2673745155334473, + "learning_rate": 7.822108345534407e-07, + "loss": 0.9571, + "mean_token_accuracy": 0.7037025094032288, + "num_tokens": 55060467.0, + "step": 2138 + }, + { + "epoch": 0.23490006589062157, + "grad_norm": 2.246171236038208, + "learning_rate": 7.825768667642753e-07, + "loss": 1.0965, + "mean_token_accuracy": 0.6741610765457153, + "num_tokens": 55087375.0, + "step": 2139 + }, + { + "epoch": 0.23500988359323524, + "grad_norm": 2.2766878604888916, + "learning_rate": 7.829428989751098e-07, + "loss": 1.0291, + "mean_token_accuracy": 0.6978285312652588, + "num_tokens": 55112810.0, + "step": 2140 + }, + { + "epoch": 0.2351197012958489, + "grad_norm": 2.5210325717926025, + "learning_rate": 7.833089311859444e-07, + "loss": 1.0209, + "mean_token_accuracy": 0.699005126953125, + "num_tokens": 55134221.0, + "step": 2141 + }, + { + "epoch": 0.23522951899846256, + "grad_norm": 2.175321102142334, + "learning_rate": 7.836749633967789e-07, + "loss": 1.0299, + "mean_token_accuracy": 0.6951518058776855, + "num_tokens": 55161282.0, + "step": 2142 + }, + { + "epoch": 0.2353393367010762, + "grad_norm": 2.3625810146331787, + "learning_rate": 7.840409956076133e-07, + "loss": 1.0512, + "mean_token_accuracy": 0.6882897615432739, + "num_tokens": 55185586.0, + "step": 2143 + }, + { + "epoch": 0.23544915440368988, + "grad_norm": 2.109564781188965, + "learning_rate": 7.844070278184481e-07, + "loss": 1.1547, + "mean_token_accuracy": 0.663184642791748, + "num_tokens": 55214146.0, + "step": 2144 + }, + { + "epoch": 0.23555897210630353, + "grad_norm": 2.522139072418213, + "learning_rate": 7.847730600292826e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.6937496066093445, + "num_tokens": 55234676.0, + "step": 2145 + }, + { + "epoch": 0.2356687898089172, + "grad_norm": 2.320432424545288, + "learning_rate": 7.85139092240117e-07, + "loss": 0.9357, + "mean_token_accuracy": 0.711510181427002, + "num_tokens": 55257852.0, + "step": 2146 + }, + { + "epoch": 0.23577860751153085, + "grad_norm": 2.460357427597046, + "learning_rate": 7.855051244509517e-07, + "loss": 1.0161, + "mean_token_accuracy": 0.693630576133728, + "num_tokens": 55280103.0, + "step": 2147 + }, + { + "epoch": 0.23588842521414452, + "grad_norm": 2.2225844860076904, + "learning_rate": 7.858711566617861e-07, + "loss": 1.0526, + "mean_token_accuracy": 0.6796634197235107, + "num_tokens": 55306689.0, + "step": 2148 + }, + { + "epoch": 0.2359982429167582, + "grad_norm": 2.1969621181488037, + "learning_rate": 7.862371888726209e-07, + "loss": 1.0883, + "mean_token_accuracy": 0.6761320233345032, + "num_tokens": 55335557.0, + "step": 2149 + }, + { + "epoch": 0.23610806061937184, + "grad_norm": 2.3852579593658447, + "learning_rate": 7.866032210834554e-07, + "loss": 1.0379, + "mean_token_accuracy": 0.6885336637496948, + "num_tokens": 55361347.0, + "step": 2150 + }, + { + "epoch": 0.2362178783219855, + "grad_norm": 2.329367160797119, + "learning_rate": 7.869692532942898e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.6883593797683716, + "num_tokens": 55386252.0, + "step": 2151 + }, + { + "epoch": 0.23632769602459916, + "grad_norm": 2.404747724533081, + "learning_rate": 7.873352855051244e-07, + "loss": 1.056, + "mean_token_accuracy": 0.6946128606796265, + "num_tokens": 55407899.0, + "step": 2152 + }, + { + "epoch": 0.23643751372721283, + "grad_norm": 2.3359878063201904, + "learning_rate": 7.877013177159589e-07, + "loss": 0.8684, + "mean_token_accuracy": 0.7300180196762085, + "num_tokens": 55430617.0, + "step": 2153 + }, + { + "epoch": 0.23654733142982648, + "grad_norm": 2.5899465084075928, + "learning_rate": 7.880673499267934e-07, + "loss": 0.857, + "mean_token_accuracy": 0.7321422100067139, + "num_tokens": 55449753.0, + "step": 2154 + }, + { + "epoch": 0.23665714913244015, + "grad_norm": 2.2224247455596924, + "learning_rate": 7.884333821376281e-07, + "loss": 1.0415, + "mean_token_accuracy": 0.6876773238182068, + "num_tokens": 55477552.0, + "step": 2155 + }, + { + "epoch": 0.2367669668350538, + "grad_norm": 2.3454113006591797, + "learning_rate": 7.887994143484626e-07, + "loss": 0.9915, + "mean_token_accuracy": 0.7019040584564209, + "num_tokens": 55500986.0, + "step": 2156 + }, + { + "epoch": 0.23687678453766747, + "grad_norm": 2.0828123092651367, + "learning_rate": 7.891654465592971e-07, + "loss": 1.0561, + "mean_token_accuracy": 0.6872153282165527, + "num_tokens": 55533557.0, + "step": 2157 + }, + { + "epoch": 0.23698660224028115, + "grad_norm": 2.1754207611083984, + "learning_rate": 7.895314787701317e-07, + "loss": 1.0505, + "mean_token_accuracy": 0.6812288761138916, + "num_tokens": 55562766.0, + "step": 2158 + }, + { + "epoch": 0.2370964199428948, + "grad_norm": 2.2568373680114746, + "learning_rate": 7.898975109809662e-07, + "loss": 1.1049, + "mean_token_accuracy": 0.6690568923950195, + "num_tokens": 55590781.0, + "step": 2159 + }, + { + "epoch": 0.23720623764550847, + "grad_norm": 2.378204345703125, + "learning_rate": 7.902635431918008e-07, + "loss": 1.0012, + "mean_token_accuracy": 0.6989362239837646, + "num_tokens": 55615755.0, + "step": 2160 + }, + { + "epoch": 0.2373160553481221, + "grad_norm": 2.3207457065582275, + "learning_rate": 7.906295754026354e-07, + "loss": 0.9894, + "mean_token_accuracy": 0.701683759689331, + "num_tokens": 55639958.0, + "step": 2161 + }, + { + "epoch": 0.2374258730507358, + "grad_norm": 2.521367311477661, + "learning_rate": 7.909956076134699e-07, + "loss": 1.0641, + "mean_token_accuracy": 0.6824328899383545, + "num_tokens": 55661257.0, + "step": 2162 + }, + { + "epoch": 0.23753569075334943, + "grad_norm": 2.194868803024292, + "learning_rate": 7.913616398243044e-07, + "loss": 0.9416, + "mean_token_accuracy": 0.7206399440765381, + "num_tokens": 55687147.0, + "step": 2163 + }, + { + "epoch": 0.2376455084559631, + "grad_norm": 2.0130627155303955, + "learning_rate": 7.91727672035139e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6841058731079102, + "num_tokens": 55718756.0, + "step": 2164 + }, + { + "epoch": 0.23775532615857675, + "grad_norm": 2.2927002906799316, + "learning_rate": 7.920937042459736e-07, + "loss": 1.1008, + "mean_token_accuracy": 0.6744195222854614, + "num_tokens": 55745692.0, + "step": 2165 + }, + { + "epoch": 0.23786514386119043, + "grad_norm": 2.225327730178833, + "learning_rate": 7.924597364568082e-07, + "loss": 0.9955, + "mean_token_accuracy": 0.7056720852851868, + "num_tokens": 55771595.0, + "step": 2166 + }, + { + "epoch": 0.23797496156380407, + "grad_norm": 2.2755370140075684, + "learning_rate": 7.928257686676427e-07, + "loss": 1.0267, + "mean_token_accuracy": 0.6883313655853271, + "num_tokens": 55795574.0, + "step": 2167 + }, + { + "epoch": 0.23808477926641775, + "grad_norm": 2.184783697128296, + "learning_rate": 7.931918008784772e-07, + "loss": 1.0622, + "mean_token_accuracy": 0.6846435070037842, + "num_tokens": 55823116.0, + "step": 2168 + }, + { + "epoch": 0.23819459696903142, + "grad_norm": 2.1769542694091797, + "learning_rate": 7.935578330893118e-07, + "loss": 1.0323, + "mean_token_accuracy": 0.6865707635879517, + "num_tokens": 55849908.0, + "step": 2169 + }, + { + "epoch": 0.23830441467164507, + "grad_norm": 2.0817923545837402, + "learning_rate": 7.939238653001463e-07, + "loss": 1.0587, + "mean_token_accuracy": 0.6843485236167908, + "num_tokens": 55879307.0, + "step": 2170 + }, + { + "epoch": 0.23841423237425874, + "grad_norm": 2.351224660873413, + "learning_rate": 7.942898975109809e-07, + "loss": 1.0978, + "mean_token_accuracy": 0.6765799522399902, + "num_tokens": 55907879.0, + "step": 2171 + }, + { + "epoch": 0.23852405007687238, + "grad_norm": 2.5105884075164795, + "learning_rate": 7.946559297218155e-07, + "loss": 1.1093, + "mean_token_accuracy": 0.6766668558120728, + "num_tokens": 55930065.0, + "step": 2172 + }, + { + "epoch": 0.23863386777948606, + "grad_norm": 2.3111605644226074, + "learning_rate": 7.9502196193265e-07, + "loss": 1.0089, + "mean_token_accuracy": 0.6953766345977783, + "num_tokens": 55954500.0, + "step": 2173 + }, + { + "epoch": 0.2387436854820997, + "grad_norm": 2.2163608074188232, + "learning_rate": 7.953879941434845e-07, + "loss": 0.9358, + "mean_token_accuracy": 0.7131621241569519, + "num_tokens": 55980294.0, + "step": 2174 + }, + { + "epoch": 0.23885350318471338, + "grad_norm": 1.9382822513580322, + "learning_rate": 7.957540263543191e-07, + "loss": 1.0224, + "mean_token_accuracy": 0.69572514295578, + "num_tokens": 56012046.0, + "step": 2175 + }, + { + "epoch": 0.23896332088732702, + "grad_norm": 2.2383530139923096, + "learning_rate": 7.961200585651537e-07, + "loss": 1.0081, + "mean_token_accuracy": 0.6977840065956116, + "num_tokens": 56039172.0, + "step": 2176 + }, + { + "epoch": 0.2390731385899407, + "grad_norm": 2.471680164337158, + "learning_rate": 7.964860907759882e-07, + "loss": 1.0008, + "mean_token_accuracy": 0.701984167098999, + "num_tokens": 56061746.0, + "step": 2177 + }, + { + "epoch": 0.23918295629255437, + "grad_norm": 2.3394649028778076, + "learning_rate": 7.968521229868228e-07, + "loss": 1.0427, + "mean_token_accuracy": 0.6817692518234253, + "num_tokens": 56086553.0, + "step": 2178 + }, + { + "epoch": 0.23929277399516802, + "grad_norm": 2.1243603229522705, + "learning_rate": 7.972181551976573e-07, + "loss": 1.1, + "mean_token_accuracy": 0.6739526987075806, + "num_tokens": 56118829.0, + "step": 2179 + }, + { + "epoch": 0.2394025916977817, + "grad_norm": 2.1831068992614746, + "learning_rate": 7.975841874084918e-07, + "loss": 1.0473, + "mean_token_accuracy": 0.6868184208869934, + "num_tokens": 56145714.0, + "step": 2180 + }, + { + "epoch": 0.23951240940039534, + "grad_norm": 2.4054200649261475, + "learning_rate": 7.979502196193265e-07, + "loss": 1.0462, + "mean_token_accuracy": 0.6837315559387207, + "num_tokens": 56170868.0, + "step": 2181 + }, + { + "epoch": 0.239622227103009, + "grad_norm": 2.3742969036102295, + "learning_rate": 7.98316251830161e-07, + "loss": 1.0208, + "mean_token_accuracy": 0.6946503520011902, + "num_tokens": 56195851.0, + "step": 2182 + }, + { + "epoch": 0.23973204480562266, + "grad_norm": 2.0330071449279785, + "learning_rate": 7.986822840409956e-07, + "loss": 1.0768, + "mean_token_accuracy": 0.6877269148826599, + "num_tokens": 56227161.0, + "step": 2183 + }, + { + "epoch": 0.23984186250823633, + "grad_norm": 2.4474267959594727, + "learning_rate": 7.990483162518301e-07, + "loss": 1.087, + "mean_token_accuracy": 0.677788257598877, + "num_tokens": 56249842.0, + "step": 2184 + }, + { + "epoch": 0.23995168021084998, + "grad_norm": 2.230593204498291, + "learning_rate": 7.994143484626646e-07, + "loss": 1.0423, + "mean_token_accuracy": 0.6887885332107544, + "num_tokens": 56276810.0, + "step": 2185 + }, + { + "epoch": 0.24006149791346365, + "grad_norm": 2.2437398433685303, + "learning_rate": 7.997803806734992e-07, + "loss": 0.9921, + "mean_token_accuracy": 0.6977744102478027, + "num_tokens": 56301375.0, + "step": 2186 + }, + { + "epoch": 0.24017131561607732, + "grad_norm": 2.2056562900543213, + "learning_rate": 8.001464128843338e-07, + "loss": 1.0512, + "mean_token_accuracy": 0.683505654335022, + "num_tokens": 56327061.0, + "step": 2187 + }, + { + "epoch": 0.24028113331869097, + "grad_norm": 2.103224277496338, + "learning_rate": 8.005124450951683e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.6945438981056213, + "num_tokens": 56357185.0, + "step": 2188 + }, + { + "epoch": 0.24039095102130464, + "grad_norm": 2.390155076980591, + "learning_rate": 8.008784773060029e-07, + "loss": 1.0445, + "mean_token_accuracy": 0.6864049434661865, + "num_tokens": 56380411.0, + "step": 2189 + }, + { + "epoch": 0.2405007687239183, + "grad_norm": 2.5178701877593994, + "learning_rate": 8.012445095168374e-07, + "loss": 1.0018, + "mean_token_accuracy": 0.703677237033844, + "num_tokens": 56403539.0, + "step": 2190 + }, + { + "epoch": 0.24061058642653196, + "grad_norm": 2.215627908706665, + "learning_rate": 8.016105417276719e-07, + "loss": 1.0087, + "mean_token_accuracy": 0.7009602785110474, + "num_tokens": 56430207.0, + "step": 2191 + }, + { + "epoch": 0.2407204041291456, + "grad_norm": 2.0596415996551514, + "learning_rate": 8.019765739385066e-07, + "loss": 1.0833, + "mean_token_accuracy": 0.674303412437439, + "num_tokens": 56460386.0, + "step": 2192 + }, + { + "epoch": 0.24083022183175928, + "grad_norm": 2.179203987121582, + "learning_rate": 8.023426061493411e-07, + "loss": 0.9326, + "mean_token_accuracy": 0.7174067497253418, + "num_tokens": 56487140.0, + "step": 2193 + }, + { + "epoch": 0.24094003953437293, + "grad_norm": 2.5659573078155518, + "learning_rate": 8.027086383601756e-07, + "loss": 1.1057, + "mean_token_accuracy": 0.6705483198165894, + "num_tokens": 56508781.0, + "step": 2194 + }, + { + "epoch": 0.2410498572369866, + "grad_norm": 2.372081756591797, + "learning_rate": 8.030746705710102e-07, + "loss": 1.0403, + "mean_token_accuracy": 0.6869010925292969, + "num_tokens": 56532472.0, + "step": 2195 + }, + { + "epoch": 0.24115967493960028, + "grad_norm": 1.9416017532348633, + "learning_rate": 8.034407027818447e-07, + "loss": 1.0747, + "mean_token_accuracy": 0.683014452457428, + "num_tokens": 56567185.0, + "step": 2196 + }, + { + "epoch": 0.24126949264221392, + "grad_norm": 2.0333855152130127, + "learning_rate": 8.038067349926793e-07, + "loss": 0.9838, + "mean_token_accuracy": 0.7069873213768005, + "num_tokens": 56596186.0, + "step": 2197 + }, + { + "epoch": 0.2413793103448276, + "grad_norm": 2.243401527404785, + "learning_rate": 8.041727672035139e-07, + "loss": 0.9672, + "mean_token_accuracy": 0.7011184692382812, + "num_tokens": 56621026.0, + "step": 2198 + }, + { + "epoch": 0.24148912804744124, + "grad_norm": 2.515585422515869, + "learning_rate": 8.045387994143484e-07, + "loss": 0.9937, + "mean_token_accuracy": 0.700768232345581, + "num_tokens": 56641149.0, + "step": 2199 + }, + { + "epoch": 0.24159894575005492, + "grad_norm": 2.3755786418914795, + "learning_rate": 8.04904831625183e-07, + "loss": 0.9816, + "mean_token_accuracy": 0.7022809386253357, + "num_tokens": 56663743.0, + "step": 2200 + }, + { + "epoch": 0.24170876345266856, + "grad_norm": 2.1024818420410156, + "learning_rate": 8.052708638360175e-07, + "loss": 0.9649, + "mean_token_accuracy": 0.7159525752067566, + "num_tokens": 56690487.0, + "step": 2201 + }, + { + "epoch": 0.24181858115528224, + "grad_norm": 2.3819003105163574, + "learning_rate": 8.05636896046852e-07, + "loss": 0.9846, + "mean_token_accuracy": 0.6959570646286011, + "num_tokens": 56714174.0, + "step": 2202 + }, + { + "epoch": 0.24192839885789588, + "grad_norm": 2.2138216495513916, + "learning_rate": 8.060029282576867e-07, + "loss": 1.0904, + "mean_token_accuracy": 0.6769012808799744, + "num_tokens": 56743225.0, + "step": 2203 + }, + { + "epoch": 0.24203821656050956, + "grad_norm": 2.3205149173736572, + "learning_rate": 8.063689604685212e-07, + "loss": 1.0518, + "mean_token_accuracy": 0.6813576221466064, + "num_tokens": 56769588.0, + "step": 2204 + }, + { + "epoch": 0.2421480342631232, + "grad_norm": 2.3137972354888916, + "learning_rate": 8.067349926793557e-07, + "loss": 1.0165, + "mean_token_accuracy": 0.694313645362854, + "num_tokens": 56793483.0, + "step": 2205 + }, + { + "epoch": 0.24225785196573688, + "grad_norm": 2.0831239223480225, + "learning_rate": 8.071010248901903e-07, + "loss": 1.0639, + "mean_token_accuracy": 0.6893069744110107, + "num_tokens": 56821789.0, + "step": 2206 + }, + { + "epoch": 0.24236766966835055, + "grad_norm": 2.2404518127441406, + "learning_rate": 8.074670571010248e-07, + "loss": 0.9567, + "mean_token_accuracy": 0.706379771232605, + "num_tokens": 56847788.0, + "step": 2207 + }, + { + "epoch": 0.2424774873709642, + "grad_norm": 2.261255979537964, + "learning_rate": 8.078330893118594e-07, + "loss": 0.9674, + "mean_token_accuracy": 0.7090303301811218, + "num_tokens": 56874038.0, + "step": 2208 + }, + { + "epoch": 0.24258730507357787, + "grad_norm": 2.4859251976013184, + "learning_rate": 8.08199121522694e-07, + "loss": 1.0404, + "mean_token_accuracy": 0.6898172497749329, + "num_tokens": 56895943.0, + "step": 2209 + }, + { + "epoch": 0.24269712277619152, + "grad_norm": 2.333634853363037, + "learning_rate": 8.085651537335285e-07, + "loss": 0.9836, + "mean_token_accuracy": 0.70143723487854, + "num_tokens": 56920459.0, + "step": 2210 + }, + { + "epoch": 0.2428069404788052, + "grad_norm": 2.2548978328704834, + "learning_rate": 8.08931185944363e-07, + "loss": 1.1119, + "mean_token_accuracy": 0.672156572341919, + "num_tokens": 56947307.0, + "step": 2211 + }, + { + "epoch": 0.24291675818141883, + "grad_norm": 2.194898843765259, + "learning_rate": 8.092972181551976e-07, + "loss": 1.0384, + "mean_token_accuracy": 0.6902453899383545, + "num_tokens": 56974344.0, + "step": 2212 + }, + { + "epoch": 0.2430265758840325, + "grad_norm": 2.0801851749420166, + "learning_rate": 8.096632503660322e-07, + "loss": 1.1601, + "mean_token_accuracy": 0.668372392654419, + "num_tokens": 57005194.0, + "step": 2213 + }, + { + "epoch": 0.24313639358664615, + "grad_norm": 2.4653217792510986, + "learning_rate": 8.100292825768668e-07, + "loss": 0.9571, + "mean_token_accuracy": 0.7031634449958801, + "num_tokens": 57027359.0, + "step": 2214 + }, + { + "epoch": 0.24324621128925983, + "grad_norm": 2.4320485591888428, + "learning_rate": 8.103953147877013e-07, + "loss": 0.9921, + "mean_token_accuracy": 0.7001507878303528, + "num_tokens": 57050898.0, + "step": 2215 + }, + { + "epoch": 0.2433560289918735, + "grad_norm": 2.3581595420837402, + "learning_rate": 8.107613469985358e-07, + "loss": 0.9635, + "mean_token_accuracy": 0.7011392116546631, + "num_tokens": 57075058.0, + "step": 2216 + }, + { + "epoch": 0.24346584669448715, + "grad_norm": 2.2680375576019287, + "learning_rate": 8.111273792093704e-07, + "loss": 0.9591, + "mean_token_accuracy": 0.7086721658706665, + "num_tokens": 57100386.0, + "step": 2217 + }, + { + "epoch": 0.24357566439710082, + "grad_norm": 2.1893858909606934, + "learning_rate": 8.114934114202049e-07, + "loss": 0.9978, + "mean_token_accuracy": 0.6979248523712158, + "num_tokens": 57126772.0, + "step": 2218 + }, + { + "epoch": 0.24368548209971447, + "grad_norm": 2.281113386154175, + "learning_rate": 8.118594436310395e-07, + "loss": 1.0437, + "mean_token_accuracy": 0.6889897584915161, + "num_tokens": 57152044.0, + "step": 2219 + }, + { + "epoch": 0.24379529980232814, + "grad_norm": 2.2502121925354004, + "learning_rate": 8.122254758418741e-07, + "loss": 1.0448, + "mean_token_accuracy": 0.6916360855102539, + "num_tokens": 57176202.0, + "step": 2220 + }, + { + "epoch": 0.2439051175049418, + "grad_norm": 2.4642879962921143, + "learning_rate": 8.125915080527086e-07, + "loss": 0.9752, + "mean_token_accuracy": 0.7030249238014221, + "num_tokens": 57198532.0, + "step": 2221 + }, + { + "epoch": 0.24401493520755546, + "grad_norm": 2.1635100841522217, + "learning_rate": 8.129575402635431e-07, + "loss": 1.0295, + "mean_token_accuracy": 0.6905651092529297, + "num_tokens": 57228410.0, + "step": 2222 + }, + { + "epoch": 0.2441247529101691, + "grad_norm": 2.030136823654175, + "learning_rate": 8.133235724743777e-07, + "loss": 0.9787, + "mean_token_accuracy": 0.7027027606964111, + "num_tokens": 57257091.0, + "step": 2223 + }, + { + "epoch": 0.24423457061278278, + "grad_norm": 2.246030807495117, + "learning_rate": 8.136896046852123e-07, + "loss": 0.9904, + "mean_token_accuracy": 0.6980575323104858, + "num_tokens": 57282667.0, + "step": 2224 + }, + { + "epoch": 0.24434438831539645, + "grad_norm": 2.5017712116241455, + "learning_rate": 8.140556368960468e-07, + "loss": 0.9391, + "mean_token_accuracy": 0.7113152742385864, + "num_tokens": 57301885.0, + "step": 2225 + }, + { + "epoch": 0.2444542060180101, + "grad_norm": 2.5803353786468506, + "learning_rate": 8.144216691068814e-07, + "loss": 1.027, + "mean_token_accuracy": 0.7010786533355713, + "num_tokens": 57322637.0, + "step": 2226 + }, + { + "epoch": 0.24456402372062377, + "grad_norm": 2.479475736618042, + "learning_rate": 8.147877013177159e-07, + "loss": 1.0655, + "mean_token_accuracy": 0.6874811053276062, + "num_tokens": 57346366.0, + "step": 2227 + }, + { + "epoch": 0.24467384142323742, + "grad_norm": 2.2593319416046143, + "learning_rate": 8.151537335285504e-07, + "loss": 1.0484, + "mean_token_accuracy": 0.6953960061073303, + "num_tokens": 57373168.0, + "step": 2228 + }, + { + "epoch": 0.2447836591258511, + "grad_norm": 2.1535823345184326, + "learning_rate": 8.155197657393851e-07, + "loss": 0.9944, + "mean_token_accuracy": 0.6960897445678711, + "num_tokens": 57402621.0, + "step": 2229 + }, + { + "epoch": 0.24489347682846474, + "grad_norm": 2.4154984951019287, + "learning_rate": 8.158857979502196e-07, + "loss": 1.046, + "mean_token_accuracy": 0.6954717636108398, + "num_tokens": 57426230.0, + "step": 2230 + }, + { + "epoch": 0.2450032945310784, + "grad_norm": 2.226860284805298, + "learning_rate": 8.162518301610542e-07, + "loss": 0.9916, + "mean_token_accuracy": 0.7025713920593262, + "num_tokens": 57453725.0, + "step": 2231 + }, + { + "epoch": 0.24511311223369206, + "grad_norm": 2.0753870010375977, + "learning_rate": 8.166178623718887e-07, + "loss": 1.054, + "mean_token_accuracy": 0.6848155856132507, + "num_tokens": 57484306.0, + "step": 2232 + }, + { + "epoch": 0.24522292993630573, + "grad_norm": 2.272854804992676, + "learning_rate": 8.169838945827232e-07, + "loss": 1.0395, + "mean_token_accuracy": 0.6912485957145691, + "num_tokens": 57510966.0, + "step": 2233 + }, + { + "epoch": 0.2453327476389194, + "grad_norm": 2.1889519691467285, + "learning_rate": 8.173499267935578e-07, + "loss": 1.0224, + "mean_token_accuracy": 0.6913892030715942, + "num_tokens": 57538533.0, + "step": 2234 + }, + { + "epoch": 0.24544256534153305, + "grad_norm": 2.2935543060302734, + "learning_rate": 8.177159590043924e-07, + "loss": 1.0117, + "mean_token_accuracy": 0.6923278570175171, + "num_tokens": 57564275.0, + "step": 2235 + }, + { + "epoch": 0.24555238304414673, + "grad_norm": 2.0468807220458984, + "learning_rate": 8.180819912152269e-07, + "loss": 1.0018, + "mean_token_accuracy": 0.7002115249633789, + "num_tokens": 57594837.0, + "step": 2236 + }, + { + "epoch": 0.24566220074676037, + "grad_norm": 2.246760606765747, + "learning_rate": 8.184480234260615e-07, + "loss": 0.9774, + "mean_token_accuracy": 0.7134082317352295, + "num_tokens": 57620478.0, + "step": 2237 + }, + { + "epoch": 0.24577201844937405, + "grad_norm": 2.245450258255005, + "learning_rate": 8.18814055636896e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.6829566955566406, + "num_tokens": 57647752.0, + "step": 2238 + }, + { + "epoch": 0.2458818361519877, + "grad_norm": 2.318263053894043, + "learning_rate": 8.191800878477305e-07, + "loss": 0.9957, + "mean_token_accuracy": 0.6996004581451416, + "num_tokens": 57672591.0, + "step": 2239 + }, + { + "epoch": 0.24599165385460137, + "grad_norm": 2.267477512359619, + "learning_rate": 8.195461200585652e-07, + "loss": 0.9835, + "mean_token_accuracy": 0.7022160887718201, + "num_tokens": 57696420.0, + "step": 2240 + }, + { + "epoch": 0.246101471557215, + "grad_norm": 2.3693437576293945, + "learning_rate": 8.199121522693997e-07, + "loss": 1.015, + "mean_token_accuracy": 0.6914263367652893, + "num_tokens": 57720386.0, + "step": 2241 + }, + { + "epoch": 0.2462112892598287, + "grad_norm": 2.5244646072387695, + "learning_rate": 8.202781844802342e-07, + "loss": 1.0319, + "mean_token_accuracy": 0.6950505375862122, + "num_tokens": 57746173.0, + "step": 2242 + }, + { + "epoch": 0.24632110696244233, + "grad_norm": 2.3772668838500977, + "learning_rate": 8.206442166910688e-07, + "loss": 1.0281, + "mean_token_accuracy": 0.6906523704528809, + "num_tokens": 57770085.0, + "step": 2243 + }, + { + "epoch": 0.246430924665056, + "grad_norm": 2.0378897190093994, + "learning_rate": 8.210102489019033e-07, + "loss": 0.9405, + "mean_token_accuracy": 0.7110406756401062, + "num_tokens": 57799026.0, + "step": 2244 + }, + { + "epoch": 0.24654074236766968, + "grad_norm": 2.2495689392089844, + "learning_rate": 8.213762811127379e-07, + "loss": 1.0169, + "mean_token_accuracy": 0.6945355534553528, + "num_tokens": 57824626.0, + "step": 2245 + }, + { + "epoch": 0.24665056007028333, + "grad_norm": 2.496551513671875, + "learning_rate": 8.217423133235725e-07, + "loss": 1.0366, + "mean_token_accuracy": 0.6816045045852661, + "num_tokens": 57847648.0, + "step": 2246 + }, + { + "epoch": 0.246760377772897, + "grad_norm": 2.3594367504119873, + "learning_rate": 8.22108345534407e-07, + "loss": 1.0323, + "mean_token_accuracy": 0.687142014503479, + "num_tokens": 57870239.0, + "step": 2247 + }, + { + "epoch": 0.24687019547551065, + "grad_norm": 2.1064412593841553, + "learning_rate": 8.224743777452416e-07, + "loss": 1.0073, + "mean_token_accuracy": 0.6927852630615234, + "num_tokens": 57899702.0, + "step": 2248 + }, + { + "epoch": 0.24698001317812432, + "grad_norm": 2.792523145675659, + "learning_rate": 8.228404099560761e-07, + "loss": 0.9861, + "mean_token_accuracy": 0.7054699063301086, + "num_tokens": 57918400.0, + "step": 2249 + }, + { + "epoch": 0.24708983088073797, + "grad_norm": 2.4784579277038574, + "learning_rate": 8.232064421669106e-07, + "loss": 1.0151, + "mean_token_accuracy": 0.6928202509880066, + "num_tokens": 57940924.0, + "step": 2250 + }, + { + "epoch": 0.24719964858335164, + "grad_norm": 2.6823482513427734, + "learning_rate": 8.235724743777453e-07, + "loss": 0.9377, + "mean_token_accuracy": 0.7207510471343994, + "num_tokens": 57959917.0, + "step": 2251 + }, + { + "epoch": 0.24730946628596528, + "grad_norm": 2.065821409225464, + "learning_rate": 8.239385065885798e-07, + "loss": 1.0614, + "mean_token_accuracy": 0.6895511150360107, + "num_tokens": 57989935.0, + "step": 2252 + }, + { + "epoch": 0.24741928398857896, + "grad_norm": 2.6068530082702637, + "learning_rate": 8.243045387994143e-07, + "loss": 0.9698, + "mean_token_accuracy": 0.7127977013587952, + "num_tokens": 58009541.0, + "step": 2253 + }, + { + "epoch": 0.24752910169119263, + "grad_norm": 2.073906421661377, + "learning_rate": 8.246705710102489e-07, + "loss": 1.0099, + "mean_token_accuracy": 0.6917431354522705, + "num_tokens": 58042087.0, + "step": 2254 + }, + { + "epoch": 0.24763891939380628, + "grad_norm": 2.177420139312744, + "learning_rate": 8.250366032210834e-07, + "loss": 0.9605, + "mean_token_accuracy": 0.7152930498123169, + "num_tokens": 58070764.0, + "step": 2255 + }, + { + "epoch": 0.24774873709641995, + "grad_norm": 2.6803367137908936, + "learning_rate": 8.25402635431918e-07, + "loss": 0.9773, + "mean_token_accuracy": 0.70455402135849, + "num_tokens": 58090387.0, + "step": 2256 + }, + { + "epoch": 0.2478585547990336, + "grad_norm": 2.359783411026001, + "learning_rate": 8.257686676427526e-07, + "loss": 1.0643, + "mean_token_accuracy": 0.6860730648040771, + "num_tokens": 58113755.0, + "step": 2257 + }, + { + "epoch": 0.24796837250164727, + "grad_norm": 2.49082612991333, + "learning_rate": 8.261346998535871e-07, + "loss": 0.9423, + "mean_token_accuracy": 0.7167620062828064, + "num_tokens": 58135562.0, + "step": 2258 + }, + { + "epoch": 0.24807819020426092, + "grad_norm": 2.3117270469665527, + "learning_rate": 8.265007320644216e-07, + "loss": 1.0525, + "mean_token_accuracy": 0.6820728778839111, + "num_tokens": 58161079.0, + "step": 2259 + }, + { + "epoch": 0.2481880079068746, + "grad_norm": 2.2937471866607666, + "learning_rate": 8.268667642752562e-07, + "loss": 0.947, + "mean_token_accuracy": 0.7184575200080872, + "num_tokens": 58186104.0, + "step": 2260 + }, + { + "epoch": 0.24829782560948824, + "grad_norm": 1.991825819015503, + "learning_rate": 8.272327964860908e-07, + "loss": 1.0838, + "mean_token_accuracy": 0.6784193515777588, + "num_tokens": 58221477.0, + "step": 2261 + }, + { + "epoch": 0.2484076433121019, + "grad_norm": 2.2645723819732666, + "learning_rate": 8.275988286969253e-07, + "loss": 1.0683, + "mean_token_accuracy": 0.6842431426048279, + "num_tokens": 58247976.0, + "step": 2262 + }, + { + "epoch": 0.24851746101471558, + "grad_norm": 2.5921123027801514, + "learning_rate": 8.279648609077599e-07, + "loss": 0.9544, + "mean_token_accuracy": 0.7120132446289062, + "num_tokens": 58267331.0, + "step": 2263 + }, + { + "epoch": 0.24862727871732923, + "grad_norm": 2.331697940826416, + "learning_rate": 8.283308931185944e-07, + "loss": 1.0449, + "mean_token_accuracy": 0.6835658550262451, + "num_tokens": 58292582.0, + "step": 2264 + }, + { + "epoch": 0.2487370964199429, + "grad_norm": 2.2711453437805176, + "learning_rate": 8.28696925329429e-07, + "loss": 1.1316, + "mean_token_accuracy": 0.6617821455001831, + "num_tokens": 58318884.0, + "step": 2265 + }, + { + "epoch": 0.24884691412255655, + "grad_norm": 2.305629253387451, + "learning_rate": 8.290629575402635e-07, + "loss": 1.0179, + "mean_token_accuracy": 0.6912456750869751, + "num_tokens": 58344826.0, + "step": 2266 + }, + { + "epoch": 0.24895673182517022, + "grad_norm": 2.2961668968200684, + "learning_rate": 8.294289897510981e-07, + "loss": 1.0243, + "mean_token_accuracy": 0.6905766129493713, + "num_tokens": 58368859.0, + "step": 2267 + }, + { + "epoch": 0.24906654952778387, + "grad_norm": 2.209723472595215, + "learning_rate": 8.297950219619327e-07, + "loss": 0.9944, + "mean_token_accuracy": 0.7035040855407715, + "num_tokens": 58397377.0, + "step": 2268 + }, + { + "epoch": 0.24917636723039754, + "grad_norm": 2.3495590686798096, + "learning_rate": 8.301610541727672e-07, + "loss": 0.8879, + "mean_token_accuracy": 0.7305992841720581, + "num_tokens": 58419411.0, + "step": 2269 + }, + { + "epoch": 0.2492861849330112, + "grad_norm": 2.103715419769287, + "learning_rate": 8.305270863836017e-07, + "loss": 0.9737, + "mean_token_accuracy": 0.7018101215362549, + "num_tokens": 58447935.0, + "step": 2270 + }, + { + "epoch": 0.24939600263562486, + "grad_norm": 2.118075370788574, + "learning_rate": 8.308931185944363e-07, + "loss": 0.9951, + "mean_token_accuracy": 0.6982691287994385, + "num_tokens": 58474769.0, + "step": 2271 + }, + { + "epoch": 0.24950582033823854, + "grad_norm": 2.2352116107940674, + "learning_rate": 8.312591508052709e-07, + "loss": 1.1362, + "mean_token_accuracy": 0.6629729270935059, + "num_tokens": 58504789.0, + "step": 2272 + }, + { + "epoch": 0.24961563804085218, + "grad_norm": 2.6431188583374023, + "learning_rate": 8.316251830161054e-07, + "loss": 0.9785, + "mean_token_accuracy": 0.7003538608551025, + "num_tokens": 58526700.0, + "step": 2273 + }, + { + "epoch": 0.24972545574346586, + "grad_norm": 2.08164644241333, + "learning_rate": 8.3199121522694e-07, + "loss": 1.0282, + "mean_token_accuracy": 0.6865013837814331, + "num_tokens": 58555555.0, + "step": 2274 + }, + { + "epoch": 0.2498352734460795, + "grad_norm": 2.34816837310791, + "learning_rate": 8.323572474377745e-07, + "loss": 0.8659, + "mean_token_accuracy": 0.7353020906448364, + "num_tokens": 58577667.0, + "step": 2275 + }, + { + "epoch": 0.24994509114869318, + "grad_norm": 2.0974671840667725, + "learning_rate": 8.32723279648609e-07, + "loss": 1.0365, + "mean_token_accuracy": 0.6926345825195312, + "num_tokens": 58606031.0, + "step": 2276 + }, + { + "epoch": 0.2500549088513068, + "grad_norm": 2.342052698135376, + "learning_rate": 8.330893118594437e-07, + "loss": 0.9762, + "mean_token_accuracy": 0.705540120601654, + "num_tokens": 58631967.0, + "step": 2277 + }, + { + "epoch": 0.25016472655392047, + "grad_norm": 2.1182658672332764, + "learning_rate": 8.334553440702782e-07, + "loss": 1.0436, + "mean_token_accuracy": 0.6861743927001953, + "num_tokens": 58662429.0, + "step": 2278 + }, + { + "epoch": 0.25027454425653417, + "grad_norm": 2.250556707382202, + "learning_rate": 8.338213762811127e-07, + "loss": 1.0719, + "mean_token_accuracy": 0.6728646755218506, + "num_tokens": 58689786.0, + "step": 2279 + }, + { + "epoch": 0.2503843619591478, + "grad_norm": 2.402358055114746, + "learning_rate": 8.341874084919473e-07, + "loss": 0.9759, + "mean_token_accuracy": 0.7036428451538086, + "num_tokens": 58711364.0, + "step": 2280 + }, + { + "epoch": 0.25049417966176146, + "grad_norm": 2.1674606800079346, + "learning_rate": 8.345534407027818e-07, + "loss": 1.0422, + "mean_token_accuracy": 0.6883289813995361, + "num_tokens": 58737840.0, + "step": 2281 + }, + { + "epoch": 0.25060399736437516, + "grad_norm": 2.1460347175598145, + "learning_rate": 8.349194729136164e-07, + "loss": 1.0947, + "mean_token_accuracy": 0.67780601978302, + "num_tokens": 58766062.0, + "step": 2282 + }, + { + "epoch": 0.2507138150669888, + "grad_norm": 2.328315258026123, + "learning_rate": 8.35285505124451e-07, + "loss": 1.0345, + "mean_token_accuracy": 0.6844587326049805, + "num_tokens": 58792625.0, + "step": 2283 + }, + { + "epoch": 0.25082363276960246, + "grad_norm": 2.386348247528076, + "learning_rate": 8.356515373352855e-07, + "loss": 1.1555, + "mean_token_accuracy": 0.6628126502037048, + "num_tokens": 58817547.0, + "step": 2284 + }, + { + "epoch": 0.2509334504722161, + "grad_norm": 2.083252191543579, + "learning_rate": 8.360175695461201e-07, + "loss": 1.0336, + "mean_token_accuracy": 0.6880732178688049, + "num_tokens": 58845967.0, + "step": 2285 + }, + { + "epoch": 0.2510432681748298, + "grad_norm": 2.063361406326294, + "learning_rate": 8.363836017569546e-07, + "loss": 1.0206, + "mean_token_accuracy": 0.6961078643798828, + "num_tokens": 58876861.0, + "step": 2286 + }, + { + "epoch": 0.25115308587744345, + "grad_norm": 2.364124298095703, + "learning_rate": 8.36749633967789e-07, + "loss": 1.042, + "mean_token_accuracy": 0.6881845593452454, + "num_tokens": 58900848.0, + "step": 2287 + }, + { + "epoch": 0.2512629035800571, + "grad_norm": 2.3223299980163574, + "learning_rate": 8.371156661786238e-07, + "loss": 0.9784, + "mean_token_accuracy": 0.7035680413246155, + "num_tokens": 58922927.0, + "step": 2288 + }, + { + "epoch": 0.25137272128267074, + "grad_norm": 2.13277268409729, + "learning_rate": 8.374816983894583e-07, + "loss": 1.019, + "mean_token_accuracy": 0.6876029372215271, + "num_tokens": 58951123.0, + "step": 2289 + }, + { + "epoch": 0.25148253898528444, + "grad_norm": 2.5982420444488525, + "learning_rate": 8.378477306002928e-07, + "loss": 1.0486, + "mean_token_accuracy": 0.6863189935684204, + "num_tokens": 58970496.0, + "step": 2290 + }, + { + "epoch": 0.2515923566878981, + "grad_norm": 2.2497498989105225, + "learning_rate": 8.382137628111274e-07, + "loss": 1.0425, + "mean_token_accuracy": 0.6904568076133728, + "num_tokens": 58994989.0, + "step": 2291 + }, + { + "epoch": 0.25170217439051173, + "grad_norm": 2.5578668117523193, + "learning_rate": 8.385797950219619e-07, + "loss": 0.967, + "mean_token_accuracy": 0.7058486342430115, + "num_tokens": 59015235.0, + "step": 2292 + }, + { + "epoch": 0.25181199209312544, + "grad_norm": 2.3423643112182617, + "learning_rate": 8.389458272327965e-07, + "loss": 0.9742, + "mean_token_accuracy": 0.6998773217201233, + "num_tokens": 59038421.0, + "step": 2293 + }, + { + "epoch": 0.2519218097957391, + "grad_norm": 2.2714731693267822, + "learning_rate": 8.393118594436311e-07, + "loss": 1.0654, + "mean_token_accuracy": 0.6756994724273682, + "num_tokens": 59064410.0, + "step": 2294 + }, + { + "epoch": 0.25203162749835273, + "grad_norm": 2.402616262435913, + "learning_rate": 8.396778916544656e-07, + "loss": 1.0373, + "mean_token_accuracy": 0.6895198822021484, + "num_tokens": 59087245.0, + "step": 2295 + }, + { + "epoch": 0.2521414452009664, + "grad_norm": 2.3038504123687744, + "learning_rate": 8.400439238653e-07, + "loss": 1.0254, + "mean_token_accuracy": 0.6897454261779785, + "num_tokens": 59111988.0, + "step": 2296 + }, + { + "epoch": 0.2522512629035801, + "grad_norm": 2.473167657852173, + "learning_rate": 8.404099560761346e-07, + "loss": 1.024, + "mean_token_accuracy": 0.6993914842605591, + "num_tokens": 59137371.0, + "step": 2297 + }, + { + "epoch": 0.2523610806061937, + "grad_norm": 2.3040456771850586, + "learning_rate": 8.407759882869691e-07, + "loss": 1.0192, + "mean_token_accuracy": 0.6964040994644165, + "num_tokens": 59163108.0, + "step": 2298 + }, + { + "epoch": 0.25247089830880737, + "grad_norm": 2.2118098735809326, + "learning_rate": 8.411420204978039e-07, + "loss": 1.0758, + "mean_token_accuracy": 0.6770771145820618, + "num_tokens": 59191057.0, + "step": 2299 + }, + { + "epoch": 0.252580716011421, + "grad_norm": 2.1293745040893555, + "learning_rate": 8.415080527086383e-07, + "loss": 1.0831, + "mean_token_accuracy": 0.6780662536621094, + "num_tokens": 59220690.0, + "step": 2300 + }, + { + "epoch": 0.2526905337140347, + "grad_norm": 2.1602783203125, + "learning_rate": 8.418740849194728e-07, + "loss": 0.9866, + "mean_token_accuracy": 0.7060353755950928, + "num_tokens": 59247310.0, + "step": 2301 + }, + { + "epoch": 0.25280035141664836, + "grad_norm": 2.3089752197265625, + "learning_rate": 8.422401171303074e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.7008684277534485, + "num_tokens": 59271797.0, + "step": 2302 + }, + { + "epoch": 0.252910169119262, + "grad_norm": 2.245051860809326, + "learning_rate": 8.426061493411419e-07, + "loss": 0.9933, + "mean_token_accuracy": 0.7015406489372253, + "num_tokens": 59299996.0, + "step": 2303 + }, + { + "epoch": 0.2530199868218757, + "grad_norm": 2.219271183013916, + "learning_rate": 8.429721815519765e-07, + "loss": 1.0737, + "mean_token_accuracy": 0.6810920238494873, + "num_tokens": 59326639.0, + "step": 2304 + }, + { + "epoch": 0.25312980452448935, + "grad_norm": 2.293030023574829, + "learning_rate": 8.433382137628111e-07, + "loss": 1.0073, + "mean_token_accuracy": 0.7039121389389038, + "num_tokens": 59353176.0, + "step": 2305 + }, + { + "epoch": 0.253239622227103, + "grad_norm": 2.110360622406006, + "learning_rate": 8.437042459736456e-07, + "loss": 1.0134, + "mean_token_accuracy": 0.698331356048584, + "num_tokens": 59381590.0, + "step": 2306 + }, + { + "epoch": 0.25334943992971665, + "grad_norm": 1.988448977470398, + "learning_rate": 8.440702781844801e-07, + "loss": 0.9765, + "mean_token_accuracy": 0.705001950263977, + "num_tokens": 59411696.0, + "step": 2307 + }, + { + "epoch": 0.25345925763233035, + "grad_norm": 2.212639331817627, + "learning_rate": 8.444363103953147e-07, + "loss": 1.0493, + "mean_token_accuracy": 0.6800693273544312, + "num_tokens": 59438691.0, + "step": 2308 + }, + { + "epoch": 0.253569075334944, + "grad_norm": 2.1144511699676514, + "learning_rate": 8.448023426061493e-07, + "loss": 1.0053, + "mean_token_accuracy": 0.696395993232727, + "num_tokens": 59467636.0, + "step": 2309 + }, + { + "epoch": 0.25367889303755764, + "grad_norm": 2.089320421218872, + "learning_rate": 8.451683748169838e-07, + "loss": 1.0419, + "mean_token_accuracy": 0.6895480155944824, + "num_tokens": 59496548.0, + "step": 2310 + }, + { + "epoch": 0.25378871074017134, + "grad_norm": 2.454511880874634, + "learning_rate": 8.455344070278184e-07, + "loss": 1.0388, + "mean_token_accuracy": 0.6896618604660034, + "num_tokens": 59518194.0, + "step": 2311 + }, + { + "epoch": 0.253898528442785, + "grad_norm": 2.4463248252868652, + "learning_rate": 8.459004392386529e-07, + "loss": 0.9084, + "mean_token_accuracy": 0.7222687602043152, + "num_tokens": 59540317.0, + "step": 2312 + }, + { + "epoch": 0.25400834614539863, + "grad_norm": 2.210131883621216, + "learning_rate": 8.462664714494874e-07, + "loss": 1.0744, + "mean_token_accuracy": 0.6747274398803711, + "num_tokens": 59569101.0, + "step": 2313 + }, + { + "epoch": 0.2541181638480123, + "grad_norm": 2.2271616458892822, + "learning_rate": 8.46632503660322e-07, + "loss": 1.0639, + "mean_token_accuracy": 0.6805738210678101, + "num_tokens": 59597709.0, + "step": 2314 + }, + { + "epoch": 0.254227981550626, + "grad_norm": 2.139704465866089, + "learning_rate": 8.469985358711566e-07, + "loss": 1.0145, + "mean_token_accuracy": 0.692929208278656, + "num_tokens": 59626571.0, + "step": 2315 + }, + { + "epoch": 0.2543377992532396, + "grad_norm": 2.289351463317871, + "learning_rate": 8.473645680819912e-07, + "loss": 1.0189, + "mean_token_accuracy": 0.6853028535842896, + "num_tokens": 59653507.0, + "step": 2316 + }, + { + "epoch": 0.2544476169558533, + "grad_norm": 2.2878577709198, + "learning_rate": 8.477306002928257e-07, + "loss": 0.9821, + "mean_token_accuracy": 0.710645854473114, + "num_tokens": 59682791.0, + "step": 2317 + }, + { + "epoch": 0.2545574346584669, + "grad_norm": 2.392786741256714, + "learning_rate": 8.480966325036602e-07, + "loss": 1.115, + "mean_token_accuracy": 0.6687904596328735, + "num_tokens": 59708535.0, + "step": 2318 + }, + { + "epoch": 0.2546672523610806, + "grad_norm": 2.7898175716400146, + "learning_rate": 8.484626647144948e-07, + "loss": 0.9821, + "mean_token_accuracy": 0.7040555477142334, + "num_tokens": 59728949.0, + "step": 2319 + }, + { + "epoch": 0.25477707006369427, + "grad_norm": 2.431230306625366, + "learning_rate": 8.488286969253294e-07, + "loss": 0.975, + "mean_token_accuracy": 0.7089655995368958, + "num_tokens": 59750730.0, + "step": 2320 + }, + { + "epoch": 0.2548868877663079, + "grad_norm": 2.354792594909668, + "learning_rate": 8.491947291361639e-07, + "loss": 1.007, + "mean_token_accuracy": 0.6984936594963074, + "num_tokens": 59773449.0, + "step": 2321 + }, + { + "epoch": 0.2549967054689216, + "grad_norm": 2.3096673488616943, + "learning_rate": 8.495607613469985e-07, + "loss": 1.048, + "mean_token_accuracy": 0.6830718517303467, + "num_tokens": 59797987.0, + "step": 2322 + }, + { + "epoch": 0.25510652317153526, + "grad_norm": 2.4719109535217285, + "learning_rate": 8.49926793557833e-07, + "loss": 0.9672, + "mean_token_accuracy": 0.7012922763824463, + "num_tokens": 59819138.0, + "step": 2323 + }, + { + "epoch": 0.2552163408741489, + "grad_norm": 2.4108433723449707, + "learning_rate": 8.502928257686675e-07, + "loss": 1.0626, + "mean_token_accuracy": 0.6912804841995239, + "num_tokens": 59842604.0, + "step": 2324 + }, + { + "epoch": 0.25532615857676255, + "grad_norm": 2.1934845447540283, + "learning_rate": 8.506588579795022e-07, + "loss": 1.0279, + "mean_token_accuracy": 0.6923414468765259, + "num_tokens": 59869413.0, + "step": 2325 + }, + { + "epoch": 0.25543597627937625, + "grad_norm": 2.255873680114746, + "learning_rate": 8.510248901903367e-07, + "loss": 1.0411, + "mean_token_accuracy": 0.6854395866394043, + "num_tokens": 59896911.0, + "step": 2326 + }, + { + "epoch": 0.2555457939819899, + "grad_norm": 2.338046073913574, + "learning_rate": 8.513909224011712e-07, + "loss": 0.9694, + "mean_token_accuracy": 0.7062954306602478, + "num_tokens": 59920209.0, + "step": 2327 + }, + { + "epoch": 0.25565561168460355, + "grad_norm": 2.2612905502319336, + "learning_rate": 8.517569546120058e-07, + "loss": 1.0593, + "mean_token_accuracy": 0.6827596426010132, + "num_tokens": 59946626.0, + "step": 2328 + }, + { + "epoch": 0.25576542938721725, + "grad_norm": 2.1314167976379395, + "learning_rate": 8.521229868228403e-07, + "loss": 1.0552, + "mean_token_accuracy": 0.6895726919174194, + "num_tokens": 59975887.0, + "step": 2329 + }, + { + "epoch": 0.2558752470898309, + "grad_norm": 2.29848051071167, + "learning_rate": 8.524890190336748e-07, + "loss": 0.9416, + "mean_token_accuracy": 0.7140660881996155, + "num_tokens": 60000691.0, + "step": 2330 + }, + { + "epoch": 0.25598506479244454, + "grad_norm": 2.3039662837982178, + "learning_rate": 8.528550512445095e-07, + "loss": 1.0181, + "mean_token_accuracy": 0.697019100189209, + "num_tokens": 60025315.0, + "step": 2331 + }, + { + "epoch": 0.2560948824950582, + "grad_norm": 2.1014046669006348, + "learning_rate": 8.53221083455344e-07, + "loss": 1.0892, + "mean_token_accuracy": 0.6868646144866943, + "num_tokens": 60057120.0, + "step": 2332 + }, + { + "epoch": 0.2562047001976719, + "grad_norm": 2.3415722846984863, + "learning_rate": 8.535871156661786e-07, + "loss": 1.0224, + "mean_token_accuracy": 0.6991190910339355, + "num_tokens": 60084156.0, + "step": 2333 + }, + { + "epoch": 0.25631451790028553, + "grad_norm": 2.325474977493286, + "learning_rate": 8.539531478770131e-07, + "loss": 1.0778, + "mean_token_accuracy": 0.6813520193099976, + "num_tokens": 60109948.0, + "step": 2334 + }, + { + "epoch": 0.2564243356028992, + "grad_norm": 2.64703106880188, + "learning_rate": 8.543191800878476e-07, + "loss": 0.9353, + "mean_token_accuracy": 0.7097787857055664, + "num_tokens": 60129571.0, + "step": 2335 + }, + { + "epoch": 0.2565341533055128, + "grad_norm": 2.1947171688079834, + "learning_rate": 8.546852122986823e-07, + "loss": 1.0558, + "mean_token_accuracy": 0.6833662986755371, + "num_tokens": 60157568.0, + "step": 2336 + }, + { + "epoch": 0.2566439710081265, + "grad_norm": 2.308866500854492, + "learning_rate": 8.550512445095168e-07, + "loss": 0.999, + "mean_token_accuracy": 0.6974608898162842, + "num_tokens": 60185034.0, + "step": 2337 + }, + { + "epoch": 0.25675378871074017, + "grad_norm": 1.9721802473068237, + "learning_rate": 8.554172767203513e-07, + "loss": 1.0285, + "mean_token_accuracy": 0.6937969923019409, + "num_tokens": 60218705.0, + "step": 2338 + }, + { + "epoch": 0.2568636064133538, + "grad_norm": 2.268324136734009, + "learning_rate": 8.557833089311859e-07, + "loss": 0.9389, + "mean_token_accuracy": 0.7169396877288818, + "num_tokens": 60243522.0, + "step": 2339 + }, + { + "epoch": 0.2569734241159675, + "grad_norm": 2.2245678901672363, + "learning_rate": 8.561493411420204e-07, + "loss": 1.0088, + "mean_token_accuracy": 0.6997082233428955, + "num_tokens": 60271606.0, + "step": 2340 + }, + { + "epoch": 0.25708324181858117, + "grad_norm": 2.111938238143921, + "learning_rate": 8.56515373352855e-07, + "loss": 1.0343, + "mean_token_accuracy": 0.6933759450912476, + "num_tokens": 60300332.0, + "step": 2341 + }, + { + "epoch": 0.2571930595211948, + "grad_norm": 2.1276662349700928, + "learning_rate": 8.568814055636896e-07, + "loss": 0.897, + "mean_token_accuracy": 0.727488100528717, + "num_tokens": 60328449.0, + "step": 2342 + }, + { + "epoch": 0.25730287722380846, + "grad_norm": 2.214752435684204, + "learning_rate": 8.572474377745241e-07, + "loss": 1.0114, + "mean_token_accuracy": 0.6992231607437134, + "num_tokens": 60354910.0, + "step": 2343 + }, + { + "epoch": 0.25741269492642216, + "grad_norm": 2.208919048309326, + "learning_rate": 8.576134699853586e-07, + "loss": 0.9908, + "mean_token_accuracy": 0.7043824195861816, + "num_tokens": 60380063.0, + "step": 2344 + }, + { + "epoch": 0.2575225126290358, + "grad_norm": 2.3845150470733643, + "learning_rate": 8.579795021961932e-07, + "loss": 1.0468, + "mean_token_accuracy": 0.6883711218833923, + "num_tokens": 60404706.0, + "step": 2345 + }, + { + "epoch": 0.25763233033164945, + "grad_norm": 2.1634721755981445, + "learning_rate": 8.583455344070277e-07, + "loss": 0.9966, + "mean_token_accuracy": 0.7010533809661865, + "num_tokens": 60433576.0, + "step": 2346 + }, + { + "epoch": 0.2577421480342631, + "grad_norm": 2.172494649887085, + "learning_rate": 8.587115666178624e-07, + "loss": 1.0017, + "mean_token_accuracy": 0.697603166103363, + "num_tokens": 60458975.0, + "step": 2347 + }, + { + "epoch": 0.2578519657368768, + "grad_norm": 2.107342004776001, + "learning_rate": 8.590775988286969e-07, + "loss": 1.0161, + "mean_token_accuracy": 0.7055054903030396, + "num_tokens": 60490043.0, + "step": 2348 + }, + { + "epoch": 0.25796178343949044, + "grad_norm": 2.351569175720215, + "learning_rate": 8.594436310395314e-07, + "loss": 1.0359, + "mean_token_accuracy": 0.6882972121238708, + "num_tokens": 60512267.0, + "step": 2349 + }, + { + "epoch": 0.2580716011421041, + "grad_norm": 2.2585411071777344, + "learning_rate": 8.59809663250366e-07, + "loss": 1.0456, + "mean_token_accuracy": 0.6901679039001465, + "num_tokens": 60537869.0, + "step": 2350 + }, + { + "epoch": 0.2581814188447178, + "grad_norm": 2.214498281478882, + "learning_rate": 8.601756954612005e-07, + "loss": 1.006, + "mean_token_accuracy": 0.6989957690238953, + "num_tokens": 60563603.0, + "step": 2351 + }, + { + "epoch": 0.25829123654733144, + "grad_norm": 2.434593915939331, + "learning_rate": 8.605417276720351e-07, + "loss": 1.0127, + "mean_token_accuracy": 0.6935616731643677, + "num_tokens": 60587192.0, + "step": 2352 + }, + { + "epoch": 0.2584010542499451, + "grad_norm": 2.352015972137451, + "learning_rate": 8.609077598828697e-07, + "loss": 0.963, + "mean_token_accuracy": 0.7020860910415649, + "num_tokens": 60610638.0, + "step": 2353 + }, + { + "epoch": 0.25851087195255873, + "grad_norm": 2.341670513153076, + "learning_rate": 8.612737920937042e-07, + "loss": 1.0692, + "mean_token_accuracy": 0.6875171661376953, + "num_tokens": 60636598.0, + "step": 2354 + }, + { + "epoch": 0.25862068965517243, + "grad_norm": 2.0582356452941895, + "learning_rate": 8.616398243045387e-07, + "loss": 1.0596, + "mean_token_accuracy": 0.6865442991256714, + "num_tokens": 60666496.0, + "step": 2355 + }, + { + "epoch": 0.2587305073577861, + "grad_norm": 2.007148027420044, + "learning_rate": 8.620058565153733e-07, + "loss": 1.0664, + "mean_token_accuracy": 0.6824562549591064, + "num_tokens": 60696995.0, + "step": 2356 + }, + { + "epoch": 0.2588403250603997, + "grad_norm": 2.6151480674743652, + "learning_rate": 8.623718887262079e-07, + "loss": 0.9604, + "mean_token_accuracy": 0.712653398513794, + "num_tokens": 60717345.0, + "step": 2357 + }, + { + "epoch": 0.2589501427630134, + "grad_norm": 2.304546594619751, + "learning_rate": 8.627379209370424e-07, + "loss": 1.0557, + "mean_token_accuracy": 0.6937141418457031, + "num_tokens": 60744639.0, + "step": 2358 + }, + { + "epoch": 0.25905996046562707, + "grad_norm": 2.102567434310913, + "learning_rate": 8.63103953147877e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.6819736957550049, + "num_tokens": 60773675.0, + "step": 2359 + }, + { + "epoch": 0.2591697781682407, + "grad_norm": 2.1832313537597656, + "learning_rate": 8.634699853587115e-07, + "loss": 1.0106, + "mean_token_accuracy": 0.6982138156890869, + "num_tokens": 60799408.0, + "step": 2360 + }, + { + "epoch": 0.25927959587085436, + "grad_norm": 2.427799701690674, + "learning_rate": 8.63836017569546e-07, + "loss": 1.0612, + "mean_token_accuracy": 0.6838030815124512, + "num_tokens": 60823031.0, + "step": 2361 + }, + { + "epoch": 0.25938941357346806, + "grad_norm": 2.1428685188293457, + "learning_rate": 8.642020497803806e-07, + "loss": 1.0377, + "mean_token_accuracy": 0.6925236582756042, + "num_tokens": 60851807.0, + "step": 2362 + }, + { + "epoch": 0.2594992312760817, + "grad_norm": 2.8973381519317627, + "learning_rate": 8.645680819912152e-07, + "loss": 1.0632, + "mean_token_accuracy": 0.6868890523910522, + "num_tokens": 60871805.0, + "step": 2363 + }, + { + "epoch": 0.25960904897869536, + "grad_norm": 1.9486896991729736, + "learning_rate": 8.649341142020498e-07, + "loss": 1.1604, + "mean_token_accuracy": 0.6571602821350098, + "num_tokens": 60909599.0, + "step": 2364 + }, + { + "epoch": 0.259718866681309, + "grad_norm": 2.2291154861450195, + "learning_rate": 8.653001464128843e-07, + "loss": 1.0774, + "mean_token_accuracy": 0.6832300424575806, + "num_tokens": 60936870.0, + "step": 2365 + }, + { + "epoch": 0.2598286843839227, + "grad_norm": 2.391807794570923, + "learning_rate": 8.656661786237188e-07, + "loss": 1.0079, + "mean_token_accuracy": 0.6967015266418457, + "num_tokens": 60958742.0, + "step": 2366 + }, + { + "epoch": 0.25993850208653635, + "grad_norm": 2.1060967445373535, + "learning_rate": 8.660322108345534e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.6969621181488037, + "num_tokens": 60987581.0, + "step": 2367 + }, + { + "epoch": 0.26004831978915, + "grad_norm": 2.1604251861572266, + "learning_rate": 8.66398243045388e-07, + "loss": 1.0694, + "mean_token_accuracy": 0.681567907333374, + "num_tokens": 61015388.0, + "step": 2368 + }, + { + "epoch": 0.2601581374917637, + "grad_norm": 2.0237154960632324, + "learning_rate": 8.667642752562225e-07, + "loss": 1.0578, + "mean_token_accuracy": 0.6860626935958862, + "num_tokens": 61046490.0, + "step": 2369 + }, + { + "epoch": 0.26026795519437734, + "grad_norm": 2.1361734867095947, + "learning_rate": 8.671303074670571e-07, + "loss": 1.1068, + "mean_token_accuracy": 0.6724743843078613, + "num_tokens": 61076806.0, + "step": 2370 + }, + { + "epoch": 0.260377772896991, + "grad_norm": 2.044992208480835, + "learning_rate": 8.674963396778916e-07, + "loss": 1.1091, + "mean_token_accuracy": 0.6780107021331787, + "num_tokens": 61109820.0, + "step": 2371 + }, + { + "epoch": 0.26048759059960463, + "grad_norm": 2.361302614212036, + "learning_rate": 8.678623718887261e-07, + "loss": 0.9913, + "mean_token_accuracy": 0.6988619565963745, + "num_tokens": 61131765.0, + "step": 2372 + }, + { + "epoch": 0.26059740830221834, + "grad_norm": 2.094019889831543, + "learning_rate": 8.682284040995608e-07, + "loss": 1.0389, + "mean_token_accuracy": 0.6991971135139465, + "num_tokens": 61160748.0, + "step": 2373 + }, + { + "epoch": 0.260707226004832, + "grad_norm": 2.061222791671753, + "learning_rate": 8.685944363103953e-07, + "loss": 0.9974, + "mean_token_accuracy": 0.7020915746688843, + "num_tokens": 61189525.0, + "step": 2374 + }, + { + "epoch": 0.26081704370744563, + "grad_norm": 2.227816104888916, + "learning_rate": 8.689604685212298e-07, + "loss": 1.0345, + "mean_token_accuracy": 0.6950889825820923, + "num_tokens": 61217544.0, + "step": 2375 + }, + { + "epoch": 0.2609268614100593, + "grad_norm": 2.183948278427124, + "learning_rate": 8.693265007320644e-07, + "loss": 1.0143, + "mean_token_accuracy": 0.6951106786727905, + "num_tokens": 61244972.0, + "step": 2376 + }, + { + "epoch": 0.261036679112673, + "grad_norm": 2.219731569290161, + "learning_rate": 8.696925329428989e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6724863052368164, + "num_tokens": 61272586.0, + "step": 2377 + }, + { + "epoch": 0.2611464968152866, + "grad_norm": 2.4318220615386963, + "learning_rate": 8.700585651537334e-07, + "loss": 0.891, + "mean_token_accuracy": 0.7247675657272339, + "num_tokens": 61294173.0, + "step": 2378 + }, + { + "epoch": 0.26125631451790027, + "grad_norm": 2.137065887451172, + "learning_rate": 8.704245973645681e-07, + "loss": 1.042, + "mean_token_accuracy": 0.6896307468414307, + "num_tokens": 61321813.0, + "step": 2379 + }, + { + "epoch": 0.26136613222051397, + "grad_norm": 2.1232032775878906, + "learning_rate": 8.707906295754026e-07, + "loss": 1.158, + "mean_token_accuracy": 0.6584106087684631, + "num_tokens": 61351003.0, + "step": 2380 + }, + { + "epoch": 0.2614759499231276, + "grad_norm": 2.28849196434021, + "learning_rate": 8.711566617862372e-07, + "loss": 1.0801, + "mean_token_accuracy": 0.6824487447738647, + "num_tokens": 61376842.0, + "step": 2381 + }, + { + "epoch": 0.26158576762574126, + "grad_norm": 1.968575119972229, + "learning_rate": 8.715226939970717e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.6962355375289917, + "num_tokens": 61409980.0, + "step": 2382 + }, + { + "epoch": 0.2616955853283549, + "grad_norm": 2.006788730621338, + "learning_rate": 8.718887262079062e-07, + "loss": 1.0622, + "mean_token_accuracy": 0.6895346641540527, + "num_tokens": 61442720.0, + "step": 2383 + }, + { + "epoch": 0.2618054030309686, + "grad_norm": 2.3398640155792236, + "learning_rate": 8.722547584187409e-07, + "loss": 0.9956, + "mean_token_accuracy": 0.701826274394989, + "num_tokens": 61468322.0, + "step": 2384 + }, + { + "epoch": 0.26191522073358225, + "grad_norm": 2.4816648960113525, + "learning_rate": 8.726207906295754e-07, + "loss": 1.0241, + "mean_token_accuracy": 0.686070442199707, + "num_tokens": 61490210.0, + "step": 2385 + }, + { + "epoch": 0.2620250384361959, + "grad_norm": 2.4072813987731934, + "learning_rate": 8.729868228404099e-07, + "loss": 1.0236, + "mean_token_accuracy": 0.6891489624977112, + "num_tokens": 61513905.0, + "step": 2386 + }, + { + "epoch": 0.2621348561388096, + "grad_norm": 2.529019832611084, + "learning_rate": 8.733528550512445e-07, + "loss": 1.0113, + "mean_token_accuracy": 0.6912796497344971, + "num_tokens": 61533984.0, + "step": 2387 + }, + { + "epoch": 0.26224467384142325, + "grad_norm": 2.165011405944824, + "learning_rate": 8.73718887262079e-07, + "loss": 0.9776, + "mean_token_accuracy": 0.7073149085044861, + "num_tokens": 61559133.0, + "step": 2388 + }, + { + "epoch": 0.2623544915440369, + "grad_norm": 2.2551050186157227, + "learning_rate": 8.740849194729136e-07, + "loss": 0.9968, + "mean_token_accuracy": 0.701492428779602, + "num_tokens": 61583791.0, + "step": 2389 + }, + { + "epoch": 0.26246430924665054, + "grad_norm": 2.568418025970459, + "learning_rate": 8.744509516837482e-07, + "loss": 0.9759, + "mean_token_accuracy": 0.707082986831665, + "num_tokens": 61604249.0, + "step": 2390 + }, + { + "epoch": 0.26257412694926424, + "grad_norm": 2.4294543266296387, + "learning_rate": 8.748169838945827e-07, + "loss": 0.984, + "mean_token_accuracy": 0.7062879800796509, + "num_tokens": 61627052.0, + "step": 2391 + }, + { + "epoch": 0.2626839446518779, + "grad_norm": 2.0870091915130615, + "learning_rate": 8.751830161054172e-07, + "loss": 1.0852, + "mean_token_accuracy": 0.6833881139755249, + "num_tokens": 61655979.0, + "step": 2392 + }, + { + "epoch": 0.26279376235449153, + "grad_norm": 2.2078628540039062, + "learning_rate": 8.755490483162518e-07, + "loss": 1.1031, + "mean_token_accuracy": 0.6699906587600708, + "num_tokens": 61684326.0, + "step": 2393 + }, + { + "epoch": 0.2629035800571052, + "grad_norm": 2.3105030059814453, + "learning_rate": 8.759150805270863e-07, + "loss": 1.0778, + "mean_token_accuracy": 0.679216742515564, + "num_tokens": 61712239.0, + "step": 2394 + }, + { + "epoch": 0.2630133977597189, + "grad_norm": 2.17193865776062, + "learning_rate": 8.76281112737921e-07, + "loss": 1.0415, + "mean_token_accuracy": 0.6897361278533936, + "num_tokens": 61740619.0, + "step": 2395 + }, + { + "epoch": 0.2631232154623325, + "grad_norm": 2.3003876209259033, + "learning_rate": 8.766471449487555e-07, + "loss": 1.0616, + "mean_token_accuracy": 0.6800096035003662, + "num_tokens": 61766457.0, + "step": 2396 + }, + { + "epoch": 0.2632330331649462, + "grad_norm": 2.3000824451446533, + "learning_rate": 8.7701317715959e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.6960592269897461, + "num_tokens": 61792109.0, + "step": 2397 + }, + { + "epoch": 0.2633428508675599, + "grad_norm": 2.050384759902954, + "learning_rate": 8.773792093704246e-07, + "loss": 1.0044, + "mean_token_accuracy": 0.6962990164756775, + "num_tokens": 61821867.0, + "step": 2398 + }, + { + "epoch": 0.2634526685701735, + "grad_norm": 2.050497531890869, + "learning_rate": 8.777452415812591e-07, + "loss": 0.9805, + "mean_token_accuracy": 0.7038292288780212, + "num_tokens": 61851076.0, + "step": 2399 + }, + { + "epoch": 0.26356248627278717, + "grad_norm": 2.078920602798462, + "learning_rate": 8.781112737920937e-07, + "loss": 0.9797, + "mean_token_accuracy": 0.7002360224723816, + "num_tokens": 61883955.0, + "step": 2400 + }, + { + "epoch": 0.2636723039754008, + "grad_norm": 2.153931140899658, + "learning_rate": 8.784773060029283e-07, + "loss": 0.9237, + "mean_token_accuracy": 0.7157310247421265, + "num_tokens": 61909733.0, + "step": 2401 + }, + { + "epoch": 0.2637821216780145, + "grad_norm": 2.1658029556274414, + "learning_rate": 8.788433382137628e-07, + "loss": 1.0275, + "mean_token_accuracy": 0.6890308260917664, + "num_tokens": 61937771.0, + "step": 2402 + }, + { + "epoch": 0.26389193938062816, + "grad_norm": 2.6416893005371094, + "learning_rate": 8.792093704245973e-07, + "loss": 0.9997, + "mean_token_accuracy": 0.700023353099823, + "num_tokens": 61958207.0, + "step": 2403 + }, + { + "epoch": 0.2640017570832418, + "grad_norm": 2.1425893306732178, + "learning_rate": 8.795754026354319e-07, + "loss": 1.0508, + "mean_token_accuracy": 0.6806877851486206, + "num_tokens": 61987806.0, + "step": 2404 + }, + { + "epoch": 0.2641115747858555, + "grad_norm": 2.0992000102996826, + "learning_rate": 8.799414348462665e-07, + "loss": 1.1131, + "mean_token_accuracy": 0.6700534820556641, + "num_tokens": 62019461.0, + "step": 2405 + }, + { + "epoch": 0.26422139248846915, + "grad_norm": 2.0113515853881836, + "learning_rate": 8.80307467057101e-07, + "loss": 1.1091, + "mean_token_accuracy": 0.6733056306838989, + "num_tokens": 62052648.0, + "step": 2406 + }, + { + "epoch": 0.2643312101910828, + "grad_norm": 2.3175504207611084, + "learning_rate": 8.806734992679356e-07, + "loss": 1.0194, + "mean_token_accuracy": 0.7013658881187439, + "num_tokens": 62077698.0, + "step": 2407 + }, + { + "epoch": 0.26444102789369645, + "grad_norm": 2.2272212505340576, + "learning_rate": 8.810395314787701e-07, + "loss": 1.047, + "mean_token_accuracy": 0.6971811056137085, + "num_tokens": 62105774.0, + "step": 2408 + }, + { + "epoch": 0.26455084559631015, + "grad_norm": 2.4328267574310303, + "learning_rate": 8.814055636896046e-07, + "loss": 1.0267, + "mean_token_accuracy": 0.6916912794113159, + "num_tokens": 62128899.0, + "step": 2409 + }, + { + "epoch": 0.2646606632989238, + "grad_norm": 2.3680005073547363, + "learning_rate": 8.817715959004392e-07, + "loss": 1.0676, + "mean_token_accuracy": 0.6778907775878906, + "num_tokens": 62152471.0, + "step": 2410 + }, + { + "epoch": 0.26477048100153744, + "grad_norm": 2.5702552795410156, + "learning_rate": 8.821376281112738e-07, + "loss": 1.0397, + "mean_token_accuracy": 0.6918016672134399, + "num_tokens": 62174310.0, + "step": 2411 + }, + { + "epoch": 0.2648802987041511, + "grad_norm": 2.676788806915283, + "learning_rate": 8.825036603221084e-07, + "loss": 1.0205, + "mean_token_accuracy": 0.7067110538482666, + "num_tokens": 62193956.0, + "step": 2412 + }, + { + "epoch": 0.2649901164067648, + "grad_norm": 2.17669939994812, + "learning_rate": 8.828696925329429e-07, + "loss": 1.0517, + "mean_token_accuracy": 0.6806479096412659, + "num_tokens": 62220723.0, + "step": 2413 + }, + { + "epoch": 0.26509993410937843, + "grad_norm": 2.2530877590179443, + "learning_rate": 8.832357247437774e-07, + "loss": 1.0442, + "mean_token_accuracy": 0.6889869570732117, + "num_tokens": 62246768.0, + "step": 2414 + }, + { + "epoch": 0.2652097518119921, + "grad_norm": 2.126307487487793, + "learning_rate": 8.83601756954612e-07, + "loss": 0.9775, + "mean_token_accuracy": 0.7082647085189819, + "num_tokens": 62274476.0, + "step": 2415 + }, + { + "epoch": 0.2653195695146058, + "grad_norm": 2.164825201034546, + "learning_rate": 8.839677891654466e-07, + "loss": 0.9723, + "mean_token_accuracy": 0.7079282402992249, + "num_tokens": 62300250.0, + "step": 2416 + }, + { + "epoch": 0.2654293872172194, + "grad_norm": 2.1639959812164307, + "learning_rate": 8.843338213762811e-07, + "loss": 0.9747, + "mean_token_accuracy": 0.7101179957389832, + "num_tokens": 62325482.0, + "step": 2417 + }, + { + "epoch": 0.26553920491983307, + "grad_norm": 2.4291534423828125, + "learning_rate": 8.846998535871157e-07, + "loss": 1.0325, + "mean_token_accuracy": 0.6949779987335205, + "num_tokens": 62348385.0, + "step": 2418 + }, + { + "epoch": 0.2656490226224467, + "grad_norm": 2.3397979736328125, + "learning_rate": 8.850658857979502e-07, + "loss": 0.9944, + "mean_token_accuracy": 0.7006866335868835, + "num_tokens": 62371942.0, + "step": 2419 + }, + { + "epoch": 0.2657588403250604, + "grad_norm": 2.29647159576416, + "learning_rate": 8.854319180087847e-07, + "loss": 0.9864, + "mean_token_accuracy": 0.7049825191497803, + "num_tokens": 62398163.0, + "step": 2420 + }, + { + "epoch": 0.26586865802767407, + "grad_norm": 2.5699291229248047, + "learning_rate": 8.857979502196194e-07, + "loss": 1.0283, + "mean_token_accuracy": 0.6864852905273438, + "num_tokens": 62419047.0, + "step": 2421 + }, + { + "epoch": 0.2659784757302877, + "grad_norm": 2.183701515197754, + "learning_rate": 8.861639824304539e-07, + "loss": 1.1135, + "mean_token_accuracy": 0.6817928552627563, + "num_tokens": 62446956.0, + "step": 2422 + }, + { + "epoch": 0.26608829343290136, + "grad_norm": 2.507927894592285, + "learning_rate": 8.865300146412884e-07, + "loss": 1.0472, + "mean_token_accuracy": 0.6899815201759338, + "num_tokens": 62468806.0, + "step": 2423 + }, + { + "epoch": 0.26619811113551506, + "grad_norm": 2.3407485485076904, + "learning_rate": 8.86896046852123e-07, + "loss": 1.0552, + "mean_token_accuracy": 0.6868053674697876, + "num_tokens": 62493696.0, + "step": 2424 + }, + { + "epoch": 0.2663079288381287, + "grad_norm": 2.8556900024414062, + "learning_rate": 8.872620790629575e-07, + "loss": 0.9241, + "mean_token_accuracy": 0.7141544818878174, + "num_tokens": 62510545.0, + "step": 2425 + }, + { + "epoch": 0.26641774654074235, + "grad_norm": 2.0946645736694336, + "learning_rate": 8.87628111273792e-07, + "loss": 1.0107, + "mean_token_accuracy": 0.7030627131462097, + "num_tokens": 62538602.0, + "step": 2426 + }, + { + "epoch": 0.26652756424335605, + "grad_norm": 2.8130509853363037, + "learning_rate": 8.879941434846267e-07, + "loss": 0.9242, + "mean_token_accuracy": 0.7165362238883972, + "num_tokens": 62556858.0, + "step": 2427 + }, + { + "epoch": 0.2666373819459697, + "grad_norm": 2.3167879581451416, + "learning_rate": 8.883601756954612e-07, + "loss": 1.0034, + "mean_token_accuracy": 0.7013217806816101, + "num_tokens": 62580368.0, + "step": 2428 + }, + { + "epoch": 0.26674719964858334, + "grad_norm": 2.221212863922119, + "learning_rate": 8.887262079062958e-07, + "loss": 1.0145, + "mean_token_accuracy": 0.6952276229858398, + "num_tokens": 62605952.0, + "step": 2429 + }, + { + "epoch": 0.266857017351197, + "grad_norm": 2.07908034324646, + "learning_rate": 8.890922401171303e-07, + "loss": 1.0346, + "mean_token_accuracy": 0.6882150173187256, + "num_tokens": 62635903.0, + "step": 2430 + }, + { + "epoch": 0.2669668350538107, + "grad_norm": 1.890231728553772, + "learning_rate": 8.894582723279648e-07, + "loss": 1.1069, + "mean_token_accuracy": 0.6686638593673706, + "num_tokens": 62675543.0, + "step": 2431 + }, + { + "epoch": 0.26707665275642434, + "grad_norm": 2.0954835414886475, + "learning_rate": 8.898243045387995e-07, + "loss": 1.0777, + "mean_token_accuracy": 0.6757506132125854, + "num_tokens": 62705144.0, + "step": 2432 + }, + { + "epoch": 0.267186470459038, + "grad_norm": 2.3976657390594482, + "learning_rate": 8.90190336749634e-07, + "loss": 1.0069, + "mean_token_accuracy": 0.6934142708778381, + "num_tokens": 62730740.0, + "step": 2433 + }, + { + "epoch": 0.2672962881616517, + "grad_norm": 2.286581039428711, + "learning_rate": 8.905563689604685e-07, + "loss": 1.0613, + "mean_token_accuracy": 0.6847360134124756, + "num_tokens": 62755879.0, + "step": 2434 + }, + { + "epoch": 0.26740610586426533, + "grad_norm": 2.0529797077178955, + "learning_rate": 8.909224011713031e-07, + "loss": 0.9942, + "mean_token_accuracy": 0.7019824385643005, + "num_tokens": 62786905.0, + "step": 2435 + }, + { + "epoch": 0.267515923566879, + "grad_norm": 2.079305648803711, + "learning_rate": 8.912884333821376e-07, + "loss": 1.0409, + "mean_token_accuracy": 0.6922109723091125, + "num_tokens": 62815251.0, + "step": 2436 + }, + { + "epoch": 0.2676257412694926, + "grad_norm": 2.2549307346343994, + "learning_rate": 8.916544655929722e-07, + "loss": 0.9787, + "mean_token_accuracy": 0.7124942541122437, + "num_tokens": 62840071.0, + "step": 2437 + }, + { + "epoch": 0.2677355589721063, + "grad_norm": 2.3523762226104736, + "learning_rate": 8.920204978038068e-07, + "loss": 0.9425, + "mean_token_accuracy": 0.7119571566581726, + "num_tokens": 62863599.0, + "step": 2438 + }, + { + "epoch": 0.26784537667471997, + "grad_norm": 2.3797807693481445, + "learning_rate": 8.923865300146413e-07, + "loss": 0.9877, + "mean_token_accuracy": 0.6995692849159241, + "num_tokens": 62886063.0, + "step": 2439 + }, + { + "epoch": 0.2679551943773336, + "grad_norm": 2.3193061351776123, + "learning_rate": 8.927525622254757e-07, + "loss": 1.013, + "mean_token_accuracy": 0.7040001749992371, + "num_tokens": 62911701.0, + "step": 2440 + }, + { + "epoch": 0.26806501207994726, + "grad_norm": 2.670470714569092, + "learning_rate": 8.931185944363104e-07, + "loss": 0.9625, + "mean_token_accuracy": 0.706550121307373, + "num_tokens": 62931389.0, + "step": 2441 + }, + { + "epoch": 0.26817482978256096, + "grad_norm": 2.178034543991089, + "learning_rate": 8.934846266471448e-07, + "loss": 1.0918, + "mean_token_accuracy": 0.6746820211410522, + "num_tokens": 62959298.0, + "step": 2442 + }, + { + "epoch": 0.2682846474851746, + "grad_norm": 2.125549554824829, + "learning_rate": 8.938506588579794e-07, + "loss": 0.9923, + "mean_token_accuracy": 0.7065426111221313, + "num_tokens": 62987228.0, + "step": 2443 + }, + { + "epoch": 0.26839446518778826, + "grad_norm": 2.0443763732910156, + "learning_rate": 8.94216691068814e-07, + "loss": 0.9993, + "mean_token_accuracy": 0.6957262754440308, + "num_tokens": 63016848.0, + "step": 2444 + }, + { + "epoch": 0.26850428289040196, + "grad_norm": 2.1532864570617676, + "learning_rate": 8.945827232796485e-07, + "loss": 1.1018, + "mean_token_accuracy": 0.6751294136047363, + "num_tokens": 63047314.0, + "step": 2445 + }, + { + "epoch": 0.2686141005930156, + "grad_norm": 1.9805080890655518, + "learning_rate": 8.949487554904832e-07, + "loss": 1.1053, + "mean_token_accuracy": 0.6734870076179504, + "num_tokens": 63081054.0, + "step": 2446 + }, + { + "epoch": 0.26872391829562925, + "grad_norm": 2.532351016998291, + "learning_rate": 8.953147877013176e-07, + "loss": 1.0073, + "mean_token_accuracy": 0.6970974206924438, + "num_tokens": 63102763.0, + "step": 2447 + }, + { + "epoch": 0.2688337359982429, + "grad_norm": 2.577040672302246, + "learning_rate": 8.956808199121522e-07, + "loss": 1.0141, + "mean_token_accuracy": 0.6936377286911011, + "num_tokens": 63124068.0, + "step": 2448 + }, + { + "epoch": 0.2689435537008566, + "grad_norm": 2.355825185775757, + "learning_rate": 8.960468521229869e-07, + "loss": 0.9631, + "mean_token_accuracy": 0.7078412771224976, + "num_tokens": 63145050.0, + "step": 2449 + }, + { + "epoch": 0.26905337140347024, + "grad_norm": 2.3840181827545166, + "learning_rate": 8.964128843338213e-07, + "loss": 0.9739, + "mean_token_accuracy": 0.7127476930618286, + "num_tokens": 63167810.0, + "step": 2450 + }, + { + "epoch": 0.2691631891060839, + "grad_norm": 2.091240167617798, + "learning_rate": 8.967789165446558e-07, + "loss": 0.9834, + "mean_token_accuracy": 0.7058261036872864, + "num_tokens": 63197274.0, + "step": 2451 + }, + { + "epoch": 0.26927300680869753, + "grad_norm": 2.2186009883880615, + "learning_rate": 8.971449487554904e-07, + "loss": 1.0375, + "mean_token_accuracy": 0.6879631280899048, + "num_tokens": 63221700.0, + "step": 2452 + }, + { + "epoch": 0.26938282451131124, + "grad_norm": 2.2975423336029053, + "learning_rate": 8.97510980966325e-07, + "loss": 1.0598, + "mean_token_accuracy": 0.6800534129142761, + "num_tokens": 63248117.0, + "step": 2453 + }, + { + "epoch": 0.2694926422139249, + "grad_norm": 2.3676915168762207, + "learning_rate": 8.978770131771595e-07, + "loss": 0.9253, + "mean_token_accuracy": 0.7169880270957947, + "num_tokens": 63271257.0, + "step": 2454 + }, + { + "epoch": 0.26960245991653853, + "grad_norm": 2.4581892490386963, + "learning_rate": 8.982430453879941e-07, + "loss": 1.0217, + "mean_token_accuracy": 0.6880698204040527, + "num_tokens": 63293129.0, + "step": 2455 + }, + { + "epoch": 0.26971227761915223, + "grad_norm": 2.6748769283294678, + "learning_rate": 8.986090775988286e-07, + "loss": 1.0059, + "mean_token_accuracy": 0.698165774345398, + "num_tokens": 63315767.0, + "step": 2456 + }, + { + "epoch": 0.2698220953217659, + "grad_norm": 2.13130784034729, + "learning_rate": 8.989751098096631e-07, + "loss": 1.0269, + "mean_token_accuracy": 0.6962118148803711, + "num_tokens": 63345045.0, + "step": 2457 + }, + { + "epoch": 0.2699319130243795, + "grad_norm": 2.287626266479492, + "learning_rate": 8.993411420204977e-07, + "loss": 0.9954, + "mean_token_accuracy": 0.6976020336151123, + "num_tokens": 63370935.0, + "step": 2458 + }, + { + "epoch": 0.27004173072699317, + "grad_norm": 2.3400206565856934, + "learning_rate": 8.997071742313323e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.6882246732711792, + "num_tokens": 63395023.0, + "step": 2459 + }, + { + "epoch": 0.27015154842960687, + "grad_norm": 1.9690861701965332, + "learning_rate": 9.000732064421668e-07, + "loss": 1.0997, + "mean_token_accuracy": 0.6743000745773315, + "num_tokens": 63429774.0, + "step": 2460 + }, + { + "epoch": 0.2702613661322205, + "grad_norm": 2.19236159324646, + "learning_rate": 9.004392386530014e-07, + "loss": 0.9567, + "mean_token_accuracy": 0.7072882652282715, + "num_tokens": 63454711.0, + "step": 2461 + }, + { + "epoch": 0.27037118383483416, + "grad_norm": 2.2741305828094482, + "learning_rate": 9.008052708638359e-07, + "loss": 0.9861, + "mean_token_accuracy": 0.6990161538124084, + "num_tokens": 63481256.0, + "step": 2462 + }, + { + "epoch": 0.27048100153744786, + "grad_norm": 2.4148290157318115, + "learning_rate": 9.011713030746705e-07, + "loss": 1.0554, + "mean_token_accuracy": 0.6841435432434082, + "num_tokens": 63507253.0, + "step": 2463 + }, + { + "epoch": 0.2705908192400615, + "grad_norm": 2.3157265186309814, + "learning_rate": 9.015373352855051e-07, + "loss": 1.0022, + "mean_token_accuracy": 0.6884528398513794, + "num_tokens": 63532003.0, + "step": 2464 + }, + { + "epoch": 0.27070063694267515, + "grad_norm": 2.461268663406372, + "learning_rate": 9.019033674963396e-07, + "loss": 0.9907, + "mean_token_accuracy": 0.6983447074890137, + "num_tokens": 63554003.0, + "step": 2465 + }, + { + "epoch": 0.2708104546452888, + "grad_norm": 2.525879144668579, + "learning_rate": 9.022693997071742e-07, + "loss": 0.8854, + "mean_token_accuracy": 0.7259849905967712, + "num_tokens": 63574058.0, + "step": 2466 + }, + { + "epoch": 0.2709202723479025, + "grad_norm": 2.4645040035247803, + "learning_rate": 9.026354319180087e-07, + "loss": 0.9622, + "mean_token_accuracy": 0.7045747637748718, + "num_tokens": 63596234.0, + "step": 2467 + }, + { + "epoch": 0.27103009005051615, + "grad_norm": 2.044118881225586, + "learning_rate": 9.030014641288432e-07, + "loss": 1.0566, + "mean_token_accuracy": 0.6850253343582153, + "num_tokens": 63627837.0, + "step": 2468 + }, + { + "epoch": 0.2711399077531298, + "grad_norm": 2.06913685798645, + "learning_rate": 9.033674963396779e-07, + "loss": 1.1236, + "mean_token_accuracy": 0.6673113703727722, + "num_tokens": 63661751.0, + "step": 2469 + }, + { + "epoch": 0.27124972545574344, + "grad_norm": 2.3415870666503906, + "learning_rate": 9.037335285505124e-07, + "loss": 1.0433, + "mean_token_accuracy": 0.6902722716331482, + "num_tokens": 63686230.0, + "step": 2470 + }, + { + "epoch": 0.27135954315835714, + "grad_norm": 2.3794801235198975, + "learning_rate": 9.040995607613469e-07, + "loss": 1.0935, + "mean_token_accuracy": 0.6723988652229309, + "num_tokens": 63710974.0, + "step": 2471 + }, + { + "epoch": 0.2714693608609708, + "grad_norm": 2.0748708248138428, + "learning_rate": 9.044655929721815e-07, + "loss": 1.0846, + "mean_token_accuracy": 0.6751030087471008, + "num_tokens": 63741057.0, + "step": 2472 + }, + { + "epoch": 0.27157917856358443, + "grad_norm": 2.306274652481079, + "learning_rate": 9.04831625183016e-07, + "loss": 1.0863, + "mean_token_accuracy": 0.6765711307525635, + "num_tokens": 63766625.0, + "step": 2473 + }, + { + "epoch": 0.27168899626619814, + "grad_norm": 2.1510682106018066, + "learning_rate": 9.051976573938505e-07, + "loss": 1.0753, + "mean_token_accuracy": 0.6823874115943909, + "num_tokens": 63792679.0, + "step": 2474 + }, + { + "epoch": 0.2717988139688118, + "grad_norm": 2.3207693099975586, + "learning_rate": 9.055636896046852e-07, + "loss": 1.0353, + "mean_token_accuracy": 0.6901476383209229, + "num_tokens": 63818259.0, + "step": 2475 + }, + { + "epoch": 0.2719086316714254, + "grad_norm": 2.4100775718688965, + "learning_rate": 9.059297218155197e-07, + "loss": 0.971, + "mean_token_accuracy": 0.701837420463562, + "num_tokens": 63839630.0, + "step": 2476 + }, + { + "epoch": 0.2720184493740391, + "grad_norm": 2.0438334941864014, + "learning_rate": 9.062957540263542e-07, + "loss": 0.98, + "mean_token_accuracy": 0.7033220529556274, + "num_tokens": 63868019.0, + "step": 2477 + }, + { + "epoch": 0.2721282670766528, + "grad_norm": 2.033534049987793, + "learning_rate": 9.066617862371888e-07, + "loss": 1.0159, + "mean_token_accuracy": 0.7019431591033936, + "num_tokens": 63896304.0, + "step": 2478 + }, + { + "epoch": 0.2722380847792664, + "grad_norm": 2.058943271636963, + "learning_rate": 9.070278184480233e-07, + "loss": 0.9485, + "mean_token_accuracy": 0.7087342143058777, + "num_tokens": 63925436.0, + "step": 2479 + }, + { + "epoch": 0.27234790248188007, + "grad_norm": 2.2888965606689453, + "learning_rate": 9.07393850658858e-07, + "loss": 0.9982, + "mean_token_accuracy": 0.6999489068984985, + "num_tokens": 63949487.0, + "step": 2480 + }, + { + "epoch": 0.27245772018449377, + "grad_norm": 2.2653985023498535, + "learning_rate": 9.077598828696925e-07, + "loss": 0.9266, + "mean_token_accuracy": 0.7220295071601868, + "num_tokens": 63972779.0, + "step": 2481 + }, + { + "epoch": 0.2725675378871074, + "grad_norm": 2.8826544284820557, + "learning_rate": 9.08125915080527e-07, + "loss": 0.9916, + "mean_token_accuracy": 0.6998679637908936, + "num_tokens": 63990229.0, + "step": 2482 + }, + { + "epoch": 0.27267735558972106, + "grad_norm": 2.4028358459472656, + "learning_rate": 9.084919472913616e-07, + "loss": 1.0855, + "mean_token_accuracy": 0.6734484434127808, + "num_tokens": 64014964.0, + "step": 2483 + }, + { + "epoch": 0.2727871732923347, + "grad_norm": 2.1989200115203857, + "learning_rate": 9.088579795021961e-07, + "loss": 0.9937, + "mean_token_accuracy": 0.6988005638122559, + "num_tokens": 64041383.0, + "step": 2484 + }, + { + "epoch": 0.2728969909949484, + "grad_norm": 2.2226336002349854, + "learning_rate": 9.092240117130307e-07, + "loss": 1.0584, + "mean_token_accuracy": 0.6798248887062073, + "num_tokens": 64070705.0, + "step": 2485 + }, + { + "epoch": 0.27300680869756205, + "grad_norm": 2.104130268096924, + "learning_rate": 9.095900439238653e-07, + "loss": 0.9206, + "mean_token_accuracy": 0.7146039605140686, + "num_tokens": 64098527.0, + "step": 2486 + }, + { + "epoch": 0.2731166264001757, + "grad_norm": 1.9727129936218262, + "learning_rate": 9.099560761346998e-07, + "loss": 0.9736, + "mean_token_accuracy": 0.7018811702728271, + "num_tokens": 64129915.0, + "step": 2487 + }, + { + "epoch": 0.27322644410278935, + "grad_norm": 2.4030699729919434, + "learning_rate": 9.103221083455343e-07, + "loss": 1.0299, + "mean_token_accuracy": 0.6911640167236328, + "num_tokens": 64154983.0, + "step": 2488 + }, + { + "epoch": 0.27333626180540305, + "grad_norm": 2.0579004287719727, + "learning_rate": 9.106881405563689e-07, + "loss": 1.1405, + "mean_token_accuracy": 0.673547089099884, + "num_tokens": 64186278.0, + "step": 2489 + }, + { + "epoch": 0.2734460795080167, + "grad_norm": 2.344285488128662, + "learning_rate": 9.110541727672034e-07, + "loss": 1.0396, + "mean_token_accuracy": 0.6896350383758545, + "num_tokens": 64211406.0, + "step": 2490 + }, + { + "epoch": 0.27355589721063034, + "grad_norm": 2.3772737979888916, + "learning_rate": 9.11420204978038e-07, + "loss": 0.9514, + "mean_token_accuracy": 0.7104822993278503, + "num_tokens": 64236208.0, + "step": 2491 + }, + { + "epoch": 0.27366571491324404, + "grad_norm": 2.015620708465576, + "learning_rate": 9.117862371888726e-07, + "loss": 1.1192, + "mean_token_accuracy": 0.6678762435913086, + "num_tokens": 64269029.0, + "step": 2492 + }, + { + "epoch": 0.2737755326158577, + "grad_norm": 2.258979082107544, + "learning_rate": 9.121522693997071e-07, + "loss": 1.0829, + "mean_token_accuracy": 0.670850396156311, + "num_tokens": 64295333.0, + "step": 2493 + }, + { + "epoch": 0.27388535031847133, + "grad_norm": 2.1413323879241943, + "learning_rate": 9.125183016105416e-07, + "loss": 1.0414, + "mean_token_accuracy": 0.685373842716217, + "num_tokens": 64325395.0, + "step": 2494 + }, + { + "epoch": 0.273995168021085, + "grad_norm": 2.1652517318725586, + "learning_rate": 9.128843338213762e-07, + "loss": 1.031, + "mean_token_accuracy": 0.6864521503448486, + "num_tokens": 64353858.0, + "step": 2495 + }, + { + "epoch": 0.2741049857236987, + "grad_norm": 2.366197347640991, + "learning_rate": 9.132503660322108e-07, + "loss": 1.0545, + "mean_token_accuracy": 0.6816405057907104, + "num_tokens": 64378320.0, + "step": 2496 + }, + { + "epoch": 0.2742148034263123, + "grad_norm": 2.379110097885132, + "learning_rate": 9.136163982430454e-07, + "loss": 0.9569, + "mean_token_accuracy": 0.710149884223938, + "num_tokens": 64402520.0, + "step": 2497 + }, + { + "epoch": 0.27432462112892597, + "grad_norm": 2.285081386566162, + "learning_rate": 9.139824304538799e-07, + "loss": 0.9075, + "mean_token_accuracy": 0.7224372625350952, + "num_tokens": 64424227.0, + "step": 2498 + }, + { + "epoch": 0.2744344388315396, + "grad_norm": 2.300689220428467, + "learning_rate": 9.143484626647144e-07, + "loss": 1.1135, + "mean_token_accuracy": 0.675136387348175, + "num_tokens": 64449016.0, + "step": 2499 + }, + { + "epoch": 0.2745442565341533, + "grad_norm": 2.2875752449035645, + "learning_rate": 9.14714494875549e-07, + "loss": 0.8579, + "mean_token_accuracy": 0.7315891981124878, + "num_tokens": 64471600.0, + "step": 2500 + }, + { + "epoch": 0.27465407423676697, + "grad_norm": 2.13348388671875, + "learning_rate": 9.150805270863836e-07, + "loss": 1.0929, + "mean_token_accuracy": 0.6796666979789734, + "num_tokens": 64501456.0, + "step": 2501 + }, + { + "epoch": 0.2747638919393806, + "grad_norm": 2.3574376106262207, + "learning_rate": 9.154465592972181e-07, + "loss": 1.0078, + "mean_token_accuracy": 0.6993120908737183, + "num_tokens": 64525896.0, + "step": 2502 + }, + { + "epoch": 0.2748737096419943, + "grad_norm": 2.1094179153442383, + "learning_rate": 9.158125915080527e-07, + "loss": 1.0636, + "mean_token_accuracy": 0.6867114305496216, + "num_tokens": 64556903.0, + "step": 2503 + }, + { + "epoch": 0.27498352734460796, + "grad_norm": 2.06178617477417, + "learning_rate": 9.161786237188872e-07, + "loss": 1.0564, + "mean_token_accuracy": 0.6805382966995239, + "num_tokens": 64586040.0, + "step": 2504 + }, + { + "epoch": 0.2750933450472216, + "grad_norm": 2.2370707988739014, + "learning_rate": 9.165446559297217e-07, + "loss": 1.0761, + "mean_token_accuracy": 0.6837582588195801, + "num_tokens": 64611675.0, + "step": 2505 + }, + { + "epoch": 0.27520316274983525, + "grad_norm": 2.1856849193573, + "learning_rate": 9.169106881405563e-07, + "loss": 0.9444, + "mean_token_accuracy": 0.7161985635757446, + "num_tokens": 64637574.0, + "step": 2506 + }, + { + "epoch": 0.27531298045244895, + "grad_norm": 2.4117820262908936, + "learning_rate": 9.172767203513909e-07, + "loss": 0.8862, + "mean_token_accuracy": 0.7256633639335632, + "num_tokens": 64659812.0, + "step": 2507 + }, + { + "epoch": 0.2754227981550626, + "grad_norm": 2.490934133529663, + "learning_rate": 9.176427525622254e-07, + "loss": 1.0891, + "mean_token_accuracy": 0.6780577898025513, + "num_tokens": 64683019.0, + "step": 2508 + }, + { + "epoch": 0.27553261585767624, + "grad_norm": 2.342721939086914, + "learning_rate": 9.1800878477306e-07, + "loss": 1.0298, + "mean_token_accuracy": 0.6964131593704224, + "num_tokens": 64708468.0, + "step": 2509 + }, + { + "epoch": 0.27564243356028995, + "grad_norm": 2.0073060989379883, + "learning_rate": 9.183748169838945e-07, + "loss": 1.0494, + "mean_token_accuracy": 0.6801457405090332, + "num_tokens": 64740161.0, + "step": 2510 + }, + { + "epoch": 0.2757522512629036, + "grad_norm": 2.5747039318084717, + "learning_rate": 9.18740849194729e-07, + "loss": 0.9658, + "mean_token_accuracy": 0.7069753408432007, + "num_tokens": 64759273.0, + "step": 2511 + }, + { + "epoch": 0.27586206896551724, + "grad_norm": 2.013700246810913, + "learning_rate": 9.191068814055637e-07, + "loss": 0.992, + "mean_token_accuracy": 0.6991157531738281, + "num_tokens": 64788926.0, + "step": 2512 + }, + { + "epoch": 0.2759718866681309, + "grad_norm": 2.219805955886841, + "learning_rate": 9.194729136163982e-07, + "loss": 1.0487, + "mean_token_accuracy": 0.6884034276008606, + "num_tokens": 64817940.0, + "step": 2513 + }, + { + "epoch": 0.2760817043707446, + "grad_norm": 1.9294956922531128, + "learning_rate": 9.198389458272328e-07, + "loss": 1.1315, + "mean_token_accuracy": 0.6795989274978638, + "num_tokens": 64855373.0, + "step": 2514 + }, + { + "epoch": 0.27619152207335823, + "grad_norm": 2.1391775608062744, + "learning_rate": 9.202049780380673e-07, + "loss": 1.0038, + "mean_token_accuracy": 0.6937045454978943, + "num_tokens": 64881957.0, + "step": 2515 + }, + { + "epoch": 0.2763013397759719, + "grad_norm": 2.163236618041992, + "learning_rate": 9.205710102489018e-07, + "loss": 0.9529, + "mean_token_accuracy": 0.7078603506088257, + "num_tokens": 64908585.0, + "step": 2516 + }, + { + "epoch": 0.2764111574785855, + "grad_norm": 2.5423784255981445, + "learning_rate": 9.209370424597365e-07, + "loss": 0.9632, + "mean_token_accuracy": 0.7061168551445007, + "num_tokens": 64930242.0, + "step": 2517 + }, + { + "epoch": 0.2765209751811992, + "grad_norm": 2.0130512714385986, + "learning_rate": 9.21303074670571e-07, + "loss": 0.9794, + "mean_token_accuracy": 0.7023200988769531, + "num_tokens": 64959196.0, + "step": 2518 + }, + { + "epoch": 0.27663079288381287, + "grad_norm": 2.145984172821045, + "learning_rate": 9.216691068814055e-07, + "loss": 1.0111, + "mean_token_accuracy": 0.6919271945953369, + "num_tokens": 64987422.0, + "step": 2519 + }, + { + "epoch": 0.2767406105864265, + "grad_norm": 2.1448028087615967, + "learning_rate": 9.220351390922401e-07, + "loss": 1.0898, + "mean_token_accuracy": 0.6831426620483398, + "num_tokens": 65015218.0, + "step": 2520 + }, + { + "epoch": 0.2768504282890402, + "grad_norm": 2.20833420753479, + "learning_rate": 9.224011713030746e-07, + "loss": 1.0146, + "mean_token_accuracy": 0.6979796886444092, + "num_tokens": 65042379.0, + "step": 2521 + }, + { + "epoch": 0.27696024599165386, + "grad_norm": 2.2184653282165527, + "learning_rate": 9.227672035139091e-07, + "loss": 1.0406, + "mean_token_accuracy": 0.6857427358627319, + "num_tokens": 65069290.0, + "step": 2522 + }, + { + "epoch": 0.2770700636942675, + "grad_norm": 2.2056493759155273, + "learning_rate": 9.231332357247438e-07, + "loss": 0.9788, + "mean_token_accuracy": 0.699761152267456, + "num_tokens": 65094108.0, + "step": 2523 + }, + { + "epoch": 0.27717988139688116, + "grad_norm": 2.1010990142822266, + "learning_rate": 9.234992679355783e-07, + "loss": 0.9712, + "mean_token_accuracy": 0.7069534063339233, + "num_tokens": 65122212.0, + "step": 2524 + }, + { + "epoch": 0.27728969909949486, + "grad_norm": 2.1044294834136963, + "learning_rate": 9.238653001464128e-07, + "loss": 1.068, + "mean_token_accuracy": 0.6854128837585449, + "num_tokens": 65153331.0, + "step": 2525 + }, + { + "epoch": 0.2773995168021085, + "grad_norm": 2.2653090953826904, + "learning_rate": 9.242313323572474e-07, + "loss": 0.9833, + "mean_token_accuracy": 0.7060443162918091, + "num_tokens": 65178441.0, + "step": 2526 + }, + { + "epoch": 0.27750933450472215, + "grad_norm": 2.067258358001709, + "learning_rate": 9.245973645680819e-07, + "loss": 0.948, + "mean_token_accuracy": 0.7082788944244385, + "num_tokens": 65207757.0, + "step": 2527 + }, + { + "epoch": 0.2776191522073358, + "grad_norm": 2.22263765335083, + "learning_rate": 9.249633967789166e-07, + "loss": 1.063, + "mean_token_accuracy": 0.6882146596908569, + "num_tokens": 65236719.0, + "step": 2528 + }, + { + "epoch": 0.2777289699099495, + "grad_norm": 2.19158935546875, + "learning_rate": 9.253294289897511e-07, + "loss": 0.9303, + "mean_token_accuracy": 0.7154924273490906, + "num_tokens": 65260911.0, + "step": 2529 + }, + { + "epoch": 0.27783878761256314, + "grad_norm": 2.610502243041992, + "learning_rate": 9.256954612005856e-07, + "loss": 0.9502, + "mean_token_accuracy": 0.7097010612487793, + "num_tokens": 65281523.0, + "step": 2530 + }, + { + "epoch": 0.2779486053151768, + "grad_norm": 2.396914005279541, + "learning_rate": 9.260614934114202e-07, + "loss": 1.0186, + "mean_token_accuracy": 0.6903395652770996, + "num_tokens": 65304936.0, + "step": 2531 + }, + { + "epoch": 0.2780584230177905, + "grad_norm": 2.166492462158203, + "learning_rate": 9.264275256222547e-07, + "loss": 0.8908, + "mean_token_accuracy": 0.7231855392456055, + "num_tokens": 65329771.0, + "step": 2532 + }, + { + "epoch": 0.27816824072040414, + "grad_norm": 2.4328455924987793, + "learning_rate": 9.267935578330893e-07, + "loss": 1.1028, + "mean_token_accuracy": 0.6751400232315063, + "num_tokens": 65352739.0, + "step": 2533 + }, + { + "epoch": 0.2782780584230178, + "grad_norm": 2.1094560623168945, + "learning_rate": 9.271595900439239e-07, + "loss": 1.0484, + "mean_token_accuracy": 0.6975799202919006, + "num_tokens": 65381469.0, + "step": 2534 + }, + { + "epoch": 0.27838787612563143, + "grad_norm": 2.4921681880950928, + "learning_rate": 9.275256222547584e-07, + "loss": 0.9499, + "mean_token_accuracy": 0.7106345295906067, + "num_tokens": 65402894.0, + "step": 2535 + }, + { + "epoch": 0.27849769382824513, + "grad_norm": 2.112710475921631, + "learning_rate": 9.278916544655929e-07, + "loss": 1.0883, + "mean_token_accuracy": 0.6840447187423706, + "num_tokens": 65433634.0, + "step": 2536 + }, + { + "epoch": 0.2786075115308588, + "grad_norm": 2.752638101577759, + "learning_rate": 9.282576866764275e-07, + "loss": 0.8889, + "mean_token_accuracy": 0.723065197467804, + "num_tokens": 65452164.0, + "step": 2537 + }, + { + "epoch": 0.2787173292334724, + "grad_norm": 2.1940295696258545, + "learning_rate": 9.28623718887262e-07, + "loss": 1.0733, + "mean_token_accuracy": 0.6878766417503357, + "num_tokens": 65478967.0, + "step": 2538 + }, + { + "epoch": 0.2788271469360861, + "grad_norm": 2.6250126361846924, + "learning_rate": 9.289897510980966e-07, + "loss": 0.9463, + "mean_token_accuracy": 0.7114288210868835, + "num_tokens": 65499590.0, + "step": 2539 + }, + { + "epoch": 0.27893696463869977, + "grad_norm": 2.352426052093506, + "learning_rate": 9.293557833089312e-07, + "loss": 0.9594, + "mean_token_accuracy": 0.7059093713760376, + "num_tokens": 65525619.0, + "step": 2540 + }, + { + "epoch": 0.2790467823413134, + "grad_norm": 2.4566256999969482, + "learning_rate": 9.297218155197657e-07, + "loss": 0.971, + "mean_token_accuracy": 0.7061828374862671, + "num_tokens": 65547328.0, + "step": 2541 + }, + { + "epoch": 0.27915660004392706, + "grad_norm": 2.23038911819458, + "learning_rate": 9.300878477306002e-07, + "loss": 1.0786, + "mean_token_accuracy": 0.678942084312439, + "num_tokens": 65574613.0, + "step": 2542 + }, + { + "epoch": 0.27926641774654076, + "grad_norm": 2.1297767162323, + "learning_rate": 9.304538799414348e-07, + "loss": 0.9611, + "mean_token_accuracy": 0.7033512592315674, + "num_tokens": 65603136.0, + "step": 2543 + }, + { + "epoch": 0.2793762354491544, + "grad_norm": 2.336867332458496, + "learning_rate": 9.308199121522694e-07, + "loss": 1.0, + "mean_token_accuracy": 0.6971574425697327, + "num_tokens": 65628228.0, + "step": 2544 + }, + { + "epoch": 0.27948605315176805, + "grad_norm": 2.287759304046631, + "learning_rate": 9.31185944363104e-07, + "loss": 1.0293, + "mean_token_accuracy": 0.689262866973877, + "num_tokens": 65655181.0, + "step": 2545 + }, + { + "epoch": 0.2795958708543817, + "grad_norm": 2.054311990737915, + "learning_rate": 9.315519765739385e-07, + "loss": 1.1155, + "mean_token_accuracy": 0.6732683181762695, + "num_tokens": 65684935.0, + "step": 2546 + }, + { + "epoch": 0.2797056885569954, + "grad_norm": 2.3805131912231445, + "learning_rate": 9.31918008784773e-07, + "loss": 0.9992, + "mean_token_accuracy": 0.694020688533783, + "num_tokens": 65708943.0, + "step": 2547 + }, + { + "epoch": 0.27981550625960905, + "grad_norm": 2.1331944465637207, + "learning_rate": 9.322840409956076e-07, + "loss": 0.9858, + "mean_token_accuracy": 0.7098878026008606, + "num_tokens": 65737123.0, + "step": 2548 + }, + { + "epoch": 0.2799253239622227, + "grad_norm": 2.4031310081481934, + "learning_rate": 9.326500732064422e-07, + "loss": 1.0342, + "mean_token_accuracy": 0.6962919235229492, + "num_tokens": 65760927.0, + "step": 2549 + }, + { + "epoch": 0.2800351416648364, + "grad_norm": 2.175689935684204, + "learning_rate": 9.330161054172767e-07, + "loss": 1.0288, + "mean_token_accuracy": 0.6904036998748779, + "num_tokens": 65788313.0, + "step": 2550 + }, + { + "epoch": 0.28014495936745004, + "grad_norm": 2.221029281616211, + "learning_rate": 9.333821376281113e-07, + "loss": 1.0172, + "mean_token_accuracy": 0.6920634508132935, + "num_tokens": 65812857.0, + "step": 2551 + }, + { + "epoch": 0.2802547770700637, + "grad_norm": 2.3106181621551514, + "learning_rate": 9.337481698389458e-07, + "loss": 0.9769, + "mean_token_accuracy": 0.708446741104126, + "num_tokens": 65836955.0, + "step": 2552 + }, + { + "epoch": 0.28036459477267733, + "grad_norm": 2.1737663745880127, + "learning_rate": 9.341142020497803e-07, + "loss": 1.0067, + "mean_token_accuracy": 0.6981669664382935, + "num_tokens": 65863603.0, + "step": 2553 + }, + { + "epoch": 0.28047441247529104, + "grad_norm": 2.3431007862091064, + "learning_rate": 9.344802342606149e-07, + "loss": 1.0377, + "mean_token_accuracy": 0.6931903958320618, + "num_tokens": 65890120.0, + "step": 2554 + }, + { + "epoch": 0.2805842301779047, + "grad_norm": 1.9381438493728638, + "learning_rate": 9.348462664714495e-07, + "loss": 1.0319, + "mean_token_accuracy": 0.6957899332046509, + "num_tokens": 65926236.0, + "step": 2555 + }, + { + "epoch": 0.2806940478805183, + "grad_norm": 2.456831932067871, + "learning_rate": 9.35212298682284e-07, + "loss": 0.9502, + "mean_token_accuracy": 0.7107003331184387, + "num_tokens": 65947643.0, + "step": 2556 + }, + { + "epoch": 0.28080386558313203, + "grad_norm": 2.4781486988067627, + "learning_rate": 9.355783308931186e-07, + "loss": 0.975, + "mean_token_accuracy": 0.7111519575119019, + "num_tokens": 65970667.0, + "step": 2557 + }, + { + "epoch": 0.2809136832857457, + "grad_norm": 2.265005588531494, + "learning_rate": 9.359443631039531e-07, + "loss": 1.0521, + "mean_token_accuracy": 0.6861891150474548, + "num_tokens": 65996991.0, + "step": 2558 + }, + { + "epoch": 0.2810235009883593, + "grad_norm": 2.08974027633667, + "learning_rate": 9.363103953147876e-07, + "loss": 1.0169, + "mean_token_accuracy": 0.6939053535461426, + "num_tokens": 66027766.0, + "step": 2559 + }, + { + "epoch": 0.28113331869097297, + "grad_norm": 2.636850595474243, + "learning_rate": 9.366764275256223e-07, + "loss": 0.919, + "mean_token_accuracy": 0.7154794335365295, + "num_tokens": 66046417.0, + "step": 2560 + }, + { + "epoch": 0.28124313639358667, + "grad_norm": 2.124711036682129, + "learning_rate": 9.370424597364568e-07, + "loss": 0.964, + "mean_token_accuracy": 0.707051694393158, + "num_tokens": 66073856.0, + "step": 2561 + }, + { + "epoch": 0.2813529540962003, + "grad_norm": 2.2387380599975586, + "learning_rate": 9.374084919472914e-07, + "loss": 1.0055, + "mean_token_accuracy": 0.6990051865577698, + "num_tokens": 66101429.0, + "step": 2562 + }, + { + "epoch": 0.28146277179881396, + "grad_norm": 2.2173373699188232, + "learning_rate": 9.377745241581259e-07, + "loss": 1.0982, + "mean_token_accuracy": 0.6724188327789307, + "num_tokens": 66129613.0, + "step": 2563 + }, + { + "epoch": 0.2815725895014276, + "grad_norm": 2.141000747680664, + "learning_rate": 9.381405563689604e-07, + "loss": 1.0233, + "mean_token_accuracy": 0.689111053943634, + "num_tokens": 66159666.0, + "step": 2564 + }, + { + "epoch": 0.2816824072040413, + "grad_norm": 2.7069318294525146, + "learning_rate": 9.385065885797951e-07, + "loss": 0.9957, + "mean_token_accuracy": 0.7007836103439331, + "num_tokens": 66181171.0, + "step": 2565 + }, + { + "epoch": 0.28179222490665495, + "grad_norm": 2.2367501258850098, + "learning_rate": 9.388726207906296e-07, + "loss": 0.931, + "mean_token_accuracy": 0.7205237150192261, + "num_tokens": 66205013.0, + "step": 2566 + }, + { + "epoch": 0.2819020426092686, + "grad_norm": 2.205528974533081, + "learning_rate": 9.392386530014641e-07, + "loss": 1.0005, + "mean_token_accuracy": 0.7026245594024658, + "num_tokens": 66230664.0, + "step": 2567 + }, + { + "epoch": 0.2820118603118823, + "grad_norm": 2.033639669418335, + "learning_rate": 9.396046852122987e-07, + "loss": 0.9995, + "mean_token_accuracy": 0.6995269060134888, + "num_tokens": 66260812.0, + "step": 2568 + }, + { + "epoch": 0.28212167801449595, + "grad_norm": 2.426039457321167, + "learning_rate": 9.399707174231332e-07, + "loss": 1.0279, + "mean_token_accuracy": 0.6858097314834595, + "num_tokens": 66282638.0, + "step": 2569 + }, + { + "epoch": 0.2822314957171096, + "grad_norm": 1.9384636878967285, + "learning_rate": 9.403367496339677e-07, + "loss": 1.0546, + "mean_token_accuracy": 0.6874921321868896, + "num_tokens": 66315689.0, + "step": 2570 + }, + { + "epoch": 0.28234131341972324, + "grad_norm": 2.055361032485962, + "learning_rate": 9.407027818448024e-07, + "loss": 0.9669, + "mean_token_accuracy": 0.7068215608596802, + "num_tokens": 66343772.0, + "step": 2571 + }, + { + "epoch": 0.28245113112233694, + "grad_norm": 2.2524971961975098, + "learning_rate": 9.410688140556369e-07, + "loss": 0.8996, + "mean_token_accuracy": 0.7202589511871338, + "num_tokens": 66366792.0, + "step": 2572 + }, + { + "epoch": 0.2825609488249506, + "grad_norm": 2.4353857040405273, + "learning_rate": 9.414348462664714e-07, + "loss": 0.9671, + "mean_token_accuracy": 0.7040020823478699, + "num_tokens": 66389908.0, + "step": 2573 + }, + { + "epoch": 0.28267076652756423, + "grad_norm": 2.0176069736480713, + "learning_rate": 9.41800878477306e-07, + "loss": 1.0522, + "mean_token_accuracy": 0.688022255897522, + "num_tokens": 66423245.0, + "step": 2574 + }, + { + "epoch": 0.2827805842301779, + "grad_norm": 2.252678871154785, + "learning_rate": 9.421669106881405e-07, + "loss": 1.0815, + "mean_token_accuracy": 0.6739844083786011, + "num_tokens": 66453024.0, + "step": 2575 + }, + { + "epoch": 0.2828904019327916, + "grad_norm": 2.206031322479248, + "learning_rate": 9.425329428989752e-07, + "loss": 1.0156, + "mean_token_accuracy": 0.6962989568710327, + "num_tokens": 66479736.0, + "step": 2576 + }, + { + "epoch": 0.2830002196354052, + "grad_norm": 2.1648874282836914, + "learning_rate": 9.428989751098097e-07, + "loss": 1.0583, + "mean_token_accuracy": 0.6794898509979248, + "num_tokens": 66509498.0, + "step": 2577 + }, + { + "epoch": 0.28311003733801887, + "grad_norm": 2.3364813327789307, + "learning_rate": 9.432650073206442e-07, + "loss": 1.0116, + "mean_token_accuracy": 0.696961522102356, + "num_tokens": 66534943.0, + "step": 2578 + }, + { + "epoch": 0.2832198550406326, + "grad_norm": 2.1236164569854736, + "learning_rate": 9.436310395314788e-07, + "loss": 0.9958, + "mean_token_accuracy": 0.6984900236129761, + "num_tokens": 66561088.0, + "step": 2579 + }, + { + "epoch": 0.2833296727432462, + "grad_norm": 2.3231942653656006, + "learning_rate": 9.439970717423133e-07, + "loss": 0.8793, + "mean_token_accuracy": 0.7304589748382568, + "num_tokens": 66584705.0, + "step": 2580 + }, + { + "epoch": 0.28343949044585987, + "grad_norm": 2.0918850898742676, + "learning_rate": 9.443631039531479e-07, + "loss": 1.0605, + "mean_token_accuracy": 0.6822526454925537, + "num_tokens": 66617156.0, + "step": 2581 + }, + { + "epoch": 0.2835493081484735, + "grad_norm": 2.329127788543701, + "learning_rate": 9.447291361639825e-07, + "loss": 1.0804, + "mean_token_accuracy": 0.6731151342391968, + "num_tokens": 66644407.0, + "step": 2582 + }, + { + "epoch": 0.2836591258510872, + "grad_norm": 2.136164903640747, + "learning_rate": 9.45095168374817e-07, + "loss": 1.1097, + "mean_token_accuracy": 0.6759012937545776, + "num_tokens": 66680089.0, + "step": 2583 + }, + { + "epoch": 0.28376894355370086, + "grad_norm": 1.8849552869796753, + "learning_rate": 9.454612005856515e-07, + "loss": 1.0214, + "mean_token_accuracy": 0.6927334070205688, + "num_tokens": 66716109.0, + "step": 2584 + }, + { + "epoch": 0.2838787612563145, + "grad_norm": 2.404611825942993, + "learning_rate": 9.458272327964861e-07, + "loss": 1.0454, + "mean_token_accuracy": 0.6920219659805298, + "num_tokens": 66740383.0, + "step": 2585 + }, + { + "epoch": 0.2839885789589282, + "grad_norm": 1.9607499837875366, + "learning_rate": 9.461932650073206e-07, + "loss": 1.0428, + "mean_token_accuracy": 0.6938774585723877, + "num_tokens": 66774475.0, + "step": 2586 + }, + { + "epoch": 0.28409839666154185, + "grad_norm": 2.0533084869384766, + "learning_rate": 9.465592972181552e-07, + "loss": 0.9328, + "mean_token_accuracy": 0.7174223065376282, + "num_tokens": 66802774.0, + "step": 2587 + }, + { + "epoch": 0.2842082143641555, + "grad_norm": 2.6839113235473633, + "learning_rate": 9.469253294289898e-07, + "loss": 0.9779, + "mean_token_accuracy": 0.6968852281570435, + "num_tokens": 66822965.0, + "step": 2588 + }, + { + "epoch": 0.28431803206676914, + "grad_norm": 2.143176794052124, + "learning_rate": 9.472913616398243e-07, + "loss": 1.0464, + "mean_token_accuracy": 0.6909154653549194, + "num_tokens": 66852734.0, + "step": 2589 + }, + { + "epoch": 0.28442784976938285, + "grad_norm": 2.250211000442505, + "learning_rate": 9.476573938506587e-07, + "loss": 0.9326, + "mean_token_accuracy": 0.718364953994751, + "num_tokens": 66877006.0, + "step": 2590 + }, + { + "epoch": 0.2845376674719965, + "grad_norm": 2.329521894454956, + "learning_rate": 9.480234260614933e-07, + "loss": 1.0556, + "mean_token_accuracy": 0.6885672807693481, + "num_tokens": 66902300.0, + "step": 2591 + }, + { + "epoch": 0.28464748517461014, + "grad_norm": 2.159374475479126, + "learning_rate": 9.48389458272328e-07, + "loss": 0.9274, + "mean_token_accuracy": 0.7193210124969482, + "num_tokens": 66930991.0, + "step": 2592 + }, + { + "epoch": 0.2847573028772238, + "grad_norm": 2.2639987468719482, + "learning_rate": 9.487554904831626e-07, + "loss": 1.0199, + "mean_token_accuracy": 0.6967711448669434, + "num_tokens": 66957004.0, + "step": 2593 + }, + { + "epoch": 0.2848671205798375, + "grad_norm": 2.518287181854248, + "learning_rate": 9.49121522693997e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.7031047344207764, + "num_tokens": 66980931.0, + "step": 2594 + }, + { + "epoch": 0.28497693828245113, + "grad_norm": 2.343151807785034, + "learning_rate": 9.494875549048315e-07, + "loss": 0.9832, + "mean_token_accuracy": 0.7085639834403992, + "num_tokens": 67005148.0, + "step": 2595 + }, + { + "epoch": 0.2850867559850648, + "grad_norm": 2.6339175701141357, + "learning_rate": 9.498535871156661e-07, + "loss": 1.0134, + "mean_token_accuracy": 0.6817078590393066, + "num_tokens": 67024780.0, + "step": 2596 + }, + { + "epoch": 0.2851965736876785, + "grad_norm": 2.2542028427124023, + "learning_rate": 9.502196193265007e-07, + "loss": 1.0349, + "mean_token_accuracy": 0.6945260167121887, + "num_tokens": 67051088.0, + "step": 2597 + }, + { + "epoch": 0.2853063913902921, + "grad_norm": 2.5782179832458496, + "learning_rate": 9.505856515373352e-07, + "loss": 1.0031, + "mean_token_accuracy": 0.7009164094924927, + "num_tokens": 67073148.0, + "step": 2598 + }, + { + "epoch": 0.28541620909290577, + "grad_norm": 2.181640625, + "learning_rate": 9.509516837481698e-07, + "loss": 1.0767, + "mean_token_accuracy": 0.6857804656028748, + "num_tokens": 67103289.0, + "step": 2599 + }, + { + "epoch": 0.2855260267955194, + "grad_norm": 2.172370672225952, + "learning_rate": 9.513177159590043e-07, + "loss": 1.1288, + "mean_token_accuracy": 0.6646380424499512, + "num_tokens": 67134940.0, + "step": 2600 + }, + { + "epoch": 0.2856358444981331, + "grad_norm": 2.133312940597534, + "learning_rate": 9.516837481698388e-07, + "loss": 1.035, + "mean_token_accuracy": 0.6913837790489197, + "num_tokens": 67162861.0, + "step": 2601 + }, + { + "epoch": 0.28574566220074676, + "grad_norm": 2.4984548091888428, + "learning_rate": 9.520497803806734e-07, + "loss": 0.9814, + "mean_token_accuracy": 0.7015656232833862, + "num_tokens": 67183291.0, + "step": 2602 + }, + { + "epoch": 0.2858554799033604, + "grad_norm": 2.2988157272338867, + "learning_rate": 9.52415812591508e-07, + "loss": 1.0169, + "mean_token_accuracy": 0.6940992474555969, + "num_tokens": 67208977.0, + "step": 2603 + }, + { + "epoch": 0.28596529760597406, + "grad_norm": 2.2771754264831543, + "learning_rate": 9.527818448023425e-07, + "loss": 0.9887, + "mean_token_accuracy": 0.7026540040969849, + "num_tokens": 67233401.0, + "step": 2604 + }, + { + "epoch": 0.28607511530858776, + "grad_norm": 2.239436626434326, + "learning_rate": 9.531478770131771e-07, + "loss": 1.0355, + "mean_token_accuracy": 0.6989148855209351, + "num_tokens": 67258898.0, + "step": 2605 + }, + { + "epoch": 0.2861849330112014, + "grad_norm": 2.0178020000457764, + "learning_rate": 9.535139092240116e-07, + "loss": 0.9922, + "mean_token_accuracy": 0.7024275660514832, + "num_tokens": 67288247.0, + "step": 2606 + }, + { + "epoch": 0.28629475071381505, + "grad_norm": 2.295494556427002, + "learning_rate": 9.538799414348462e-07, + "loss": 1.0075, + "mean_token_accuracy": 0.6974903345108032, + "num_tokens": 67312209.0, + "step": 2607 + }, + { + "epoch": 0.28640456841642875, + "grad_norm": 2.2244272232055664, + "learning_rate": 9.542459736456808e-07, + "loss": 1.0055, + "mean_token_accuracy": 0.7013359069824219, + "num_tokens": 67336959.0, + "step": 2608 + }, + { + "epoch": 0.2865143861190424, + "grad_norm": 2.11763858795166, + "learning_rate": 9.546120058565154e-07, + "loss": 0.9988, + "mean_token_accuracy": 0.6951291561126709, + "num_tokens": 67365192.0, + "step": 2609 + }, + { + "epoch": 0.28662420382165604, + "grad_norm": 2.2658581733703613, + "learning_rate": 9.549780380673498e-07, + "loss": 1.0004, + "mean_token_accuracy": 0.6951194405555725, + "num_tokens": 67392262.0, + "step": 2610 + }, + { + "epoch": 0.2867340215242697, + "grad_norm": 2.1534008979797363, + "learning_rate": 9.553440702781844e-07, + "loss": 1.0039, + "mean_token_accuracy": 0.7016963958740234, + "num_tokens": 67419572.0, + "step": 2611 + }, + { + "epoch": 0.2868438392268834, + "grad_norm": 2.4897913932800293, + "learning_rate": 9.55710102489019e-07, + "loss": 0.9455, + "mean_token_accuracy": 0.7169423699378967, + "num_tokens": 67440259.0, + "step": 2612 + }, + { + "epoch": 0.28695365692949704, + "grad_norm": 2.4561638832092285, + "learning_rate": 9.560761346998536e-07, + "loss": 0.9465, + "mean_token_accuracy": 0.7072044610977173, + "num_tokens": 67464103.0, + "step": 2613 + }, + { + "epoch": 0.2870634746321107, + "grad_norm": 1.9406682252883911, + "learning_rate": 9.56442166910688e-07, + "loss": 1.095, + "mean_token_accuracy": 0.6734815835952759, + "num_tokens": 67498304.0, + "step": 2614 + }, + { + "epoch": 0.2871732923347244, + "grad_norm": 2.545159339904785, + "learning_rate": 9.568081991215226e-07, + "loss": 1.0774, + "mean_token_accuracy": 0.6785274147987366, + "num_tokens": 67521821.0, + "step": 2615 + }, + { + "epoch": 0.28728311003733803, + "grad_norm": 2.5848679542541504, + "learning_rate": 9.571742313323572e-07, + "loss": 0.9643, + "mean_token_accuracy": 0.7086870670318604, + "num_tokens": 67541624.0, + "step": 2616 + }, + { + "epoch": 0.2873929277399517, + "grad_norm": 2.3948473930358887, + "learning_rate": 9.575402635431916e-07, + "loss": 0.9852, + "mean_token_accuracy": 0.7107275128364563, + "num_tokens": 67563863.0, + "step": 2617 + }, + { + "epoch": 0.2875027454425653, + "grad_norm": 2.3477399349212646, + "learning_rate": 9.579062957540262e-07, + "loss": 1.0739, + "mean_token_accuracy": 0.6756486296653748, + "num_tokens": 67588335.0, + "step": 2618 + }, + { + "epoch": 0.287612563145179, + "grad_norm": 2.500983238220215, + "learning_rate": 9.582723279648608e-07, + "loss": 0.9331, + "mean_token_accuracy": 0.7125064134597778, + "num_tokens": 67610513.0, + "step": 2619 + }, + { + "epoch": 0.28772238084779267, + "grad_norm": 2.024230480194092, + "learning_rate": 9.586383601756954e-07, + "loss": 1.1315, + "mean_token_accuracy": 0.6655651330947876, + "num_tokens": 67643857.0, + "step": 2620 + }, + { + "epoch": 0.2878321985504063, + "grad_norm": 2.4292452335357666, + "learning_rate": 9.5900439238653e-07, + "loss": 0.9808, + "mean_token_accuracy": 0.6978029012680054, + "num_tokens": 67665589.0, + "step": 2621 + }, + { + "epoch": 0.28794201625301996, + "grad_norm": 2.0352909564971924, + "learning_rate": 9.593704245973644e-07, + "loss": 1.0175, + "mean_token_accuracy": 0.6960277557373047, + "num_tokens": 67695418.0, + "step": 2622 + }, + { + "epoch": 0.28805183395563366, + "grad_norm": 2.4364404678344727, + "learning_rate": 9.59736456808199e-07, + "loss": 0.9845, + "mean_token_accuracy": 0.6981973052024841, + "num_tokens": 67716503.0, + "step": 2623 + }, + { + "epoch": 0.2881616516582473, + "grad_norm": 2.128634214401245, + "learning_rate": 9.601024890190336e-07, + "loss": 0.9935, + "mean_token_accuracy": 0.7036351561546326, + "num_tokens": 67743036.0, + "step": 2624 + }, + { + "epoch": 0.28827146936086095, + "grad_norm": 1.9877231121063232, + "learning_rate": 9.604685212298682e-07, + "loss": 1.0729, + "mean_token_accuracy": 0.6799983978271484, + "num_tokens": 67775477.0, + "step": 2625 + }, + { + "epoch": 0.28838128706347466, + "grad_norm": 2.603290319442749, + "learning_rate": 9.608345534407028e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.6935309767723083, + "num_tokens": 67796246.0, + "step": 2626 + }, + { + "epoch": 0.2884911047660883, + "grad_norm": 2.4273996353149414, + "learning_rate": 9.612005856515372e-07, + "loss": 0.983, + "mean_token_accuracy": 0.7081029415130615, + "num_tokens": 67820758.0, + "step": 2627 + }, + { + "epoch": 0.28860092246870195, + "grad_norm": 2.314307689666748, + "learning_rate": 9.615666178623718e-07, + "loss": 1.0078, + "mean_token_accuracy": 0.6977823972702026, + "num_tokens": 67844105.0, + "step": 2628 + }, + { + "epoch": 0.2887107401713156, + "grad_norm": 2.0451059341430664, + "learning_rate": 9.619326500732064e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6736538410186768, + "num_tokens": 67876536.0, + "step": 2629 + }, + { + "epoch": 0.2888205578739293, + "grad_norm": 2.117231607437134, + "learning_rate": 9.62298682284041e-07, + "loss": 0.9761, + "mean_token_accuracy": 0.7098772525787354, + "num_tokens": 67906726.0, + "step": 2630 + }, + { + "epoch": 0.28893037557654294, + "grad_norm": 2.203791618347168, + "learning_rate": 9.626647144948754e-07, + "loss": 1.027, + "mean_token_accuracy": 0.6825027465820312, + "num_tokens": 67934853.0, + "step": 2631 + }, + { + "epoch": 0.2890401932791566, + "grad_norm": 2.2449982166290283, + "learning_rate": 9.6303074670571e-07, + "loss": 1.0164, + "mean_token_accuracy": 0.7034356594085693, + "num_tokens": 67960487.0, + "step": 2632 + }, + { + "epoch": 0.2891500109817703, + "grad_norm": 2.246037244796753, + "learning_rate": 9.633967789165446e-07, + "loss": 0.9967, + "mean_token_accuracy": 0.6961237192153931, + "num_tokens": 67985099.0, + "step": 2633 + }, + { + "epoch": 0.28925982868438394, + "grad_norm": 2.6340997219085693, + "learning_rate": 9.63762811127379e-07, + "loss": 1.1005, + "mean_token_accuracy": 0.6745730638504028, + "num_tokens": 68005907.0, + "step": 2634 + }, + { + "epoch": 0.2893696463869976, + "grad_norm": 2.5215237140655518, + "learning_rate": 9.641288433382138e-07, + "loss": 0.9632, + "mean_token_accuracy": 0.7113520503044128, + "num_tokens": 68025604.0, + "step": 2635 + }, + { + "epoch": 0.2894794640896112, + "grad_norm": 2.1085686683654785, + "learning_rate": 9.644948755490482e-07, + "loss": 0.8923, + "mean_token_accuracy": 0.7266351580619812, + "num_tokens": 68053834.0, + "step": 2636 + }, + { + "epoch": 0.28958928179222493, + "grad_norm": 2.450920343399048, + "learning_rate": 9.648609077598828e-07, + "loss": 1.0182, + "mean_token_accuracy": 0.6910839080810547, + "num_tokens": 68075654.0, + "step": 2637 + }, + { + "epoch": 0.2896990994948386, + "grad_norm": 2.2957730293273926, + "learning_rate": 9.652269399707174e-07, + "loss": 0.9959, + "mean_token_accuracy": 0.701550304889679, + "num_tokens": 68100679.0, + "step": 2638 + }, + { + "epoch": 0.2898089171974522, + "grad_norm": 2.2809786796569824, + "learning_rate": 9.655929721815518e-07, + "loss": 1.0433, + "mean_token_accuracy": 0.6926870942115784, + "num_tokens": 68126152.0, + "step": 2639 + }, + { + "epoch": 0.28991873490006587, + "grad_norm": 1.8635330200195312, + "learning_rate": 9.659590043923866e-07, + "loss": 1.028, + "mean_token_accuracy": 0.6938780546188354, + "num_tokens": 68160904.0, + "step": 2640 + }, + { + "epoch": 0.29002855260267957, + "grad_norm": 2.0309765338897705, + "learning_rate": 9.66325036603221e-07, + "loss": 1.1072, + "mean_token_accuracy": 0.6692895889282227, + "num_tokens": 68194352.0, + "step": 2641 + }, + { + "epoch": 0.2901383703052932, + "grad_norm": 2.259702682495117, + "learning_rate": 9.666910688140556e-07, + "loss": 1.0263, + "mean_token_accuracy": 0.6963697671890259, + "num_tokens": 68220095.0, + "step": 2642 + }, + { + "epoch": 0.29024818800790686, + "grad_norm": 2.2767622470855713, + "learning_rate": 9.670571010248902e-07, + "loss": 0.9867, + "mean_token_accuracy": 0.7031564712524414, + "num_tokens": 68245461.0, + "step": 2643 + }, + { + "epoch": 0.29035800571052056, + "grad_norm": 2.2722342014312744, + "learning_rate": 9.674231332357246e-07, + "loss": 0.9957, + "mean_token_accuracy": 0.6934914588928223, + "num_tokens": 68270879.0, + "step": 2644 + }, + { + "epoch": 0.2904678234131342, + "grad_norm": 2.4367966651916504, + "learning_rate": 9.677891654465592e-07, + "loss": 0.9283, + "mean_token_accuracy": 0.7106821537017822, + "num_tokens": 68292893.0, + "step": 2645 + }, + { + "epoch": 0.29057764111574785, + "grad_norm": 2.142404079437256, + "learning_rate": 9.681551976573938e-07, + "loss": 1.0376, + "mean_token_accuracy": 0.6836990118026733, + "num_tokens": 68319075.0, + "step": 2646 + }, + { + "epoch": 0.2906874588183615, + "grad_norm": 2.4736099243164062, + "learning_rate": 9.685212298682284e-07, + "loss": 0.945, + "mean_token_accuracy": 0.7108451128005981, + "num_tokens": 68342098.0, + "step": 2647 + }, + { + "epoch": 0.2907972765209752, + "grad_norm": 2.183861255645752, + "learning_rate": 9.688872620790628e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.6719247102737427, + "num_tokens": 68371225.0, + "step": 2648 + }, + { + "epoch": 0.29090709422358885, + "grad_norm": 2.3119678497314453, + "learning_rate": 9.692532942898974e-07, + "loss": 1.0452, + "mean_token_accuracy": 0.6855568289756775, + "num_tokens": 68398188.0, + "step": 2649 + }, + { + "epoch": 0.2910169119262025, + "grad_norm": 2.229978561401367, + "learning_rate": 9.69619326500732e-07, + "loss": 0.9624, + "mean_token_accuracy": 0.707917332649231, + "num_tokens": 68425520.0, + "step": 2650 + }, + { + "epoch": 0.29112672962881614, + "grad_norm": 1.8240798711776733, + "learning_rate": 9.699853587115666e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.6781541109085083, + "num_tokens": 68462456.0, + "step": 2651 + }, + { + "epoch": 0.29123654733142984, + "grad_norm": 2.2227518558502197, + "learning_rate": 9.703513909224012e-07, + "loss": 1.0856, + "mean_token_accuracy": 0.6777976751327515, + "num_tokens": 68489331.0, + "step": 2652 + }, + { + "epoch": 0.2913463650340435, + "grad_norm": 2.2159085273742676, + "learning_rate": 9.707174231332356e-07, + "loss": 1.0311, + "mean_token_accuracy": 0.6884282827377319, + "num_tokens": 68516051.0, + "step": 2653 + }, + { + "epoch": 0.29145618273665713, + "grad_norm": 2.298194408416748, + "learning_rate": 9.710834553440702e-07, + "loss": 1.0347, + "mean_token_accuracy": 0.6937570571899414, + "num_tokens": 68539805.0, + "step": 2654 + }, + { + "epoch": 0.29156600043927083, + "grad_norm": 2.2887446880340576, + "learning_rate": 9.714494875549048e-07, + "loss": 0.9995, + "mean_token_accuracy": 0.6990783214569092, + "num_tokens": 68565221.0, + "step": 2655 + }, + { + "epoch": 0.2916758181418845, + "grad_norm": 2.173250436782837, + "learning_rate": 9.718155197657394e-07, + "loss": 0.9885, + "mean_token_accuracy": 0.7030911445617676, + "num_tokens": 68593703.0, + "step": 2656 + }, + { + "epoch": 0.2917856358444981, + "grad_norm": 2.340853691101074, + "learning_rate": 9.72181551976574e-07, + "loss": 0.986, + "mean_token_accuracy": 0.7007484436035156, + "num_tokens": 68615507.0, + "step": 2657 + }, + { + "epoch": 0.29189545354711177, + "grad_norm": 2.1384568214416504, + "learning_rate": 9.725475841874084e-07, + "loss": 1.0613, + "mean_token_accuracy": 0.6984245777130127, + "num_tokens": 68644423.0, + "step": 2658 + }, + { + "epoch": 0.2920052712497255, + "grad_norm": 2.2012290954589844, + "learning_rate": 9.72913616398243e-07, + "loss": 1.0232, + "mean_token_accuracy": 0.6956584453582764, + "num_tokens": 68670329.0, + "step": 2659 + }, + { + "epoch": 0.2921150889523391, + "grad_norm": 2.6119883060455322, + "learning_rate": 9.732796486090776e-07, + "loss": 1.0359, + "mean_token_accuracy": 0.6985217332839966, + "num_tokens": 68691542.0, + "step": 2660 + }, + { + "epoch": 0.29222490665495277, + "grad_norm": 2.4481046199798584, + "learning_rate": 9.736456808199122e-07, + "loss": 0.9832, + "mean_token_accuracy": 0.7018170952796936, + "num_tokens": 68713333.0, + "step": 2661 + }, + { + "epoch": 0.29233472435756647, + "grad_norm": 2.3623719215393066, + "learning_rate": 9.740117130307466e-07, + "loss": 0.9866, + "mean_token_accuracy": 0.7075522541999817, + "num_tokens": 68735818.0, + "step": 2662 + }, + { + "epoch": 0.2924445420601801, + "grad_norm": 2.1721391677856445, + "learning_rate": 9.743777452415812e-07, + "loss": 1.0159, + "mean_token_accuracy": 0.6948463916778564, + "num_tokens": 68762899.0, + "step": 2663 + }, + { + "epoch": 0.29255435976279376, + "grad_norm": 2.3620572090148926, + "learning_rate": 9.747437774524158e-07, + "loss": 1.0305, + "mean_token_accuracy": 0.6985598802566528, + "num_tokens": 68786750.0, + "step": 2664 + }, + { + "epoch": 0.2926641774654074, + "grad_norm": 2.4017579555511475, + "learning_rate": 9.751098096632502e-07, + "loss": 0.9802, + "mean_token_accuracy": 0.7032058238983154, + "num_tokens": 68810131.0, + "step": 2665 + }, + { + "epoch": 0.2927739951680211, + "grad_norm": 2.295124053955078, + "learning_rate": 9.754758418740848e-07, + "loss": 1.0909, + "mean_token_accuracy": 0.6784361600875854, + "num_tokens": 68837302.0, + "step": 2666 + }, + { + "epoch": 0.29288381287063475, + "grad_norm": 2.1128971576690674, + "learning_rate": 9.758418740849194e-07, + "loss": 1.0122, + "mean_token_accuracy": 0.7019174098968506, + "num_tokens": 68866848.0, + "step": 2667 + }, + { + "epoch": 0.2929936305732484, + "grad_norm": 2.3780105113983154, + "learning_rate": 9.76207906295754e-07, + "loss": 1.069, + "mean_token_accuracy": 0.683986485004425, + "num_tokens": 68892221.0, + "step": 2668 + }, + { + "epoch": 0.29310344827586204, + "grad_norm": 2.087470054626465, + "learning_rate": 9.765739385065886e-07, + "loss": 0.977, + "mean_token_accuracy": 0.7033562064170837, + "num_tokens": 68920022.0, + "step": 2669 + }, + { + "epoch": 0.29321326597847575, + "grad_norm": 2.5218374729156494, + "learning_rate": 9.76939970717423e-07, + "loss": 0.9778, + "mean_token_accuracy": 0.7096598148345947, + "num_tokens": 68941852.0, + "step": 2670 + }, + { + "epoch": 0.2933230836810894, + "grad_norm": 2.1031758785247803, + "learning_rate": 9.773060029282576e-07, + "loss": 1.0602, + "mean_token_accuracy": 0.6811301708221436, + "num_tokens": 68973292.0, + "step": 2671 + }, + { + "epoch": 0.29343290138370304, + "grad_norm": 2.5442121028900146, + "learning_rate": 9.776720351390922e-07, + "loss": 1.0487, + "mean_token_accuracy": 0.6899944543838501, + "num_tokens": 68995369.0, + "step": 2672 + }, + { + "epoch": 0.29354271908631674, + "grad_norm": 2.271013021469116, + "learning_rate": 9.780380673499268e-07, + "loss": 1.0183, + "mean_token_accuracy": 0.6855603456497192, + "num_tokens": 69018419.0, + "step": 2673 + }, + { + "epoch": 0.2936525367889304, + "grad_norm": 2.660414934158325, + "learning_rate": 9.784040995607614e-07, + "loss": 0.9705, + "mean_token_accuracy": 0.7110034227371216, + "num_tokens": 69039570.0, + "step": 2674 + }, + { + "epoch": 0.29376235449154403, + "grad_norm": 2.2443454265594482, + "learning_rate": 9.787701317715958e-07, + "loss": 0.9688, + "mean_token_accuracy": 0.709706723690033, + "num_tokens": 69065295.0, + "step": 2675 + }, + { + "epoch": 0.2938721721941577, + "grad_norm": 2.405879020690918, + "learning_rate": 9.791361639824304e-07, + "loss": 0.9889, + "mean_token_accuracy": 0.7084033489227295, + "num_tokens": 69088292.0, + "step": 2676 + }, + { + "epoch": 0.2939819898967714, + "grad_norm": 2.50042986869812, + "learning_rate": 9.79502196193265e-07, + "loss": 0.9745, + "mean_token_accuracy": 0.6989266872406006, + "num_tokens": 69109009.0, + "step": 2677 + }, + { + "epoch": 0.294091807599385, + "grad_norm": 2.137540578842163, + "learning_rate": 9.798682284040996e-07, + "loss": 0.9684, + "mean_token_accuracy": 0.7125083208084106, + "num_tokens": 69135296.0, + "step": 2678 + }, + { + "epoch": 0.29420162530199867, + "grad_norm": 2.0588443279266357, + "learning_rate": 9.80234260614934e-07, + "loss": 1.0632, + "mean_token_accuracy": 0.6791002154350281, + "num_tokens": 69167474.0, + "step": 2679 + }, + { + "epoch": 0.2943114430046123, + "grad_norm": 2.0134243965148926, + "learning_rate": 9.806002928257686e-07, + "loss": 1.0446, + "mean_token_accuracy": 0.6908745765686035, + "num_tokens": 69199833.0, + "step": 2680 + }, + { + "epoch": 0.294421260707226, + "grad_norm": 2.127913236618042, + "learning_rate": 9.809663250366032e-07, + "loss": 1.0454, + "mean_token_accuracy": 0.6859221458435059, + "num_tokens": 69228030.0, + "step": 2681 + }, + { + "epoch": 0.29453107840983966, + "grad_norm": 2.2489614486694336, + "learning_rate": 9.813323572474376e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.6808000206947327, + "num_tokens": 69253176.0, + "step": 2682 + }, + { + "epoch": 0.2946408961124533, + "grad_norm": 2.195676803588867, + "learning_rate": 9.816983894582724e-07, + "loss": 0.9609, + "mean_token_accuracy": 0.7094383835792542, + "num_tokens": 69279049.0, + "step": 2683 + }, + { + "epoch": 0.294750713815067, + "grad_norm": 1.9518990516662598, + "learning_rate": 9.820644216691068e-07, + "loss": 1.0486, + "mean_token_accuracy": 0.6866904497146606, + "num_tokens": 69311963.0, + "step": 2684 + }, + { + "epoch": 0.29486053151768066, + "grad_norm": 2.202749252319336, + "learning_rate": 9.824304538799414e-07, + "loss": 0.9492, + "mean_token_accuracy": 0.7129898071289062, + "num_tokens": 69338490.0, + "step": 2685 + }, + { + "epoch": 0.2949703492202943, + "grad_norm": 2.470165729522705, + "learning_rate": 9.82796486090776e-07, + "loss": 0.9795, + "mean_token_accuracy": 0.6985303163528442, + "num_tokens": 69360211.0, + "step": 2686 + }, + { + "epoch": 0.29508016692290795, + "grad_norm": 2.28065824508667, + "learning_rate": 9.831625183016104e-07, + "loss": 1.1454, + "mean_token_accuracy": 0.6730860471725464, + "num_tokens": 69389732.0, + "step": 2687 + }, + { + "epoch": 0.29518998462552165, + "grad_norm": 2.428617238998413, + "learning_rate": 9.835285505124452e-07, + "loss": 0.9146, + "mean_token_accuracy": 0.7168680429458618, + "num_tokens": 69412227.0, + "step": 2688 + }, + { + "epoch": 0.2952998023281353, + "grad_norm": 2.31239652633667, + "learning_rate": 9.838945827232796e-07, + "loss": 1.0335, + "mean_token_accuracy": 0.6912139058113098, + "num_tokens": 69435342.0, + "step": 2689 + }, + { + "epoch": 0.29540962003074894, + "grad_norm": 1.9640334844589233, + "learning_rate": 9.842606149341142e-07, + "loss": 1.0375, + "mean_token_accuracy": 0.6844738721847534, + "num_tokens": 69466875.0, + "step": 2690 + }, + { + "epoch": 0.29551943773336264, + "grad_norm": 2.224968910217285, + "learning_rate": 9.846266471449488e-07, + "loss": 1.0357, + "mean_token_accuracy": 0.7005177736282349, + "num_tokens": 69493073.0, + "step": 2691 + }, + { + "epoch": 0.2956292554359763, + "grad_norm": 2.16896915435791, + "learning_rate": 9.849926793557832e-07, + "loss": 0.964, + "mean_token_accuracy": 0.7049350738525391, + "num_tokens": 69520144.0, + "step": 2692 + }, + { + "epoch": 0.29573907313858994, + "grad_norm": 2.233616352081299, + "learning_rate": 9.853587115666178e-07, + "loss": 0.9917, + "mean_token_accuracy": 0.6986651420593262, + "num_tokens": 69545286.0, + "step": 2693 + }, + { + "epoch": 0.2958488908412036, + "grad_norm": 2.126495122909546, + "learning_rate": 9.857247437774524e-07, + "loss": 1.1238, + "mean_token_accuracy": 0.6733127236366272, + "num_tokens": 69575666.0, + "step": 2694 + }, + { + "epoch": 0.2959587085438173, + "grad_norm": 2.366623640060425, + "learning_rate": 9.86090775988287e-07, + "loss": 0.9856, + "mean_token_accuracy": 0.6988163590431213, + "num_tokens": 69597333.0, + "step": 2695 + }, + { + "epoch": 0.29606852624643093, + "grad_norm": 2.7827746868133545, + "learning_rate": 9.864568081991214e-07, + "loss": 0.957, + "mean_token_accuracy": 0.7057133913040161, + "num_tokens": 69615995.0, + "step": 2696 + }, + { + "epoch": 0.2961783439490446, + "grad_norm": 2.332331895828247, + "learning_rate": 9.86822840409956e-07, + "loss": 1.0994, + "mean_token_accuracy": 0.6766530275344849, + "num_tokens": 69640263.0, + "step": 2697 + }, + { + "epoch": 0.2962881616516582, + "grad_norm": 2.4079017639160156, + "learning_rate": 9.871888726207906e-07, + "loss": 1.006, + "mean_token_accuracy": 0.7052060961723328, + "num_tokens": 69662277.0, + "step": 2698 + }, + { + "epoch": 0.2963979793542719, + "grad_norm": 1.9755194187164307, + "learning_rate": 9.875549048316252e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.6781561374664307, + "num_tokens": 69694745.0, + "step": 2699 + }, + { + "epoch": 0.29650779705688557, + "grad_norm": 2.2780539989471436, + "learning_rate": 9.879209370424598e-07, + "loss": 1.0134, + "mean_token_accuracy": 0.6956132650375366, + "num_tokens": 69720316.0, + "step": 2700 + }, + { + "epoch": 0.2966176147594992, + "grad_norm": 2.148362874984741, + "learning_rate": 9.882869692532942e-07, + "loss": 1.0342, + "mean_token_accuracy": 0.6885716319084167, + "num_tokens": 69750100.0, + "step": 2701 + }, + { + "epoch": 0.2967274324621129, + "grad_norm": 2.1811254024505615, + "learning_rate": 9.886530014641288e-07, + "loss": 1.0462, + "mean_token_accuracy": 0.6889955997467041, + "num_tokens": 69777792.0, + "step": 2702 + }, + { + "epoch": 0.29683725016472656, + "grad_norm": 2.2349164485931396, + "learning_rate": 9.890190336749634e-07, + "loss": 1.0669, + "mean_token_accuracy": 0.6749979257583618, + "num_tokens": 69804220.0, + "step": 2703 + }, + { + "epoch": 0.2969470678673402, + "grad_norm": 2.076106548309326, + "learning_rate": 9.89385065885798e-07, + "loss": 1.0673, + "mean_token_accuracy": 0.6745750904083252, + "num_tokens": 69837893.0, + "step": 2704 + }, + { + "epoch": 0.29705688556995385, + "grad_norm": 2.306227207183838, + "learning_rate": 9.897510980966326e-07, + "loss": 0.9817, + "mean_token_accuracy": 0.7099364995956421, + "num_tokens": 69862269.0, + "step": 2705 + }, + { + "epoch": 0.29716670327256756, + "grad_norm": 2.0741591453552246, + "learning_rate": 9.90117130307467e-07, + "loss": 1.0806, + "mean_token_accuracy": 0.674550473690033, + "num_tokens": 69893396.0, + "step": 2706 + }, + { + "epoch": 0.2972765209751812, + "grad_norm": 2.1917598247528076, + "learning_rate": 9.904831625183016e-07, + "loss": 0.9881, + "mean_token_accuracy": 0.697969913482666, + "num_tokens": 69920648.0, + "step": 2707 + }, + { + "epoch": 0.29738633867779485, + "grad_norm": 2.1522910594940186, + "learning_rate": 9.908491947291362e-07, + "loss": 0.9968, + "mean_token_accuracy": 0.6986153721809387, + "num_tokens": 69947077.0, + "step": 2708 + }, + { + "epoch": 0.29749615638040855, + "grad_norm": 2.169537305831909, + "learning_rate": 9.912152269399708e-07, + "loss": 1.0757, + "mean_token_accuracy": 0.6833815574645996, + "num_tokens": 69977260.0, + "step": 2709 + }, + { + "epoch": 0.2976059740830222, + "grad_norm": 2.0514261722564697, + "learning_rate": 9.915812591508052e-07, + "loss": 0.9972, + "mean_token_accuracy": 0.7023543119430542, + "num_tokens": 70006314.0, + "step": 2710 + }, + { + "epoch": 0.29771579178563584, + "grad_norm": 2.172773838043213, + "learning_rate": 9.919472913616398e-07, + "loss": 0.9911, + "mean_token_accuracy": 0.6999436020851135, + "num_tokens": 70033098.0, + "step": 2711 + }, + { + "epoch": 0.2978256094882495, + "grad_norm": 2.170665979385376, + "learning_rate": 9.923133235724744e-07, + "loss": 1.0075, + "mean_token_accuracy": 0.7043900489807129, + "num_tokens": 70062200.0, + "step": 2712 + }, + { + "epoch": 0.2979354271908632, + "grad_norm": 2.169872760772705, + "learning_rate": 9.926793557833088e-07, + "loss": 1.0667, + "mean_token_accuracy": 0.6953064799308777, + "num_tokens": 70089711.0, + "step": 2713 + }, + { + "epoch": 0.29804524489347684, + "grad_norm": 2.3210315704345703, + "learning_rate": 9.930453879941434e-07, + "loss": 0.9237, + "mean_token_accuracy": 0.7204326391220093, + "num_tokens": 70113978.0, + "step": 2714 + }, + { + "epoch": 0.2981550625960905, + "grad_norm": 2.2660446166992188, + "learning_rate": 9.93411420204978e-07, + "loss": 0.9835, + "mean_token_accuracy": 0.7000664472579956, + "num_tokens": 70138238.0, + "step": 2715 + }, + { + "epoch": 0.2982648802987041, + "grad_norm": 2.388376474380493, + "learning_rate": 9.937774524158126e-07, + "loss": 0.9634, + "mean_token_accuracy": 0.7082678079605103, + "num_tokens": 70161906.0, + "step": 2716 + }, + { + "epoch": 0.29837469800131783, + "grad_norm": 2.0736913681030273, + "learning_rate": 9.941434846266472e-07, + "loss": 0.9597, + "mean_token_accuracy": 0.7063441276550293, + "num_tokens": 70192723.0, + "step": 2717 + }, + { + "epoch": 0.2984845157039315, + "grad_norm": 2.1220645904541016, + "learning_rate": 9.945095168374816e-07, + "loss": 0.9891, + "mean_token_accuracy": 0.7014169692993164, + "num_tokens": 70221730.0, + "step": 2718 + }, + { + "epoch": 0.2985943334065451, + "grad_norm": 2.156407356262207, + "learning_rate": 9.948755490483162e-07, + "loss": 0.9695, + "mean_token_accuracy": 0.7089258432388306, + "num_tokens": 70249072.0, + "step": 2719 + }, + { + "epoch": 0.2987041511091588, + "grad_norm": 2.5050876140594482, + "learning_rate": 9.952415812591508e-07, + "loss": 0.9675, + "mean_token_accuracy": 0.7041623592376709, + "num_tokens": 70270451.0, + "step": 2720 + }, + { + "epoch": 0.29881396881177247, + "grad_norm": 2.5383832454681396, + "learning_rate": 9.956076134699854e-07, + "loss": 0.9151, + "mean_token_accuracy": 0.7243280410766602, + "num_tokens": 70290888.0, + "step": 2721 + }, + { + "epoch": 0.2989237865143861, + "grad_norm": 2.42425274848938, + "learning_rate": 9.9597364568082e-07, + "loss": 1.0862, + "mean_token_accuracy": 0.6754619479179382, + "num_tokens": 70314777.0, + "step": 2722 + }, + { + "epoch": 0.29903360421699976, + "grad_norm": 2.0854458808898926, + "learning_rate": 9.963396778916544e-07, + "loss": 0.9776, + "mean_token_accuracy": 0.7058759331703186, + "num_tokens": 70343692.0, + "step": 2723 + }, + { + "epoch": 0.29914342191961346, + "grad_norm": 2.2198433876037598, + "learning_rate": 9.96705710102489e-07, + "loss": 0.993, + "mean_token_accuracy": 0.6952611804008484, + "num_tokens": 70367561.0, + "step": 2724 + }, + { + "epoch": 0.2992532396222271, + "grad_norm": 2.4132978916168213, + "learning_rate": 9.970717423133236e-07, + "loss": 0.9263, + "mean_token_accuracy": 0.7109625339508057, + "num_tokens": 70389900.0, + "step": 2725 + }, + { + "epoch": 0.29936305732484075, + "grad_norm": 2.375779628753662, + "learning_rate": 9.974377745241582e-07, + "loss": 0.9668, + "mean_token_accuracy": 0.7089971899986267, + "num_tokens": 70415989.0, + "step": 2726 + }, + { + "epoch": 0.2994728750274544, + "grad_norm": 1.9589409828186035, + "learning_rate": 9.978038067349926e-07, + "loss": 1.0582, + "mean_token_accuracy": 0.6826830506324768, + "num_tokens": 70448872.0, + "step": 2727 + }, + { + "epoch": 0.2995826927300681, + "grad_norm": 2.045923948287964, + "learning_rate": 9.981698389458272e-07, + "loss": 1.0276, + "mean_token_accuracy": 0.692943811416626, + "num_tokens": 70479507.0, + "step": 2728 + }, + { + "epoch": 0.29969251043268175, + "grad_norm": 2.0197594165802, + "learning_rate": 9.985358711566618e-07, + "loss": 0.9735, + "mean_token_accuracy": 0.7015440464019775, + "num_tokens": 70510493.0, + "step": 2729 + }, + { + "epoch": 0.2998023281352954, + "grad_norm": 2.5519773960113525, + "learning_rate": 9.989019033674961e-07, + "loss": 0.9331, + "mean_token_accuracy": 0.7116044163703918, + "num_tokens": 70532115.0, + "step": 2730 + }, + { + "epoch": 0.2999121458379091, + "grad_norm": 1.9601178169250488, + "learning_rate": 9.99267935578331e-07, + "loss": 0.9075, + "mean_token_accuracy": 0.7114285826683044, + "num_tokens": 70563634.0, + "step": 2731 + }, + { + "epoch": 0.30002196354052274, + "grad_norm": 2.1698544025421143, + "learning_rate": 9.996339677891654e-07, + "loss": 0.9946, + "mean_token_accuracy": 0.7061415910720825, + "num_tokens": 70591530.0, + "step": 2732 + }, + { + "epoch": 0.3001317812431364, + "grad_norm": 2.6030757427215576, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6906983852386475, + "num_tokens": 70611898.0, + "step": 2733 + }, + { + "epoch": 0.30024159894575003, + "grad_norm": 2.5532631874084473, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7200855016708374, + "num_tokens": 70631779.0, + "step": 2734 + }, + { + "epoch": 0.30035141664836373, + "grad_norm": 2.0352096557617188, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6969099640846252, + "num_tokens": 70660246.0, + "step": 2735 + }, + { + "epoch": 0.3004612343509774, + "grad_norm": 2.062602996826172, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.691136360168457, + "num_tokens": 70691353.0, + "step": 2736 + }, + { + "epoch": 0.300571052053591, + "grad_norm": 2.5229716300964355, + "learning_rate": 1e-06, + "loss": 1.0888, + "mean_token_accuracy": 0.6730800867080688, + "num_tokens": 70714143.0, + "step": 2737 + }, + { + "epoch": 0.3006808697562047, + "grad_norm": 2.2652294635772705, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6881372928619385, + "num_tokens": 70738976.0, + "step": 2738 + }, + { + "epoch": 0.3007906874588184, + "grad_norm": 2.4426419734954834, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7152851223945618, + "num_tokens": 70760448.0, + "step": 2739 + }, + { + "epoch": 0.300900505161432, + "grad_norm": 2.144124746322632, + "learning_rate": 1e-06, + "loss": 1.0912, + "mean_token_accuracy": 0.6698315739631653, + "num_tokens": 70786483.0, + "step": 2740 + }, + { + "epoch": 0.30101032286404567, + "grad_norm": 2.353247880935669, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.710724949836731, + "num_tokens": 70809208.0, + "step": 2741 + }, + { + "epoch": 0.30112014056665937, + "grad_norm": 2.3668510913848877, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7082432508468628, + "num_tokens": 70834608.0, + "step": 2742 + }, + { + "epoch": 0.301229958269273, + "grad_norm": 2.2379753589630127, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7112662196159363, + "num_tokens": 70861261.0, + "step": 2743 + }, + { + "epoch": 0.30133977597188666, + "grad_norm": 2.215268611907959, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6864985227584839, + "num_tokens": 70889551.0, + "step": 2744 + }, + { + "epoch": 0.3014495936745003, + "grad_norm": 2.2173893451690674, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6955307722091675, + "num_tokens": 70916162.0, + "step": 2745 + }, + { + "epoch": 0.301559411377114, + "grad_norm": 2.1480531692504883, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7093936204910278, + "num_tokens": 70944163.0, + "step": 2746 + }, + { + "epoch": 0.30166922907972765, + "grad_norm": 2.147045612335205, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7061982750892639, + "num_tokens": 70970646.0, + "step": 2747 + }, + { + "epoch": 0.3017790467823413, + "grad_norm": 1.9532780647277832, + "learning_rate": 1e-06, + "loss": 1.0744, + "mean_token_accuracy": 0.6788488626480103, + "num_tokens": 71003040.0, + "step": 2748 + }, + { + "epoch": 0.301888864484955, + "grad_norm": 1.9920622110366821, + "learning_rate": 1e-06, + "loss": 1.0667, + "mean_token_accuracy": 0.6805837750434875, + "num_tokens": 71034419.0, + "step": 2749 + }, + { + "epoch": 0.30199868218756865, + "grad_norm": 2.287336587905884, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.6947891116142273, + "num_tokens": 71060266.0, + "step": 2750 + }, + { + "epoch": 0.3021084998901823, + "grad_norm": 2.2777209281921387, + "learning_rate": 1e-06, + "loss": 1.1108, + "mean_token_accuracy": 0.6671314239501953, + "num_tokens": 71085939.0, + "step": 2751 + }, + { + "epoch": 0.30221831759279594, + "grad_norm": 2.2398009300231934, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.693156361579895, + "num_tokens": 71110518.0, + "step": 2752 + }, + { + "epoch": 0.30232813529540964, + "grad_norm": 2.4073309898376465, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7086651921272278, + "num_tokens": 71132356.0, + "step": 2753 + }, + { + "epoch": 0.3024379529980233, + "grad_norm": 2.3206276893615723, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6936534643173218, + "num_tokens": 71155990.0, + "step": 2754 + }, + { + "epoch": 0.30254777070063693, + "grad_norm": 2.4664785861968994, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7188900709152222, + "num_tokens": 71176187.0, + "step": 2755 + }, + { + "epoch": 0.3026575884032506, + "grad_norm": 2.2650814056396484, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6996750831604004, + "num_tokens": 71200764.0, + "step": 2756 + }, + { + "epoch": 0.3027674061058643, + "grad_norm": 2.2769742012023926, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.694259762763977, + "num_tokens": 71228226.0, + "step": 2757 + }, + { + "epoch": 0.3028772238084779, + "grad_norm": 2.2407281398773193, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6924159526824951, + "num_tokens": 71252338.0, + "step": 2758 + }, + { + "epoch": 0.30298704151109157, + "grad_norm": 2.4407944679260254, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6830337047576904, + "num_tokens": 71276029.0, + "step": 2759 + }, + { + "epoch": 0.30309685921370527, + "grad_norm": 2.3228166103363037, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7171829342842102, + "num_tokens": 71300798.0, + "step": 2760 + }, + { + "epoch": 0.3032066769163189, + "grad_norm": 1.9746551513671875, + "learning_rate": 1e-06, + "loss": 1.0927, + "mean_token_accuracy": 0.6725797057151794, + "num_tokens": 71335051.0, + "step": 2761 + }, + { + "epoch": 0.30331649461893256, + "grad_norm": 2.2769570350646973, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.690750241279602, + "num_tokens": 71361661.0, + "step": 2762 + }, + { + "epoch": 0.3034263123215462, + "grad_norm": 2.6287975311279297, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7011345624923706, + "num_tokens": 71382572.0, + "step": 2763 + }, + { + "epoch": 0.3035361300241599, + "grad_norm": 2.4575233459472656, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7162296772003174, + "num_tokens": 71403425.0, + "step": 2764 + }, + { + "epoch": 0.30364594772677356, + "grad_norm": 2.2100350856781006, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6952389478683472, + "num_tokens": 71428908.0, + "step": 2765 + }, + { + "epoch": 0.3037557654293872, + "grad_norm": 2.1744606494903564, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7165393829345703, + "num_tokens": 71454772.0, + "step": 2766 + }, + { + "epoch": 0.3038655831320009, + "grad_norm": 2.2838010787963867, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6850720047950745, + "num_tokens": 71482693.0, + "step": 2767 + }, + { + "epoch": 0.30397540083461455, + "grad_norm": 2.4163553714752197, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.715385913848877, + "num_tokens": 71504932.0, + "step": 2768 + }, + { + "epoch": 0.3040852185372282, + "grad_norm": 2.670367479324341, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7078539729118347, + "num_tokens": 71524792.0, + "step": 2769 + }, + { + "epoch": 0.30419503623984184, + "grad_norm": 2.1299359798431396, + "learning_rate": 1e-06, + "loss": 1.1276, + "mean_token_accuracy": 0.6637427806854248, + "num_tokens": 71555938.0, + "step": 2770 + }, + { + "epoch": 0.30430485394245554, + "grad_norm": 2.3501484394073486, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.684821367263794, + "num_tokens": 71582278.0, + "step": 2771 + }, + { + "epoch": 0.3044146716450692, + "grad_norm": 2.2600724697113037, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6947888135910034, + "num_tokens": 71611249.0, + "step": 2772 + }, + { + "epoch": 0.30452448934768284, + "grad_norm": 2.13834547996521, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6958944201469421, + "num_tokens": 71640985.0, + "step": 2773 + }, + { + "epoch": 0.3046343070502965, + "grad_norm": 2.423631191253662, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7042292356491089, + "num_tokens": 71663987.0, + "step": 2774 + }, + { + "epoch": 0.3047441247529102, + "grad_norm": 2.423217535018921, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6817339658737183, + "num_tokens": 71688019.0, + "step": 2775 + }, + { + "epoch": 0.30485394245552383, + "grad_norm": 2.106966495513916, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7148206233978271, + "num_tokens": 71715223.0, + "step": 2776 + }, + { + "epoch": 0.3049637601581375, + "grad_norm": 2.2145490646362305, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6778223514556885, + "num_tokens": 71742935.0, + "step": 2777 + }, + { + "epoch": 0.3050735778607512, + "grad_norm": 2.128706693649292, + "learning_rate": 1e-06, + "loss": 1.1264, + "mean_token_accuracy": 0.6684524416923523, + "num_tokens": 71774364.0, + "step": 2778 + }, + { + "epoch": 0.3051833955633648, + "grad_norm": 1.9839247465133667, + "learning_rate": 1e-06, + "loss": 1.1279, + "mean_token_accuracy": 0.6694551110267639, + "num_tokens": 71808027.0, + "step": 2779 + }, + { + "epoch": 0.30529321326597847, + "grad_norm": 2.367640733718872, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7097848653793335, + "num_tokens": 71830722.0, + "step": 2780 + }, + { + "epoch": 0.3054030309685921, + "grad_norm": 2.3934686183929443, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.7005793452262878, + "num_tokens": 71852928.0, + "step": 2781 + }, + { + "epoch": 0.3055128486712058, + "grad_norm": 2.1254053115844727, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6922355890274048, + "num_tokens": 71882433.0, + "step": 2782 + }, + { + "epoch": 0.30562266637381946, + "grad_norm": 1.962071180343628, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6867620944976807, + "num_tokens": 71914594.0, + "step": 2783 + }, + { + "epoch": 0.3057324840764331, + "grad_norm": 2.3300790786743164, + "learning_rate": 1e-06, + "loss": 1.061, + "mean_token_accuracy": 0.6821752786636353, + "num_tokens": 71938171.0, + "step": 2784 + }, + { + "epoch": 0.30584230177904675, + "grad_norm": 2.1926987171173096, + "learning_rate": 1e-06, + "loss": 1.0847, + "mean_token_accuracy": 0.6897628903388977, + "num_tokens": 71966619.0, + "step": 2785 + }, + { + "epoch": 0.30595211948166046, + "grad_norm": 2.172145128250122, + "learning_rate": 1e-06, + "loss": 1.1087, + "mean_token_accuracy": 0.6680939197540283, + "num_tokens": 71996236.0, + "step": 2786 + }, + { + "epoch": 0.3060619371842741, + "grad_norm": 2.4403510093688965, + "learning_rate": 1e-06, + "loss": 1.088, + "mean_token_accuracy": 0.6764043569564819, + "num_tokens": 72020186.0, + "step": 2787 + }, + { + "epoch": 0.30617175488688775, + "grad_norm": 2.6310269832611084, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.70548015832901, + "num_tokens": 72045722.0, + "step": 2788 + }, + { + "epoch": 0.30628157258950145, + "grad_norm": 2.3110716342926025, + "learning_rate": 1e-06, + "loss": 1.1027, + "mean_token_accuracy": 0.6735934019088745, + "num_tokens": 72072531.0, + "step": 2789 + }, + { + "epoch": 0.3063913902921151, + "grad_norm": 2.208059310913086, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6970080733299255, + "num_tokens": 72098928.0, + "step": 2790 + }, + { + "epoch": 0.30650120799472874, + "grad_norm": 2.391710042953491, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.6873553395271301, + "num_tokens": 72123697.0, + "step": 2791 + }, + { + "epoch": 0.3066110256973424, + "grad_norm": 2.6498947143554688, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7060103416442871, + "num_tokens": 72144223.0, + "step": 2792 + }, + { + "epoch": 0.3067208433999561, + "grad_norm": 2.3573215007781982, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7077444791793823, + "num_tokens": 72167845.0, + "step": 2793 + }, + { + "epoch": 0.30683066110256974, + "grad_norm": 2.3171546459198, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6996273994445801, + "num_tokens": 72190607.0, + "step": 2794 + }, + { + "epoch": 0.3069404788051834, + "grad_norm": 2.2329771518707275, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6910985708236694, + "num_tokens": 72216993.0, + "step": 2795 + }, + { + "epoch": 0.3070502965077971, + "grad_norm": 2.369483470916748, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7061307430267334, + "num_tokens": 72240619.0, + "step": 2796 + }, + { + "epoch": 0.30716011421041073, + "grad_norm": 2.5360889434814453, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6888361573219299, + "num_tokens": 72261970.0, + "step": 2797 + }, + { + "epoch": 0.3072699319130244, + "grad_norm": 2.503462076187134, + "learning_rate": 1e-06, + "loss": 1.0838, + "mean_token_accuracy": 0.6800782680511475, + "num_tokens": 72286977.0, + "step": 2798 + }, + { + "epoch": 0.307379749615638, + "grad_norm": 3.03521728515625, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6909676790237427, + "num_tokens": 72312612.0, + "step": 2799 + }, + { + "epoch": 0.3074895673182517, + "grad_norm": 2.0670948028564453, + "learning_rate": 1e-06, + "loss": 1.0707, + "mean_token_accuracy": 0.685041069984436, + "num_tokens": 72342397.0, + "step": 2800 + }, + { + "epoch": 0.30759938502086537, + "grad_norm": 2.0286617279052734, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6858212351799011, + "num_tokens": 72373431.0, + "step": 2801 + }, + { + "epoch": 0.307709202723479, + "grad_norm": 2.2415096759796143, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6962478160858154, + "num_tokens": 72399898.0, + "step": 2802 + }, + { + "epoch": 0.30781902042609266, + "grad_norm": 2.3296942710876465, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7527475357055664, + "num_tokens": 72421490.0, + "step": 2803 + }, + { + "epoch": 0.30792883812870636, + "grad_norm": 2.4962656497955322, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7182456254959106, + "num_tokens": 72441399.0, + "step": 2804 + }, + { + "epoch": 0.30803865583132, + "grad_norm": 2.5049922466278076, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7205519676208496, + "num_tokens": 72465359.0, + "step": 2805 + }, + { + "epoch": 0.30814847353393365, + "grad_norm": 2.1725199222564697, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7064976692199707, + "num_tokens": 72492832.0, + "step": 2806 + }, + { + "epoch": 0.30825829123654735, + "grad_norm": 2.286738872528076, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6920400857925415, + "num_tokens": 72517740.0, + "step": 2807 + }, + { + "epoch": 0.308368108939161, + "grad_norm": 2.06439208984375, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7184773683547974, + "num_tokens": 72546634.0, + "step": 2808 + }, + { + "epoch": 0.30847792664177465, + "grad_norm": 2.2177376747131348, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.697704553604126, + "num_tokens": 72574884.0, + "step": 2809 + }, + { + "epoch": 0.3085877443443883, + "grad_norm": 2.3182315826416016, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7120234966278076, + "num_tokens": 72599568.0, + "step": 2810 + }, + { + "epoch": 0.308697562047002, + "grad_norm": 2.3967485427856445, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7060660123825073, + "num_tokens": 72623881.0, + "step": 2811 + }, + { + "epoch": 0.30880737974961564, + "grad_norm": 1.9955501556396484, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6964418888092041, + "num_tokens": 72656182.0, + "step": 2812 + }, + { + "epoch": 0.3089171974522293, + "grad_norm": 2.1598432064056396, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7147977352142334, + "num_tokens": 72681467.0, + "step": 2813 + }, + { + "epoch": 0.309027015154843, + "grad_norm": 2.4780070781707764, + "learning_rate": 1e-06, + "loss": 1.0852, + "mean_token_accuracy": 0.6846796274185181, + "num_tokens": 72704525.0, + "step": 2814 + }, + { + "epoch": 0.30913683285745663, + "grad_norm": 2.2765989303588867, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6874451041221619, + "num_tokens": 72729940.0, + "step": 2815 + }, + { + "epoch": 0.3092466505600703, + "grad_norm": 2.3852598667144775, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7088407278060913, + "num_tokens": 72754018.0, + "step": 2816 + }, + { + "epoch": 0.3093564682626839, + "grad_norm": 2.2917428016662598, + "learning_rate": 1e-06, + "loss": 1.0741, + "mean_token_accuracy": 0.6850155591964722, + "num_tokens": 72780723.0, + "step": 2817 + }, + { + "epoch": 0.3094662859652976, + "grad_norm": 2.1806247234344482, + "learning_rate": 1e-06, + "loss": 1.0603, + "mean_token_accuracy": 0.6809555292129517, + "num_tokens": 72809617.0, + "step": 2818 + }, + { + "epoch": 0.3095761036679113, + "grad_norm": 2.5949933528900146, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7361770868301392, + "num_tokens": 72829097.0, + "step": 2819 + }, + { + "epoch": 0.3096859213705249, + "grad_norm": 2.152435064315796, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7048335671424866, + "num_tokens": 72856357.0, + "step": 2820 + }, + { + "epoch": 0.30979573907313857, + "grad_norm": 2.251028299331665, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7078832387924194, + "num_tokens": 72882233.0, + "step": 2821 + }, + { + "epoch": 0.30990555677575227, + "grad_norm": 2.3461854457855225, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7114623785018921, + "num_tokens": 72905153.0, + "step": 2822 + }, + { + "epoch": 0.3100153744783659, + "grad_norm": 2.574882984161377, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.718224823474884, + "num_tokens": 72926296.0, + "step": 2823 + }, + { + "epoch": 0.31012519218097956, + "grad_norm": 2.1161112785339355, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6910467147827148, + "num_tokens": 72955950.0, + "step": 2824 + }, + { + "epoch": 0.31023500988359326, + "grad_norm": 2.3430376052856445, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6819729804992676, + "num_tokens": 72982794.0, + "step": 2825 + }, + { + "epoch": 0.3103448275862069, + "grad_norm": 2.617885112762451, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7004318237304688, + "num_tokens": 73007331.0, + "step": 2826 + }, + { + "epoch": 0.31045464528882055, + "grad_norm": 2.5668628215789795, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6966067552566528, + "num_tokens": 73027103.0, + "step": 2827 + }, + { + "epoch": 0.3105644629914342, + "grad_norm": 2.3952786922454834, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6959429979324341, + "num_tokens": 73049063.0, + "step": 2828 + }, + { + "epoch": 0.3106742806940479, + "grad_norm": 2.027296304702759, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6947770118713379, + "num_tokens": 73078869.0, + "step": 2829 + }, + { + "epoch": 0.31078409839666155, + "grad_norm": 2.340458869934082, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6980283260345459, + "num_tokens": 73105191.0, + "step": 2830 + }, + { + "epoch": 0.3108939160992752, + "grad_norm": 2.214291572570801, + "learning_rate": 1e-06, + "loss": 1.0606, + "mean_token_accuracy": 0.6888599395751953, + "num_tokens": 73131768.0, + "step": 2831 + }, + { + "epoch": 0.31100373380188884, + "grad_norm": 2.2616024017333984, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6992628574371338, + "num_tokens": 73158912.0, + "step": 2832 + }, + { + "epoch": 0.31111355150450254, + "grad_norm": 2.344287157058716, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.701494038105011, + "num_tokens": 73183322.0, + "step": 2833 + }, + { + "epoch": 0.3112233692071162, + "grad_norm": 2.0510294437408447, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.700211763381958, + "num_tokens": 73212338.0, + "step": 2834 + }, + { + "epoch": 0.31133318690972983, + "grad_norm": 2.284680128097534, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.6904090046882629, + "num_tokens": 73241951.0, + "step": 2835 + }, + { + "epoch": 0.31144300461234353, + "grad_norm": 2.1868810653686523, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7179754972457886, + "num_tokens": 73269042.0, + "step": 2836 + }, + { + "epoch": 0.3115528223149572, + "grad_norm": 2.2472667694091797, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7000734806060791, + "num_tokens": 73295104.0, + "step": 2837 + }, + { + "epoch": 0.3116626400175708, + "grad_norm": 2.1607141494750977, + "learning_rate": 1e-06, + "loss": 1.0641, + "mean_token_accuracy": 0.6844555735588074, + "num_tokens": 73322207.0, + "step": 2838 + }, + { + "epoch": 0.31177245772018447, + "grad_norm": 2.440340757369995, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.691348671913147, + "num_tokens": 73344542.0, + "step": 2839 + }, + { + "epoch": 0.31188227542279817, + "grad_norm": 2.5084426403045654, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7039819955825806, + "num_tokens": 73366839.0, + "step": 2840 + }, + { + "epoch": 0.3119920931254118, + "grad_norm": 2.372661828994751, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.697403609752655, + "num_tokens": 73390568.0, + "step": 2841 + }, + { + "epoch": 0.31210191082802546, + "grad_norm": 2.466508388519287, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7102663516998291, + "num_tokens": 73411547.0, + "step": 2842 + }, + { + "epoch": 0.31221172853063917, + "grad_norm": 2.4805243015289307, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6960940957069397, + "num_tokens": 73436034.0, + "step": 2843 + }, + { + "epoch": 0.3123215462332528, + "grad_norm": 2.320453405380249, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6942951679229736, + "num_tokens": 73461319.0, + "step": 2844 + }, + { + "epoch": 0.31243136393586646, + "grad_norm": 2.397242546081543, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6958357691764832, + "num_tokens": 73483607.0, + "step": 2845 + }, + { + "epoch": 0.3125411816384801, + "grad_norm": 2.201517105102539, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6813859939575195, + "num_tokens": 73510148.0, + "step": 2846 + }, + { + "epoch": 0.3126509993410938, + "grad_norm": 2.8790249824523926, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7389792799949646, + "num_tokens": 73527606.0, + "step": 2847 + }, + { + "epoch": 0.31276081704370745, + "grad_norm": 2.476264238357544, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.70220947265625, + "num_tokens": 73548554.0, + "step": 2848 + }, + { + "epoch": 0.3128706347463211, + "grad_norm": 2.0871737003326416, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7248125672340393, + "num_tokens": 73575796.0, + "step": 2849 + }, + { + "epoch": 0.31298045244893474, + "grad_norm": 2.1163666248321533, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6995006799697876, + "num_tokens": 73605313.0, + "step": 2850 + }, + { + "epoch": 0.31309027015154844, + "grad_norm": 2.3376498222351074, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6900674104690552, + "num_tokens": 73630136.0, + "step": 2851 + }, + { + "epoch": 0.3132000878541621, + "grad_norm": 2.527273178100586, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7127583026885986, + "num_tokens": 73650701.0, + "step": 2852 + }, + { + "epoch": 0.31330990555677574, + "grad_norm": 2.6761527061462402, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7183845043182373, + "num_tokens": 73675451.0, + "step": 2853 + }, + { + "epoch": 0.31341972325938944, + "grad_norm": 2.3875255584716797, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6934593915939331, + "num_tokens": 73699631.0, + "step": 2854 + }, + { + "epoch": 0.3135295409620031, + "grad_norm": 2.0533015727996826, + "learning_rate": 1e-06, + "loss": 1.1429, + "mean_token_accuracy": 0.6696893572807312, + "num_tokens": 73734841.0, + "step": 2855 + }, + { + "epoch": 0.31363935866461673, + "grad_norm": 2.0893492698669434, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.6776345372200012, + "num_tokens": 73766927.0, + "step": 2856 + }, + { + "epoch": 0.3137491763672304, + "grad_norm": 2.200838804244995, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6969180107116699, + "num_tokens": 73794957.0, + "step": 2857 + }, + { + "epoch": 0.3138589940698441, + "grad_norm": 2.6132102012634277, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7015772461891174, + "num_tokens": 73814952.0, + "step": 2858 + }, + { + "epoch": 0.3139688117724577, + "grad_norm": 2.1412343978881836, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.69328773021698, + "num_tokens": 73840972.0, + "step": 2859 + }, + { + "epoch": 0.31407862947507137, + "grad_norm": 2.056715726852417, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6817313432693481, + "num_tokens": 73871707.0, + "step": 2860 + }, + { + "epoch": 0.314188447177685, + "grad_norm": 2.3461215496063232, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7062157392501831, + "num_tokens": 73896564.0, + "step": 2861 + }, + { + "epoch": 0.3142982648802987, + "grad_norm": 2.209724187850952, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7044140100479126, + "num_tokens": 73923668.0, + "step": 2862 + }, + { + "epoch": 0.31440808258291236, + "grad_norm": 2.170976400375366, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6876309514045715, + "num_tokens": 73949813.0, + "step": 2863 + }, + { + "epoch": 0.314517900285526, + "grad_norm": 2.490293264389038, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.6887708902359009, + "num_tokens": 73972394.0, + "step": 2864 + }, + { + "epoch": 0.3146277179881397, + "grad_norm": 2.249941349029541, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6966354250907898, + "num_tokens": 73996907.0, + "step": 2865 + }, + { + "epoch": 0.31473753569075336, + "grad_norm": 2.116176128387451, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7253651022911072, + "num_tokens": 74023279.0, + "step": 2866 + }, + { + "epoch": 0.314847353393367, + "grad_norm": 2.0450663566589355, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6992558240890503, + "num_tokens": 74055139.0, + "step": 2867 + }, + { + "epoch": 0.31495717109598065, + "grad_norm": 2.4856064319610596, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7032881379127502, + "num_tokens": 74077381.0, + "step": 2868 + }, + { + "epoch": 0.31506698879859435, + "grad_norm": 2.375068187713623, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.697799563407898, + "num_tokens": 74102507.0, + "step": 2869 + }, + { + "epoch": 0.315176806501208, + "grad_norm": 2.1709325313568115, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6826997399330139, + "num_tokens": 74131402.0, + "step": 2870 + }, + { + "epoch": 0.31528662420382164, + "grad_norm": 2.6731317043304443, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7163121104240417, + "num_tokens": 74150490.0, + "step": 2871 + }, + { + "epoch": 0.31539644190643534, + "grad_norm": 2.214905023574829, + "learning_rate": 1e-06, + "loss": 1.0642, + "mean_token_accuracy": 0.679731011390686, + "num_tokens": 74177232.0, + "step": 2872 + }, + { + "epoch": 0.315506259609049, + "grad_norm": 2.339500665664673, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7170981168746948, + "num_tokens": 74201663.0, + "step": 2873 + }, + { + "epoch": 0.31561607731166264, + "grad_norm": 2.3103909492492676, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7020046710968018, + "num_tokens": 74225721.0, + "step": 2874 + }, + { + "epoch": 0.3157258950142763, + "grad_norm": 2.0705113410949707, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.7013416290283203, + "num_tokens": 74254833.0, + "step": 2875 + }, + { + "epoch": 0.31583571271689, + "grad_norm": 2.2670137882232666, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6984462141990662, + "num_tokens": 74280086.0, + "step": 2876 + }, + { + "epoch": 0.31594553041950363, + "grad_norm": 2.183583974838257, + "learning_rate": 1e-06, + "loss": 1.1228, + "mean_token_accuracy": 0.6803502440452576, + "num_tokens": 74308202.0, + "step": 2877 + }, + { + "epoch": 0.3160553481221173, + "grad_norm": 1.9654664993286133, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6965497136116028, + "num_tokens": 74340124.0, + "step": 2878 + }, + { + "epoch": 0.3161651658247309, + "grad_norm": 2.450181245803833, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.7047119140625, + "num_tokens": 74363775.0, + "step": 2879 + }, + { + "epoch": 0.3162749835273446, + "grad_norm": 2.3993680477142334, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7148064374923706, + "num_tokens": 74385059.0, + "step": 2880 + }, + { + "epoch": 0.31638480122995827, + "grad_norm": 2.026306390762329, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6925089359283447, + "num_tokens": 74415274.0, + "step": 2881 + }, + { + "epoch": 0.3164946189325719, + "grad_norm": 2.7727692127227783, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.697201669216156, + "num_tokens": 74433509.0, + "step": 2882 + }, + { + "epoch": 0.3166044366351856, + "grad_norm": 2.2430083751678467, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.7095212936401367, + "num_tokens": 74456649.0, + "step": 2883 + }, + { + "epoch": 0.31671425433779926, + "grad_norm": 2.5176119804382324, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.6970747709274292, + "num_tokens": 74478535.0, + "step": 2884 + }, + { + "epoch": 0.3168240720404129, + "grad_norm": 2.339869737625122, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.710760772228241, + "num_tokens": 74501778.0, + "step": 2885 + }, + { + "epoch": 0.31693388974302655, + "grad_norm": 2.14589524269104, + "learning_rate": 1e-06, + "loss": 1.0886, + "mean_token_accuracy": 0.6748849153518677, + "num_tokens": 74530204.0, + "step": 2886 + }, + { + "epoch": 0.31704370744564025, + "grad_norm": 2.358659029006958, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7372862696647644, + "num_tokens": 74552726.0, + "step": 2887 + }, + { + "epoch": 0.3171535251482539, + "grad_norm": 2.42226505279541, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.7054919004440308, + "num_tokens": 74574343.0, + "step": 2888 + }, + { + "epoch": 0.31726334285086755, + "grad_norm": 2.3391387462615967, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7069790363311768, + "num_tokens": 74597489.0, + "step": 2889 + }, + { + "epoch": 0.31737316055348125, + "grad_norm": 1.9012235403060913, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6992372870445251, + "num_tokens": 74630000.0, + "step": 2890 + }, + { + "epoch": 0.3174829782560949, + "grad_norm": 2.2773468494415283, + "learning_rate": 1e-06, + "loss": 1.0957, + "mean_token_accuracy": 0.6791690587997437, + "num_tokens": 74656838.0, + "step": 2891 + }, + { + "epoch": 0.31759279595870854, + "grad_norm": 2.26869535446167, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.692357063293457, + "num_tokens": 74680459.0, + "step": 2892 + }, + { + "epoch": 0.3177026136613222, + "grad_norm": 2.08111834526062, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7071111798286438, + "num_tokens": 74709266.0, + "step": 2893 + }, + { + "epoch": 0.3178124313639359, + "grad_norm": 2.0877621173858643, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7170765995979309, + "num_tokens": 74737699.0, + "step": 2894 + }, + { + "epoch": 0.31792224906654953, + "grad_norm": 2.3840553760528564, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.707516074180603, + "num_tokens": 74761079.0, + "step": 2895 + }, + { + "epoch": 0.3180320667691632, + "grad_norm": 2.0219149589538574, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7020126581192017, + "num_tokens": 74790938.0, + "step": 2896 + }, + { + "epoch": 0.3181418844717768, + "grad_norm": 2.0738067626953125, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7079988718032837, + "num_tokens": 74819771.0, + "step": 2897 + }, + { + "epoch": 0.3182517021743905, + "grad_norm": 2.2982468605041504, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6983941793441772, + "num_tokens": 74845767.0, + "step": 2898 + }, + { + "epoch": 0.3183615198770042, + "grad_norm": 2.12863826751709, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7057243585586548, + "num_tokens": 74873917.0, + "step": 2899 + }, + { + "epoch": 0.3184713375796178, + "grad_norm": 2.178619623184204, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.715438961982727, + "num_tokens": 74898683.0, + "step": 2900 + }, + { + "epoch": 0.3185811552822315, + "grad_norm": 2.2349936962127686, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6923836469650269, + "num_tokens": 74923942.0, + "step": 2901 + }, + { + "epoch": 0.31869097298484517, + "grad_norm": 2.290609836578369, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.709144115447998, + "num_tokens": 74950771.0, + "step": 2902 + }, + { + "epoch": 0.3188007906874588, + "grad_norm": 2.4662833213806152, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6952986121177673, + "num_tokens": 74973253.0, + "step": 2903 + }, + { + "epoch": 0.31891060839007246, + "grad_norm": 2.274219512939453, + "learning_rate": 1e-06, + "loss": 1.0784, + "mean_token_accuracy": 0.6797120571136475, + "num_tokens": 74999470.0, + "step": 2904 + }, + { + "epoch": 0.31902042609268616, + "grad_norm": 2.6376540660858154, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6887922286987305, + "num_tokens": 75019191.0, + "step": 2905 + }, + { + "epoch": 0.3191302437952998, + "grad_norm": 2.1019186973571777, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6863033175468445, + "num_tokens": 75047634.0, + "step": 2906 + }, + { + "epoch": 0.31924006149791345, + "grad_norm": 2.1025326251983643, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7055878639221191, + "num_tokens": 75077206.0, + "step": 2907 + }, + { + "epoch": 0.3193498792005271, + "grad_norm": 2.338132858276367, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6995351314544678, + "num_tokens": 75103531.0, + "step": 2908 + }, + { + "epoch": 0.3194596969031408, + "grad_norm": 2.007936954498291, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.713421106338501, + "num_tokens": 75135218.0, + "step": 2909 + }, + { + "epoch": 0.31956951460575445, + "grad_norm": 2.2511210441589355, + "learning_rate": 1e-06, + "loss": 1.072, + "mean_token_accuracy": 0.6723315715789795, + "num_tokens": 75162189.0, + "step": 2910 + }, + { + "epoch": 0.3196793323083681, + "grad_norm": 2.435441732406616, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7269498109817505, + "num_tokens": 75183522.0, + "step": 2911 + }, + { + "epoch": 0.3197891500109818, + "grad_norm": 2.0880374908447266, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7037169337272644, + "num_tokens": 75211860.0, + "step": 2912 + }, + { + "epoch": 0.31989896771359544, + "grad_norm": 1.8262044191360474, + "learning_rate": 1e-06, + "loss": 1.0495, + "mean_token_accuracy": 0.6841772198677063, + "num_tokens": 75250506.0, + "step": 2913 + }, + { + "epoch": 0.3200087854162091, + "grad_norm": 1.947271704673767, + "learning_rate": 1e-06, + "loss": 1.1088, + "mean_token_accuracy": 0.673751711845398, + "num_tokens": 75285783.0, + "step": 2914 + }, + { + "epoch": 0.32011860311882273, + "grad_norm": 1.9888923168182373, + "learning_rate": 1e-06, + "loss": 1.14, + "mean_token_accuracy": 0.6567166447639465, + "num_tokens": 75318059.0, + "step": 2915 + }, + { + "epoch": 0.32022842082143643, + "grad_norm": 2.324803113937378, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6955757737159729, + "num_tokens": 75345039.0, + "step": 2916 + }, + { + "epoch": 0.3203382385240501, + "grad_norm": 2.0824670791625977, + "learning_rate": 1e-06, + "loss": 1.0884, + "mean_token_accuracy": 0.6833487749099731, + "num_tokens": 75374543.0, + "step": 2917 + }, + { + "epoch": 0.3204480562266637, + "grad_norm": 2.4788994789123535, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.7021666765213013, + "num_tokens": 75397039.0, + "step": 2918 + }, + { + "epoch": 0.3205578739292774, + "grad_norm": 2.26521372795105, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6903501152992249, + "num_tokens": 75422381.0, + "step": 2919 + }, + { + "epoch": 0.32066769163189107, + "grad_norm": 2.3919801712036133, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7215748429298401, + "num_tokens": 75446255.0, + "step": 2920 + }, + { + "epoch": 0.3207775093345047, + "grad_norm": 2.39884352684021, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7209700345993042, + "num_tokens": 75468639.0, + "step": 2921 + }, + { + "epoch": 0.32088732703711836, + "grad_norm": 2.2158572673797607, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7214499711990356, + "num_tokens": 75492994.0, + "step": 2922 + }, + { + "epoch": 0.32099714473973207, + "grad_norm": 2.173862934112549, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6910949945449829, + "num_tokens": 75520922.0, + "step": 2923 + }, + { + "epoch": 0.3211069624423457, + "grad_norm": 2.12805438041687, + "learning_rate": 1e-06, + "loss": 1.0679, + "mean_token_accuracy": 0.6887800693511963, + "num_tokens": 75548894.0, + "step": 2924 + }, + { + "epoch": 0.32121678014495936, + "grad_norm": 2.4564120769500732, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6945276856422424, + "num_tokens": 75569737.0, + "step": 2925 + }, + { + "epoch": 0.321326597847573, + "grad_norm": 2.1772427558898926, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7005420327186584, + "num_tokens": 75594651.0, + "step": 2926 + }, + { + "epoch": 0.3214364155501867, + "grad_norm": 2.1838817596435547, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6989789009094238, + "num_tokens": 75620646.0, + "step": 2927 + }, + { + "epoch": 0.32154623325280035, + "grad_norm": 2.396801710128784, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7148655652999878, + "num_tokens": 75642082.0, + "step": 2928 + }, + { + "epoch": 0.321656050955414, + "grad_norm": 2.0994787216186523, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.689552366733551, + "num_tokens": 75671471.0, + "step": 2929 + }, + { + "epoch": 0.3217658686580277, + "grad_norm": 2.2250823974609375, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6870473623275757, + "num_tokens": 75698969.0, + "step": 2930 + }, + { + "epoch": 0.32187568636064134, + "grad_norm": 2.120962381362915, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7259553670883179, + "num_tokens": 75726030.0, + "step": 2931 + }, + { + "epoch": 0.321985504063255, + "grad_norm": 2.329983949661255, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.711445689201355, + "num_tokens": 75749806.0, + "step": 2932 + }, + { + "epoch": 0.32209532176586864, + "grad_norm": 2.469494342803955, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7023214101791382, + "num_tokens": 75774356.0, + "step": 2933 + }, + { + "epoch": 0.32220513946848234, + "grad_norm": 2.24924898147583, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6865265369415283, + "num_tokens": 75801477.0, + "step": 2934 + }, + { + "epoch": 0.322314957171096, + "grad_norm": 2.252578020095825, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7119007110595703, + "num_tokens": 75826278.0, + "step": 2935 + }, + { + "epoch": 0.32242477487370963, + "grad_norm": 2.2000527381896973, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7000764608383179, + "num_tokens": 75853384.0, + "step": 2936 + }, + { + "epoch": 0.3225345925763233, + "grad_norm": 2.3762781620025635, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7162686586380005, + "num_tokens": 75875598.0, + "step": 2937 + }, + { + "epoch": 0.322644410278937, + "grad_norm": 2.2082138061523438, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6999216079711914, + "num_tokens": 75901014.0, + "step": 2938 + }, + { + "epoch": 0.3227542279815506, + "grad_norm": 2.2194323539733887, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7064104080200195, + "num_tokens": 75927081.0, + "step": 2939 + }, + { + "epoch": 0.32286404568416427, + "grad_norm": 2.4856598377227783, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7376431226730347, + "num_tokens": 75947874.0, + "step": 2940 + }, + { + "epoch": 0.32297386338677797, + "grad_norm": 2.218071222305298, + "learning_rate": 1e-06, + "loss": 1.0864, + "mean_token_accuracy": 0.6779295206069946, + "num_tokens": 75974473.0, + "step": 2941 + }, + { + "epoch": 0.3230836810893916, + "grad_norm": 2.426743507385254, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7044892907142639, + "num_tokens": 75997171.0, + "step": 2942 + }, + { + "epoch": 0.32319349879200526, + "grad_norm": 2.432302951812744, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7046432495117188, + "num_tokens": 76020257.0, + "step": 2943 + }, + { + "epoch": 0.3233033164946189, + "grad_norm": 2.381195068359375, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7050540447235107, + "num_tokens": 76042908.0, + "step": 2944 + }, + { + "epoch": 0.3234131341972326, + "grad_norm": 2.1227707862854004, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6990813612937927, + "num_tokens": 76071996.0, + "step": 2945 + }, + { + "epoch": 0.32352295189984626, + "grad_norm": 2.2512705326080322, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7235026359558105, + "num_tokens": 76095474.0, + "step": 2946 + }, + { + "epoch": 0.3236327696024599, + "grad_norm": 2.3811614513397217, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.711423397064209, + "num_tokens": 76119182.0, + "step": 2947 + }, + { + "epoch": 0.3237425873050736, + "grad_norm": 2.319498062133789, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7143780589103699, + "num_tokens": 76141518.0, + "step": 2948 + }, + { + "epoch": 0.32385240500768725, + "grad_norm": 2.155513286590576, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6985107064247131, + "num_tokens": 76166749.0, + "step": 2949 + }, + { + "epoch": 0.3239622227103009, + "grad_norm": 2.139237403869629, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7209344506263733, + "num_tokens": 76194031.0, + "step": 2950 + }, + { + "epoch": 0.32407204041291454, + "grad_norm": 2.3538818359375, + "learning_rate": 1e-06, + "loss": 1.1273, + "mean_token_accuracy": 0.6728742122650146, + "num_tokens": 76220946.0, + "step": 2951 + }, + { + "epoch": 0.32418185811552824, + "grad_norm": 2.188746690750122, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.7004429697990417, + "num_tokens": 76247852.0, + "step": 2952 + }, + { + "epoch": 0.3242916758181419, + "grad_norm": 2.687541961669922, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7464184761047363, + "num_tokens": 76265952.0, + "step": 2953 + }, + { + "epoch": 0.32440149352075554, + "grad_norm": 2.080868721008301, + "learning_rate": 1e-06, + "loss": 1.0655, + "mean_token_accuracy": 0.6880608201026917, + "num_tokens": 76295479.0, + "step": 2954 + }, + { + "epoch": 0.3245113112233692, + "grad_norm": 2.421152114868164, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7133134007453918, + "num_tokens": 76319832.0, + "step": 2955 + }, + { + "epoch": 0.3246211289259829, + "grad_norm": 2.2516000270843506, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6994572281837463, + "num_tokens": 76346144.0, + "step": 2956 + }, + { + "epoch": 0.32473094662859653, + "grad_norm": 2.2567296028137207, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7137391567230225, + "num_tokens": 76373994.0, + "step": 2957 + }, + { + "epoch": 0.3248407643312102, + "grad_norm": 2.141228199005127, + "learning_rate": 1e-06, + "loss": 1.1143, + "mean_token_accuracy": 0.6684913635253906, + "num_tokens": 76404158.0, + "step": 2958 + }, + { + "epoch": 0.3249505820338239, + "grad_norm": 1.8933594226837158, + "learning_rate": 1e-06, + "loss": 1.0589, + "mean_token_accuracy": 0.6902651786804199, + "num_tokens": 76440125.0, + "step": 2959 + }, + { + "epoch": 0.3250603997364375, + "grad_norm": 2.4139792919158936, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6833874583244324, + "num_tokens": 76464590.0, + "step": 2960 + }, + { + "epoch": 0.32517021743905117, + "grad_norm": 2.33268666267395, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6859805583953857, + "num_tokens": 76488566.0, + "step": 2961 + }, + { + "epoch": 0.3252800351416648, + "grad_norm": 2.1062779426574707, + "learning_rate": 1e-06, + "loss": 1.1039, + "mean_token_accuracy": 0.6823970079421997, + "num_tokens": 76517572.0, + "step": 2962 + }, + { + "epoch": 0.3253898528442785, + "grad_norm": 2.4149680137634277, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6910901665687561, + "num_tokens": 76539776.0, + "step": 2963 + }, + { + "epoch": 0.32549967054689216, + "grad_norm": 2.1565561294555664, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.720868706703186, + "num_tokens": 76566605.0, + "step": 2964 + }, + { + "epoch": 0.3256094882495058, + "grad_norm": 2.2122368812561035, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.711363673210144, + "num_tokens": 76591486.0, + "step": 2965 + }, + { + "epoch": 0.3257193059521195, + "grad_norm": 2.476783037185669, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7051964998245239, + "num_tokens": 76611866.0, + "step": 2966 + }, + { + "epoch": 0.32582912365473315, + "grad_norm": 2.2285521030426025, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6988598704338074, + "num_tokens": 76637126.0, + "step": 2967 + }, + { + "epoch": 0.3259389413573468, + "grad_norm": 2.4744699001312256, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6995570659637451, + "num_tokens": 76659032.0, + "step": 2968 + }, + { + "epoch": 0.32604875905996045, + "grad_norm": 2.5150258541107178, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7033417820930481, + "num_tokens": 76681382.0, + "step": 2969 + }, + { + "epoch": 0.32615857676257415, + "grad_norm": 2.46466326713562, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7020208835601807, + "num_tokens": 76703872.0, + "step": 2970 + }, + { + "epoch": 0.3262683944651878, + "grad_norm": 2.1173009872436523, + "learning_rate": 1e-06, + "loss": 1.1139, + "mean_token_accuracy": 0.6681435108184814, + "num_tokens": 76732208.0, + "step": 2971 + }, + { + "epoch": 0.32637821216780144, + "grad_norm": 2.070439338684082, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6892127394676208, + "num_tokens": 76761485.0, + "step": 2972 + }, + { + "epoch": 0.3264880298704151, + "grad_norm": 2.0949838161468506, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6991931200027466, + "num_tokens": 76790799.0, + "step": 2973 + }, + { + "epoch": 0.3265978475730288, + "grad_norm": 2.3650271892547607, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6938376426696777, + "num_tokens": 76814140.0, + "step": 2974 + }, + { + "epoch": 0.32670766527564243, + "grad_norm": 2.3988635540008545, + "learning_rate": 1e-06, + "loss": 1.0494, + "mean_token_accuracy": 0.6924896836280823, + "num_tokens": 76836439.0, + "step": 2975 + }, + { + "epoch": 0.3268174829782561, + "grad_norm": 2.444016218185425, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7072182893753052, + "num_tokens": 76857722.0, + "step": 2976 + }, + { + "epoch": 0.3269273006808698, + "grad_norm": 2.3487133979797363, + "learning_rate": 1e-06, + "loss": 1.1364, + "mean_token_accuracy": 0.6658016443252563, + "num_tokens": 76883370.0, + "step": 2977 + }, + { + "epoch": 0.3270371183834834, + "grad_norm": 2.3743247985839844, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6778336763381958, + "num_tokens": 76907397.0, + "step": 2978 + }, + { + "epoch": 0.3271469360860971, + "grad_norm": 2.2997639179229736, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6966186165809631, + "num_tokens": 76931225.0, + "step": 2979 + }, + { + "epoch": 0.3272567537887107, + "grad_norm": 2.2849059104919434, + "learning_rate": 1e-06, + "loss": 1.0869, + "mean_token_accuracy": 0.6708048582077026, + "num_tokens": 76956296.0, + "step": 2980 + }, + { + "epoch": 0.3273665714913244, + "grad_norm": 2.223268985748291, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7152222394943237, + "num_tokens": 76979668.0, + "step": 2981 + }, + { + "epoch": 0.32747638919393807, + "grad_norm": 2.7469825744628906, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7012408971786499, + "num_tokens": 76999757.0, + "step": 2982 + }, + { + "epoch": 0.3275862068965517, + "grad_norm": 1.8809641599655151, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7115830779075623, + "num_tokens": 77031664.0, + "step": 2983 + }, + { + "epoch": 0.32769602459916536, + "grad_norm": 2.6412534713745117, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6994761228561401, + "num_tokens": 77050012.0, + "step": 2984 + }, + { + "epoch": 0.32780584230177906, + "grad_norm": 2.4514529705047607, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.6964367628097534, + "num_tokens": 77071442.0, + "step": 2985 + }, + { + "epoch": 0.3279156600043927, + "grad_norm": 2.2862253189086914, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.691569447517395, + "num_tokens": 77098835.0, + "step": 2986 + }, + { + "epoch": 0.32802547770700635, + "grad_norm": 2.1683285236358643, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7119444608688354, + "num_tokens": 77125012.0, + "step": 2987 + }, + { + "epoch": 0.32813529540962005, + "grad_norm": 2.32991361618042, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7264134883880615, + "num_tokens": 77148158.0, + "step": 2988 + }, + { + "epoch": 0.3282451131122337, + "grad_norm": 2.2095017433166504, + "learning_rate": 1e-06, + "loss": 1.0635, + "mean_token_accuracy": 0.681878924369812, + "num_tokens": 77175883.0, + "step": 2989 + }, + { + "epoch": 0.32835493081484735, + "grad_norm": 2.5820398330688477, + "learning_rate": 1e-06, + "loss": 1.1079, + "mean_token_accuracy": 0.6711025238037109, + "num_tokens": 77200164.0, + "step": 2990 + }, + { + "epoch": 0.328464748517461, + "grad_norm": 2.2539024353027344, + "learning_rate": 1e-06, + "loss": 1.0904, + "mean_token_accuracy": 0.6787406802177429, + "num_tokens": 77226914.0, + "step": 2991 + }, + { + "epoch": 0.3285745662200747, + "grad_norm": 2.2157397270202637, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6961168050765991, + "num_tokens": 77253262.0, + "step": 2992 + }, + { + "epoch": 0.32868438392268834, + "grad_norm": 2.3109729290008545, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.730559766292572, + "num_tokens": 77280602.0, + "step": 2993 + }, + { + "epoch": 0.328794201625302, + "grad_norm": 2.1259660720825195, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7021744251251221, + "num_tokens": 77308632.0, + "step": 2994 + }, + { + "epoch": 0.3289040193279157, + "grad_norm": 2.538783311843872, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6951965093612671, + "num_tokens": 77331211.0, + "step": 2995 + }, + { + "epoch": 0.32901383703052933, + "grad_norm": 2.1220672130584717, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7165347337722778, + "num_tokens": 77358266.0, + "step": 2996 + }, + { + "epoch": 0.329123654733143, + "grad_norm": 2.338277816772461, + "learning_rate": 1e-06, + "loss": 1.087, + "mean_token_accuracy": 0.6883184909820557, + "num_tokens": 77382026.0, + "step": 2997 + }, + { + "epoch": 0.3292334724357566, + "grad_norm": 1.9720203876495361, + "learning_rate": 1e-06, + "loss": 1.1228, + "mean_token_accuracy": 0.6727042198181152, + "num_tokens": 77420022.0, + "step": 2998 + }, + { + "epoch": 0.3293432901383703, + "grad_norm": 2.027132034301758, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6910063028335571, + "num_tokens": 77451856.0, + "step": 2999 + }, + { + "epoch": 0.32945310784098397, + "grad_norm": 2.0824129581451416, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6842415928840637, + "num_tokens": 77481883.0, + "step": 3000 + }, + { + "epoch": 0.3295629255435976, + "grad_norm": 2.6229135990142822, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7183070182800293, + "num_tokens": 77501260.0, + "step": 3001 + }, + { + "epoch": 0.32967274324621126, + "grad_norm": 2.4445440769195557, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7054955363273621, + "num_tokens": 77522736.0, + "step": 3002 + }, + { + "epoch": 0.32978256094882497, + "grad_norm": 2.098680257797241, + "learning_rate": 1e-06, + "loss": 1.126, + "mean_token_accuracy": 0.6612862348556519, + "num_tokens": 77552652.0, + "step": 3003 + }, + { + "epoch": 0.3298923786514386, + "grad_norm": 2.189702272415161, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7252151966094971, + "num_tokens": 77577360.0, + "step": 3004 + }, + { + "epoch": 0.33000219635405226, + "grad_norm": 2.045902729034424, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7095977067947388, + "num_tokens": 77606752.0, + "step": 3005 + }, + { + "epoch": 0.33011201405666596, + "grad_norm": 2.3722939491271973, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7104741334915161, + "num_tokens": 77628709.0, + "step": 3006 + }, + { + "epoch": 0.3302218317592796, + "grad_norm": 2.1376900672912598, + "learning_rate": 1e-06, + "loss": 1.0554, + "mean_token_accuracy": 0.6809626221656799, + "num_tokens": 77656252.0, + "step": 3007 + }, + { + "epoch": 0.33033164946189325, + "grad_norm": 2.286168336868286, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.7016791701316833, + "num_tokens": 77681961.0, + "step": 3008 + }, + { + "epoch": 0.3304414671645069, + "grad_norm": 2.270369052886963, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7020524740219116, + "num_tokens": 77709226.0, + "step": 3009 + }, + { + "epoch": 0.3305512848671206, + "grad_norm": 2.6285653114318848, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7179715037345886, + "num_tokens": 77726517.0, + "step": 3010 + }, + { + "epoch": 0.33066110256973424, + "grad_norm": 2.2395522594451904, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.7008282542228699, + "num_tokens": 77752521.0, + "step": 3011 + }, + { + "epoch": 0.3307709202723479, + "grad_norm": 2.1813700199127197, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6973781585693359, + "num_tokens": 77779017.0, + "step": 3012 + }, + { + "epoch": 0.33088073797496154, + "grad_norm": 2.322791814804077, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.678852915763855, + "num_tokens": 77803179.0, + "step": 3013 + }, + { + "epoch": 0.33099055567757524, + "grad_norm": 2.107468366622925, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7156240940093994, + "num_tokens": 77830529.0, + "step": 3014 + }, + { + "epoch": 0.3311003733801889, + "grad_norm": 2.3463070392608643, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.698091983795166, + "num_tokens": 77857041.0, + "step": 3015 + }, + { + "epoch": 0.33121019108280253, + "grad_norm": 2.4453394412994385, + "learning_rate": 1e-06, + "loss": 1.0795, + "mean_token_accuracy": 0.6821399927139282, + "num_tokens": 77881178.0, + "step": 3016 + }, + { + "epoch": 0.33132000878541623, + "grad_norm": 2.183170795440674, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6825385689735413, + "num_tokens": 77910440.0, + "step": 3017 + }, + { + "epoch": 0.3314298264880299, + "grad_norm": 2.079617977142334, + "learning_rate": 1e-06, + "loss": 1.0839, + "mean_token_accuracy": 0.6763923764228821, + "num_tokens": 77941588.0, + "step": 3018 + }, + { + "epoch": 0.3315396441906435, + "grad_norm": 2.4840121269226074, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7134196162223816, + "num_tokens": 77962746.0, + "step": 3019 + }, + { + "epoch": 0.33164946189325717, + "grad_norm": 2.4800667762756348, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6889059543609619, + "num_tokens": 77985228.0, + "step": 3020 + }, + { + "epoch": 0.33175927959587087, + "grad_norm": 2.138540744781494, + "learning_rate": 1e-06, + "loss": 1.0785, + "mean_token_accuracy": 0.6759001016616821, + "num_tokens": 78015834.0, + "step": 3021 + }, + { + "epoch": 0.3318690972984845, + "grad_norm": 2.532095432281494, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.7040449380874634, + "num_tokens": 78038463.0, + "step": 3022 + }, + { + "epoch": 0.33197891500109816, + "grad_norm": 2.067568063735962, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6963317394256592, + "num_tokens": 78068067.0, + "step": 3023 + }, + { + "epoch": 0.33208873270371186, + "grad_norm": 2.1284728050231934, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.699042558670044, + "num_tokens": 78099716.0, + "step": 3024 + }, + { + "epoch": 0.3321985504063255, + "grad_norm": 2.180168390274048, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7293545007705688, + "num_tokens": 78123803.0, + "step": 3025 + }, + { + "epoch": 0.33230836810893916, + "grad_norm": 2.3273425102233887, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7092466950416565, + "num_tokens": 78148900.0, + "step": 3026 + }, + { + "epoch": 0.3324181858115528, + "grad_norm": 2.5723884105682373, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7043830156326294, + "num_tokens": 78172082.0, + "step": 3027 + }, + { + "epoch": 0.3325280035141665, + "grad_norm": 2.284304618835449, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7153723239898682, + "num_tokens": 78197140.0, + "step": 3028 + }, + { + "epoch": 0.33263782121678015, + "grad_norm": 2.024603843688965, + "learning_rate": 1e-06, + "loss": 1.0774, + "mean_token_accuracy": 0.6775999665260315, + "num_tokens": 78229344.0, + "step": 3029 + }, + { + "epoch": 0.3327476389193938, + "grad_norm": 2.121798515319824, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6953766345977783, + "num_tokens": 78257350.0, + "step": 3030 + }, + { + "epoch": 0.33285745662200744, + "grad_norm": 2.030778646469116, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7104856967926025, + "num_tokens": 78286663.0, + "step": 3031 + }, + { + "epoch": 0.33296727432462114, + "grad_norm": 2.0579729080200195, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6877710819244385, + "num_tokens": 78318532.0, + "step": 3032 + }, + { + "epoch": 0.3330770920272348, + "grad_norm": 2.284701108932495, + "learning_rate": 1e-06, + "loss": 1.093, + "mean_token_accuracy": 0.6735143661499023, + "num_tokens": 78342676.0, + "step": 3033 + }, + { + "epoch": 0.33318690972984844, + "grad_norm": 2.3481295108795166, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7043293714523315, + "num_tokens": 78364760.0, + "step": 3034 + }, + { + "epoch": 0.33329672743246214, + "grad_norm": 2.4403464794158936, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6967402696609497, + "num_tokens": 78388601.0, + "step": 3035 + }, + { + "epoch": 0.3334065451350758, + "grad_norm": 2.2979984283447266, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6962724924087524, + "num_tokens": 78415549.0, + "step": 3036 + }, + { + "epoch": 0.33351636283768943, + "grad_norm": 2.2022321224212646, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7042747735977173, + "num_tokens": 78442685.0, + "step": 3037 + }, + { + "epoch": 0.3336261805403031, + "grad_norm": 2.255260467529297, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6969360113143921, + "num_tokens": 78469472.0, + "step": 3038 + }, + { + "epoch": 0.3337359982429168, + "grad_norm": 2.0364410877227783, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7122349739074707, + "num_tokens": 78501465.0, + "step": 3039 + }, + { + "epoch": 0.3338458159455304, + "grad_norm": 2.2000575065612793, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7177809476852417, + "num_tokens": 78525782.0, + "step": 3040 + }, + { + "epoch": 0.33395563364814407, + "grad_norm": 2.255323648452759, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.699783205986023, + "num_tokens": 78550247.0, + "step": 3041 + }, + { + "epoch": 0.33406545135075777, + "grad_norm": 2.1245617866516113, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7056915760040283, + "num_tokens": 78576229.0, + "step": 3042 + }, + { + "epoch": 0.3341752690533714, + "grad_norm": 2.099001884460449, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6969267725944519, + "num_tokens": 78603794.0, + "step": 3043 + }, + { + "epoch": 0.33428508675598506, + "grad_norm": 2.4879796504974365, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.703511118888855, + "num_tokens": 78624317.0, + "step": 3044 + }, + { + "epoch": 0.3343949044585987, + "grad_norm": 2.2519540786743164, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.682349681854248, + "num_tokens": 78649693.0, + "step": 3045 + }, + { + "epoch": 0.3345047221612124, + "grad_norm": 2.072659492492676, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7278920412063599, + "num_tokens": 78676792.0, + "step": 3046 + }, + { + "epoch": 0.33461453986382605, + "grad_norm": 2.390838861465454, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7138082385063171, + "num_tokens": 78698812.0, + "step": 3047 + }, + { + "epoch": 0.3347243575664397, + "grad_norm": 2.422772169113159, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6992180347442627, + "num_tokens": 78723823.0, + "step": 3048 + }, + { + "epoch": 0.33483417526905335, + "grad_norm": 2.1286540031433105, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6893064975738525, + "num_tokens": 78751359.0, + "step": 3049 + }, + { + "epoch": 0.33494399297166705, + "grad_norm": 2.1491880416870117, + "learning_rate": 1e-06, + "loss": 1.1003, + "mean_token_accuracy": 0.6702858209609985, + "num_tokens": 78778325.0, + "step": 3050 + }, + { + "epoch": 0.3350538106742807, + "grad_norm": 2.4489686489105225, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.721811056137085, + "num_tokens": 78800096.0, + "step": 3051 + }, + { + "epoch": 0.33516362837689434, + "grad_norm": 2.6581437587738037, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7040542364120483, + "num_tokens": 78820036.0, + "step": 3052 + }, + { + "epoch": 0.33527344607950804, + "grad_norm": 2.2387495040893555, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6969193816184998, + "num_tokens": 78845369.0, + "step": 3053 + }, + { + "epoch": 0.3353832637821217, + "grad_norm": 1.9971299171447754, + "learning_rate": 1e-06, + "loss": 1.0941, + "mean_token_accuracy": 0.6748728156089783, + "num_tokens": 78878373.0, + "step": 3054 + }, + { + "epoch": 0.33549308148473533, + "grad_norm": 2.3840763568878174, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6957359313964844, + "num_tokens": 78900627.0, + "step": 3055 + }, + { + "epoch": 0.335602899187349, + "grad_norm": 1.9393892288208008, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6935455799102783, + "num_tokens": 78932358.0, + "step": 3056 + }, + { + "epoch": 0.3357127168899627, + "grad_norm": 2.353628158569336, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.7018864154815674, + "num_tokens": 78954929.0, + "step": 3057 + }, + { + "epoch": 0.3358225345925763, + "grad_norm": 2.1058154106140137, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7138864398002625, + "num_tokens": 78984236.0, + "step": 3058 + }, + { + "epoch": 0.33593235229519, + "grad_norm": 2.1397178173065186, + "learning_rate": 1e-06, + "loss": 1.042, + "mean_token_accuracy": 0.6959008574485779, + "num_tokens": 79013737.0, + "step": 3059 + }, + { + "epoch": 0.3360421699978036, + "grad_norm": 1.938673734664917, + "learning_rate": 1e-06, + "loss": 1.0697, + "mean_token_accuracy": 0.6795358657836914, + "num_tokens": 79047783.0, + "step": 3060 + }, + { + "epoch": 0.3361519877004173, + "grad_norm": 2.525111198425293, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6999320983886719, + "num_tokens": 79069512.0, + "step": 3061 + }, + { + "epoch": 0.33626180540303097, + "grad_norm": 2.0645911693573, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7239184379577637, + "num_tokens": 79098817.0, + "step": 3062 + }, + { + "epoch": 0.3363716231056446, + "grad_norm": 2.5598397254943848, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6867634057998657, + "num_tokens": 79119523.0, + "step": 3063 + }, + { + "epoch": 0.3364814408082583, + "grad_norm": 2.153447389602661, + "learning_rate": 1e-06, + "loss": 1.0916, + "mean_token_accuracy": 0.6762416362762451, + "num_tokens": 79148617.0, + "step": 3064 + }, + { + "epoch": 0.33659125851087196, + "grad_norm": 2.205281972885132, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.704389214515686, + "num_tokens": 79174100.0, + "step": 3065 + }, + { + "epoch": 0.3367010762134856, + "grad_norm": 2.1747090816497803, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7102673649787903, + "num_tokens": 79200804.0, + "step": 3066 + }, + { + "epoch": 0.33681089391609925, + "grad_norm": 2.362337589263916, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7046030163764954, + "num_tokens": 79223988.0, + "step": 3067 + }, + { + "epoch": 0.33692071161871295, + "grad_norm": 2.328676223754883, + "learning_rate": 1e-06, + "loss": 1.0822, + "mean_token_accuracy": 0.6784589886665344, + "num_tokens": 79248404.0, + "step": 3068 + }, + { + "epoch": 0.3370305293213266, + "grad_norm": 2.072249174118042, + "learning_rate": 1e-06, + "loss": 1.0808, + "mean_token_accuracy": 0.6857913136482239, + "num_tokens": 79277072.0, + "step": 3069 + }, + { + "epoch": 0.33714034702394025, + "grad_norm": 2.1636369228363037, + "learning_rate": 1e-06, + "loss": 1.0977, + "mean_token_accuracy": 0.6725550889968872, + "num_tokens": 79304757.0, + "step": 3070 + }, + { + "epoch": 0.33725016472655395, + "grad_norm": 1.9347187280654907, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6892539858818054, + "num_tokens": 79339350.0, + "step": 3071 + }, + { + "epoch": 0.3373599824291676, + "grad_norm": 2.1147427558898926, + "learning_rate": 1e-06, + "loss": 1.0499, + "mean_token_accuracy": 0.6825212836265564, + "num_tokens": 79369818.0, + "step": 3072 + }, + { + "epoch": 0.33746980013178124, + "grad_norm": 2.0969276428222656, + "learning_rate": 1e-06, + "loss": 1.0865, + "mean_token_accuracy": 0.6818598508834839, + "num_tokens": 79401881.0, + "step": 3073 + }, + { + "epoch": 0.3375796178343949, + "grad_norm": 2.498089075088501, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7043799757957458, + "num_tokens": 79420457.0, + "step": 3074 + }, + { + "epoch": 0.3376894355370086, + "grad_norm": 2.123119354248047, + "learning_rate": 1e-06, + "loss": 1.0409, + "mean_token_accuracy": 0.7003256678581238, + "num_tokens": 79450330.0, + "step": 3075 + }, + { + "epoch": 0.33779925323962223, + "grad_norm": 2.2017931938171387, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6976743936538696, + "num_tokens": 79477626.0, + "step": 3076 + }, + { + "epoch": 0.3379090709422359, + "grad_norm": 2.303701400756836, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6955447196960449, + "num_tokens": 79500703.0, + "step": 3077 + }, + { + "epoch": 0.3380188886448495, + "grad_norm": 2.2426469326019287, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6851875185966492, + "num_tokens": 79526708.0, + "step": 3078 + }, + { + "epoch": 0.3381287063474632, + "grad_norm": 1.8960222005844116, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.701757550239563, + "num_tokens": 79561806.0, + "step": 3079 + }, + { + "epoch": 0.33823852405007687, + "grad_norm": 2.2376444339752197, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7063292264938354, + "num_tokens": 79587764.0, + "step": 3080 + }, + { + "epoch": 0.3383483417526905, + "grad_norm": 2.158233880996704, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7117651700973511, + "num_tokens": 79616144.0, + "step": 3081 + }, + { + "epoch": 0.3384581594553042, + "grad_norm": 2.3615925312042236, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7079466581344604, + "num_tokens": 79638796.0, + "step": 3082 + }, + { + "epoch": 0.33856797715791787, + "grad_norm": 2.2623291015625, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7193810939788818, + "num_tokens": 79662566.0, + "step": 3083 + }, + { + "epoch": 0.3386777948605315, + "grad_norm": 2.217526435852051, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7055915594100952, + "num_tokens": 79691734.0, + "step": 3084 + }, + { + "epoch": 0.33878761256314516, + "grad_norm": 2.0853688716888428, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.7003160119056702, + "num_tokens": 79720234.0, + "step": 3085 + }, + { + "epoch": 0.33889743026575886, + "grad_norm": 2.2569265365600586, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6916005611419678, + "num_tokens": 79744274.0, + "step": 3086 + }, + { + "epoch": 0.3390072479683725, + "grad_norm": 2.144928455352783, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6964778900146484, + "num_tokens": 79771274.0, + "step": 3087 + }, + { + "epoch": 0.33911706567098615, + "grad_norm": 2.2778337001800537, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6838922500610352, + "num_tokens": 79797808.0, + "step": 3088 + }, + { + "epoch": 0.3392268833735998, + "grad_norm": 2.380842924118042, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7049227952957153, + "num_tokens": 79818490.0, + "step": 3089 + }, + { + "epoch": 0.3393367010762135, + "grad_norm": 2.264909267425537, + "learning_rate": 1e-06, + "loss": 1.0645, + "mean_token_accuracy": 0.6793058514595032, + "num_tokens": 79846912.0, + "step": 3090 + }, + { + "epoch": 0.33944651877882714, + "grad_norm": 2.9074959754943848, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7184174060821533, + "num_tokens": 79862416.0, + "step": 3091 + }, + { + "epoch": 0.3395563364814408, + "grad_norm": 2.2728259563446045, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.687519371509552, + "num_tokens": 79888186.0, + "step": 3092 + }, + { + "epoch": 0.3396661541840545, + "grad_norm": 2.209343433380127, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6927525997161865, + "num_tokens": 79914312.0, + "step": 3093 + }, + { + "epoch": 0.33977597188666814, + "grad_norm": 2.3604674339294434, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7084294557571411, + "num_tokens": 79938231.0, + "step": 3094 + }, + { + "epoch": 0.3398857895892818, + "grad_norm": 2.0512807369232178, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7163206934928894, + "num_tokens": 79965742.0, + "step": 3095 + }, + { + "epoch": 0.33999560729189543, + "grad_norm": 1.9633893966674805, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7000283598899841, + "num_tokens": 80000024.0, + "step": 3096 + }, + { + "epoch": 0.34010542499450913, + "grad_norm": 1.9773838520050049, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6982870697975159, + "num_tokens": 80032041.0, + "step": 3097 + }, + { + "epoch": 0.3402152426971228, + "grad_norm": 2.2041068077087402, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6860973238945007, + "num_tokens": 80057633.0, + "step": 3098 + }, + { + "epoch": 0.3403250603997364, + "grad_norm": 2.2360188961029053, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.696930468082428, + "num_tokens": 80083175.0, + "step": 3099 + }, + { + "epoch": 0.3404348781023501, + "grad_norm": 2.098942756652832, + "learning_rate": 1e-06, + "loss": 1.0851, + "mean_token_accuracy": 0.6802912354469299, + "num_tokens": 80112222.0, + "step": 3100 + }, + { + "epoch": 0.34054469580496377, + "grad_norm": 2.7381772994995117, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7063320875167847, + "num_tokens": 80129502.0, + "step": 3101 + }, + { + "epoch": 0.3406545135075774, + "grad_norm": 2.430572509765625, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7144830822944641, + "num_tokens": 80152719.0, + "step": 3102 + }, + { + "epoch": 0.34076433121019106, + "grad_norm": 2.1674017906188965, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7128955721855164, + "num_tokens": 80177076.0, + "step": 3103 + }, + { + "epoch": 0.34087414891280476, + "grad_norm": 2.140265464782715, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6974880695343018, + "num_tokens": 80204022.0, + "step": 3104 + }, + { + "epoch": 0.3409839666154184, + "grad_norm": 2.1867706775665283, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7121832370758057, + "num_tokens": 80229727.0, + "step": 3105 + }, + { + "epoch": 0.34109378431803206, + "grad_norm": 2.2865774631500244, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7197115421295166, + "num_tokens": 80251886.0, + "step": 3106 + }, + { + "epoch": 0.3412036020206457, + "grad_norm": 1.9898580312728882, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6820816993713379, + "num_tokens": 80283712.0, + "step": 3107 + }, + { + "epoch": 0.3413134197232594, + "grad_norm": 2.222456693649292, + "learning_rate": 1e-06, + "loss": 1.081, + "mean_token_accuracy": 0.6835242509841919, + "num_tokens": 80311758.0, + "step": 3108 + }, + { + "epoch": 0.34142323742587305, + "grad_norm": 2.4175565242767334, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6959311962127686, + "num_tokens": 80333423.0, + "step": 3109 + }, + { + "epoch": 0.3415330551284867, + "grad_norm": 2.3315577507019043, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6830146312713623, + "num_tokens": 80358027.0, + "step": 3110 + }, + { + "epoch": 0.3416428728311004, + "grad_norm": 2.1463429927825928, + "learning_rate": 1e-06, + "loss": 1.0665, + "mean_token_accuracy": 0.6981658935546875, + "num_tokens": 80385582.0, + "step": 3111 + }, + { + "epoch": 0.34175269053371404, + "grad_norm": 2.1691272258758545, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7161579132080078, + "num_tokens": 80415533.0, + "step": 3112 + }, + { + "epoch": 0.3418625082363277, + "grad_norm": 2.2877721786499023, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7166838645935059, + "num_tokens": 80439223.0, + "step": 3113 + }, + { + "epoch": 0.34197232593894134, + "grad_norm": 2.4852871894836426, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7024582624435425, + "num_tokens": 80462023.0, + "step": 3114 + }, + { + "epoch": 0.34208214364155504, + "grad_norm": 2.4631574153900146, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6956118941307068, + "num_tokens": 80484919.0, + "step": 3115 + }, + { + "epoch": 0.3421919613441687, + "grad_norm": 2.219287872314453, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7238947153091431, + "num_tokens": 80508170.0, + "step": 3116 + }, + { + "epoch": 0.34230177904678233, + "grad_norm": 2.154585123062134, + "learning_rate": 1e-06, + "loss": 1.0404, + "mean_token_accuracy": 0.6866679191589355, + "num_tokens": 80539222.0, + "step": 3117 + }, + { + "epoch": 0.34241159674939603, + "grad_norm": 2.2859385013580322, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.711068868637085, + "num_tokens": 80562836.0, + "step": 3118 + }, + { + "epoch": 0.3425214144520097, + "grad_norm": 2.1923491954803467, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.701377272605896, + "num_tokens": 80588438.0, + "step": 3119 + }, + { + "epoch": 0.3426312321546233, + "grad_norm": 2.239814519882202, + "learning_rate": 1e-06, + "loss": 1.0806, + "mean_token_accuracy": 0.6802494525909424, + "num_tokens": 80615939.0, + "step": 3120 + }, + { + "epoch": 0.34274104985723697, + "grad_norm": 2.194626569747925, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6908087730407715, + "num_tokens": 80642335.0, + "step": 3121 + }, + { + "epoch": 0.34285086755985067, + "grad_norm": 2.349241018295288, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6935924291610718, + "num_tokens": 80665781.0, + "step": 3122 + }, + { + "epoch": 0.3429606852624643, + "grad_norm": 2.025794267654419, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6982409358024597, + "num_tokens": 80696047.0, + "step": 3123 + }, + { + "epoch": 0.34307050296507796, + "grad_norm": 2.4836373329162598, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.705386757850647, + "num_tokens": 80719122.0, + "step": 3124 + }, + { + "epoch": 0.3431803206676916, + "grad_norm": 2.0825982093811035, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6990143060684204, + "num_tokens": 80747066.0, + "step": 3125 + }, + { + "epoch": 0.3432901383703053, + "grad_norm": 2.122567892074585, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.7025371789932251, + "num_tokens": 80775529.0, + "step": 3126 + }, + { + "epoch": 0.34339995607291895, + "grad_norm": 2.059997797012329, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6885445713996887, + "num_tokens": 80806422.0, + "step": 3127 + }, + { + "epoch": 0.3435097737755326, + "grad_norm": 2.235698699951172, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7009830474853516, + "num_tokens": 80829206.0, + "step": 3128 + }, + { + "epoch": 0.3436195914781463, + "grad_norm": 2.0186259746551514, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6907292604446411, + "num_tokens": 80859251.0, + "step": 3129 + }, + { + "epoch": 0.34372940918075995, + "grad_norm": 2.5110630989074707, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.707222580909729, + "num_tokens": 80879585.0, + "step": 3130 + }, + { + "epoch": 0.3438392268833736, + "grad_norm": 2.7509865760803223, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7164404392242432, + "num_tokens": 80896459.0, + "step": 3131 + }, + { + "epoch": 0.34394904458598724, + "grad_norm": 2.518198251724243, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7081202268600464, + "num_tokens": 80916349.0, + "step": 3132 + }, + { + "epoch": 0.34405886228860094, + "grad_norm": 2.4274392127990723, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7129653692245483, + "num_tokens": 80940274.0, + "step": 3133 + }, + { + "epoch": 0.3441686799912146, + "grad_norm": 2.1948001384735107, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6962416172027588, + "num_tokens": 80967458.0, + "step": 3134 + }, + { + "epoch": 0.34427849769382823, + "grad_norm": 2.60101580619812, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7098292708396912, + "num_tokens": 80987311.0, + "step": 3135 + }, + { + "epoch": 0.3443883153964419, + "grad_norm": 2.5063421726226807, + "learning_rate": 1e-06, + "loss": 1.1238, + "mean_token_accuracy": 0.6647402048110962, + "num_tokens": 81009634.0, + "step": 3136 + }, + { + "epoch": 0.3444981330990556, + "grad_norm": 2.5467514991760254, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7043356895446777, + "num_tokens": 81029324.0, + "step": 3137 + }, + { + "epoch": 0.3446079508016692, + "grad_norm": 2.0516443252563477, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6884233951568604, + "num_tokens": 81060117.0, + "step": 3138 + }, + { + "epoch": 0.3447177685042829, + "grad_norm": 2.3907268047332764, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7054236531257629, + "num_tokens": 81084163.0, + "step": 3139 + }, + { + "epoch": 0.3448275862068966, + "grad_norm": 2.1276962757110596, + "learning_rate": 1e-06, + "loss": 1.0651, + "mean_token_accuracy": 0.6837841272354126, + "num_tokens": 81115321.0, + "step": 3140 + }, + { + "epoch": 0.3449374039095102, + "grad_norm": 2.62249493598938, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7060313820838928, + "num_tokens": 81136307.0, + "step": 3141 + }, + { + "epoch": 0.34504722161212387, + "grad_norm": 2.162533760070801, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6885571479797363, + "num_tokens": 81163479.0, + "step": 3142 + }, + { + "epoch": 0.3451570393147375, + "grad_norm": 2.371734380722046, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6871792078018188, + "num_tokens": 81188105.0, + "step": 3143 + }, + { + "epoch": 0.3452668570173512, + "grad_norm": 2.0826199054718018, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6993240118026733, + "num_tokens": 81217529.0, + "step": 3144 + }, + { + "epoch": 0.34537667471996486, + "grad_norm": 2.0743601322174072, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7072267532348633, + "num_tokens": 81245442.0, + "step": 3145 + }, + { + "epoch": 0.3454864924225785, + "grad_norm": 2.1323354244232178, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7126692533493042, + "num_tokens": 81273374.0, + "step": 3146 + }, + { + "epoch": 0.3455963101251922, + "grad_norm": 2.227078676223755, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.705491840839386, + "num_tokens": 81299150.0, + "step": 3147 + }, + { + "epoch": 0.34570612782780585, + "grad_norm": 2.35914945602417, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.6840807795524597, + "num_tokens": 81323114.0, + "step": 3148 + }, + { + "epoch": 0.3458159455304195, + "grad_norm": 2.5501153469085693, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7202956676483154, + "num_tokens": 81343642.0, + "step": 3149 + }, + { + "epoch": 0.34592576323303315, + "grad_norm": 2.165602207183838, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7144187092781067, + "num_tokens": 81371186.0, + "step": 3150 + }, + { + "epoch": 0.34603558093564685, + "grad_norm": 2.1963517665863037, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7006353735923767, + "num_tokens": 81398073.0, + "step": 3151 + }, + { + "epoch": 0.3461453986382605, + "grad_norm": 2.0889179706573486, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7052386999130249, + "num_tokens": 81426255.0, + "step": 3152 + }, + { + "epoch": 0.34625521634087414, + "grad_norm": 2.2229325771331787, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.693641185760498, + "num_tokens": 81452329.0, + "step": 3153 + }, + { + "epoch": 0.3463650340434878, + "grad_norm": 2.2364988327026367, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7263468503952026, + "num_tokens": 81475750.0, + "step": 3154 + }, + { + "epoch": 0.3464748517461015, + "grad_norm": 2.548447847366333, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6987727880477905, + "num_tokens": 81499098.0, + "step": 3155 + }, + { + "epoch": 0.34658466944871513, + "grad_norm": 2.172499179840088, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.686233401298523, + "num_tokens": 81525567.0, + "step": 3156 + }, + { + "epoch": 0.3466944871513288, + "grad_norm": 2.2687649726867676, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.6728339195251465, + "num_tokens": 81552757.0, + "step": 3157 + }, + { + "epoch": 0.3468043048539425, + "grad_norm": 2.226559638977051, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6908591985702515, + "num_tokens": 81579060.0, + "step": 3158 + }, + { + "epoch": 0.3469141225565561, + "grad_norm": 2.1826364994049072, + "learning_rate": 1e-06, + "loss": 1.0553, + "mean_token_accuracy": 0.6845325231552124, + "num_tokens": 81608542.0, + "step": 3159 + }, + { + "epoch": 0.34702394025916977, + "grad_norm": 1.9755669832229614, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7027075290679932, + "num_tokens": 81639879.0, + "step": 3160 + }, + { + "epoch": 0.3471337579617834, + "grad_norm": 2.1925034523010254, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7143052220344543, + "num_tokens": 81664867.0, + "step": 3161 + }, + { + "epoch": 0.3472435756643971, + "grad_norm": 2.3732056617736816, + "learning_rate": 1e-06, + "loss": 1.077, + "mean_token_accuracy": 0.6818896532058716, + "num_tokens": 81691172.0, + "step": 3162 + }, + { + "epoch": 0.34735339336701077, + "grad_norm": 1.9615397453308105, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.7070543766021729, + "num_tokens": 81722600.0, + "step": 3163 + }, + { + "epoch": 0.3474632110696244, + "grad_norm": 2.9475724697113037, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7236332893371582, + "num_tokens": 81737797.0, + "step": 3164 + }, + { + "epoch": 0.34757302877223806, + "grad_norm": 2.5217373371124268, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.711997389793396, + "num_tokens": 81757397.0, + "step": 3165 + }, + { + "epoch": 0.34768284647485176, + "grad_norm": 2.373213529586792, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7185298204421997, + "num_tokens": 81778433.0, + "step": 3166 + }, + { + "epoch": 0.3477926641774654, + "grad_norm": 2.0669636726379395, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6931374073028564, + "num_tokens": 81808813.0, + "step": 3167 + }, + { + "epoch": 0.34790248188007905, + "grad_norm": 2.2936160564422607, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6831448674201965, + "num_tokens": 81835433.0, + "step": 3168 + }, + { + "epoch": 0.34801229958269275, + "grad_norm": 2.552480697631836, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7079712152481079, + "num_tokens": 81855762.0, + "step": 3169 + }, + { + "epoch": 0.3481221172853064, + "grad_norm": 2.3468210697174072, + "learning_rate": 1e-06, + "loss": 1.0866, + "mean_token_accuracy": 0.6732620596885681, + "num_tokens": 81880430.0, + "step": 3170 + }, + { + "epoch": 0.34823193498792004, + "grad_norm": 2.3540761470794678, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7150944471359253, + "num_tokens": 81903500.0, + "step": 3171 + }, + { + "epoch": 0.3483417526905337, + "grad_norm": 2.0961453914642334, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7126027345657349, + "num_tokens": 81935057.0, + "step": 3172 + }, + { + "epoch": 0.3484515703931474, + "grad_norm": 2.459256887435913, + "learning_rate": 1e-06, + "loss": 1.0692, + "mean_token_accuracy": 0.6759637594223022, + "num_tokens": 81959076.0, + "step": 3173 + }, + { + "epoch": 0.34856138809576104, + "grad_norm": 2.465921640396118, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7093615531921387, + "num_tokens": 81981974.0, + "step": 3174 + }, + { + "epoch": 0.3486712057983747, + "grad_norm": 1.9356497526168823, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.6866433620452881, + "num_tokens": 82015685.0, + "step": 3175 + }, + { + "epoch": 0.3487810235009884, + "grad_norm": 2.4780569076538086, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7074451446533203, + "num_tokens": 82037034.0, + "step": 3176 + }, + { + "epoch": 0.34889084120360203, + "grad_norm": 2.3559372425079346, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7088446617126465, + "num_tokens": 82060036.0, + "step": 3177 + }, + { + "epoch": 0.3490006589062157, + "grad_norm": 2.310220956802368, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6982858180999756, + "num_tokens": 82086924.0, + "step": 3178 + }, + { + "epoch": 0.3491104766088293, + "grad_norm": 2.4127490520477295, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7070664763450623, + "num_tokens": 82110692.0, + "step": 3179 + }, + { + "epoch": 0.349220294311443, + "grad_norm": 2.5010597705841064, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.6905829906463623, + "num_tokens": 82133124.0, + "step": 3180 + }, + { + "epoch": 0.34933011201405667, + "grad_norm": 2.124746322631836, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7345318794250488, + "num_tokens": 82160041.0, + "step": 3181 + }, + { + "epoch": 0.3494399297166703, + "grad_norm": 2.4356377124786377, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7047765851020813, + "num_tokens": 82183128.0, + "step": 3182 + }, + { + "epoch": 0.34954974741928396, + "grad_norm": 2.352447986602783, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7041435241699219, + "num_tokens": 82206128.0, + "step": 3183 + }, + { + "epoch": 0.34965956512189766, + "grad_norm": 2.4540584087371826, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7047394514083862, + "num_tokens": 82227187.0, + "step": 3184 + }, + { + "epoch": 0.3497693828245113, + "grad_norm": 2.3268041610717773, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7103843092918396, + "num_tokens": 82250941.0, + "step": 3185 + }, + { + "epoch": 0.34987920052712496, + "grad_norm": 2.1249446868896484, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6930930614471436, + "num_tokens": 82279220.0, + "step": 3186 + }, + { + "epoch": 0.34998901822973866, + "grad_norm": 2.134416103363037, + "learning_rate": 1e-06, + "loss": 1.0466, + "mean_token_accuracy": 0.688525915145874, + "num_tokens": 82308794.0, + "step": 3187 + }, + { + "epoch": 0.3500988359323523, + "grad_norm": 2.1674129962921143, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.7042349576950073, + "num_tokens": 82333167.0, + "step": 3188 + }, + { + "epoch": 0.35020865363496595, + "grad_norm": 2.40385103225708, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7159888744354248, + "num_tokens": 82356761.0, + "step": 3189 + }, + { + "epoch": 0.3503184713375796, + "grad_norm": 2.4006245136260986, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.704897403717041, + "num_tokens": 82377746.0, + "step": 3190 + }, + { + "epoch": 0.3504282890401933, + "grad_norm": 2.2213237285614014, + "learning_rate": 1e-06, + "loss": 1.118, + "mean_token_accuracy": 0.670147180557251, + "num_tokens": 82404722.0, + "step": 3191 + }, + { + "epoch": 0.35053810674280694, + "grad_norm": 2.1977009773254395, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.705649733543396, + "num_tokens": 82431008.0, + "step": 3192 + }, + { + "epoch": 0.3506479244454206, + "grad_norm": 2.2003962993621826, + "learning_rate": 1e-06, + "loss": 1.0466, + "mean_token_accuracy": 0.6835319399833679, + "num_tokens": 82459447.0, + "step": 3193 + }, + { + "epoch": 0.3507577421480343, + "grad_norm": 2.165105104446411, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6902946829795837, + "num_tokens": 82485581.0, + "step": 3194 + }, + { + "epoch": 0.35086755985064794, + "grad_norm": 2.0809502601623535, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7042417526245117, + "num_tokens": 82512694.0, + "step": 3195 + }, + { + "epoch": 0.3509773775532616, + "grad_norm": 2.2352406978607178, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.716319739818573, + "num_tokens": 82537247.0, + "step": 3196 + }, + { + "epoch": 0.35108719525587523, + "grad_norm": 2.423868179321289, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7176817655563354, + "num_tokens": 82558570.0, + "step": 3197 + }, + { + "epoch": 0.35119701295848893, + "grad_norm": 2.090611457824707, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6947519183158875, + "num_tokens": 82587299.0, + "step": 3198 + }, + { + "epoch": 0.3513068306611026, + "grad_norm": 1.955155849456787, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6935479640960693, + "num_tokens": 82619644.0, + "step": 3199 + }, + { + "epoch": 0.3514166483637162, + "grad_norm": 2.3845365047454834, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7079277038574219, + "num_tokens": 82644335.0, + "step": 3200 + }, + { + "epoch": 0.35152646606632987, + "grad_norm": 2.0195963382720947, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6932345032691956, + "num_tokens": 82674506.0, + "step": 3201 + }, + { + "epoch": 0.35163628376894357, + "grad_norm": 2.1480767726898193, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.68881756067276, + "num_tokens": 82702472.0, + "step": 3202 + }, + { + "epoch": 0.3517461014715572, + "grad_norm": 2.1030354499816895, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6804242134094238, + "num_tokens": 82730656.0, + "step": 3203 + }, + { + "epoch": 0.35185591917417086, + "grad_norm": 2.0651931762695312, + "learning_rate": 1e-06, + "loss": 1.0898, + "mean_token_accuracy": 0.67392498254776, + "num_tokens": 82761816.0, + "step": 3204 + }, + { + "epoch": 0.35196573687678456, + "grad_norm": 2.2537665367126465, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7124799489974976, + "num_tokens": 82785818.0, + "step": 3205 + }, + { + "epoch": 0.3520755545793982, + "grad_norm": 2.1501007080078125, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7033542990684509, + "num_tokens": 82813666.0, + "step": 3206 + }, + { + "epoch": 0.35218537228201185, + "grad_norm": 2.24294376373291, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.7056417465209961, + "num_tokens": 82840296.0, + "step": 3207 + }, + { + "epoch": 0.3522951899846255, + "grad_norm": 2.8406646251678467, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7048428058624268, + "num_tokens": 82857720.0, + "step": 3208 + }, + { + "epoch": 0.3524050076872392, + "grad_norm": 2.3855397701263428, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6846311688423157, + "num_tokens": 82882989.0, + "step": 3209 + }, + { + "epoch": 0.35251482538985285, + "grad_norm": 2.1706366539001465, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7257144451141357, + "num_tokens": 82908373.0, + "step": 3210 + }, + { + "epoch": 0.3526246430924665, + "grad_norm": 1.9521323442459106, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7044033408164978, + "num_tokens": 82939675.0, + "step": 3211 + }, + { + "epoch": 0.35273446079508014, + "grad_norm": 2.445702075958252, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.7069679498672485, + "num_tokens": 82964508.0, + "step": 3212 + }, + { + "epoch": 0.35284427849769384, + "grad_norm": 2.3364949226379395, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7228312492370605, + "num_tokens": 82986542.0, + "step": 3213 + }, + { + "epoch": 0.3529540962003075, + "grad_norm": 2.2970924377441406, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.7041052579879761, + "num_tokens": 83010692.0, + "step": 3214 + }, + { + "epoch": 0.35306391390292113, + "grad_norm": 2.180535316467285, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7180328965187073, + "num_tokens": 83037047.0, + "step": 3215 + }, + { + "epoch": 0.35317373160553484, + "grad_norm": 2.149299144744873, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7003516554832458, + "num_tokens": 83063122.0, + "step": 3216 + }, + { + "epoch": 0.3532835493081485, + "grad_norm": 2.1369287967681885, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6972562670707703, + "num_tokens": 83091684.0, + "step": 3217 + }, + { + "epoch": 0.3533933670107621, + "grad_norm": 2.1102242469787598, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6959885954856873, + "num_tokens": 83119244.0, + "step": 3218 + }, + { + "epoch": 0.3535031847133758, + "grad_norm": 2.4282174110412598, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.7110227346420288, + "num_tokens": 83141800.0, + "step": 3219 + }, + { + "epoch": 0.3536130024159895, + "grad_norm": 2.1466829776763916, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7003023624420166, + "num_tokens": 83170081.0, + "step": 3220 + }, + { + "epoch": 0.3537228201186031, + "grad_norm": 2.5085105895996094, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7021868228912354, + "num_tokens": 83190689.0, + "step": 3221 + }, + { + "epoch": 0.35383263782121677, + "grad_norm": 2.088087558746338, + "learning_rate": 1e-06, + "loss": 1.0757, + "mean_token_accuracy": 0.6767551898956299, + "num_tokens": 83223907.0, + "step": 3222 + }, + { + "epoch": 0.35394245552383047, + "grad_norm": 2.5068788528442383, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6997190713882446, + "num_tokens": 83245462.0, + "step": 3223 + }, + { + "epoch": 0.3540522732264441, + "grad_norm": 2.467400550842285, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7071214914321899, + "num_tokens": 83265293.0, + "step": 3224 + }, + { + "epoch": 0.35416209092905776, + "grad_norm": 2.2569501399993896, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7125564217567444, + "num_tokens": 83291068.0, + "step": 3225 + }, + { + "epoch": 0.3542719086316714, + "grad_norm": 2.23321533203125, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6917891502380371, + "num_tokens": 83318510.0, + "step": 3226 + }, + { + "epoch": 0.3543817263342851, + "grad_norm": 1.8764371871948242, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6797689199447632, + "num_tokens": 83353576.0, + "step": 3227 + }, + { + "epoch": 0.35449154403689875, + "grad_norm": 2.389758825302124, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6942360401153564, + "num_tokens": 83376074.0, + "step": 3228 + }, + { + "epoch": 0.3546013617395124, + "grad_norm": 2.1200220584869385, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7009314298629761, + "num_tokens": 83401544.0, + "step": 3229 + }, + { + "epoch": 0.35471117944212605, + "grad_norm": 1.945949673652649, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6967397928237915, + "num_tokens": 83433777.0, + "step": 3230 + }, + { + "epoch": 0.35482099714473975, + "grad_norm": 2.1931746006011963, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6917572617530823, + "num_tokens": 83459923.0, + "step": 3231 + }, + { + "epoch": 0.3549308148473534, + "grad_norm": 2.2078731060028076, + "learning_rate": 1e-06, + "loss": 1.0599, + "mean_token_accuracy": 0.6777613759040833, + "num_tokens": 83487301.0, + "step": 3232 + }, + { + "epoch": 0.35504063254996704, + "grad_norm": 2.5072684288024902, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7113038301467896, + "num_tokens": 83507683.0, + "step": 3233 + }, + { + "epoch": 0.35515045025258074, + "grad_norm": 2.2690601348876953, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6838583946228027, + "num_tokens": 83533092.0, + "step": 3234 + }, + { + "epoch": 0.3552602679551944, + "grad_norm": 2.146190881729126, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6871073246002197, + "num_tokens": 83562221.0, + "step": 3235 + }, + { + "epoch": 0.35537008565780803, + "grad_norm": 2.215914249420166, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7053350806236267, + "num_tokens": 83587927.0, + "step": 3236 + }, + { + "epoch": 0.3554799033604217, + "grad_norm": 1.944357991218567, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6953542232513428, + "num_tokens": 83619553.0, + "step": 3237 + }, + { + "epoch": 0.3555897210630354, + "grad_norm": 1.926609754562378, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7242616415023804, + "num_tokens": 83652859.0, + "step": 3238 + }, + { + "epoch": 0.355699538765649, + "grad_norm": 2.1462924480438232, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6931299567222595, + "num_tokens": 83679314.0, + "step": 3239 + }, + { + "epoch": 0.35580935646826267, + "grad_norm": 2.2057628631591797, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6982043981552124, + "num_tokens": 83704588.0, + "step": 3240 + }, + { + "epoch": 0.3559191741708763, + "grad_norm": 2.031981945037842, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7032662630081177, + "num_tokens": 83732073.0, + "step": 3241 + }, + { + "epoch": 0.35602899187349, + "grad_norm": 2.071467638015747, + "learning_rate": 1e-06, + "loss": 1.0591, + "mean_token_accuracy": 0.6813948154449463, + "num_tokens": 83761233.0, + "step": 3242 + }, + { + "epoch": 0.35613880957610367, + "grad_norm": 2.232316017150879, + "learning_rate": 1e-06, + "loss": 1.0648, + "mean_token_accuracy": 0.6756671667098999, + "num_tokens": 83789309.0, + "step": 3243 + }, + { + "epoch": 0.3562486272787173, + "grad_norm": 2.081914186477661, + "learning_rate": 1e-06, + "loss": 1.0773, + "mean_token_accuracy": 0.6841238141059875, + "num_tokens": 83822253.0, + "step": 3244 + }, + { + "epoch": 0.356358444981331, + "grad_norm": 2.103947162628174, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7060599327087402, + "num_tokens": 83850718.0, + "step": 3245 + }, + { + "epoch": 0.35646826268394466, + "grad_norm": 2.282409906387329, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.697910726070404, + "num_tokens": 83874658.0, + "step": 3246 + }, + { + "epoch": 0.3565780803865583, + "grad_norm": 2.271681785583496, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7221094965934753, + "num_tokens": 83899291.0, + "step": 3247 + }, + { + "epoch": 0.35668789808917195, + "grad_norm": 2.3285961151123047, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7207983732223511, + "num_tokens": 83922590.0, + "step": 3248 + }, + { + "epoch": 0.35679771579178565, + "grad_norm": 2.0890440940856934, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6999472379684448, + "num_tokens": 83950687.0, + "step": 3249 + }, + { + "epoch": 0.3569075334943993, + "grad_norm": 2.154392719268799, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.695846438407898, + "num_tokens": 83977564.0, + "step": 3250 + }, + { + "epoch": 0.35701735119701294, + "grad_norm": 2.3680968284606934, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6825435757637024, + "num_tokens": 84003184.0, + "step": 3251 + }, + { + "epoch": 0.35712716889962665, + "grad_norm": 2.029148817062378, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.7010875940322876, + "num_tokens": 84031308.0, + "step": 3252 + }, + { + "epoch": 0.3572369866022403, + "grad_norm": 2.2364351749420166, + "learning_rate": 1e-06, + "loss": 1.0629, + "mean_token_accuracy": 0.676332950592041, + "num_tokens": 84057005.0, + "step": 3253 + }, + { + "epoch": 0.35734680430485394, + "grad_norm": 2.2482500076293945, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6872901916503906, + "num_tokens": 84084931.0, + "step": 3254 + }, + { + "epoch": 0.3574566220074676, + "grad_norm": 2.384514331817627, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6841747760772705, + "num_tokens": 84108380.0, + "step": 3255 + }, + { + "epoch": 0.3575664397100813, + "grad_norm": 2.4514615535736084, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7222016453742981, + "num_tokens": 84130087.0, + "step": 3256 + }, + { + "epoch": 0.35767625741269493, + "grad_norm": 2.2137210369110107, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7063616514205933, + "num_tokens": 84154892.0, + "step": 3257 + }, + { + "epoch": 0.3577860751153086, + "grad_norm": 2.4465560913085938, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7092547416687012, + "num_tokens": 84176496.0, + "step": 3258 + }, + { + "epoch": 0.3578958928179222, + "grad_norm": 2.3929150104522705, + "learning_rate": 1e-06, + "loss": 1.0415, + "mean_token_accuracy": 0.6872941851615906, + "num_tokens": 84200797.0, + "step": 3259 + }, + { + "epoch": 0.3580057105205359, + "grad_norm": 2.2553272247314453, + "learning_rate": 1e-06, + "loss": 1.0879, + "mean_token_accuracy": 0.688791036605835, + "num_tokens": 84226155.0, + "step": 3260 + }, + { + "epoch": 0.35811552822314957, + "grad_norm": 2.3221850395202637, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7193241119384766, + "num_tokens": 84249673.0, + "step": 3261 + }, + { + "epoch": 0.3582253459257632, + "grad_norm": 2.377760648727417, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.7005267143249512, + "num_tokens": 84273724.0, + "step": 3262 + }, + { + "epoch": 0.3583351636283769, + "grad_norm": 2.3504934310913086, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6924147605895996, + "num_tokens": 84299628.0, + "step": 3263 + }, + { + "epoch": 0.35844498133099056, + "grad_norm": 2.2743046283721924, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7125413417816162, + "num_tokens": 84327527.0, + "step": 3264 + }, + { + "epoch": 0.3585547990336042, + "grad_norm": 2.2649500370025635, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6909389495849609, + "num_tokens": 84354889.0, + "step": 3265 + }, + { + "epoch": 0.35866461673621786, + "grad_norm": 2.2813212871551514, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6980130076408386, + "num_tokens": 84380522.0, + "step": 3266 + }, + { + "epoch": 0.35877443443883156, + "grad_norm": 2.273224353790283, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7123720645904541, + "num_tokens": 84405266.0, + "step": 3267 + }, + { + "epoch": 0.3588842521414452, + "grad_norm": 2.3082334995269775, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7189358472824097, + "num_tokens": 84428101.0, + "step": 3268 + }, + { + "epoch": 0.35899406984405885, + "grad_norm": 2.2353665828704834, + "learning_rate": 1e-06, + "loss": 1.0799, + "mean_token_accuracy": 0.6785385608673096, + "num_tokens": 84454829.0, + "step": 3269 + }, + { + "epoch": 0.35910388754667255, + "grad_norm": 2.3597524166107178, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7246158719062805, + "num_tokens": 84475796.0, + "step": 3270 + }, + { + "epoch": 0.3592137052492862, + "grad_norm": 1.9882222414016724, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6750792860984802, + "num_tokens": 84506712.0, + "step": 3271 + }, + { + "epoch": 0.35932352295189984, + "grad_norm": 2.5794830322265625, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7072175741195679, + "num_tokens": 84527613.0, + "step": 3272 + }, + { + "epoch": 0.3594333406545135, + "grad_norm": 2.2672643661499023, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6958171725273132, + "num_tokens": 84552379.0, + "step": 3273 + }, + { + "epoch": 0.3595431583571272, + "grad_norm": 1.9694784879684448, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6803798675537109, + "num_tokens": 84583957.0, + "step": 3274 + }, + { + "epoch": 0.35965297605974084, + "grad_norm": 2.1055691242218018, + "learning_rate": 1e-06, + "loss": 1.08, + "mean_token_accuracy": 0.6728924512863159, + "num_tokens": 84616061.0, + "step": 3275 + }, + { + "epoch": 0.3597627937623545, + "grad_norm": 2.3716397285461426, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6944897174835205, + "num_tokens": 84639966.0, + "step": 3276 + }, + { + "epoch": 0.35987261146496813, + "grad_norm": 2.015211582183838, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.7007525563240051, + "num_tokens": 84670805.0, + "step": 3277 + }, + { + "epoch": 0.35998242916758183, + "grad_norm": 2.3292458057403564, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6971184611320496, + "num_tokens": 84695492.0, + "step": 3278 + }, + { + "epoch": 0.3600922468701955, + "grad_norm": 2.205960988998413, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7252385020256042, + "num_tokens": 84721170.0, + "step": 3279 + }, + { + "epoch": 0.3602020645728091, + "grad_norm": 2.2344470024108887, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6932923793792725, + "num_tokens": 84746589.0, + "step": 3280 + }, + { + "epoch": 0.3603118822754228, + "grad_norm": 2.2583107948303223, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6966372132301331, + "num_tokens": 84770710.0, + "step": 3281 + }, + { + "epoch": 0.36042169997803647, + "grad_norm": 1.8953628540039062, + "learning_rate": 1e-06, + "loss": 1.1258, + "mean_token_accuracy": 0.665616512298584, + "num_tokens": 84806101.0, + "step": 3282 + }, + { + "epoch": 0.3605315176806501, + "grad_norm": 2.092935085296631, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7106930017471313, + "num_tokens": 84835710.0, + "step": 3283 + }, + { + "epoch": 0.36064133538326376, + "grad_norm": 2.265566349029541, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6935133934020996, + "num_tokens": 84863843.0, + "step": 3284 + }, + { + "epoch": 0.36075115308587746, + "grad_norm": 2.367544412612915, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.687884509563446, + "num_tokens": 84886682.0, + "step": 3285 + }, + { + "epoch": 0.3608609707884911, + "grad_norm": 2.551464319229126, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6955804824829102, + "num_tokens": 84907655.0, + "step": 3286 + }, + { + "epoch": 0.36097078849110475, + "grad_norm": 2.217040538787842, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6938531398773193, + "num_tokens": 84936194.0, + "step": 3287 + }, + { + "epoch": 0.3610806061937184, + "grad_norm": 2.0502641201019287, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6867678165435791, + "num_tokens": 84964512.0, + "step": 3288 + }, + { + "epoch": 0.3611904238963321, + "grad_norm": 2.198617696762085, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.698491096496582, + "num_tokens": 84991502.0, + "step": 3289 + }, + { + "epoch": 0.36130024159894575, + "grad_norm": 2.4446523189544678, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6886889338493347, + "num_tokens": 85013129.0, + "step": 3290 + }, + { + "epoch": 0.3614100593015594, + "grad_norm": 2.324418783187866, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7164101004600525, + "num_tokens": 85036620.0, + "step": 3291 + }, + { + "epoch": 0.3615198770041731, + "grad_norm": 2.3721935749053955, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6937062740325928, + "num_tokens": 85060171.0, + "step": 3292 + }, + { + "epoch": 0.36162969470678674, + "grad_norm": 2.1720099449157715, + "learning_rate": 1e-06, + "loss": 1.0415, + "mean_token_accuracy": 0.6864640116691589, + "num_tokens": 85086553.0, + "step": 3293 + }, + { + "epoch": 0.3617395124094004, + "grad_norm": 2.078345775604248, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6935330629348755, + "num_tokens": 85118680.0, + "step": 3294 + }, + { + "epoch": 0.36184933011201403, + "grad_norm": 2.3065223693847656, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7129853963851929, + "num_tokens": 85142298.0, + "step": 3295 + }, + { + "epoch": 0.36195914781462774, + "grad_norm": 2.066176176071167, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7113521695137024, + "num_tokens": 85170484.0, + "step": 3296 + }, + { + "epoch": 0.3620689655172414, + "grad_norm": 2.196538209915161, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7206929922103882, + "num_tokens": 85196017.0, + "step": 3297 + }, + { + "epoch": 0.362178783219855, + "grad_norm": 2.393195867538452, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7273392677307129, + "num_tokens": 85216845.0, + "step": 3298 + }, + { + "epoch": 0.36228860092246873, + "grad_norm": 2.1160213947296143, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7124605774879456, + "num_tokens": 85243824.0, + "step": 3299 + }, + { + "epoch": 0.3623984186250824, + "grad_norm": 2.1114115715026855, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.7054271697998047, + "num_tokens": 85274198.0, + "step": 3300 + }, + { + "epoch": 0.362508236327696, + "grad_norm": 2.1604530811309814, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6987314224243164, + "num_tokens": 85303239.0, + "step": 3301 + }, + { + "epoch": 0.36261805403030967, + "grad_norm": 2.0519442558288574, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7086690068244934, + "num_tokens": 85330757.0, + "step": 3302 + }, + { + "epoch": 0.36272787173292337, + "grad_norm": 2.3101067543029785, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7334262132644653, + "num_tokens": 85354065.0, + "step": 3303 + }, + { + "epoch": 0.362837689435537, + "grad_norm": 2.250488758087158, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7399371862411499, + "num_tokens": 85377501.0, + "step": 3304 + }, + { + "epoch": 0.36294750713815066, + "grad_norm": 2.1478893756866455, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6946730613708496, + "num_tokens": 85405009.0, + "step": 3305 + }, + { + "epoch": 0.3630573248407643, + "grad_norm": 2.462932586669922, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.724704384803772, + "num_tokens": 85425905.0, + "step": 3306 + }, + { + "epoch": 0.363167142543378, + "grad_norm": 2.2382614612579346, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7078512907028198, + "num_tokens": 85452029.0, + "step": 3307 + }, + { + "epoch": 0.36327696024599165, + "grad_norm": 2.338992118835449, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7037255764007568, + "num_tokens": 85478947.0, + "step": 3308 + }, + { + "epoch": 0.3633867779486053, + "grad_norm": 2.1085102558135986, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.699212372303009, + "num_tokens": 85510887.0, + "step": 3309 + }, + { + "epoch": 0.363496595651219, + "grad_norm": 2.190581798553467, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6936172842979431, + "num_tokens": 85539105.0, + "step": 3310 + }, + { + "epoch": 0.36360641335383265, + "grad_norm": 2.430769681930542, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7175524234771729, + "num_tokens": 85561827.0, + "step": 3311 + }, + { + "epoch": 0.3637162310564463, + "grad_norm": 1.9850385189056396, + "learning_rate": 1e-06, + "loss": 1.091, + "mean_token_accuracy": 0.6722692251205444, + "num_tokens": 85594193.0, + "step": 3312 + }, + { + "epoch": 0.36382604875905994, + "grad_norm": 2.2068870067596436, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6930347681045532, + "num_tokens": 85619805.0, + "step": 3313 + }, + { + "epoch": 0.36393586646167364, + "grad_norm": 2.581279754638672, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7025335431098938, + "num_tokens": 85640275.0, + "step": 3314 + }, + { + "epoch": 0.3640456841642873, + "grad_norm": 2.30938458442688, + "learning_rate": 1e-06, + "loss": 1.0494, + "mean_token_accuracy": 0.6898033022880554, + "num_tokens": 85665126.0, + "step": 3315 + }, + { + "epoch": 0.36415550186690093, + "grad_norm": 2.398786783218384, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.7030512094497681, + "num_tokens": 85689650.0, + "step": 3316 + }, + { + "epoch": 0.3642653195695146, + "grad_norm": 1.994136095046997, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6947767734527588, + "num_tokens": 85720567.0, + "step": 3317 + }, + { + "epoch": 0.3643751372721283, + "grad_norm": 2.341732978820801, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6934903860092163, + "num_tokens": 85744555.0, + "step": 3318 + }, + { + "epoch": 0.3644849549747419, + "grad_norm": 2.192168712615967, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7008118033409119, + "num_tokens": 85769197.0, + "step": 3319 + }, + { + "epoch": 0.36459477267735557, + "grad_norm": 2.1566081047058105, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7056691646575928, + "num_tokens": 85794927.0, + "step": 3320 + }, + { + "epoch": 0.3647045903799693, + "grad_norm": 2.165804147720337, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7059681415557861, + "num_tokens": 85820362.0, + "step": 3321 + }, + { + "epoch": 0.3648144080825829, + "grad_norm": 2.3875889778137207, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7235710620880127, + "num_tokens": 85841839.0, + "step": 3322 + }, + { + "epoch": 0.36492422578519657, + "grad_norm": 2.2274062633514404, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6907871961593628, + "num_tokens": 85867377.0, + "step": 3323 + }, + { + "epoch": 0.3650340434878102, + "grad_norm": 2.227569341659546, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6820789575576782, + "num_tokens": 85894728.0, + "step": 3324 + }, + { + "epoch": 0.3651438611904239, + "grad_norm": 2.179175615310669, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7153077125549316, + "num_tokens": 85920460.0, + "step": 3325 + }, + { + "epoch": 0.36525367889303756, + "grad_norm": 2.205244302749634, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6895307302474976, + "num_tokens": 85947417.0, + "step": 3326 + }, + { + "epoch": 0.3653634965956512, + "grad_norm": 2.0508620738983154, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6864743232727051, + "num_tokens": 85978475.0, + "step": 3327 + }, + { + "epoch": 0.3654733142982649, + "grad_norm": 2.0969061851501465, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.695431113243103, + "num_tokens": 86005268.0, + "step": 3328 + }, + { + "epoch": 0.36558313200087855, + "grad_norm": 2.161865472793579, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7134942412376404, + "num_tokens": 86032447.0, + "step": 3329 + }, + { + "epoch": 0.3656929497034922, + "grad_norm": 2.3742148876190186, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6849013566970825, + "num_tokens": 86055965.0, + "step": 3330 + }, + { + "epoch": 0.36580276740610584, + "grad_norm": 2.136157989501953, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6791602969169617, + "num_tokens": 86083407.0, + "step": 3331 + }, + { + "epoch": 0.36591258510871955, + "grad_norm": 2.2929723262786865, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7072094678878784, + "num_tokens": 86107077.0, + "step": 3332 + }, + { + "epoch": 0.3660224028113332, + "grad_norm": 2.3277475833892822, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.691476047039032, + "num_tokens": 86131381.0, + "step": 3333 + }, + { + "epoch": 0.36613222051394684, + "grad_norm": 2.549745798110962, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.692789614200592, + "num_tokens": 86154122.0, + "step": 3334 + }, + { + "epoch": 0.3662420382165605, + "grad_norm": 2.2689707279205322, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.693354606628418, + "num_tokens": 86180267.0, + "step": 3335 + }, + { + "epoch": 0.3663518559191742, + "grad_norm": 1.8850297927856445, + "learning_rate": 1e-06, + "loss": 1.0757, + "mean_token_accuracy": 0.67949378490448, + "num_tokens": 86215594.0, + "step": 3336 + }, + { + "epoch": 0.36646167362178783, + "grad_norm": 1.9965801239013672, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.692554771900177, + "num_tokens": 86246071.0, + "step": 3337 + }, + { + "epoch": 0.3665714913244015, + "grad_norm": 2.091017723083496, + "learning_rate": 1e-06, + "loss": 1.0991, + "mean_token_accuracy": 0.671552300453186, + "num_tokens": 86277975.0, + "step": 3338 + }, + { + "epoch": 0.3666813090270152, + "grad_norm": 2.1532225608825684, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6954286098480225, + "num_tokens": 86304784.0, + "step": 3339 + }, + { + "epoch": 0.3667911267296288, + "grad_norm": 2.1386795043945312, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7122111320495605, + "num_tokens": 86334727.0, + "step": 3340 + }, + { + "epoch": 0.36690094443224247, + "grad_norm": 2.2795674800872803, + "learning_rate": 1e-06, + "loss": 1.0836, + "mean_token_accuracy": 0.6953946352005005, + "num_tokens": 86360301.0, + "step": 3341 + }, + { + "epoch": 0.3670107621348561, + "grad_norm": 2.182264566421509, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7111125588417053, + "num_tokens": 86384913.0, + "step": 3342 + }, + { + "epoch": 0.3671205798374698, + "grad_norm": 2.186349630355835, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.7018957138061523, + "num_tokens": 86410825.0, + "step": 3343 + }, + { + "epoch": 0.36723039754008346, + "grad_norm": 2.2170064449310303, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6947817802429199, + "num_tokens": 86436639.0, + "step": 3344 + }, + { + "epoch": 0.3673402152426971, + "grad_norm": 2.083707332611084, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.70087730884552, + "num_tokens": 86465164.0, + "step": 3345 + }, + { + "epoch": 0.3674500329453108, + "grad_norm": 2.3820338249206543, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6938060522079468, + "num_tokens": 86487983.0, + "step": 3346 + }, + { + "epoch": 0.36755985064792446, + "grad_norm": 2.0513453483581543, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7066284418106079, + "num_tokens": 86515243.0, + "step": 3347 + }, + { + "epoch": 0.3676696683505381, + "grad_norm": 2.2874557971954346, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7090954780578613, + "num_tokens": 86540172.0, + "step": 3348 + }, + { + "epoch": 0.36777948605315175, + "grad_norm": 2.0552191734313965, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6815671920776367, + "num_tokens": 86571138.0, + "step": 3349 + }, + { + "epoch": 0.36788930375576545, + "grad_norm": 1.9647547006607056, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6864038705825806, + "num_tokens": 86604797.0, + "step": 3350 + }, + { + "epoch": 0.3679991214583791, + "grad_norm": 2.2461326122283936, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7127271890640259, + "num_tokens": 86628636.0, + "step": 3351 + }, + { + "epoch": 0.36810893916099274, + "grad_norm": 1.8931399583816528, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.694940447807312, + "num_tokens": 86663120.0, + "step": 3352 + }, + { + "epoch": 0.3682187568636064, + "grad_norm": 2.1405820846557617, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7193476557731628, + "num_tokens": 86689552.0, + "step": 3353 + }, + { + "epoch": 0.3683285745662201, + "grad_norm": 2.4111671447753906, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.700566291809082, + "num_tokens": 86712863.0, + "step": 3354 + }, + { + "epoch": 0.36843839226883374, + "grad_norm": 2.2983269691467285, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6929484605789185, + "num_tokens": 86736563.0, + "step": 3355 + }, + { + "epoch": 0.3685482099714474, + "grad_norm": 2.0801162719726562, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7045639753341675, + "num_tokens": 86764256.0, + "step": 3356 + }, + { + "epoch": 0.3686580276740611, + "grad_norm": 2.2906367778778076, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7195597887039185, + "num_tokens": 86787797.0, + "step": 3357 + }, + { + "epoch": 0.36876784537667473, + "grad_norm": 2.099743127822876, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7134901285171509, + "num_tokens": 86815390.0, + "step": 3358 + }, + { + "epoch": 0.3688776630792884, + "grad_norm": 1.9647157192230225, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.734890878200531, + "num_tokens": 86846028.0, + "step": 3359 + }, + { + "epoch": 0.368987480781902, + "grad_norm": 2.426198720932007, + "learning_rate": 1e-06, + "loss": 1.0634, + "mean_token_accuracy": 0.6842612028121948, + "num_tokens": 86869114.0, + "step": 3360 + }, + { + "epoch": 0.3690972984845157, + "grad_norm": 2.3446621894836426, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6945258378982544, + "num_tokens": 86893563.0, + "step": 3361 + }, + { + "epoch": 0.36920711618712937, + "grad_norm": 2.3721649646759033, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.704498291015625, + "num_tokens": 86916057.0, + "step": 3362 + }, + { + "epoch": 0.369316933889743, + "grad_norm": 2.1581788063049316, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6884677410125732, + "num_tokens": 86944193.0, + "step": 3363 + }, + { + "epoch": 0.36942675159235666, + "grad_norm": 2.0971596240997314, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6963300108909607, + "num_tokens": 86973130.0, + "step": 3364 + }, + { + "epoch": 0.36953656929497036, + "grad_norm": 2.314357042312622, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7178045511245728, + "num_tokens": 86997045.0, + "step": 3365 + }, + { + "epoch": 0.369646386997584, + "grad_norm": 2.30098557472229, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.688927173614502, + "num_tokens": 87021295.0, + "step": 3366 + }, + { + "epoch": 0.36975620470019765, + "grad_norm": 2.1243417263031006, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.7020313739776611, + "num_tokens": 87046413.0, + "step": 3367 + }, + { + "epoch": 0.36986602240281136, + "grad_norm": 2.4211502075195312, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6942015886306763, + "num_tokens": 87073882.0, + "step": 3368 + }, + { + "epoch": 0.369975840105425, + "grad_norm": 2.275653123855591, + "learning_rate": 1e-06, + "loss": 1.0756, + "mean_token_accuracy": 0.6855958104133606, + "num_tokens": 87100385.0, + "step": 3369 + }, + { + "epoch": 0.37008565780803865, + "grad_norm": 2.2317535877227783, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7143098711967468, + "num_tokens": 87126593.0, + "step": 3370 + }, + { + "epoch": 0.3701954755106523, + "grad_norm": 1.8628978729248047, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6892256736755371, + "num_tokens": 87161071.0, + "step": 3371 + }, + { + "epoch": 0.370305293213266, + "grad_norm": 2.185849905014038, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6984485983848572, + "num_tokens": 87186272.0, + "step": 3372 + }, + { + "epoch": 0.37041511091587964, + "grad_norm": 2.1899373531341553, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.709092915058136, + "num_tokens": 87212120.0, + "step": 3373 + }, + { + "epoch": 0.3705249286184933, + "grad_norm": 2.552349805831909, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7098040580749512, + "num_tokens": 87231883.0, + "step": 3374 + }, + { + "epoch": 0.370634746321107, + "grad_norm": 2.06240177154541, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.686701774597168, + "num_tokens": 87260370.0, + "step": 3375 + }, + { + "epoch": 0.37074456402372064, + "grad_norm": 2.145745277404785, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6968179941177368, + "num_tokens": 87286894.0, + "step": 3376 + }, + { + "epoch": 0.3708543817263343, + "grad_norm": 1.9730300903320312, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.6828985810279846, + "num_tokens": 87321398.0, + "step": 3377 + }, + { + "epoch": 0.3709641994289479, + "grad_norm": 2.381434202194214, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7160549163818359, + "num_tokens": 87345279.0, + "step": 3378 + }, + { + "epoch": 0.37107401713156163, + "grad_norm": 2.5244498252868652, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7051742076873779, + "num_tokens": 87369573.0, + "step": 3379 + }, + { + "epoch": 0.3711838348341753, + "grad_norm": 2.301992177963257, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.683978259563446, + "num_tokens": 87396176.0, + "step": 3380 + }, + { + "epoch": 0.3712936525367889, + "grad_norm": 1.966528296470642, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6970386505126953, + "num_tokens": 87426871.0, + "step": 3381 + }, + { + "epoch": 0.37140347023940257, + "grad_norm": 2.1876282691955566, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.7037215232849121, + "num_tokens": 87453349.0, + "step": 3382 + }, + { + "epoch": 0.37151328794201627, + "grad_norm": 1.9500433206558228, + "learning_rate": 1e-06, + "loss": 1.1011, + "mean_token_accuracy": 0.6976771354675293, + "num_tokens": 87484742.0, + "step": 3383 + }, + { + "epoch": 0.3716231056446299, + "grad_norm": 2.4678966999053955, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7152572870254517, + "num_tokens": 87506704.0, + "step": 3384 + }, + { + "epoch": 0.37173292334724356, + "grad_norm": 2.357618570327759, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6931384205818176, + "num_tokens": 87529839.0, + "step": 3385 + }, + { + "epoch": 0.37184274104985726, + "grad_norm": 2.1694834232330322, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6926568746566772, + "num_tokens": 87558485.0, + "step": 3386 + }, + { + "epoch": 0.3719525587524709, + "grad_norm": 2.183310031890869, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7077705264091492, + "num_tokens": 87586352.0, + "step": 3387 + }, + { + "epoch": 0.37206237645508455, + "grad_norm": 2.1679129600524902, + "learning_rate": 1e-06, + "loss": 1.0855, + "mean_token_accuracy": 0.688461184501648, + "num_tokens": 87614672.0, + "step": 3388 + }, + { + "epoch": 0.3721721941576982, + "grad_norm": 2.035391092300415, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7155239582061768, + "num_tokens": 87643313.0, + "step": 3389 + }, + { + "epoch": 0.3722820118603119, + "grad_norm": 2.139786958694458, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6983263492584229, + "num_tokens": 87670451.0, + "step": 3390 + }, + { + "epoch": 0.37239182956292555, + "grad_norm": 2.5093822479248047, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6869298815727234, + "num_tokens": 87692044.0, + "step": 3391 + }, + { + "epoch": 0.3725016472655392, + "grad_norm": 1.9274742603302002, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7030304074287415, + "num_tokens": 87721089.0, + "step": 3392 + }, + { + "epoch": 0.37261146496815284, + "grad_norm": 2.311614751815796, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.7021079063415527, + "num_tokens": 87743663.0, + "step": 3393 + }, + { + "epoch": 0.37272128267076654, + "grad_norm": 2.5010645389556885, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7108855843544006, + "num_tokens": 87764986.0, + "step": 3394 + }, + { + "epoch": 0.3728311003733802, + "grad_norm": 2.2405683994293213, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7081848978996277, + "num_tokens": 87790380.0, + "step": 3395 + }, + { + "epoch": 0.37294091807599383, + "grad_norm": 2.317086696624756, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6901869177818298, + "num_tokens": 87814012.0, + "step": 3396 + }, + { + "epoch": 0.37305073577860753, + "grad_norm": 2.27608585357666, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6905070543289185, + "num_tokens": 87838307.0, + "step": 3397 + }, + { + "epoch": 0.3731605534812212, + "grad_norm": 2.2250559329986572, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7020050287246704, + "num_tokens": 87864332.0, + "step": 3398 + }, + { + "epoch": 0.3732703711838348, + "grad_norm": 2.6504411697387695, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6932108402252197, + "num_tokens": 87884903.0, + "step": 3399 + }, + { + "epoch": 0.37338018888644847, + "grad_norm": 2.2688729763031006, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.6971598863601685, + "num_tokens": 87908543.0, + "step": 3400 + }, + { + "epoch": 0.3734900065890622, + "grad_norm": 2.424797773361206, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7080845832824707, + "num_tokens": 87930022.0, + "step": 3401 + }, + { + "epoch": 0.3735998242916758, + "grad_norm": 1.9627985954284668, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6916685104370117, + "num_tokens": 87961748.0, + "step": 3402 + }, + { + "epoch": 0.37370964199428947, + "grad_norm": 1.886224627494812, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7076574563980103, + "num_tokens": 87995355.0, + "step": 3403 + }, + { + "epoch": 0.37381945969690317, + "grad_norm": 1.9729117155075073, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.6809518933296204, + "num_tokens": 88026996.0, + "step": 3404 + }, + { + "epoch": 0.3739292773995168, + "grad_norm": 2.171665668487549, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7223004102706909, + "num_tokens": 88052140.0, + "step": 3405 + }, + { + "epoch": 0.37403909510213046, + "grad_norm": 2.40944242477417, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.690436840057373, + "num_tokens": 88073640.0, + "step": 3406 + }, + { + "epoch": 0.3741489128047441, + "grad_norm": 2.2734909057617188, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.699600338935852, + "num_tokens": 88096965.0, + "step": 3407 + }, + { + "epoch": 0.3742587305073578, + "grad_norm": 1.917035460472107, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6932065486907959, + "num_tokens": 88128966.0, + "step": 3408 + }, + { + "epoch": 0.37436854820997145, + "grad_norm": 2.2169179916381836, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7130520939826965, + "num_tokens": 88152934.0, + "step": 3409 + }, + { + "epoch": 0.3744783659125851, + "grad_norm": 2.5428507328033447, + "learning_rate": 1e-06, + "loss": 1.056, + "mean_token_accuracy": 0.6852928400039673, + "num_tokens": 88175382.0, + "step": 3410 + }, + { + "epoch": 0.37458818361519874, + "grad_norm": 2.1405837535858154, + "learning_rate": 1e-06, + "loss": 1.0817, + "mean_token_accuracy": 0.6859754323959351, + "num_tokens": 88204938.0, + "step": 3411 + }, + { + "epoch": 0.37469800131781245, + "grad_norm": 2.45394229888916, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.700993537902832, + "num_tokens": 88226259.0, + "step": 3412 + }, + { + "epoch": 0.3748078190204261, + "grad_norm": 2.2493577003479004, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.6733479499816895, + "num_tokens": 88251893.0, + "step": 3413 + }, + { + "epoch": 0.37491763672303974, + "grad_norm": 2.2144439220428467, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7071022987365723, + "num_tokens": 88277786.0, + "step": 3414 + }, + { + "epoch": 0.37502745442565344, + "grad_norm": 2.4174413681030273, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7038275003433228, + "num_tokens": 88300143.0, + "step": 3415 + }, + { + "epoch": 0.3751372721282671, + "grad_norm": 2.426701307296753, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7054135799407959, + "num_tokens": 88323457.0, + "step": 3416 + }, + { + "epoch": 0.37524708983088073, + "grad_norm": 2.6008152961730957, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7136090993881226, + "num_tokens": 88343643.0, + "step": 3417 + }, + { + "epoch": 0.3753569075334944, + "grad_norm": 2.0486438274383545, + "learning_rate": 1e-06, + "loss": 1.0823, + "mean_token_accuracy": 0.6752936244010925, + "num_tokens": 88373513.0, + "step": 3418 + }, + { + "epoch": 0.3754667252361081, + "grad_norm": 2.075321674346924, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6919224858283997, + "num_tokens": 88402245.0, + "step": 3419 + }, + { + "epoch": 0.3755765429387217, + "grad_norm": 2.130197048187256, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6944167613983154, + "num_tokens": 88430919.0, + "step": 3420 + }, + { + "epoch": 0.37568636064133537, + "grad_norm": 2.0584187507629395, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7115572690963745, + "num_tokens": 88460174.0, + "step": 3421 + }, + { + "epoch": 0.37579617834394907, + "grad_norm": 2.1246583461761475, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6908191442489624, + "num_tokens": 88488063.0, + "step": 3422 + }, + { + "epoch": 0.3759059960465627, + "grad_norm": 2.0766422748565674, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6943503618240356, + "num_tokens": 88520294.0, + "step": 3423 + }, + { + "epoch": 0.37601581374917636, + "grad_norm": 1.9227750301361084, + "learning_rate": 1e-06, + "loss": 1.0651, + "mean_token_accuracy": 0.6782054901123047, + "num_tokens": 88554544.0, + "step": 3424 + }, + { + "epoch": 0.37612563145179, + "grad_norm": 2.241102695465088, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7088872790336609, + "num_tokens": 88578934.0, + "step": 3425 + }, + { + "epoch": 0.3762354491544037, + "grad_norm": 2.3563387393951416, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7076473236083984, + "num_tokens": 88600495.0, + "step": 3426 + }, + { + "epoch": 0.37634526685701736, + "grad_norm": 2.145965337753296, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7154170274734497, + "num_tokens": 88627034.0, + "step": 3427 + }, + { + "epoch": 0.376455084559631, + "grad_norm": 2.2688794136047363, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7024806141853333, + "num_tokens": 88650799.0, + "step": 3428 + }, + { + "epoch": 0.37656490226224465, + "grad_norm": 2.4832000732421875, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7080345153808594, + "num_tokens": 88670895.0, + "step": 3429 + }, + { + "epoch": 0.37667471996485835, + "grad_norm": 2.2764792442321777, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6953346729278564, + "num_tokens": 88693919.0, + "step": 3430 + }, + { + "epoch": 0.376784537667472, + "grad_norm": 2.1509416103363037, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6750772595405579, + "num_tokens": 88721418.0, + "step": 3431 + }, + { + "epoch": 0.37689435537008564, + "grad_norm": 2.2969095706939697, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.722145676612854, + "num_tokens": 88745403.0, + "step": 3432 + }, + { + "epoch": 0.37700417307269934, + "grad_norm": 1.9243184328079224, + "learning_rate": 1e-06, + "loss": 1.06, + "mean_token_accuracy": 0.6854401230812073, + "num_tokens": 88780492.0, + "step": 3433 + }, + { + "epoch": 0.377113990775313, + "grad_norm": 2.0725038051605225, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.690028190612793, + "num_tokens": 88811625.0, + "step": 3434 + }, + { + "epoch": 0.37722380847792664, + "grad_norm": 2.0311410427093506, + "learning_rate": 1e-06, + "loss": 1.1216, + "mean_token_accuracy": 0.6774294376373291, + "num_tokens": 88843377.0, + "step": 3435 + }, + { + "epoch": 0.3773336261805403, + "grad_norm": 2.1645538806915283, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7048262357711792, + "num_tokens": 88871222.0, + "step": 3436 + }, + { + "epoch": 0.377443443883154, + "grad_norm": 1.857916235923767, + "learning_rate": 1e-06, + "loss": 1.0773, + "mean_token_accuracy": 0.6756021976470947, + "num_tokens": 88906003.0, + "step": 3437 + }, + { + "epoch": 0.37755326158576763, + "grad_norm": 2.173142433166504, + "learning_rate": 1e-06, + "loss": 1.0671, + "mean_token_accuracy": 0.6847217679023743, + "num_tokens": 88934175.0, + "step": 3438 + }, + { + "epoch": 0.3776630792883813, + "grad_norm": 1.9749023914337158, + "learning_rate": 1e-06, + "loss": 1.0853, + "mean_token_accuracy": 0.6774473786354065, + "num_tokens": 88967305.0, + "step": 3439 + }, + { + "epoch": 0.3777728969909949, + "grad_norm": 2.1947410106658936, + "learning_rate": 1e-06, + "loss": 1.0886, + "mean_token_accuracy": 0.6728881597518921, + "num_tokens": 88995058.0, + "step": 3440 + }, + { + "epoch": 0.3778827146936086, + "grad_norm": 2.50858473777771, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6869018077850342, + "num_tokens": 89016825.0, + "step": 3441 + }, + { + "epoch": 0.37799253239622227, + "grad_norm": 2.245027542114258, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6997928619384766, + "num_tokens": 89044034.0, + "step": 3442 + }, + { + "epoch": 0.3781023500988359, + "grad_norm": 2.316415309906006, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7025946378707886, + "num_tokens": 89068052.0, + "step": 3443 + }, + { + "epoch": 0.3782121678014496, + "grad_norm": 2.129035711288452, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7133219838142395, + "num_tokens": 89094888.0, + "step": 3444 + }, + { + "epoch": 0.37832198550406326, + "grad_norm": 2.149656057357788, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7022757530212402, + "num_tokens": 89120885.0, + "step": 3445 + }, + { + "epoch": 0.3784318032066769, + "grad_norm": 2.035099983215332, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7170010805130005, + "num_tokens": 89148606.0, + "step": 3446 + }, + { + "epoch": 0.37854162090929055, + "grad_norm": 2.4501686096191406, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7243483066558838, + "num_tokens": 89170649.0, + "step": 3447 + }, + { + "epoch": 0.37865143861190426, + "grad_norm": 2.2467658519744873, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.690431535243988, + "num_tokens": 89196599.0, + "step": 3448 + }, + { + "epoch": 0.3787612563145179, + "grad_norm": 2.383690357208252, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.710533857345581, + "num_tokens": 89218130.0, + "step": 3449 + }, + { + "epoch": 0.37887107401713155, + "grad_norm": 2.2573800086975098, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6971402168273926, + "num_tokens": 89242478.0, + "step": 3450 + }, + { + "epoch": 0.37898089171974525, + "grad_norm": 2.28426194190979, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6952919960021973, + "num_tokens": 89267931.0, + "step": 3451 + }, + { + "epoch": 0.3790907094223589, + "grad_norm": 1.9618271589279175, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7238448858261108, + "num_tokens": 89298706.0, + "step": 3452 + }, + { + "epoch": 0.37920052712497254, + "grad_norm": 2.2362279891967773, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.7100476026535034, + "num_tokens": 89325375.0, + "step": 3453 + }, + { + "epoch": 0.3793103448275862, + "grad_norm": 2.549398183822632, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7109785079956055, + "num_tokens": 89346526.0, + "step": 3454 + }, + { + "epoch": 0.3794201625301999, + "grad_norm": 2.228080987930298, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6928515434265137, + "num_tokens": 89372830.0, + "step": 3455 + }, + { + "epoch": 0.37952998023281354, + "grad_norm": 2.2993004322052, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6986883282661438, + "num_tokens": 89396497.0, + "step": 3456 + }, + { + "epoch": 0.3796397979354272, + "grad_norm": 2.2166950702667236, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.707627534866333, + "num_tokens": 89422444.0, + "step": 3457 + }, + { + "epoch": 0.3797496156380408, + "grad_norm": 2.2404870986938477, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6837743520736694, + "num_tokens": 89447722.0, + "step": 3458 + }, + { + "epoch": 0.37985943334065453, + "grad_norm": 2.1928586959838867, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7232624292373657, + "num_tokens": 89472695.0, + "step": 3459 + }, + { + "epoch": 0.3799692510432682, + "grad_norm": 1.991197109222412, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6927632689476013, + "num_tokens": 89504156.0, + "step": 3460 + }, + { + "epoch": 0.3800790687458818, + "grad_norm": 2.0151937007904053, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6928607821464539, + "num_tokens": 89533887.0, + "step": 3461 + }, + { + "epoch": 0.3801888864484955, + "grad_norm": 2.364013195037842, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7050462961196899, + "num_tokens": 89558518.0, + "step": 3462 + }, + { + "epoch": 0.38029870415110917, + "grad_norm": 2.5481462478637695, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6920375823974609, + "num_tokens": 89579890.0, + "step": 3463 + }, + { + "epoch": 0.3804085218537228, + "grad_norm": 2.3029096126556396, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7046068906784058, + "num_tokens": 89604732.0, + "step": 3464 + }, + { + "epoch": 0.38051833955633646, + "grad_norm": 2.488347053527832, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7120071053504944, + "num_tokens": 89625793.0, + "step": 3465 + }, + { + "epoch": 0.38062815725895016, + "grad_norm": 2.322580575942993, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7140179872512817, + "num_tokens": 89648859.0, + "step": 3466 + }, + { + "epoch": 0.3807379749615638, + "grad_norm": 1.996881365776062, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6990599632263184, + "num_tokens": 89679663.0, + "step": 3467 + }, + { + "epoch": 0.38084779266417745, + "grad_norm": 2.1974618434906006, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.6829662322998047, + "num_tokens": 89705345.0, + "step": 3468 + }, + { + "epoch": 0.3809576103667911, + "grad_norm": 2.2325587272644043, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6930285692214966, + "num_tokens": 89730700.0, + "step": 3469 + }, + { + "epoch": 0.3810674280694048, + "grad_norm": 1.9599058628082275, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7120883464813232, + "num_tokens": 89763152.0, + "step": 3470 + }, + { + "epoch": 0.38117724577201845, + "grad_norm": 2.3442089557647705, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.692742109298706, + "num_tokens": 89786840.0, + "step": 3471 + }, + { + "epoch": 0.3812870634746321, + "grad_norm": 2.219726085662842, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7138283252716064, + "num_tokens": 89811531.0, + "step": 3472 + }, + { + "epoch": 0.3813968811772458, + "grad_norm": 2.187114953994751, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6945134401321411, + "num_tokens": 89837392.0, + "step": 3473 + }, + { + "epoch": 0.38150669887985944, + "grad_norm": 2.293778896331787, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.715643048286438, + "num_tokens": 89861657.0, + "step": 3474 + }, + { + "epoch": 0.3816165165824731, + "grad_norm": 2.2304279804229736, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7004347443580627, + "num_tokens": 89888089.0, + "step": 3475 + }, + { + "epoch": 0.38172633428508673, + "grad_norm": 1.879531741142273, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7014685273170471, + "num_tokens": 89922136.0, + "step": 3476 + }, + { + "epoch": 0.38183615198770043, + "grad_norm": 2.218838930130005, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7084540724754333, + "num_tokens": 89946651.0, + "step": 3477 + }, + { + "epoch": 0.3819459696903141, + "grad_norm": 2.4049980640411377, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7040989995002747, + "num_tokens": 89970358.0, + "step": 3478 + }, + { + "epoch": 0.3820557873929277, + "grad_norm": 2.3925535678863525, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6917771100997925, + "num_tokens": 89994077.0, + "step": 3479 + }, + { + "epoch": 0.3821656050955414, + "grad_norm": 2.432232141494751, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.6875146627426147, + "num_tokens": 90017250.0, + "step": 3480 + }, + { + "epoch": 0.3822754227981551, + "grad_norm": 2.1252706050872803, + "learning_rate": 1e-06, + "loss": 1.0639, + "mean_token_accuracy": 0.6807325482368469, + "num_tokens": 90046866.0, + "step": 3481 + }, + { + "epoch": 0.3823852405007687, + "grad_norm": 2.3950891494750977, + "learning_rate": 1e-06, + "loss": 1.0904, + "mean_token_accuracy": 0.6809586882591248, + "num_tokens": 90070820.0, + "step": 3482 + }, + { + "epoch": 0.38249505820338237, + "grad_norm": 1.9814778566360474, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6842586994171143, + "num_tokens": 90103266.0, + "step": 3483 + }, + { + "epoch": 0.38260487590599607, + "grad_norm": 2.3199214935302734, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.7071367502212524, + "num_tokens": 90128277.0, + "step": 3484 + }, + { + "epoch": 0.3827146936086097, + "grad_norm": 2.080543279647827, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6922218203544617, + "num_tokens": 90158627.0, + "step": 3485 + }, + { + "epoch": 0.38282451131122336, + "grad_norm": 2.1854889392852783, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6785917282104492, + "num_tokens": 90188177.0, + "step": 3486 + }, + { + "epoch": 0.382934329013837, + "grad_norm": 2.0116209983825684, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6894952058792114, + "num_tokens": 90219370.0, + "step": 3487 + }, + { + "epoch": 0.3830441467164507, + "grad_norm": 2.5029988288879395, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7269309759140015, + "num_tokens": 90239794.0, + "step": 3488 + }, + { + "epoch": 0.38315396441906435, + "grad_norm": 2.067430019378662, + "learning_rate": 1e-06, + "loss": 1.0662, + "mean_token_accuracy": 0.679092288017273, + "num_tokens": 90270780.0, + "step": 3489 + }, + { + "epoch": 0.383263782121678, + "grad_norm": 2.045177698135376, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7170854806900024, + "num_tokens": 90299170.0, + "step": 3490 + }, + { + "epoch": 0.3833735998242917, + "grad_norm": 2.0249390602111816, + "learning_rate": 1e-06, + "loss": 1.0904, + "mean_token_accuracy": 0.6770050525665283, + "num_tokens": 90330360.0, + "step": 3491 + }, + { + "epoch": 0.38348341752690535, + "grad_norm": 2.6024394035339355, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7076466679573059, + "num_tokens": 90350344.0, + "step": 3492 + }, + { + "epoch": 0.383593235229519, + "grad_norm": 2.3903443813323975, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7176019549369812, + "num_tokens": 90374020.0, + "step": 3493 + }, + { + "epoch": 0.38370305293213264, + "grad_norm": 2.3008651733398438, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7028203010559082, + "num_tokens": 90398968.0, + "step": 3494 + }, + { + "epoch": 0.38381287063474634, + "grad_norm": 2.287567138671875, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7348897457122803, + "num_tokens": 90424099.0, + "step": 3495 + }, + { + "epoch": 0.38392268833736, + "grad_norm": 2.3014543056488037, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6984984874725342, + "num_tokens": 90447219.0, + "step": 3496 + }, + { + "epoch": 0.38403250603997363, + "grad_norm": 2.0848381519317627, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.708411455154419, + "num_tokens": 90475098.0, + "step": 3497 + }, + { + "epoch": 0.38414232374258733, + "grad_norm": 2.565642833709717, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7384375333786011, + "num_tokens": 90493385.0, + "step": 3498 + }, + { + "epoch": 0.384252141445201, + "grad_norm": 2.1089792251586914, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.703862190246582, + "num_tokens": 90524253.0, + "step": 3499 + }, + { + "epoch": 0.3843619591478146, + "grad_norm": 2.059297800064087, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6926709413528442, + "num_tokens": 90554973.0, + "step": 3500 + }, + { + "epoch": 0.38447177685042827, + "grad_norm": 2.0692670345306396, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.7037011384963989, + "num_tokens": 90584515.0, + "step": 3501 + }, + { + "epoch": 0.38458159455304197, + "grad_norm": 2.354339599609375, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7008670568466187, + "num_tokens": 90608481.0, + "step": 3502 + }, + { + "epoch": 0.3846914122556556, + "grad_norm": 2.1858811378479004, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6949841380119324, + "num_tokens": 90633367.0, + "step": 3503 + }, + { + "epoch": 0.38480122995826926, + "grad_norm": 2.3402159214019775, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7023200392723083, + "num_tokens": 90657309.0, + "step": 3504 + }, + { + "epoch": 0.3849110476608829, + "grad_norm": 2.4151298999786377, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7014299035072327, + "num_tokens": 90681447.0, + "step": 3505 + }, + { + "epoch": 0.3850208653634966, + "grad_norm": 2.3229315280914307, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6887948513031006, + "num_tokens": 90705941.0, + "step": 3506 + }, + { + "epoch": 0.38513068306611026, + "grad_norm": 1.9987798929214478, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7126832008361816, + "num_tokens": 90734216.0, + "step": 3507 + }, + { + "epoch": 0.3852405007687239, + "grad_norm": 2.1600375175476074, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7056833505630493, + "num_tokens": 90761487.0, + "step": 3508 + }, + { + "epoch": 0.3853503184713376, + "grad_norm": 2.1509764194488525, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6913402080535889, + "num_tokens": 90789622.0, + "step": 3509 + }, + { + "epoch": 0.38546013617395125, + "grad_norm": 2.050119400024414, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6929596662521362, + "num_tokens": 90819427.0, + "step": 3510 + }, + { + "epoch": 0.3855699538765649, + "grad_norm": 2.198925018310547, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6875860691070557, + "num_tokens": 90846329.0, + "step": 3511 + }, + { + "epoch": 0.38567977157917854, + "grad_norm": 2.0567617416381836, + "learning_rate": 1e-06, + "loss": 1.0946, + "mean_token_accuracy": 0.6713924407958984, + "num_tokens": 90875571.0, + "step": 3512 + }, + { + "epoch": 0.38578958928179224, + "grad_norm": 2.5478084087371826, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7330259084701538, + "num_tokens": 90895193.0, + "step": 3513 + }, + { + "epoch": 0.3858994069844059, + "grad_norm": 1.9929797649383545, + "learning_rate": 1e-06, + "loss": 1.0898, + "mean_token_accuracy": 0.6782416701316833, + "num_tokens": 90926516.0, + "step": 3514 + }, + { + "epoch": 0.38600922468701954, + "grad_norm": 2.2034664154052734, + "learning_rate": 1e-06, + "loss": 1.0596, + "mean_token_accuracy": 0.6772213578224182, + "num_tokens": 90955106.0, + "step": 3515 + }, + { + "epoch": 0.3861190423896332, + "grad_norm": 2.2314436435699463, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.7020263671875, + "num_tokens": 90980991.0, + "step": 3516 + }, + { + "epoch": 0.3862288600922469, + "grad_norm": 2.4456238746643066, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6978292465209961, + "num_tokens": 91004316.0, + "step": 3517 + }, + { + "epoch": 0.38633867779486053, + "grad_norm": 2.238295555114746, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7261958122253418, + "num_tokens": 91027761.0, + "step": 3518 + }, + { + "epoch": 0.3864484954974742, + "grad_norm": 2.367985725402832, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6864590048789978, + "num_tokens": 91050506.0, + "step": 3519 + }, + { + "epoch": 0.3865583132000879, + "grad_norm": 1.9702991247177124, + "learning_rate": 1e-06, + "loss": 1.0569, + "mean_token_accuracy": 0.6784712076187134, + "num_tokens": 91082101.0, + "step": 3520 + }, + { + "epoch": 0.3866681309027015, + "grad_norm": 2.4644908905029297, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7194606065750122, + "num_tokens": 91104010.0, + "step": 3521 + }, + { + "epoch": 0.38677794860531517, + "grad_norm": 2.2047014236450195, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.7046807408332825, + "num_tokens": 91129990.0, + "step": 3522 + }, + { + "epoch": 0.3868877663079288, + "grad_norm": 2.1281790733337402, + "learning_rate": 1e-06, + "loss": 1.0922, + "mean_token_accuracy": 0.6753151416778564, + "num_tokens": 91158919.0, + "step": 3523 + }, + { + "epoch": 0.3869975840105425, + "grad_norm": 2.355445623397827, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7073631286621094, + "num_tokens": 91180887.0, + "step": 3524 + }, + { + "epoch": 0.38710740171315616, + "grad_norm": 2.0293526649475098, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6921019554138184, + "num_tokens": 91211256.0, + "step": 3525 + }, + { + "epoch": 0.3872172194157698, + "grad_norm": 2.2862203121185303, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.700677752494812, + "num_tokens": 91235786.0, + "step": 3526 + }, + { + "epoch": 0.3873270371183835, + "grad_norm": 2.384659767150879, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7002922296524048, + "num_tokens": 91259530.0, + "step": 3527 + }, + { + "epoch": 0.38743685482099716, + "grad_norm": 2.334530830383301, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6828173398971558, + "num_tokens": 91284855.0, + "step": 3528 + }, + { + "epoch": 0.3875466725236108, + "grad_norm": 2.112677812576294, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7245563268661499, + "num_tokens": 91311673.0, + "step": 3529 + }, + { + "epoch": 0.38765649022622445, + "grad_norm": 2.2310361862182617, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6945230960845947, + "num_tokens": 91337540.0, + "step": 3530 + }, + { + "epoch": 0.38776630792883815, + "grad_norm": 2.2072534561157227, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6962061524391174, + "num_tokens": 91363407.0, + "step": 3531 + }, + { + "epoch": 0.3878761256314518, + "grad_norm": 2.6692521572113037, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.700694739818573, + "num_tokens": 91382935.0, + "step": 3532 + }, + { + "epoch": 0.38798594333406544, + "grad_norm": 2.325491189956665, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6994235515594482, + "num_tokens": 91407494.0, + "step": 3533 + }, + { + "epoch": 0.3880957610366791, + "grad_norm": 2.1170461177825928, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7202534675598145, + "num_tokens": 91434532.0, + "step": 3534 + }, + { + "epoch": 0.3882055787392928, + "grad_norm": 2.723174810409546, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7294647693634033, + "num_tokens": 91453366.0, + "step": 3535 + }, + { + "epoch": 0.38831539644190644, + "grad_norm": 2.204735517501831, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6976852416992188, + "num_tokens": 91479978.0, + "step": 3536 + }, + { + "epoch": 0.3884252141445201, + "grad_norm": 1.969370722770691, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6926288604736328, + "num_tokens": 91512790.0, + "step": 3537 + }, + { + "epoch": 0.3885350318471338, + "grad_norm": 2.118227005004883, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7001551985740662, + "num_tokens": 91539101.0, + "step": 3538 + }, + { + "epoch": 0.38864484954974743, + "grad_norm": 2.5018985271453857, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6942470073699951, + "num_tokens": 91561374.0, + "step": 3539 + }, + { + "epoch": 0.3887546672523611, + "grad_norm": 1.9809138774871826, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7082159519195557, + "num_tokens": 91589580.0, + "step": 3540 + }, + { + "epoch": 0.3888644849549747, + "grad_norm": 2.3820996284484863, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6931548118591309, + "num_tokens": 91612399.0, + "step": 3541 + }, + { + "epoch": 0.3889743026575884, + "grad_norm": 1.997899055480957, + "learning_rate": 1e-06, + "loss": 1.0745, + "mean_token_accuracy": 0.6815651655197144, + "num_tokens": 91644091.0, + "step": 3542 + }, + { + "epoch": 0.38908412036020207, + "grad_norm": 2.1207046508789062, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7004848718643188, + "num_tokens": 91670530.0, + "step": 3543 + }, + { + "epoch": 0.3891939380628157, + "grad_norm": 2.391664981842041, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.7060239315032959, + "num_tokens": 91694669.0, + "step": 3544 + }, + { + "epoch": 0.38930375576542936, + "grad_norm": 1.9582738876342773, + "learning_rate": 1e-06, + "loss": 1.0409, + "mean_token_accuracy": 0.6859198808670044, + "num_tokens": 91728597.0, + "step": 3545 + }, + { + "epoch": 0.38941357346804306, + "grad_norm": 2.0455009937286377, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6981750130653381, + "num_tokens": 91756801.0, + "step": 3546 + }, + { + "epoch": 0.3895233911706567, + "grad_norm": 2.5066723823547363, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6888402104377747, + "num_tokens": 91778500.0, + "step": 3547 + }, + { + "epoch": 0.38963320887327035, + "grad_norm": 2.4852988719940186, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.697590708732605, + "num_tokens": 91804944.0, + "step": 3548 + }, + { + "epoch": 0.38974302657588406, + "grad_norm": 2.1147141456604004, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7290257215499878, + "num_tokens": 91833600.0, + "step": 3549 + }, + { + "epoch": 0.3898528442784977, + "grad_norm": 2.214475393295288, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7291994094848633, + "num_tokens": 91856897.0, + "step": 3550 + }, + { + "epoch": 0.38996266198111135, + "grad_norm": 2.3605637550354004, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6976670026779175, + "num_tokens": 91878769.0, + "step": 3551 + }, + { + "epoch": 0.390072479683725, + "grad_norm": 2.329681873321533, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7114741802215576, + "num_tokens": 91901129.0, + "step": 3552 + }, + { + "epoch": 0.3901822973863387, + "grad_norm": 2.4688613414764404, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7262765765190125, + "num_tokens": 91921789.0, + "step": 3553 + }, + { + "epoch": 0.39029211508895234, + "grad_norm": 2.1783876419067383, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6928052306175232, + "num_tokens": 91949252.0, + "step": 3554 + }, + { + "epoch": 0.390401932791566, + "grad_norm": 2.3782501220703125, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7160876393318176, + "num_tokens": 91970893.0, + "step": 3555 + }, + { + "epoch": 0.3905117504941797, + "grad_norm": 2.033618211746216, + "learning_rate": 1e-06, + "loss": 1.0858, + "mean_token_accuracy": 0.6826561689376831, + "num_tokens": 92003459.0, + "step": 3556 + }, + { + "epoch": 0.39062156819679333, + "grad_norm": 2.1896913051605225, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.6837859153747559, + "num_tokens": 92032949.0, + "step": 3557 + }, + { + "epoch": 0.390731385899407, + "grad_norm": 2.267002582550049, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7004753351211548, + "num_tokens": 92058753.0, + "step": 3558 + }, + { + "epoch": 0.3908412036020206, + "grad_norm": 2.3057944774627686, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.676413893699646, + "num_tokens": 92086817.0, + "step": 3559 + }, + { + "epoch": 0.3909510213046343, + "grad_norm": 2.150648832321167, + "learning_rate": 1e-06, + "loss": 1.072, + "mean_token_accuracy": 0.6923211812973022, + "num_tokens": 92114863.0, + "step": 3560 + }, + { + "epoch": 0.391060839007248, + "grad_norm": 2.169379949569702, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6930480599403381, + "num_tokens": 92143174.0, + "step": 3561 + }, + { + "epoch": 0.3911706567098616, + "grad_norm": 2.2090697288513184, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7069700956344604, + "num_tokens": 92169271.0, + "step": 3562 + }, + { + "epoch": 0.39128047441247527, + "grad_norm": 2.0988638401031494, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.7007673978805542, + "num_tokens": 92197530.0, + "step": 3563 + }, + { + "epoch": 0.39139029211508897, + "grad_norm": 2.0849757194519043, + "learning_rate": 1e-06, + "loss": 1.0866, + "mean_token_accuracy": 0.6671055555343628, + "num_tokens": 92226522.0, + "step": 3564 + }, + { + "epoch": 0.3915001098177026, + "grad_norm": 2.0245728492736816, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6903603672981262, + "num_tokens": 92255728.0, + "step": 3565 + }, + { + "epoch": 0.39160992752031626, + "grad_norm": 1.9344797134399414, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7076014280319214, + "num_tokens": 92288415.0, + "step": 3566 + }, + { + "epoch": 0.39171974522292996, + "grad_norm": 2.2940421104431152, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7074259519577026, + "num_tokens": 92310582.0, + "step": 3567 + }, + { + "epoch": 0.3918295629255436, + "grad_norm": 2.2343740463256836, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6912243366241455, + "num_tokens": 92337192.0, + "step": 3568 + }, + { + "epoch": 0.39193938062815725, + "grad_norm": 2.171004295349121, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.6761401295661926, + "num_tokens": 92366075.0, + "step": 3569 + }, + { + "epoch": 0.3920491983307709, + "grad_norm": 2.2754838466644287, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.695537805557251, + "num_tokens": 92391433.0, + "step": 3570 + }, + { + "epoch": 0.3921590160333846, + "grad_norm": 2.2602837085723877, + "learning_rate": 1e-06, + "loss": 1.0976, + "mean_token_accuracy": 0.6750729084014893, + "num_tokens": 92419463.0, + "step": 3571 + }, + { + "epoch": 0.39226883373599825, + "grad_norm": 2.377466917037964, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7041165828704834, + "num_tokens": 92441456.0, + "step": 3572 + }, + { + "epoch": 0.3923786514386119, + "grad_norm": 2.0891129970550537, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7218363881111145, + "num_tokens": 92466494.0, + "step": 3573 + }, + { + "epoch": 0.3924884691412256, + "grad_norm": 2.179676055908203, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6874912977218628, + "num_tokens": 92491099.0, + "step": 3574 + }, + { + "epoch": 0.39259828684383924, + "grad_norm": 2.2576348781585693, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.723015546798706, + "num_tokens": 92518006.0, + "step": 3575 + }, + { + "epoch": 0.3927081045464529, + "grad_norm": 2.2053470611572266, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7071401476860046, + "num_tokens": 92542700.0, + "step": 3576 + }, + { + "epoch": 0.39281792224906653, + "grad_norm": 2.3025929927825928, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6946216821670532, + "num_tokens": 92566279.0, + "step": 3577 + }, + { + "epoch": 0.39292773995168023, + "grad_norm": 1.9924243688583374, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6865537166595459, + "num_tokens": 92599121.0, + "step": 3578 + }, + { + "epoch": 0.3930375576542939, + "grad_norm": 2.457331418991089, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.7019580602645874, + "num_tokens": 92624001.0, + "step": 3579 + }, + { + "epoch": 0.3931473753569075, + "grad_norm": 2.1063435077667236, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.702279269695282, + "num_tokens": 92652425.0, + "step": 3580 + }, + { + "epoch": 0.39325719305952117, + "grad_norm": 2.3927531242370605, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.698296844959259, + "num_tokens": 92674724.0, + "step": 3581 + }, + { + "epoch": 0.39336701076213487, + "grad_norm": 2.258366823196411, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6970287561416626, + "num_tokens": 92699667.0, + "step": 3582 + }, + { + "epoch": 0.3934768284647485, + "grad_norm": 2.1740493774414062, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6897591352462769, + "num_tokens": 92727517.0, + "step": 3583 + }, + { + "epoch": 0.39358664616736216, + "grad_norm": 2.582773208618164, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7117640376091003, + "num_tokens": 92747373.0, + "step": 3584 + }, + { + "epoch": 0.39369646386997587, + "grad_norm": 2.0992400646209717, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7086181640625, + "num_tokens": 92775832.0, + "step": 3585 + }, + { + "epoch": 0.3938062815725895, + "grad_norm": 1.9019960165023804, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6979423761367798, + "num_tokens": 92807378.0, + "step": 3586 + }, + { + "epoch": 0.39391609927520316, + "grad_norm": 2.315178394317627, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7054519057273865, + "num_tokens": 92832155.0, + "step": 3587 + }, + { + "epoch": 0.3940259169778168, + "grad_norm": 2.324922800064087, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7031739950180054, + "num_tokens": 92856290.0, + "step": 3588 + }, + { + "epoch": 0.3941357346804305, + "grad_norm": 2.2756707668304443, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6976761221885681, + "num_tokens": 92881250.0, + "step": 3589 + }, + { + "epoch": 0.39424555238304415, + "grad_norm": 2.3656866550445557, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6939530372619629, + "num_tokens": 92904024.0, + "step": 3590 + }, + { + "epoch": 0.3943553700856578, + "grad_norm": 2.0377261638641357, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7023268938064575, + "num_tokens": 92932387.0, + "step": 3591 + }, + { + "epoch": 0.39446518778827144, + "grad_norm": 2.1806766986846924, + "learning_rate": 1e-06, + "loss": 1.0571, + "mean_token_accuracy": 0.682860255241394, + "num_tokens": 92959209.0, + "step": 3592 + }, + { + "epoch": 0.39457500549088514, + "grad_norm": 2.689812660217285, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7164056897163391, + "num_tokens": 92977796.0, + "step": 3593 + }, + { + "epoch": 0.3946848231934988, + "grad_norm": 2.309030771255493, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7197588682174683, + "num_tokens": 93002954.0, + "step": 3594 + }, + { + "epoch": 0.39479464089611244, + "grad_norm": 2.0035152435302734, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6905473470687866, + "num_tokens": 93033062.0, + "step": 3595 + }, + { + "epoch": 0.39490445859872614, + "grad_norm": 2.690244674682617, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.69374680519104, + "num_tokens": 93059527.0, + "step": 3596 + }, + { + "epoch": 0.3950142763013398, + "grad_norm": 2.1607043743133545, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7171316146850586, + "num_tokens": 93085994.0, + "step": 3597 + }, + { + "epoch": 0.39512409400395343, + "grad_norm": 2.19693660736084, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6914833784103394, + "num_tokens": 93110793.0, + "step": 3598 + }, + { + "epoch": 0.3952339117065671, + "grad_norm": 2.1663942337036133, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7124947309494019, + "num_tokens": 93137040.0, + "step": 3599 + }, + { + "epoch": 0.3953437294091808, + "grad_norm": 2.3396289348602295, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7034991979598999, + "num_tokens": 93160651.0, + "step": 3600 + }, + { + "epoch": 0.3954535471117944, + "grad_norm": 2.0799710750579834, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7083456516265869, + "num_tokens": 93189203.0, + "step": 3601 + }, + { + "epoch": 0.39556336481440807, + "grad_norm": 2.2231132984161377, + "learning_rate": 1e-06, + "loss": 1.0666, + "mean_token_accuracy": 0.6839388608932495, + "num_tokens": 93215152.0, + "step": 3602 + }, + { + "epoch": 0.39567318251702177, + "grad_norm": 2.057919502258301, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7101924419403076, + "num_tokens": 93242481.0, + "step": 3603 + }, + { + "epoch": 0.3957830002196354, + "grad_norm": 1.9217840433120728, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6788734793663025, + "num_tokens": 93273852.0, + "step": 3604 + }, + { + "epoch": 0.39589281792224906, + "grad_norm": 2.2832157611846924, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.7003492712974548, + "num_tokens": 93297600.0, + "step": 3605 + }, + { + "epoch": 0.3960026356248627, + "grad_norm": 2.2056243419647217, + "learning_rate": 1e-06, + "loss": 1.1049, + "mean_token_accuracy": 0.6714211106300354, + "num_tokens": 93326205.0, + "step": 3606 + }, + { + "epoch": 0.3961124533274764, + "grad_norm": 2.162351131439209, + "learning_rate": 1e-06, + "loss": 1.0891, + "mean_token_accuracy": 0.6768156886100769, + "num_tokens": 93356719.0, + "step": 3607 + }, + { + "epoch": 0.39622227103009006, + "grad_norm": 2.3463730812072754, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7213855981826782, + "num_tokens": 93379681.0, + "step": 3608 + }, + { + "epoch": 0.3963320887327037, + "grad_norm": 2.1399271488189697, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.694731593132019, + "num_tokens": 93407884.0, + "step": 3609 + }, + { + "epoch": 0.39644190643531735, + "grad_norm": 2.1122353076934814, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7008212804794312, + "num_tokens": 93434797.0, + "step": 3610 + }, + { + "epoch": 0.39655172413793105, + "grad_norm": 2.107081651687622, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7006625533103943, + "num_tokens": 93461110.0, + "step": 3611 + }, + { + "epoch": 0.3966615418405447, + "grad_norm": 2.0349183082580566, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.694274365901947, + "num_tokens": 93493577.0, + "step": 3612 + }, + { + "epoch": 0.39677135954315834, + "grad_norm": 2.3786566257476807, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7103958129882812, + "num_tokens": 93516458.0, + "step": 3613 + }, + { + "epoch": 0.39688117724577204, + "grad_norm": 2.1212077140808105, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7128639817237854, + "num_tokens": 93543958.0, + "step": 3614 + }, + { + "epoch": 0.3969909949483857, + "grad_norm": 2.916733980178833, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7095600366592407, + "num_tokens": 93560899.0, + "step": 3615 + }, + { + "epoch": 0.39710081265099934, + "grad_norm": 2.5916943550109863, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7164835929870605, + "num_tokens": 93579658.0, + "step": 3616 + }, + { + "epoch": 0.397210630353613, + "grad_norm": 2.055234909057617, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7443205118179321, + "num_tokens": 93605222.0, + "step": 3617 + }, + { + "epoch": 0.3973204480562267, + "grad_norm": 2.239698886871338, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7055826187133789, + "num_tokens": 93631550.0, + "step": 3618 + }, + { + "epoch": 0.39743026575884033, + "grad_norm": 2.000429630279541, + "learning_rate": 1e-06, + "loss": 1.037, + "mean_token_accuracy": 0.691943883895874, + "num_tokens": 93661110.0, + "step": 3619 + }, + { + "epoch": 0.397540083461454, + "grad_norm": 2.617344379425049, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7045038342475891, + "num_tokens": 93680548.0, + "step": 3620 + }, + { + "epoch": 0.3976499011640676, + "grad_norm": 2.1019585132598877, + "learning_rate": 1e-06, + "loss": 1.1538, + "mean_token_accuracy": 0.6582701802253723, + "num_tokens": 93710057.0, + "step": 3621 + }, + { + "epoch": 0.3977597188666813, + "grad_norm": 2.0142714977264404, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6934882402420044, + "num_tokens": 93742099.0, + "step": 3622 + }, + { + "epoch": 0.39786953656929497, + "grad_norm": 2.120194911956787, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7239657044410706, + "num_tokens": 93767255.0, + "step": 3623 + }, + { + "epoch": 0.3979793542719086, + "grad_norm": 2.56862735748291, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7070105075836182, + "num_tokens": 93787179.0, + "step": 3624 + }, + { + "epoch": 0.3980891719745223, + "grad_norm": 1.9863886833190918, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7197585105895996, + "num_tokens": 93818656.0, + "step": 3625 + }, + { + "epoch": 0.39819898967713596, + "grad_norm": 2.284374237060547, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6979841589927673, + "num_tokens": 93842689.0, + "step": 3626 + }, + { + "epoch": 0.3983088073797496, + "grad_norm": 2.3631176948547363, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7031928300857544, + "num_tokens": 93866051.0, + "step": 3627 + }, + { + "epoch": 0.39841862508236325, + "grad_norm": 1.9925752878189087, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7108966112136841, + "num_tokens": 93894339.0, + "step": 3628 + }, + { + "epoch": 0.39852844278497696, + "grad_norm": 2.4708545207977295, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.6875523924827576, + "num_tokens": 93917128.0, + "step": 3629 + }, + { + "epoch": 0.3986382604875906, + "grad_norm": 2.452448606491089, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7201517820358276, + "num_tokens": 93937068.0, + "step": 3630 + }, + { + "epoch": 0.39874807819020425, + "grad_norm": 2.32833194732666, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7077181935310364, + "num_tokens": 93961825.0, + "step": 3631 + }, + { + "epoch": 0.39885789589281795, + "grad_norm": 2.4205238819122314, + "learning_rate": 1e-06, + "loss": 1.0777, + "mean_token_accuracy": 0.6750181913375854, + "num_tokens": 93986820.0, + "step": 3632 + }, + { + "epoch": 0.3989677135954316, + "grad_norm": 2.475037097930908, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6853063702583313, + "num_tokens": 94009183.0, + "step": 3633 + }, + { + "epoch": 0.39907753129804524, + "grad_norm": 2.3182942867279053, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7083616256713867, + "num_tokens": 94035169.0, + "step": 3634 + }, + { + "epoch": 0.3991873490006589, + "grad_norm": 2.180070400238037, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.701626181602478, + "num_tokens": 94058915.0, + "step": 3635 + }, + { + "epoch": 0.3992971667032726, + "grad_norm": 2.0950732231140137, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7049031257629395, + "num_tokens": 94087673.0, + "step": 3636 + }, + { + "epoch": 0.39940698440588623, + "grad_norm": 2.033928394317627, + "learning_rate": 1e-06, + "loss": 1.0724, + "mean_token_accuracy": 0.679751992225647, + "num_tokens": 94117111.0, + "step": 3637 + }, + { + "epoch": 0.3995168021084999, + "grad_norm": 2.342139959335327, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7128797769546509, + "num_tokens": 94142061.0, + "step": 3638 + }, + { + "epoch": 0.3996266198111135, + "grad_norm": 2.250293493270874, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7246742248535156, + "num_tokens": 94166473.0, + "step": 3639 + }, + { + "epoch": 0.3997364375137272, + "grad_norm": 1.9169938564300537, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7220585942268372, + "num_tokens": 94195339.0, + "step": 3640 + }, + { + "epoch": 0.3998462552163409, + "grad_norm": 2.3166449069976807, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7063034772872925, + "num_tokens": 94220333.0, + "step": 3641 + }, + { + "epoch": 0.3999560729189545, + "grad_norm": 2.0466134548187256, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7025748491287231, + "num_tokens": 94247191.0, + "step": 3642 + }, + { + "epoch": 0.4000658906215682, + "grad_norm": 2.269758939743042, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6814839243888855, + "num_tokens": 94272074.0, + "step": 3643 + }, + { + "epoch": 0.40017570832418187, + "grad_norm": 2.429934024810791, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7238715887069702, + "num_tokens": 94292812.0, + "step": 3644 + }, + { + "epoch": 0.4002855260267955, + "grad_norm": 2.408646583557129, + "learning_rate": 1e-06, + "loss": 1.0317, + "mean_token_accuracy": 0.6935325860977173, + "num_tokens": 94317623.0, + "step": 3645 + }, + { + "epoch": 0.40039534372940916, + "grad_norm": 2.0501296520233154, + "learning_rate": 1e-06, + "loss": 1.0811, + "mean_token_accuracy": 0.6925410032272339, + "num_tokens": 94348024.0, + "step": 3646 + }, + { + "epoch": 0.40050516143202286, + "grad_norm": 2.3934919834136963, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6971808671951294, + "num_tokens": 94370827.0, + "step": 3647 + }, + { + "epoch": 0.4006149791346365, + "grad_norm": 2.4517345428466797, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6870061159133911, + "num_tokens": 94395141.0, + "step": 3648 + }, + { + "epoch": 0.40072479683725015, + "grad_norm": 1.9089605808258057, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6884713172912598, + "num_tokens": 94427195.0, + "step": 3649 + }, + { + "epoch": 0.4008346145398638, + "grad_norm": 2.078667640686035, + "learning_rate": 1e-06, + "loss": 1.0503, + "mean_token_accuracy": 0.6849402785301208, + "num_tokens": 94456986.0, + "step": 3650 + }, + { + "epoch": 0.4009444322424775, + "grad_norm": 2.1795196533203125, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6939027309417725, + "num_tokens": 94486412.0, + "step": 3651 + }, + { + "epoch": 0.40105424994509115, + "grad_norm": 2.1916663646698, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7010031342506409, + "num_tokens": 94513341.0, + "step": 3652 + }, + { + "epoch": 0.4011640676477048, + "grad_norm": 2.292377233505249, + "learning_rate": 1e-06, + "loss": 1.0706, + "mean_token_accuracy": 0.6910340785980225, + "num_tokens": 94539325.0, + "step": 3653 + }, + { + "epoch": 0.4012738853503185, + "grad_norm": 2.1006686687469482, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6917765736579895, + "num_tokens": 94569341.0, + "step": 3654 + }, + { + "epoch": 0.40138370305293214, + "grad_norm": 2.3940839767456055, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7039269208908081, + "num_tokens": 94590125.0, + "step": 3655 + }, + { + "epoch": 0.4014935207555458, + "grad_norm": 2.2123448848724365, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6867184042930603, + "num_tokens": 94618712.0, + "step": 3656 + }, + { + "epoch": 0.40160333845815943, + "grad_norm": 2.3529670238494873, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7211942672729492, + "num_tokens": 94639634.0, + "step": 3657 + }, + { + "epoch": 0.40171315616077313, + "grad_norm": 2.2820370197296143, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6827238202095032, + "num_tokens": 94663874.0, + "step": 3658 + }, + { + "epoch": 0.4018229738633868, + "grad_norm": 2.145350456237793, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6873301863670349, + "num_tokens": 94692854.0, + "step": 3659 + }, + { + "epoch": 0.4019327915660004, + "grad_norm": 2.625131607055664, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7086825370788574, + "num_tokens": 94711951.0, + "step": 3660 + }, + { + "epoch": 0.4020426092686141, + "grad_norm": 2.265904188156128, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6987419128417969, + "num_tokens": 94738139.0, + "step": 3661 + }, + { + "epoch": 0.40215242697122777, + "grad_norm": 2.08150577545166, + "learning_rate": 1e-06, + "loss": 1.1139, + "mean_token_accuracy": 0.6696346998214722, + "num_tokens": 94769276.0, + "step": 3662 + }, + { + "epoch": 0.4022622446738414, + "grad_norm": 2.351816415786743, + "learning_rate": 1e-06, + "loss": 1.1054, + "mean_token_accuracy": 0.6725724935531616, + "num_tokens": 94793987.0, + "step": 3663 + }, + { + "epoch": 0.40237206237645506, + "grad_norm": 2.4113500118255615, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7069607973098755, + "num_tokens": 94815453.0, + "step": 3664 + }, + { + "epoch": 0.40248188007906877, + "grad_norm": 1.9819056987762451, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7163145542144775, + "num_tokens": 94842720.0, + "step": 3665 + }, + { + "epoch": 0.4025916977816824, + "grad_norm": 2.263232469558716, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6885538101196289, + "num_tokens": 94869987.0, + "step": 3666 + }, + { + "epoch": 0.40270151548429606, + "grad_norm": 2.5447535514831543, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7017427682876587, + "num_tokens": 94891386.0, + "step": 3667 + }, + { + "epoch": 0.4028113331869097, + "grad_norm": 2.301887273788452, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6973294019699097, + "num_tokens": 94915405.0, + "step": 3668 + }, + { + "epoch": 0.4029211508895234, + "grad_norm": 2.2304604053497314, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6990518569946289, + "num_tokens": 94940529.0, + "step": 3669 + }, + { + "epoch": 0.40303096859213705, + "grad_norm": 2.300412893295288, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7044541239738464, + "num_tokens": 94963891.0, + "step": 3670 + }, + { + "epoch": 0.4031407862947507, + "grad_norm": 2.4251821041107178, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6963751316070557, + "num_tokens": 94986480.0, + "step": 3671 + }, + { + "epoch": 0.4032506039973644, + "grad_norm": 2.2567806243896484, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6840552091598511, + "num_tokens": 95012838.0, + "step": 3672 + }, + { + "epoch": 0.40336042169997804, + "grad_norm": 2.3984270095825195, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6960139870643616, + "num_tokens": 95034266.0, + "step": 3673 + }, + { + "epoch": 0.4034702394025917, + "grad_norm": 2.449115753173828, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6904380917549133, + "num_tokens": 95055943.0, + "step": 3674 + }, + { + "epoch": 0.40358005710520534, + "grad_norm": 2.1865336894989014, + "learning_rate": 1e-06, + "loss": 1.0921, + "mean_token_accuracy": 0.6812910437583923, + "num_tokens": 95083037.0, + "step": 3675 + }, + { + "epoch": 0.40368987480781904, + "grad_norm": 2.2267255783081055, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.690196692943573, + "num_tokens": 95109674.0, + "step": 3676 + }, + { + "epoch": 0.4037996925104327, + "grad_norm": 2.5988376140594482, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.703020453453064, + "num_tokens": 95128170.0, + "step": 3677 + }, + { + "epoch": 0.40390951021304633, + "grad_norm": 2.0695128440856934, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6917877197265625, + "num_tokens": 95158536.0, + "step": 3678 + }, + { + "epoch": 0.40401932791566003, + "grad_norm": 1.9564272165298462, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.7058278322219849, + "num_tokens": 95188713.0, + "step": 3679 + }, + { + "epoch": 0.4041291456182737, + "grad_norm": 2.3117258548736572, + "learning_rate": 1e-06, + "loss": 1.0811, + "mean_token_accuracy": 0.6740509867668152, + "num_tokens": 95214405.0, + "step": 3680 + }, + { + "epoch": 0.4042389633208873, + "grad_norm": 1.9800721406936646, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7200382947921753, + "num_tokens": 95244165.0, + "step": 3681 + }, + { + "epoch": 0.40434878102350097, + "grad_norm": 2.2180798053741455, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.71561598777771, + "num_tokens": 95267477.0, + "step": 3682 + }, + { + "epoch": 0.40445859872611467, + "grad_norm": 2.1472318172454834, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6895561814308167, + "num_tokens": 95295044.0, + "step": 3683 + }, + { + "epoch": 0.4045684164287283, + "grad_norm": 2.3790173530578613, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6976538896560669, + "num_tokens": 95319298.0, + "step": 3684 + }, + { + "epoch": 0.40467823413134196, + "grad_norm": 2.2360429763793945, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7121343612670898, + "num_tokens": 95344447.0, + "step": 3685 + }, + { + "epoch": 0.4047880518339556, + "grad_norm": 2.5460846424102783, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.703261137008667, + "num_tokens": 95364424.0, + "step": 3686 + }, + { + "epoch": 0.4048978695365693, + "grad_norm": 2.4484031200408936, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6934525966644287, + "num_tokens": 95387166.0, + "step": 3687 + }, + { + "epoch": 0.40500768723918296, + "grad_norm": 2.190009593963623, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.691972553730011, + "num_tokens": 95414783.0, + "step": 3688 + }, + { + "epoch": 0.4051175049417966, + "grad_norm": 2.233776092529297, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.698595404624939, + "num_tokens": 95440817.0, + "step": 3689 + }, + { + "epoch": 0.4052273226444103, + "grad_norm": 2.4094581604003906, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7027693390846252, + "num_tokens": 95463800.0, + "step": 3690 + }, + { + "epoch": 0.40533714034702395, + "grad_norm": 1.9988882541656494, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7101327180862427, + "num_tokens": 95493112.0, + "step": 3691 + }, + { + "epoch": 0.4054469580496376, + "grad_norm": 2.129181385040283, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7075955867767334, + "num_tokens": 95519677.0, + "step": 3692 + }, + { + "epoch": 0.40555677575225124, + "grad_norm": 2.011685371398926, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6951509714126587, + "num_tokens": 95549291.0, + "step": 3693 + }, + { + "epoch": 0.40566659345486494, + "grad_norm": 2.2112503051757812, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7178111672401428, + "num_tokens": 95575323.0, + "step": 3694 + }, + { + "epoch": 0.4057764111574786, + "grad_norm": 2.1906187534332275, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7106038331985474, + "num_tokens": 95601929.0, + "step": 3695 + }, + { + "epoch": 0.40588622886009224, + "grad_norm": 2.0028183460235596, + "learning_rate": 1e-06, + "loss": 1.07, + "mean_token_accuracy": 0.683966338634491, + "num_tokens": 95632256.0, + "step": 3696 + }, + { + "epoch": 0.4059960465627059, + "grad_norm": 2.2093048095703125, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7125232815742493, + "num_tokens": 95657228.0, + "step": 3697 + }, + { + "epoch": 0.4061058642653196, + "grad_norm": 2.2007453441619873, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6818084716796875, + "num_tokens": 95683329.0, + "step": 3698 + }, + { + "epoch": 0.40621568196793323, + "grad_norm": 2.0846896171569824, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6923611760139465, + "num_tokens": 95712366.0, + "step": 3699 + }, + { + "epoch": 0.4063254996705469, + "grad_norm": 2.0259485244750977, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6921786665916443, + "num_tokens": 95741651.0, + "step": 3700 + }, + { + "epoch": 0.4064353173731606, + "grad_norm": 2.0023181438446045, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.7021015882492065, + "num_tokens": 95773832.0, + "step": 3701 + }, + { + "epoch": 0.4065451350757742, + "grad_norm": 2.152834892272949, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7034412622451782, + "num_tokens": 95801416.0, + "step": 3702 + }, + { + "epoch": 0.40665495277838787, + "grad_norm": 2.261134386062622, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7023491263389587, + "num_tokens": 95826058.0, + "step": 3703 + }, + { + "epoch": 0.4067647704810015, + "grad_norm": 1.920411229133606, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6909659504890442, + "num_tokens": 95859076.0, + "step": 3704 + }, + { + "epoch": 0.4068745881836152, + "grad_norm": 2.2293899059295654, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7041099667549133, + "num_tokens": 95884058.0, + "step": 3705 + }, + { + "epoch": 0.40698440588622886, + "grad_norm": 2.382540225982666, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7137662172317505, + "num_tokens": 95906494.0, + "step": 3706 + }, + { + "epoch": 0.4070942235888425, + "grad_norm": 2.0771353244781494, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6916465759277344, + "num_tokens": 95933524.0, + "step": 3707 + }, + { + "epoch": 0.4072040412914562, + "grad_norm": 2.2557926177978516, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7101011276245117, + "num_tokens": 95957723.0, + "step": 3708 + }, + { + "epoch": 0.40731385899406986, + "grad_norm": 2.3489460945129395, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7010425925254822, + "num_tokens": 95980974.0, + "step": 3709 + }, + { + "epoch": 0.4074236766966835, + "grad_norm": 2.06077241897583, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6992042064666748, + "num_tokens": 96010658.0, + "step": 3710 + }, + { + "epoch": 0.40753349439929715, + "grad_norm": 2.2707602977752686, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6812319159507751, + "num_tokens": 96036358.0, + "step": 3711 + }, + { + "epoch": 0.40764331210191085, + "grad_norm": 2.1389613151550293, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6955463886260986, + "num_tokens": 96062798.0, + "step": 3712 + }, + { + "epoch": 0.4077531298045245, + "grad_norm": 2.137240409851074, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6903754472732544, + "num_tokens": 96089971.0, + "step": 3713 + }, + { + "epoch": 0.40786294750713814, + "grad_norm": 2.168363332748413, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.689679741859436, + "num_tokens": 96116747.0, + "step": 3714 + }, + { + "epoch": 0.4079727652097518, + "grad_norm": 2.27048397064209, + "learning_rate": 1e-06, + "loss": 1.0391, + "mean_token_accuracy": 0.6880025863647461, + "num_tokens": 96142612.0, + "step": 3715 + }, + { + "epoch": 0.4080825829123655, + "grad_norm": 2.391396999359131, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6929678916931152, + "num_tokens": 96167683.0, + "step": 3716 + }, + { + "epoch": 0.40819240061497913, + "grad_norm": 2.1015079021453857, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7130012512207031, + "num_tokens": 96195221.0, + "step": 3717 + }, + { + "epoch": 0.4083022183175928, + "grad_norm": 2.1615774631500244, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7146197557449341, + "num_tokens": 96221480.0, + "step": 3718 + }, + { + "epoch": 0.4084120360202065, + "grad_norm": 2.5088753700256348, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.7000661492347717, + "num_tokens": 96242839.0, + "step": 3719 + }, + { + "epoch": 0.4085218537228201, + "grad_norm": 2.5223913192749023, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6919190287590027, + "num_tokens": 96264354.0, + "step": 3720 + }, + { + "epoch": 0.4086316714254338, + "grad_norm": 2.05765438079834, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7166048884391785, + "num_tokens": 96292974.0, + "step": 3721 + }, + { + "epoch": 0.4087414891280474, + "grad_norm": 2.3376150131225586, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7059814929962158, + "num_tokens": 96317311.0, + "step": 3722 + }, + { + "epoch": 0.4088513068306611, + "grad_norm": 2.2028913497924805, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6772358417510986, + "num_tokens": 96344098.0, + "step": 3723 + }, + { + "epoch": 0.40896112453327477, + "grad_norm": 2.272993326187134, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7002846002578735, + "num_tokens": 96368361.0, + "step": 3724 + }, + { + "epoch": 0.4090709422358884, + "grad_norm": 2.165058135986328, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7110922336578369, + "num_tokens": 96394469.0, + "step": 3725 + }, + { + "epoch": 0.40918075993850206, + "grad_norm": 2.242241859436035, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6927359104156494, + "num_tokens": 96419691.0, + "step": 3726 + }, + { + "epoch": 0.40929057764111576, + "grad_norm": 1.9248454570770264, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6991702914237976, + "num_tokens": 96453773.0, + "step": 3727 + }, + { + "epoch": 0.4094003953437294, + "grad_norm": 2.2064049243927, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7220849394798279, + "num_tokens": 96478387.0, + "step": 3728 + }, + { + "epoch": 0.40951021304634305, + "grad_norm": 2.2028653621673584, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6966758370399475, + "num_tokens": 96504213.0, + "step": 3729 + }, + { + "epoch": 0.40962003074895675, + "grad_norm": 2.0090479850769043, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6872072219848633, + "num_tokens": 96536107.0, + "step": 3730 + }, + { + "epoch": 0.4097298484515704, + "grad_norm": 2.2029435634613037, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.7105945348739624, + "num_tokens": 96561775.0, + "step": 3731 + }, + { + "epoch": 0.40983966615418405, + "grad_norm": 2.328278064727783, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7179547548294067, + "num_tokens": 96584036.0, + "step": 3732 + }, + { + "epoch": 0.4099494838567977, + "grad_norm": 2.4544105529785156, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7161283493041992, + "num_tokens": 96605831.0, + "step": 3733 + }, + { + "epoch": 0.4100593015594114, + "grad_norm": 2.4681804180145264, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7049484252929688, + "num_tokens": 96627335.0, + "step": 3734 + }, + { + "epoch": 0.41016911926202504, + "grad_norm": 2.1670243740081787, + "learning_rate": 1e-06, + "loss": 1.0686, + "mean_token_accuracy": 0.6898136734962463, + "num_tokens": 96655776.0, + "step": 3735 + }, + { + "epoch": 0.4102789369646387, + "grad_norm": 2.0082805156707764, + "learning_rate": 1e-06, + "loss": 1.0883, + "mean_token_accuracy": 0.685325026512146, + "num_tokens": 96686329.0, + "step": 3736 + }, + { + "epoch": 0.4103887546672524, + "grad_norm": 1.914404034614563, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7120888233184814, + "num_tokens": 96717402.0, + "step": 3737 + }, + { + "epoch": 0.41049857236986603, + "grad_norm": 2.1064577102661133, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7139080166816711, + "num_tokens": 96743834.0, + "step": 3738 + }, + { + "epoch": 0.4106083900724797, + "grad_norm": 2.3254547119140625, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.6979199647903442, + "num_tokens": 96766567.0, + "step": 3739 + }, + { + "epoch": 0.4107182077750933, + "grad_norm": 2.1957919597625732, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6930885910987854, + "num_tokens": 96793264.0, + "step": 3740 + }, + { + "epoch": 0.410828025477707, + "grad_norm": 2.00056529045105, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6919491291046143, + "num_tokens": 96825333.0, + "step": 3741 + }, + { + "epoch": 0.41093784318032067, + "grad_norm": 2.023847818374634, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7216097116470337, + "num_tokens": 96853811.0, + "step": 3742 + }, + { + "epoch": 0.4110476608829343, + "grad_norm": 2.278690814971924, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7018188834190369, + "num_tokens": 96878822.0, + "step": 3743 + }, + { + "epoch": 0.41115747858554796, + "grad_norm": 2.428807258605957, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7294391393661499, + "num_tokens": 96901200.0, + "step": 3744 + }, + { + "epoch": 0.41126729628816167, + "grad_norm": 2.533019542694092, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7017804980278015, + "num_tokens": 96920484.0, + "step": 3745 + }, + { + "epoch": 0.4113771139907753, + "grad_norm": 2.2924258708953857, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7040239572525024, + "num_tokens": 96944095.0, + "step": 3746 + }, + { + "epoch": 0.41148693169338896, + "grad_norm": 2.5432333946228027, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7025008201599121, + "num_tokens": 96964442.0, + "step": 3747 + }, + { + "epoch": 0.41159674939600266, + "grad_norm": 2.061984062194824, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.687529444694519, + "num_tokens": 96994310.0, + "step": 3748 + }, + { + "epoch": 0.4117065670986163, + "grad_norm": 2.5093111991882324, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7160277366638184, + "num_tokens": 97015086.0, + "step": 3749 + }, + { + "epoch": 0.41181638480122995, + "grad_norm": 2.4554526805877686, + "learning_rate": 1e-06, + "loss": 1.0663, + "mean_token_accuracy": 0.6852394342422485, + "num_tokens": 97038406.0, + "step": 3750 + }, + { + "epoch": 0.4119262025038436, + "grad_norm": 2.271260976791382, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6925705671310425, + "num_tokens": 97063504.0, + "step": 3751 + }, + { + "epoch": 0.4120360202064573, + "grad_norm": 2.453193426132202, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.6783565282821655, + "num_tokens": 97087001.0, + "step": 3752 + }, + { + "epoch": 0.41214583790907094, + "grad_norm": 2.0681936740875244, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7113053202629089, + "num_tokens": 97114967.0, + "step": 3753 + }, + { + "epoch": 0.4122556556116846, + "grad_norm": 2.3846535682678223, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7213618159294128, + "num_tokens": 97137484.0, + "step": 3754 + }, + { + "epoch": 0.4123654733142983, + "grad_norm": 2.128566265106201, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7029466032981873, + "num_tokens": 97164287.0, + "step": 3755 + }, + { + "epoch": 0.41247529101691194, + "grad_norm": 2.5152359008789062, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7234547138214111, + "num_tokens": 97185153.0, + "step": 3756 + }, + { + "epoch": 0.4125851087195256, + "grad_norm": 2.152691602706909, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.697319746017456, + "num_tokens": 97211905.0, + "step": 3757 + }, + { + "epoch": 0.41269492642213923, + "grad_norm": 2.193249464035034, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6997021436691284, + "num_tokens": 97238603.0, + "step": 3758 + }, + { + "epoch": 0.41280474412475293, + "grad_norm": 1.9022221565246582, + "learning_rate": 1e-06, + "loss": 1.0642, + "mean_token_accuracy": 0.6787528991699219, + "num_tokens": 97271073.0, + "step": 3759 + }, + { + "epoch": 0.4129145618273666, + "grad_norm": 2.207866668701172, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7003555297851562, + "num_tokens": 97297350.0, + "step": 3760 + }, + { + "epoch": 0.4130243795299802, + "grad_norm": 2.221794605255127, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7053728103637695, + "num_tokens": 97321488.0, + "step": 3761 + }, + { + "epoch": 0.41313419723259387, + "grad_norm": 2.1161861419677734, + "learning_rate": 1e-06, + "loss": 1.1036, + "mean_token_accuracy": 0.6812911033630371, + "num_tokens": 97349133.0, + "step": 3762 + }, + { + "epoch": 0.41324401493520757, + "grad_norm": 2.2510547637939453, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6990430355072021, + "num_tokens": 97373337.0, + "step": 3763 + }, + { + "epoch": 0.4133538326378212, + "grad_norm": 2.531446933746338, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7340801358222961, + "num_tokens": 97393305.0, + "step": 3764 + }, + { + "epoch": 0.41346365034043486, + "grad_norm": 2.1741549968719482, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7004194259643555, + "num_tokens": 97420834.0, + "step": 3765 + }, + { + "epoch": 0.41357346804304856, + "grad_norm": 2.2224323749542236, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6957688927650452, + "num_tokens": 97447875.0, + "step": 3766 + }, + { + "epoch": 0.4136832857456622, + "grad_norm": 2.4536516666412354, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7175174355506897, + "num_tokens": 97468715.0, + "step": 3767 + }, + { + "epoch": 0.41379310344827586, + "grad_norm": 2.405783176422119, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6992431879043579, + "num_tokens": 97491436.0, + "step": 3768 + }, + { + "epoch": 0.4139029211508895, + "grad_norm": 2.072450637817383, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6920711994171143, + "num_tokens": 97521771.0, + "step": 3769 + }, + { + "epoch": 0.4140127388535032, + "grad_norm": 2.204660415649414, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6976031064987183, + "num_tokens": 97548123.0, + "step": 3770 + }, + { + "epoch": 0.41412255655611685, + "grad_norm": 2.351646661758423, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7191941142082214, + "num_tokens": 97569764.0, + "step": 3771 + }, + { + "epoch": 0.4142323742587305, + "grad_norm": 2.074136972427368, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.6995628476142883, + "num_tokens": 97598522.0, + "step": 3772 + }, + { + "epoch": 0.41434219196134414, + "grad_norm": 2.1820576190948486, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7351526021957397, + "num_tokens": 97623658.0, + "step": 3773 + }, + { + "epoch": 0.41445200966395784, + "grad_norm": 1.9636880159378052, + "learning_rate": 1e-06, + "loss": 1.0568, + "mean_token_accuracy": 0.6879091262817383, + "num_tokens": 97654071.0, + "step": 3774 + }, + { + "epoch": 0.4145618273665715, + "grad_norm": 2.3204259872436523, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6956573724746704, + "num_tokens": 97677831.0, + "step": 3775 + }, + { + "epoch": 0.41467164506918514, + "grad_norm": 2.3440723419189453, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6862350702285767, + "num_tokens": 97702503.0, + "step": 3776 + }, + { + "epoch": 0.41478146277179884, + "grad_norm": 2.4559645652770996, + "learning_rate": 1e-06, + "loss": 1.0454, + "mean_token_accuracy": 0.6891589164733887, + "num_tokens": 97724802.0, + "step": 3777 + }, + { + "epoch": 0.4148912804744125, + "grad_norm": 2.350109338760376, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7129555940628052, + "num_tokens": 97749234.0, + "step": 3778 + }, + { + "epoch": 0.41500109817702613, + "grad_norm": 2.341597080230713, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7131034135818481, + "num_tokens": 97770698.0, + "step": 3779 + }, + { + "epoch": 0.4151109158796398, + "grad_norm": 2.2621419429779053, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7007254958152771, + "num_tokens": 97794172.0, + "step": 3780 + }, + { + "epoch": 0.4152207335822535, + "grad_norm": 2.4758992195129395, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.7023563385009766, + "num_tokens": 97815066.0, + "step": 3781 + }, + { + "epoch": 0.4153305512848671, + "grad_norm": 2.0877068042755127, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7269567251205444, + "num_tokens": 97841199.0, + "step": 3782 + }, + { + "epoch": 0.41544036898748077, + "grad_norm": 2.378342628479004, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6930016279220581, + "num_tokens": 97865472.0, + "step": 3783 + }, + { + "epoch": 0.41555018669009447, + "grad_norm": 2.40442156791687, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.712236762046814, + "num_tokens": 97886053.0, + "step": 3784 + }, + { + "epoch": 0.4156600043927081, + "grad_norm": 2.617809534072876, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7044147253036499, + "num_tokens": 97908195.0, + "step": 3785 + }, + { + "epoch": 0.41576982209532176, + "grad_norm": 2.242358684539795, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7079549431800842, + "num_tokens": 97935081.0, + "step": 3786 + }, + { + "epoch": 0.4158796397979354, + "grad_norm": 1.9791830778121948, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6794906258583069, + "num_tokens": 97966325.0, + "step": 3787 + }, + { + "epoch": 0.4159894575005491, + "grad_norm": 2.3897342681884766, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6995355486869812, + "num_tokens": 97991457.0, + "step": 3788 + }, + { + "epoch": 0.41609927520316276, + "grad_norm": 2.222947120666504, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.687820315361023, + "num_tokens": 98018341.0, + "step": 3789 + }, + { + "epoch": 0.4162090929057764, + "grad_norm": 2.128082036972046, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7075653076171875, + "num_tokens": 98045098.0, + "step": 3790 + }, + { + "epoch": 0.41631891060839005, + "grad_norm": 2.3390774726867676, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6849387288093567, + "num_tokens": 98069357.0, + "step": 3791 + }, + { + "epoch": 0.41642872831100375, + "grad_norm": 2.3024725914001465, + "learning_rate": 1e-06, + "loss": 1.0695, + "mean_token_accuracy": 0.6819208860397339, + "num_tokens": 98094848.0, + "step": 3792 + }, + { + "epoch": 0.4165385460136174, + "grad_norm": 2.3794825077056885, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6863790154457092, + "num_tokens": 98118179.0, + "step": 3793 + }, + { + "epoch": 0.41664836371623104, + "grad_norm": 2.2894937992095947, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6961486339569092, + "num_tokens": 98142211.0, + "step": 3794 + }, + { + "epoch": 0.41675818141884474, + "grad_norm": 2.2519116401672363, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6964176893234253, + "num_tokens": 98168425.0, + "step": 3795 + }, + { + "epoch": 0.4168679991214584, + "grad_norm": 2.4270784854888916, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7132683992385864, + "num_tokens": 98191140.0, + "step": 3796 + }, + { + "epoch": 0.41697781682407203, + "grad_norm": 2.020420789718628, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7119210958480835, + "num_tokens": 98221268.0, + "step": 3797 + }, + { + "epoch": 0.4170876345266857, + "grad_norm": 2.4126195907592773, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7127431035041809, + "num_tokens": 98243642.0, + "step": 3798 + }, + { + "epoch": 0.4171974522292994, + "grad_norm": 2.3072149753570557, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7310106754302979, + "num_tokens": 98265808.0, + "step": 3799 + }, + { + "epoch": 0.417307269931913, + "grad_norm": 2.050508499145508, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7114778757095337, + "num_tokens": 98293589.0, + "step": 3800 + }, + { + "epoch": 0.4174170876345267, + "grad_norm": 2.149808406829834, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6950573921203613, + "num_tokens": 98321281.0, + "step": 3801 + }, + { + "epoch": 0.4175269053371403, + "grad_norm": 2.184945583343506, + "learning_rate": 1e-06, + "loss": 1.0725, + "mean_token_accuracy": 0.6780463457107544, + "num_tokens": 98347302.0, + "step": 3802 + }, + { + "epoch": 0.417636723039754, + "grad_norm": 2.1802594661712646, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.6775192618370056, + "num_tokens": 98375310.0, + "step": 3803 + }, + { + "epoch": 0.41774654074236767, + "grad_norm": 1.9028608798980713, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7164052724838257, + "num_tokens": 98405962.0, + "step": 3804 + }, + { + "epoch": 0.4178563584449813, + "grad_norm": 2.2828829288482666, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7351672649383545, + "num_tokens": 98429344.0, + "step": 3805 + }, + { + "epoch": 0.417966176147595, + "grad_norm": 2.2161998748779297, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7030481100082397, + "num_tokens": 98455018.0, + "step": 3806 + }, + { + "epoch": 0.41807599385020866, + "grad_norm": 2.1007208824157715, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.7088344693183899, + "num_tokens": 98484221.0, + "step": 3807 + }, + { + "epoch": 0.4181858115528223, + "grad_norm": 1.9909087419509888, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6968576908111572, + "num_tokens": 98515513.0, + "step": 3808 + }, + { + "epoch": 0.41829562925543595, + "grad_norm": 2.433438539505005, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7150940895080566, + "num_tokens": 98537078.0, + "step": 3809 + }, + { + "epoch": 0.41840544695804965, + "grad_norm": 2.511526107788086, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7062598466873169, + "num_tokens": 98556703.0, + "step": 3810 + }, + { + "epoch": 0.4185152646606633, + "grad_norm": 2.104620933532715, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6896935701370239, + "num_tokens": 98586006.0, + "step": 3811 + }, + { + "epoch": 0.41862508236327695, + "grad_norm": 2.345823049545288, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6836295127868652, + "num_tokens": 98609590.0, + "step": 3812 + }, + { + "epoch": 0.41873490006589065, + "grad_norm": 2.340775489807129, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7061625123023987, + "num_tokens": 98632955.0, + "step": 3813 + }, + { + "epoch": 0.4188447177685043, + "grad_norm": 2.486255168914795, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.6997776627540588, + "num_tokens": 98654276.0, + "step": 3814 + }, + { + "epoch": 0.41895453547111794, + "grad_norm": 2.425178050994873, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7022057771682739, + "num_tokens": 98676588.0, + "step": 3815 + }, + { + "epoch": 0.4190643531737316, + "grad_norm": 2.097905397415161, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6939023733139038, + "num_tokens": 98704836.0, + "step": 3816 + }, + { + "epoch": 0.4191741708763453, + "grad_norm": 2.123450517654419, + "learning_rate": 1e-06, + "loss": 1.1001, + "mean_token_accuracy": 0.6711176633834839, + "num_tokens": 98736500.0, + "step": 3817 + }, + { + "epoch": 0.41928398857895893, + "grad_norm": 2.512317657470703, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6991147994995117, + "num_tokens": 98758217.0, + "step": 3818 + }, + { + "epoch": 0.4193938062815726, + "grad_norm": 2.238482713699341, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6808948516845703, + "num_tokens": 98782642.0, + "step": 3819 + }, + { + "epoch": 0.4195036239841862, + "grad_norm": 2.1169722080230713, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6994658708572388, + "num_tokens": 98809715.0, + "step": 3820 + }, + { + "epoch": 0.4196134416867999, + "grad_norm": 2.317883014678955, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7086127400398254, + "num_tokens": 98832396.0, + "step": 3821 + }, + { + "epoch": 0.41972325938941357, + "grad_norm": 2.375232696533203, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.7070720195770264, + "num_tokens": 98855032.0, + "step": 3822 + }, + { + "epoch": 0.4198330770920272, + "grad_norm": 2.134732961654663, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.684306263923645, + "num_tokens": 98885502.0, + "step": 3823 + }, + { + "epoch": 0.4199428947946409, + "grad_norm": 2.270250082015991, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6832031011581421, + "num_tokens": 98910833.0, + "step": 3824 + }, + { + "epoch": 0.42005271249725457, + "grad_norm": 2.1552581787109375, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.678155243396759, + "num_tokens": 98939922.0, + "step": 3825 + }, + { + "epoch": 0.4201625301998682, + "grad_norm": 2.216794729232788, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6976593732833862, + "num_tokens": 98966447.0, + "step": 3826 + }, + { + "epoch": 0.42027234790248186, + "grad_norm": 2.1953957080841064, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7016400694847107, + "num_tokens": 98990069.0, + "step": 3827 + }, + { + "epoch": 0.42038216560509556, + "grad_norm": 2.0023481845855713, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7124853730201721, + "num_tokens": 99018779.0, + "step": 3828 + }, + { + "epoch": 0.4204919833077092, + "grad_norm": 2.1544737815856934, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6834605932235718, + "num_tokens": 99046989.0, + "step": 3829 + }, + { + "epoch": 0.42060180101032285, + "grad_norm": 2.2000746726989746, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6909981966018677, + "num_tokens": 99075830.0, + "step": 3830 + }, + { + "epoch": 0.42071161871293655, + "grad_norm": 2.216564655303955, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6916559338569641, + "num_tokens": 99104043.0, + "step": 3831 + }, + { + "epoch": 0.4208214364155502, + "grad_norm": 2.067950487136841, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6919552087783813, + "num_tokens": 99134589.0, + "step": 3832 + }, + { + "epoch": 0.42093125411816384, + "grad_norm": 2.461747646331787, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6892185807228088, + "num_tokens": 99159009.0, + "step": 3833 + }, + { + "epoch": 0.4210410718207775, + "grad_norm": 2.114555835723877, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6885339021682739, + "num_tokens": 99188690.0, + "step": 3834 + }, + { + "epoch": 0.4211508895233912, + "grad_norm": 2.4430813789367676, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7206463813781738, + "num_tokens": 99209626.0, + "step": 3835 + }, + { + "epoch": 0.42126070722600484, + "grad_norm": 2.1486666202545166, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6926552057266235, + "num_tokens": 99237195.0, + "step": 3836 + }, + { + "epoch": 0.4213705249286185, + "grad_norm": 2.1000301837921143, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7172437906265259, + "num_tokens": 99265505.0, + "step": 3837 + }, + { + "epoch": 0.42148034263123213, + "grad_norm": 2.179396390914917, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7173235416412354, + "num_tokens": 99290547.0, + "step": 3838 + }, + { + "epoch": 0.42159016033384583, + "grad_norm": 2.360332727432251, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7040722370147705, + "num_tokens": 99313701.0, + "step": 3839 + }, + { + "epoch": 0.4216999780364595, + "grad_norm": 2.11175537109375, + "learning_rate": 1e-06, + "loss": 1.0763, + "mean_token_accuracy": 0.6778095960617065, + "num_tokens": 99344109.0, + "step": 3840 + }, + { + "epoch": 0.4218097957390731, + "grad_norm": 2.245110511779785, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7079733610153198, + "num_tokens": 99370893.0, + "step": 3841 + }, + { + "epoch": 0.4219196134416868, + "grad_norm": 2.2917490005493164, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7024689316749573, + "num_tokens": 99393832.0, + "step": 3842 + }, + { + "epoch": 0.42202943114430047, + "grad_norm": 2.4089608192443848, + "learning_rate": 1e-06, + "loss": 1.1187, + "mean_token_accuracy": 0.6663526892662048, + "num_tokens": 99418060.0, + "step": 3843 + }, + { + "epoch": 0.4221392488469141, + "grad_norm": 2.324911594390869, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7099337577819824, + "num_tokens": 99441684.0, + "step": 3844 + }, + { + "epoch": 0.42224906654952776, + "grad_norm": 2.1224076747894287, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7042201161384583, + "num_tokens": 99469380.0, + "step": 3845 + }, + { + "epoch": 0.42235888425214146, + "grad_norm": 2.7864630222320557, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7289431095123291, + "num_tokens": 99487015.0, + "step": 3846 + }, + { + "epoch": 0.4224687019547551, + "grad_norm": 2.265072822570801, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6902819871902466, + "num_tokens": 99513084.0, + "step": 3847 + }, + { + "epoch": 0.42257851965736876, + "grad_norm": 2.305586099624634, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6905900239944458, + "num_tokens": 99538790.0, + "step": 3848 + }, + { + "epoch": 0.4226883373599824, + "grad_norm": 2.209294557571411, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7123368382453918, + "num_tokens": 99563439.0, + "step": 3849 + }, + { + "epoch": 0.4227981550625961, + "grad_norm": 2.297370195388794, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7255086302757263, + "num_tokens": 99586655.0, + "step": 3850 + }, + { + "epoch": 0.42290797276520975, + "grad_norm": 2.2645156383514404, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.6851929426193237, + "num_tokens": 99611237.0, + "step": 3851 + }, + { + "epoch": 0.4230177904678234, + "grad_norm": 1.9466922283172607, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6944467425346375, + "num_tokens": 99645810.0, + "step": 3852 + }, + { + "epoch": 0.4231276081704371, + "grad_norm": 2.6303558349609375, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.699356198310852, + "num_tokens": 99664956.0, + "step": 3853 + }, + { + "epoch": 0.42323742587305074, + "grad_norm": 2.037339210510254, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7172360420227051, + "num_tokens": 99693325.0, + "step": 3854 + }, + { + "epoch": 0.4233472435756644, + "grad_norm": 2.4359688758850098, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7185778617858887, + "num_tokens": 99715867.0, + "step": 3855 + }, + { + "epoch": 0.42345706127827804, + "grad_norm": 2.217420816421509, + "learning_rate": 1e-06, + "loss": 1.1004, + "mean_token_accuracy": 0.6731748580932617, + "num_tokens": 99745782.0, + "step": 3856 + }, + { + "epoch": 0.42356687898089174, + "grad_norm": 2.3853719234466553, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6839874982833862, + "num_tokens": 99768440.0, + "step": 3857 + }, + { + "epoch": 0.4236766966835054, + "grad_norm": 2.5102579593658447, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6940685510635376, + "num_tokens": 99789971.0, + "step": 3858 + }, + { + "epoch": 0.42378651438611903, + "grad_norm": 2.3808186054229736, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6992372274398804, + "num_tokens": 99813570.0, + "step": 3859 + }, + { + "epoch": 0.42389633208873273, + "grad_norm": 2.4161295890808105, + "learning_rate": 1e-06, + "loss": 1.0513, + "mean_token_accuracy": 0.6872460246086121, + "num_tokens": 99837657.0, + "step": 3860 + }, + { + "epoch": 0.4240061497913464, + "grad_norm": 2.0964791774749756, + "learning_rate": 1e-06, + "loss": 1.1135, + "mean_token_accuracy": 0.666741669178009, + "num_tokens": 99868728.0, + "step": 3861 + }, + { + "epoch": 0.42411596749396, + "grad_norm": 2.214771032333374, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.6829276084899902, + "num_tokens": 99896969.0, + "step": 3862 + }, + { + "epoch": 0.42422578519657367, + "grad_norm": 2.376530885696411, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7100889682769775, + "num_tokens": 99919296.0, + "step": 3863 + }, + { + "epoch": 0.42433560289918737, + "grad_norm": 2.503135919570923, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7097131013870239, + "num_tokens": 99941534.0, + "step": 3864 + }, + { + "epoch": 0.424445420601801, + "grad_norm": 2.1891090869903564, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7070846557617188, + "num_tokens": 99968505.0, + "step": 3865 + }, + { + "epoch": 0.42455523830441466, + "grad_norm": 2.1481454372406006, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7315942049026489, + "num_tokens": 99993455.0, + "step": 3866 + }, + { + "epoch": 0.4246650560070283, + "grad_norm": 2.3486270904541016, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.7001744508743286, + "num_tokens": 100018287.0, + "step": 3867 + }, + { + "epoch": 0.424774873709642, + "grad_norm": 2.4939770698547363, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7100327610969543, + "num_tokens": 100040541.0, + "step": 3868 + }, + { + "epoch": 0.42488469141225566, + "grad_norm": 2.127152919769287, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7178745269775391, + "num_tokens": 100066442.0, + "step": 3869 + }, + { + "epoch": 0.4249945091148693, + "grad_norm": 2.365659475326538, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7501087188720703, + "num_tokens": 100086709.0, + "step": 3870 + }, + { + "epoch": 0.425104326817483, + "grad_norm": 2.3313376903533936, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.713001012802124, + "num_tokens": 100110542.0, + "step": 3871 + }, + { + "epoch": 0.42521414452009665, + "grad_norm": 2.009547233581543, + "learning_rate": 1e-06, + "loss": 1.1043, + "mean_token_accuracy": 0.6737298965454102, + "num_tokens": 100141320.0, + "step": 3872 + }, + { + "epoch": 0.4253239622227103, + "grad_norm": 2.3975424766540527, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.6860806941986084, + "num_tokens": 100164223.0, + "step": 3873 + }, + { + "epoch": 0.42543377992532394, + "grad_norm": 2.4186275005340576, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7126951813697815, + "num_tokens": 100185941.0, + "step": 3874 + }, + { + "epoch": 0.42554359762793764, + "grad_norm": 2.33866548538208, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7110280394554138, + "num_tokens": 100210732.0, + "step": 3875 + }, + { + "epoch": 0.4256534153305513, + "grad_norm": 2.149138927459717, + "learning_rate": 1e-06, + "loss": 1.0645, + "mean_token_accuracy": 0.6827512979507446, + "num_tokens": 100238137.0, + "step": 3876 + }, + { + "epoch": 0.42576323303316493, + "grad_norm": 2.150479793548584, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6913440227508545, + "num_tokens": 100264186.0, + "step": 3877 + }, + { + "epoch": 0.4258730507357786, + "grad_norm": 1.9708802700042725, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6988574266433716, + "num_tokens": 100294331.0, + "step": 3878 + }, + { + "epoch": 0.4259828684383923, + "grad_norm": 2.289970874786377, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7339788675308228, + "num_tokens": 100316130.0, + "step": 3879 + }, + { + "epoch": 0.4260926861410059, + "grad_norm": 2.3329732418060303, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.701467752456665, + "num_tokens": 100340052.0, + "step": 3880 + }, + { + "epoch": 0.4262025038436196, + "grad_norm": 2.419417381286621, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6992363333702087, + "num_tokens": 100365026.0, + "step": 3881 + }, + { + "epoch": 0.4263123215462333, + "grad_norm": 2.3841965198516846, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6957459449768066, + "num_tokens": 100386885.0, + "step": 3882 + }, + { + "epoch": 0.4264221392488469, + "grad_norm": 2.222270965576172, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6825243234634399, + "num_tokens": 100413857.0, + "step": 3883 + }, + { + "epoch": 0.42653195695146057, + "grad_norm": 1.8319604396820068, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7023136615753174, + "num_tokens": 100448443.0, + "step": 3884 + }, + { + "epoch": 0.4266417746540742, + "grad_norm": 2.204359292984009, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7288875579833984, + "num_tokens": 100473036.0, + "step": 3885 + }, + { + "epoch": 0.4267515923566879, + "grad_norm": 2.579150438308716, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6993693113327026, + "num_tokens": 100492808.0, + "step": 3886 + }, + { + "epoch": 0.42686141005930156, + "grad_norm": 2.3012919425964355, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.6868700981140137, + "num_tokens": 100516357.0, + "step": 3887 + }, + { + "epoch": 0.4269712277619152, + "grad_norm": 2.2604570388793945, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7099575996398926, + "num_tokens": 100539907.0, + "step": 3888 + }, + { + "epoch": 0.4270810454645289, + "grad_norm": 1.9241604804992676, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7151005268096924, + "num_tokens": 100570127.0, + "step": 3889 + }, + { + "epoch": 0.42719086316714255, + "grad_norm": 2.772357702255249, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7213947772979736, + "num_tokens": 100587307.0, + "step": 3890 + }, + { + "epoch": 0.4273006808697562, + "grad_norm": 2.407881021499634, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6963195204734802, + "num_tokens": 100610619.0, + "step": 3891 + }, + { + "epoch": 0.42741049857236985, + "grad_norm": 2.1246626377105713, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6948632001876831, + "num_tokens": 100639567.0, + "step": 3892 + }, + { + "epoch": 0.42752031627498355, + "grad_norm": 2.4793474674224854, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7047716379165649, + "num_tokens": 100659524.0, + "step": 3893 + }, + { + "epoch": 0.4276301339775972, + "grad_norm": 2.3588368892669678, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7091763019561768, + "num_tokens": 100681819.0, + "step": 3894 + }, + { + "epoch": 0.42773995168021084, + "grad_norm": 1.9289988279342651, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.7016409039497375, + "num_tokens": 100715905.0, + "step": 3895 + }, + { + "epoch": 0.4278497693828245, + "grad_norm": 1.986120343208313, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6947259902954102, + "num_tokens": 100751228.0, + "step": 3896 + }, + { + "epoch": 0.4279595870854382, + "grad_norm": 2.4362010955810547, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7031340003013611, + "num_tokens": 100773363.0, + "step": 3897 + }, + { + "epoch": 0.42806940478805183, + "grad_norm": 2.040515899658203, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6886497139930725, + "num_tokens": 100801727.0, + "step": 3898 + }, + { + "epoch": 0.4281792224906655, + "grad_norm": 1.9252558946609497, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7044162750244141, + "num_tokens": 100834173.0, + "step": 3899 + }, + { + "epoch": 0.4282890401932792, + "grad_norm": 2.522646188735962, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7017699480056763, + "num_tokens": 100855435.0, + "step": 3900 + }, + { + "epoch": 0.4283988578958928, + "grad_norm": 2.0638787746429443, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.691765308380127, + "num_tokens": 100884327.0, + "step": 3901 + }, + { + "epoch": 0.42850867559850647, + "grad_norm": 2.1853744983673096, + "learning_rate": 1e-06, + "loss": 1.0592, + "mean_token_accuracy": 0.6775804162025452, + "num_tokens": 100910211.0, + "step": 3902 + }, + { + "epoch": 0.4286184933011201, + "grad_norm": 2.2771308422088623, + "learning_rate": 1e-06, + "loss": 1.0613, + "mean_token_accuracy": 0.6846479773521423, + "num_tokens": 100935401.0, + "step": 3903 + }, + { + "epoch": 0.4287283110037338, + "grad_norm": 2.3206560611724854, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.713766872882843, + "num_tokens": 100958877.0, + "step": 3904 + }, + { + "epoch": 0.42883812870634747, + "grad_norm": 2.309223175048828, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7121866941452026, + "num_tokens": 100980312.0, + "step": 3905 + }, + { + "epoch": 0.4289479464089611, + "grad_norm": 2.00077748298645, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7181680202484131, + "num_tokens": 101008596.0, + "step": 3906 + }, + { + "epoch": 0.4290577641115748, + "grad_norm": 2.174680233001709, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7187559604644775, + "num_tokens": 101033996.0, + "step": 3907 + }, + { + "epoch": 0.42916758181418846, + "grad_norm": 2.2694761753082275, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7063357830047607, + "num_tokens": 101057963.0, + "step": 3908 + }, + { + "epoch": 0.4292773995168021, + "grad_norm": 2.6199898719787598, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.695752739906311, + "num_tokens": 101077804.0, + "step": 3909 + }, + { + "epoch": 0.42938721721941575, + "grad_norm": 2.564558506011963, + "learning_rate": 1e-06, + "loss": 1.0724, + "mean_token_accuracy": 0.686935305595398, + "num_tokens": 101100214.0, + "step": 3910 + }, + { + "epoch": 0.42949703492202945, + "grad_norm": 2.3375444412231445, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7273873686790466, + "num_tokens": 101124305.0, + "step": 3911 + }, + { + "epoch": 0.4296068526246431, + "grad_norm": 2.1502034664154053, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7210531234741211, + "num_tokens": 101150289.0, + "step": 3912 + }, + { + "epoch": 0.42971667032725674, + "grad_norm": 2.2058639526367188, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6913051605224609, + "num_tokens": 101178339.0, + "step": 3913 + }, + { + "epoch": 0.4298264880298704, + "grad_norm": 2.3044543266296387, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7142668962478638, + "num_tokens": 101202614.0, + "step": 3914 + }, + { + "epoch": 0.4299363057324841, + "grad_norm": 2.3149008750915527, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7219653129577637, + "num_tokens": 101224528.0, + "step": 3915 + }, + { + "epoch": 0.43004612343509774, + "grad_norm": 2.1089584827423096, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.73250412940979, + "num_tokens": 101249037.0, + "step": 3916 + }, + { + "epoch": 0.4301559411377114, + "grad_norm": 1.9784214496612549, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6958545446395874, + "num_tokens": 101282444.0, + "step": 3917 + }, + { + "epoch": 0.4302657588403251, + "grad_norm": 2.1179609298706055, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7079271674156189, + "num_tokens": 101313358.0, + "step": 3918 + }, + { + "epoch": 0.43037557654293873, + "grad_norm": 2.083448886871338, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7010596990585327, + "num_tokens": 101340581.0, + "step": 3919 + }, + { + "epoch": 0.4304853942455524, + "grad_norm": 2.4964003562927246, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.718083918094635, + "num_tokens": 101361533.0, + "step": 3920 + }, + { + "epoch": 0.430595211948166, + "grad_norm": 2.39035701751709, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7101325988769531, + "num_tokens": 101383326.0, + "step": 3921 + }, + { + "epoch": 0.4307050296507797, + "grad_norm": 2.276090145111084, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.7053691148757935, + "num_tokens": 101409081.0, + "step": 3922 + }, + { + "epoch": 0.43081484735339337, + "grad_norm": 2.092381000518799, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6980422735214233, + "num_tokens": 101438699.0, + "step": 3923 + }, + { + "epoch": 0.430924665056007, + "grad_norm": 2.0856330394744873, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7148017287254333, + "num_tokens": 101464864.0, + "step": 3924 + }, + { + "epoch": 0.43103448275862066, + "grad_norm": 2.4066224098205566, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.688666820526123, + "num_tokens": 101488078.0, + "step": 3925 + }, + { + "epoch": 0.43114430046123436, + "grad_norm": 2.2736775875091553, + "learning_rate": 1e-06, + "loss": 1.0873, + "mean_token_accuracy": 0.6817991733551025, + "num_tokens": 101514050.0, + "step": 3926 + }, + { + "epoch": 0.431254118163848, + "grad_norm": 2.4071152210235596, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7227698564529419, + "num_tokens": 101536723.0, + "step": 3927 + }, + { + "epoch": 0.43136393586646166, + "grad_norm": 2.2780864238739014, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6894123554229736, + "num_tokens": 101561970.0, + "step": 3928 + }, + { + "epoch": 0.43147375356907536, + "grad_norm": 2.089731216430664, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7041501402854919, + "num_tokens": 101589583.0, + "step": 3929 + }, + { + "epoch": 0.431583571271689, + "grad_norm": 2.0198991298675537, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.6831966638565063, + "num_tokens": 101621673.0, + "step": 3930 + }, + { + "epoch": 0.43169338897430265, + "grad_norm": 2.4245383739471436, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6925047636032104, + "num_tokens": 101644065.0, + "step": 3931 + }, + { + "epoch": 0.4318032066769163, + "grad_norm": 2.9154133796691895, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7306327819824219, + "num_tokens": 101662750.0, + "step": 3932 + }, + { + "epoch": 0.43191302437953, + "grad_norm": 2.1128735542297363, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7091822624206543, + "num_tokens": 101688990.0, + "step": 3933 + }, + { + "epoch": 0.43202284208214364, + "grad_norm": 2.434335470199585, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6975769996643066, + "num_tokens": 101710686.0, + "step": 3934 + }, + { + "epoch": 0.4321326597847573, + "grad_norm": 2.2793073654174805, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6902844905853271, + "num_tokens": 101737899.0, + "step": 3935 + }, + { + "epoch": 0.432242477487371, + "grad_norm": 2.1559202671051025, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6840397715568542, + "num_tokens": 101766454.0, + "step": 3936 + }, + { + "epoch": 0.43235229518998464, + "grad_norm": 2.247702121734619, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7148640155792236, + "num_tokens": 101791654.0, + "step": 3937 + }, + { + "epoch": 0.4324621128925983, + "grad_norm": 2.200687885284424, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6992028951644897, + "num_tokens": 101817589.0, + "step": 3938 + }, + { + "epoch": 0.43257193059521193, + "grad_norm": 2.014894485473633, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6934955716133118, + "num_tokens": 101847859.0, + "step": 3939 + }, + { + "epoch": 0.43268174829782563, + "grad_norm": 2.2962701320648193, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7006490230560303, + "num_tokens": 101871930.0, + "step": 3940 + }, + { + "epoch": 0.4327915660004393, + "grad_norm": 2.0874013900756836, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7050756216049194, + "num_tokens": 101898638.0, + "step": 3941 + }, + { + "epoch": 0.4329013837030529, + "grad_norm": 2.2146530151367188, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6926062107086182, + "num_tokens": 101926998.0, + "step": 3942 + }, + { + "epoch": 0.43301120140566657, + "grad_norm": 2.472550630569458, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6917107105255127, + "num_tokens": 101949123.0, + "step": 3943 + }, + { + "epoch": 0.43312101910828027, + "grad_norm": 2.3361339569091797, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7149654626846313, + "num_tokens": 101971407.0, + "step": 3944 + }, + { + "epoch": 0.4332308368108939, + "grad_norm": 2.195319890975952, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.705136775970459, + "num_tokens": 101997943.0, + "step": 3945 + }, + { + "epoch": 0.43334065451350756, + "grad_norm": 2.0680935382843018, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.707706093788147, + "num_tokens": 102025822.0, + "step": 3946 + }, + { + "epoch": 0.43345047221612126, + "grad_norm": 2.0889012813568115, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6899824738502502, + "num_tokens": 102053305.0, + "step": 3947 + }, + { + "epoch": 0.4335602899187349, + "grad_norm": 2.2225706577301025, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.703635036945343, + "num_tokens": 102078836.0, + "step": 3948 + }, + { + "epoch": 0.43367010762134856, + "grad_norm": 2.1929306983947754, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7225438952445984, + "num_tokens": 102103001.0, + "step": 3949 + }, + { + "epoch": 0.4337799253239622, + "grad_norm": 2.160184383392334, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6840825080871582, + "num_tokens": 102131895.0, + "step": 3950 + }, + { + "epoch": 0.4338897430265759, + "grad_norm": 2.612569808959961, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7206809520721436, + "num_tokens": 102156023.0, + "step": 3951 + }, + { + "epoch": 0.43399956072918955, + "grad_norm": 2.640720844268799, + "learning_rate": 1e-06, + "loss": 1.064, + "mean_token_accuracy": 0.6870484352111816, + "num_tokens": 102176947.0, + "step": 3952 + }, + { + "epoch": 0.4341093784318032, + "grad_norm": 2.24932861328125, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6984884738922119, + "num_tokens": 102203157.0, + "step": 3953 + }, + { + "epoch": 0.43421919613441684, + "grad_norm": 2.4227006435394287, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7013438940048218, + "num_tokens": 102226106.0, + "step": 3954 + }, + { + "epoch": 0.43432901383703054, + "grad_norm": 2.3884665966033936, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6944533586502075, + "num_tokens": 102250220.0, + "step": 3955 + }, + { + "epoch": 0.4344388315396442, + "grad_norm": 2.2218868732452393, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.6787387132644653, + "num_tokens": 102276373.0, + "step": 3956 + }, + { + "epoch": 0.43454864924225783, + "grad_norm": 2.355048418045044, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7092238664627075, + "num_tokens": 102298062.0, + "step": 3957 + }, + { + "epoch": 0.43465846694487154, + "grad_norm": 2.1950433254241943, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6999787092208862, + "num_tokens": 102324012.0, + "step": 3958 + }, + { + "epoch": 0.4347682846474852, + "grad_norm": 2.163672924041748, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6869017481803894, + "num_tokens": 102352815.0, + "step": 3959 + }, + { + "epoch": 0.4348781023500988, + "grad_norm": 2.2406795024871826, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6995242834091187, + "num_tokens": 102378014.0, + "step": 3960 + }, + { + "epoch": 0.4349879200527125, + "grad_norm": 2.240297794342041, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7291990518569946, + "num_tokens": 102402223.0, + "step": 3961 + }, + { + "epoch": 0.4350977377553262, + "grad_norm": 2.085850954055786, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7356022000312805, + "num_tokens": 102429827.0, + "step": 3962 + }, + { + "epoch": 0.4352075554579398, + "grad_norm": 2.0537619590759277, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6958494782447815, + "num_tokens": 102458769.0, + "step": 3963 + }, + { + "epoch": 0.43531737316055347, + "grad_norm": 2.074856758117676, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.698002815246582, + "num_tokens": 102487677.0, + "step": 3964 + }, + { + "epoch": 0.43542719086316717, + "grad_norm": 2.1211507320404053, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.7055408954620361, + "num_tokens": 102516194.0, + "step": 3965 + }, + { + "epoch": 0.4355370085657808, + "grad_norm": 2.6199209690093994, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7212020754814148, + "num_tokens": 102534115.0, + "step": 3966 + }, + { + "epoch": 0.43564682626839446, + "grad_norm": 2.354282855987549, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7151418924331665, + "num_tokens": 102557009.0, + "step": 3967 + }, + { + "epoch": 0.4357566439710081, + "grad_norm": 2.185746669769287, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6969698071479797, + "num_tokens": 102581438.0, + "step": 3968 + }, + { + "epoch": 0.4358664616736218, + "grad_norm": 2.3767318725585938, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7179255485534668, + "num_tokens": 102603822.0, + "step": 3969 + }, + { + "epoch": 0.43597627937623545, + "grad_norm": 2.3603360652923584, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7063208818435669, + "num_tokens": 102625582.0, + "step": 3970 + }, + { + "epoch": 0.4360860970788491, + "grad_norm": 2.173682689666748, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6887279152870178, + "num_tokens": 102655237.0, + "step": 3971 + }, + { + "epoch": 0.43619591478146275, + "grad_norm": 2.188720703125, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7071218490600586, + "num_tokens": 102681610.0, + "step": 3972 + }, + { + "epoch": 0.43630573248407645, + "grad_norm": 2.0735316276550293, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6823537349700928, + "num_tokens": 102711636.0, + "step": 3973 + }, + { + "epoch": 0.4364155501866901, + "grad_norm": 2.2530221939086914, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6807178258895874, + "num_tokens": 102740269.0, + "step": 3974 + }, + { + "epoch": 0.43652536788930374, + "grad_norm": 2.2182421684265137, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7293705940246582, + "num_tokens": 102765140.0, + "step": 3975 + }, + { + "epoch": 0.43663518559191744, + "grad_norm": 2.391786575317383, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6943376064300537, + "num_tokens": 102790550.0, + "step": 3976 + }, + { + "epoch": 0.4367450032945311, + "grad_norm": 2.133288860321045, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6926085352897644, + "num_tokens": 102818283.0, + "step": 3977 + }, + { + "epoch": 0.43685482099714473, + "grad_norm": 2.4328198432922363, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6978839635848999, + "num_tokens": 102839697.0, + "step": 3978 + }, + { + "epoch": 0.4369646386997584, + "grad_norm": 2.073482036590576, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6926337480545044, + "num_tokens": 102868038.0, + "step": 3979 + }, + { + "epoch": 0.4370744564023721, + "grad_norm": 2.421949625015259, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6968101263046265, + "num_tokens": 102890261.0, + "step": 3980 + }, + { + "epoch": 0.4371842741049857, + "grad_norm": 2.510390520095825, + "learning_rate": 1e-06, + "loss": 1.0462, + "mean_token_accuracy": 0.6846044063568115, + "num_tokens": 102912602.0, + "step": 3981 + }, + { + "epoch": 0.43729409180759937, + "grad_norm": 2.2888801097869873, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7192433476448059, + "num_tokens": 102936652.0, + "step": 3982 + }, + { + "epoch": 0.4374039095102131, + "grad_norm": 2.068164825439453, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6803573369979858, + "num_tokens": 102964564.0, + "step": 3983 + }, + { + "epoch": 0.4375137272128267, + "grad_norm": 2.207702875137329, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.723520040512085, + "num_tokens": 102991026.0, + "step": 3984 + }, + { + "epoch": 0.43762354491544037, + "grad_norm": 2.188507080078125, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6948637962341309, + "num_tokens": 103017666.0, + "step": 3985 + }, + { + "epoch": 0.437733362618054, + "grad_norm": 2.5119380950927734, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.719221830368042, + "num_tokens": 103038445.0, + "step": 3986 + }, + { + "epoch": 0.4378431803206677, + "grad_norm": 2.101364850997925, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6864023804664612, + "num_tokens": 103069019.0, + "step": 3987 + }, + { + "epoch": 0.43795299802328136, + "grad_norm": 2.508281707763672, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7081664800643921, + "num_tokens": 103090281.0, + "step": 3988 + }, + { + "epoch": 0.438062815725895, + "grad_norm": 2.039738655090332, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7315297722816467, + "num_tokens": 103118962.0, + "step": 3989 + }, + { + "epoch": 0.43817263342850865, + "grad_norm": 2.16422963142395, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.718408465385437, + "num_tokens": 103145401.0, + "step": 3990 + }, + { + "epoch": 0.43828245113112235, + "grad_norm": 2.1709136962890625, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.7013653516769409, + "num_tokens": 103172448.0, + "step": 3991 + }, + { + "epoch": 0.438392268833736, + "grad_norm": 2.286451816558838, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7070038914680481, + "num_tokens": 103198006.0, + "step": 3992 + }, + { + "epoch": 0.43850208653634964, + "grad_norm": 2.2981784343719482, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7079724669456482, + "num_tokens": 103222454.0, + "step": 3993 + }, + { + "epoch": 0.43861190423896335, + "grad_norm": 2.2074732780456543, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.709843635559082, + "num_tokens": 103249089.0, + "step": 3994 + }, + { + "epoch": 0.438721721941577, + "grad_norm": 2.395794153213501, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.729600191116333, + "num_tokens": 103270197.0, + "step": 3995 + }, + { + "epoch": 0.43883153964419064, + "grad_norm": 2.160689353942871, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.6901775002479553, + "num_tokens": 103297499.0, + "step": 3996 + }, + { + "epoch": 0.4389413573468043, + "grad_norm": 2.1799962520599365, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7153385877609253, + "num_tokens": 103324855.0, + "step": 3997 + }, + { + "epoch": 0.439051175049418, + "grad_norm": 2.172480583190918, + "learning_rate": 1e-06, + "loss": 1.0943, + "mean_token_accuracy": 0.6814744472503662, + "num_tokens": 103352855.0, + "step": 3998 + }, + { + "epoch": 0.43916099275203163, + "grad_norm": 2.2762973308563232, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.6986400485038757, + "num_tokens": 103375974.0, + "step": 3999 + }, + { + "epoch": 0.4392708104546453, + "grad_norm": 1.9920597076416016, + "learning_rate": 1e-06, + "loss": 1.103, + "mean_token_accuracy": 0.667326807975769, + "num_tokens": 103407738.0, + "step": 4000 + }, + { + "epoch": 0.4393806281572589, + "grad_norm": 2.0855135917663574, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6960691213607788, + "num_tokens": 103437027.0, + "step": 4001 + }, + { + "epoch": 0.4394904458598726, + "grad_norm": 2.342715263366699, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7060871124267578, + "num_tokens": 103460046.0, + "step": 4002 + }, + { + "epoch": 0.43960026356248627, + "grad_norm": 2.2479541301727295, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7044873237609863, + "num_tokens": 103485621.0, + "step": 4003 + }, + { + "epoch": 0.4397100812650999, + "grad_norm": 1.897674798965454, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6851648092269897, + "num_tokens": 103518984.0, + "step": 4004 + }, + { + "epoch": 0.4398198989677136, + "grad_norm": 2.427664041519165, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6932374238967896, + "num_tokens": 103541595.0, + "step": 4005 + }, + { + "epoch": 0.43992971667032726, + "grad_norm": 2.3378183841705322, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6900876760482788, + "num_tokens": 103565995.0, + "step": 4006 + }, + { + "epoch": 0.4400395343729409, + "grad_norm": 2.451307535171509, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6901692748069763, + "num_tokens": 103589961.0, + "step": 4007 + }, + { + "epoch": 0.44014935207555456, + "grad_norm": 1.9436975717544556, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.6875158548355103, + "num_tokens": 103624015.0, + "step": 4008 + }, + { + "epoch": 0.44025916977816826, + "grad_norm": 2.1890954971313477, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7171458005905151, + "num_tokens": 103648981.0, + "step": 4009 + }, + { + "epoch": 0.4403689874807819, + "grad_norm": 2.314235210418701, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7141579985618591, + "num_tokens": 103672992.0, + "step": 4010 + }, + { + "epoch": 0.44047880518339555, + "grad_norm": 2.477855682373047, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7107111215591431, + "num_tokens": 103693983.0, + "step": 4011 + }, + { + "epoch": 0.44058862288600925, + "grad_norm": 2.313319206237793, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6870108842849731, + "num_tokens": 103717413.0, + "step": 4012 + }, + { + "epoch": 0.4406984405886229, + "grad_norm": 2.4298863410949707, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.706445574760437, + "num_tokens": 103739793.0, + "step": 4013 + }, + { + "epoch": 0.44080825829123654, + "grad_norm": 2.315690040588379, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7131863832473755, + "num_tokens": 103763425.0, + "step": 4014 + }, + { + "epoch": 0.4409180759938502, + "grad_norm": 2.4823403358459473, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7270753979682922, + "num_tokens": 103784225.0, + "step": 4015 + }, + { + "epoch": 0.4410278936964639, + "grad_norm": 2.1107497215270996, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7054880261421204, + "num_tokens": 103811751.0, + "step": 4016 + }, + { + "epoch": 0.44113771139907754, + "grad_norm": 1.9380637407302856, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6922887563705444, + "num_tokens": 103843323.0, + "step": 4017 + }, + { + "epoch": 0.4412475291016912, + "grad_norm": 2.2522382736206055, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.7104579210281372, + "num_tokens": 103869122.0, + "step": 4018 + }, + { + "epoch": 0.44135734680430483, + "grad_norm": 2.2699601650238037, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6916559934616089, + "num_tokens": 103892813.0, + "step": 4019 + }, + { + "epoch": 0.44146716450691853, + "grad_norm": 1.8972212076187134, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6914312839508057, + "num_tokens": 103925498.0, + "step": 4020 + }, + { + "epoch": 0.4415769822095322, + "grad_norm": 2.316833972930908, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7153345346450806, + "num_tokens": 103948200.0, + "step": 4021 + }, + { + "epoch": 0.4416867999121458, + "grad_norm": 2.4827094078063965, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7312266826629639, + "num_tokens": 103969306.0, + "step": 4022 + }, + { + "epoch": 0.4417966176147595, + "grad_norm": 2.341756582260132, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7105696201324463, + "num_tokens": 103992620.0, + "step": 4023 + }, + { + "epoch": 0.44190643531737317, + "grad_norm": 2.1459598541259766, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.6975867748260498, + "num_tokens": 104022069.0, + "step": 4024 + }, + { + "epoch": 0.4420162530199868, + "grad_norm": 2.090446710586548, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6976785659790039, + "num_tokens": 104052384.0, + "step": 4025 + }, + { + "epoch": 0.44212607072260046, + "grad_norm": 2.14335036277771, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.6861473917961121, + "num_tokens": 104080332.0, + "step": 4026 + }, + { + "epoch": 0.44223588842521416, + "grad_norm": 2.182913064956665, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6921471953392029, + "num_tokens": 104106260.0, + "step": 4027 + }, + { + "epoch": 0.4423457061278278, + "grad_norm": 1.9685205221176147, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6968406438827515, + "num_tokens": 104135474.0, + "step": 4028 + }, + { + "epoch": 0.44245552383044146, + "grad_norm": 2.370481014251709, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.702822208404541, + "num_tokens": 104157966.0, + "step": 4029 + }, + { + "epoch": 0.4425653415330551, + "grad_norm": 2.4457077980041504, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6938090920448303, + "num_tokens": 104180901.0, + "step": 4030 + }, + { + "epoch": 0.4426751592356688, + "grad_norm": 2.0873641967773438, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7025642395019531, + "num_tokens": 104210427.0, + "step": 4031 + }, + { + "epoch": 0.44278497693828245, + "grad_norm": 2.017348527908325, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7114397883415222, + "num_tokens": 104240781.0, + "step": 4032 + }, + { + "epoch": 0.4428947946408961, + "grad_norm": 2.2637929916381836, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6990566253662109, + "num_tokens": 104265444.0, + "step": 4033 + }, + { + "epoch": 0.4430046123435098, + "grad_norm": 2.2990217208862305, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7100487947463989, + "num_tokens": 104289851.0, + "step": 4034 + }, + { + "epoch": 0.44311443004612344, + "grad_norm": 2.087270498275757, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7077721953392029, + "num_tokens": 104318477.0, + "step": 4035 + }, + { + "epoch": 0.4432242477487371, + "grad_norm": 2.225114345550537, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6890085339546204, + "num_tokens": 104344467.0, + "step": 4036 + }, + { + "epoch": 0.44333406545135073, + "grad_norm": 2.274620532989502, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6880538463592529, + "num_tokens": 104370130.0, + "step": 4037 + }, + { + "epoch": 0.44344388315396444, + "grad_norm": 2.273543357849121, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7230509519577026, + "num_tokens": 104393844.0, + "step": 4038 + }, + { + "epoch": 0.4435537008565781, + "grad_norm": 2.5871472358703613, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6985810995101929, + "num_tokens": 104413570.0, + "step": 4039 + }, + { + "epoch": 0.4436635185591917, + "grad_norm": 2.139371395111084, + "learning_rate": 1e-06, + "loss": 1.1055, + "mean_token_accuracy": 0.6687197089195251, + "num_tokens": 104443227.0, + "step": 4040 + }, + { + "epoch": 0.44377333626180543, + "grad_norm": 2.2973382472991943, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7048619985580444, + "num_tokens": 104467348.0, + "step": 4041 + }, + { + "epoch": 0.4438831539644191, + "grad_norm": 2.46431040763855, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7172635197639465, + "num_tokens": 104487199.0, + "step": 4042 + }, + { + "epoch": 0.4439929716670327, + "grad_norm": 2.4085495471954346, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7037094235420227, + "num_tokens": 104508389.0, + "step": 4043 + }, + { + "epoch": 0.44410278936964637, + "grad_norm": 1.9730520248413086, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7091149687767029, + "num_tokens": 104541411.0, + "step": 4044 + }, + { + "epoch": 0.44421260707226007, + "grad_norm": 2.089252233505249, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7142115831375122, + "num_tokens": 104568153.0, + "step": 4045 + }, + { + "epoch": 0.4443224247748737, + "grad_norm": 2.423664093017578, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6927661299705505, + "num_tokens": 104591936.0, + "step": 4046 + }, + { + "epoch": 0.44443224247748736, + "grad_norm": 2.1475064754486084, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6852209568023682, + "num_tokens": 104621841.0, + "step": 4047 + }, + { + "epoch": 0.444542060180101, + "grad_norm": 2.1538078784942627, + "learning_rate": 1e-06, + "loss": 1.0454, + "mean_token_accuracy": 0.6982542276382446, + "num_tokens": 104651647.0, + "step": 4048 + }, + { + "epoch": 0.4446518778827147, + "grad_norm": 2.393803834915161, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6886874437332153, + "num_tokens": 104677572.0, + "step": 4049 + }, + { + "epoch": 0.44476169558532835, + "grad_norm": 2.1011786460876465, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7165212035179138, + "num_tokens": 104706092.0, + "step": 4050 + }, + { + "epoch": 0.444871513287942, + "grad_norm": 2.2441494464874268, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.6852942705154419, + "num_tokens": 104732079.0, + "step": 4051 + }, + { + "epoch": 0.4449813309905557, + "grad_norm": 2.1743228435516357, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6886596083641052, + "num_tokens": 104760767.0, + "step": 4052 + }, + { + "epoch": 0.44509114869316935, + "grad_norm": 2.20461106300354, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7163894176483154, + "num_tokens": 104788430.0, + "step": 4053 + }, + { + "epoch": 0.445200966395783, + "grad_norm": 2.2190375328063965, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6905397176742554, + "num_tokens": 104817342.0, + "step": 4054 + }, + { + "epoch": 0.44531078409839664, + "grad_norm": 2.1625137329101562, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.6990331411361694, + "num_tokens": 104843385.0, + "step": 4055 + }, + { + "epoch": 0.44542060180101034, + "grad_norm": 2.242837905883789, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6769596338272095, + "num_tokens": 104868452.0, + "step": 4056 + }, + { + "epoch": 0.445530419503624, + "grad_norm": 2.331239938735962, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7228463292121887, + "num_tokens": 104890186.0, + "step": 4057 + }, + { + "epoch": 0.44564023720623763, + "grad_norm": 2.2677161693573, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6935672163963318, + "num_tokens": 104914247.0, + "step": 4058 + }, + { + "epoch": 0.44575005490885133, + "grad_norm": 2.18086576461792, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6946853995323181, + "num_tokens": 104939335.0, + "step": 4059 + }, + { + "epoch": 0.445859872611465, + "grad_norm": 2.3314502239227295, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7109223008155823, + "num_tokens": 104962635.0, + "step": 4060 + }, + { + "epoch": 0.4459696903140786, + "grad_norm": 2.3986289501190186, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.726628303527832, + "num_tokens": 104983584.0, + "step": 4061 + }, + { + "epoch": 0.44607950801669227, + "grad_norm": 2.421945333480835, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7217813730239868, + "num_tokens": 105004819.0, + "step": 4062 + }, + { + "epoch": 0.446189325719306, + "grad_norm": 2.3680665493011475, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7337489128112793, + "num_tokens": 105026420.0, + "step": 4063 + }, + { + "epoch": 0.4462991434219196, + "grad_norm": 2.391547679901123, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7046544551849365, + "num_tokens": 105048663.0, + "step": 4064 + }, + { + "epoch": 0.44640896112453327, + "grad_norm": 2.4861366748809814, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7041992545127869, + "num_tokens": 105072117.0, + "step": 4065 + }, + { + "epoch": 0.4465187788271469, + "grad_norm": 2.1909728050231934, + "learning_rate": 1e-06, + "loss": 1.1139, + "mean_token_accuracy": 0.6725166440010071, + "num_tokens": 105100997.0, + "step": 4066 + }, + { + "epoch": 0.4466285965297606, + "grad_norm": 2.215353012084961, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6913211941719055, + "num_tokens": 105126987.0, + "step": 4067 + }, + { + "epoch": 0.44673841423237426, + "grad_norm": 2.292633295059204, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7102680206298828, + "num_tokens": 105151037.0, + "step": 4068 + }, + { + "epoch": 0.4468482319349879, + "grad_norm": 2.160472869873047, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.6817511320114136, + "num_tokens": 105179052.0, + "step": 4069 + }, + { + "epoch": 0.4469580496376016, + "grad_norm": 2.2989397048950195, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7115604877471924, + "num_tokens": 105202762.0, + "step": 4070 + }, + { + "epoch": 0.44706786734021525, + "grad_norm": 2.0873143672943115, + "learning_rate": 1e-06, + "loss": 1.0609, + "mean_token_accuracy": 0.6816939115524292, + "num_tokens": 105231743.0, + "step": 4071 + }, + { + "epoch": 0.4471776850428289, + "grad_norm": 2.1389520168304443, + "learning_rate": 1e-06, + "loss": 1.0646, + "mean_token_accuracy": 0.6764951348304749, + "num_tokens": 105258755.0, + "step": 4072 + }, + { + "epoch": 0.44728750274544254, + "grad_norm": 2.30210280418396, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6933633089065552, + "num_tokens": 105283134.0, + "step": 4073 + }, + { + "epoch": 0.44739732044805625, + "grad_norm": 2.5859827995300293, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7338504791259766, + "num_tokens": 105302217.0, + "step": 4074 + }, + { + "epoch": 0.4475071381506699, + "grad_norm": 2.3853540420532227, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7011389136314392, + "num_tokens": 105324669.0, + "step": 4075 + }, + { + "epoch": 0.44761695585328354, + "grad_norm": 2.170178174972534, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6895331144332886, + "num_tokens": 105351150.0, + "step": 4076 + }, + { + "epoch": 0.4477267735558972, + "grad_norm": 1.8918755054473877, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7010406255722046, + "num_tokens": 105382344.0, + "step": 4077 + }, + { + "epoch": 0.4478365912585109, + "grad_norm": 1.9791173934936523, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7149966359138489, + "num_tokens": 105412565.0, + "step": 4078 + }, + { + "epoch": 0.44794640896112453, + "grad_norm": 2.4482858180999756, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6991035342216492, + "num_tokens": 105434373.0, + "step": 4079 + }, + { + "epoch": 0.4480562266637382, + "grad_norm": 2.374363422393799, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6933048963546753, + "num_tokens": 105457771.0, + "step": 4080 + }, + { + "epoch": 0.4481660443663519, + "grad_norm": 2.0604283809661865, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7193623781204224, + "num_tokens": 105483652.0, + "step": 4081 + }, + { + "epoch": 0.4482758620689655, + "grad_norm": 2.3442413806915283, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7214113473892212, + "num_tokens": 105506506.0, + "step": 4082 + }, + { + "epoch": 0.44838567977157917, + "grad_norm": 2.0259015560150146, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7144917845726013, + "num_tokens": 105535039.0, + "step": 4083 + }, + { + "epoch": 0.4484954974741928, + "grad_norm": 2.1339879035949707, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6972880363464355, + "num_tokens": 105561456.0, + "step": 4084 + }, + { + "epoch": 0.4486053151768065, + "grad_norm": 2.2242884635925293, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6840795278549194, + "num_tokens": 105587627.0, + "step": 4085 + }, + { + "epoch": 0.44871513287942016, + "grad_norm": 2.3241465091705322, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7070241570472717, + "num_tokens": 105613040.0, + "step": 4086 + }, + { + "epoch": 0.4488249505820338, + "grad_norm": 2.2131588459014893, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.7011896967887878, + "num_tokens": 105638821.0, + "step": 4087 + }, + { + "epoch": 0.4489347682846475, + "grad_norm": 2.3573217391967773, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6997249722480774, + "num_tokens": 105661894.0, + "step": 4088 + }, + { + "epoch": 0.44904458598726116, + "grad_norm": 1.8619545698165894, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6914661526679993, + "num_tokens": 105697736.0, + "step": 4089 + }, + { + "epoch": 0.4491544036898748, + "grad_norm": 2.4148313999176025, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.724562406539917, + "num_tokens": 105717630.0, + "step": 4090 + }, + { + "epoch": 0.44926422139248845, + "grad_norm": 2.355598211288452, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7073959708213806, + "num_tokens": 105739780.0, + "step": 4091 + }, + { + "epoch": 0.44937403909510215, + "grad_norm": 2.1186532974243164, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7154091596603394, + "num_tokens": 105767513.0, + "step": 4092 + }, + { + "epoch": 0.4494838567977158, + "grad_norm": 2.2305150032043457, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7002838253974915, + "num_tokens": 105793045.0, + "step": 4093 + }, + { + "epoch": 0.44959367450032944, + "grad_norm": 2.1332671642303467, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.69580078125, + "num_tokens": 105818510.0, + "step": 4094 + }, + { + "epoch": 0.4497034922029431, + "grad_norm": 2.6216373443603516, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7103164792060852, + "num_tokens": 105836643.0, + "step": 4095 + }, + { + "epoch": 0.4498133099055568, + "grad_norm": 1.9785407781600952, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6850390434265137, + "num_tokens": 105865818.0, + "step": 4096 + }, + { + "epoch": 0.44992312760817044, + "grad_norm": 2.2215094566345215, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7127019166946411, + "num_tokens": 105891557.0, + "step": 4097 + }, + { + "epoch": 0.4500329453107841, + "grad_norm": 2.4036154747009277, + "learning_rate": 1e-06, + "loss": 1.0366, + "mean_token_accuracy": 0.6922075748443604, + "num_tokens": 105914959.0, + "step": 4098 + }, + { + "epoch": 0.4501427630133978, + "grad_norm": 2.2504234313964844, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.696386456489563, + "num_tokens": 105940788.0, + "step": 4099 + }, + { + "epoch": 0.45025258071601143, + "grad_norm": 2.7344157695770264, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7278294563293457, + "num_tokens": 105960218.0, + "step": 4100 + }, + { + "epoch": 0.4503623984186251, + "grad_norm": 2.2939703464508057, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7088968753814697, + "num_tokens": 105985451.0, + "step": 4101 + }, + { + "epoch": 0.4504722161212387, + "grad_norm": 1.963388204574585, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7098745703697205, + "num_tokens": 106015411.0, + "step": 4102 + }, + { + "epoch": 0.4505820338238524, + "grad_norm": 2.253119468688965, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6959606409072876, + "num_tokens": 106040921.0, + "step": 4103 + }, + { + "epoch": 0.45069185152646607, + "grad_norm": 2.2599873542785645, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7080770134925842, + "num_tokens": 106065656.0, + "step": 4104 + }, + { + "epoch": 0.4508016692290797, + "grad_norm": 2.0030694007873535, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7076794505119324, + "num_tokens": 106097280.0, + "step": 4105 + }, + { + "epoch": 0.45091148693169336, + "grad_norm": 2.128709316253662, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7300194501876831, + "num_tokens": 106123047.0, + "step": 4106 + }, + { + "epoch": 0.45102130463430706, + "grad_norm": 2.116938591003418, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7134606838226318, + "num_tokens": 106151119.0, + "step": 4107 + }, + { + "epoch": 0.4511311223369207, + "grad_norm": 1.9518563747406006, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6832839846611023, + "num_tokens": 106183173.0, + "step": 4108 + }, + { + "epoch": 0.45124094003953436, + "grad_norm": 2.378096103668213, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7092189788818359, + "num_tokens": 106206388.0, + "step": 4109 + }, + { + "epoch": 0.45135075774214806, + "grad_norm": 2.0693612098693848, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7087908983230591, + "num_tokens": 106235476.0, + "step": 4110 + }, + { + "epoch": 0.4514605754447617, + "grad_norm": 2.1709468364715576, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7093507051467896, + "num_tokens": 106261962.0, + "step": 4111 + }, + { + "epoch": 0.45157039314737535, + "grad_norm": 2.303008556365967, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6828850507736206, + "num_tokens": 106288609.0, + "step": 4112 + }, + { + "epoch": 0.451680210849989, + "grad_norm": 2.0996758937835693, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7128075957298279, + "num_tokens": 106316201.0, + "step": 4113 + }, + { + "epoch": 0.4517900285526027, + "grad_norm": 1.9837424755096436, + "learning_rate": 1e-06, + "loss": 1.0572, + "mean_token_accuracy": 0.6860003471374512, + "num_tokens": 106349835.0, + "step": 4114 + }, + { + "epoch": 0.45189984625521634, + "grad_norm": 2.1726558208465576, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6890472173690796, + "num_tokens": 106376203.0, + "step": 4115 + }, + { + "epoch": 0.45200966395783, + "grad_norm": 2.1370277404785156, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.7020536661148071, + "num_tokens": 106404493.0, + "step": 4116 + }, + { + "epoch": 0.4521194816604437, + "grad_norm": 2.040092706680298, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7021463513374329, + "num_tokens": 106432988.0, + "step": 4117 + }, + { + "epoch": 0.45222929936305734, + "grad_norm": 2.589890956878662, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6959627866744995, + "num_tokens": 106453120.0, + "step": 4118 + }, + { + "epoch": 0.452339117065671, + "grad_norm": 2.2893500328063965, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7153815031051636, + "num_tokens": 106475765.0, + "step": 4119 + }, + { + "epoch": 0.4524489347682846, + "grad_norm": 2.245269775390625, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7106106281280518, + "num_tokens": 106501456.0, + "step": 4120 + }, + { + "epoch": 0.45255875247089833, + "grad_norm": 1.9908058643341064, + "learning_rate": 1e-06, + "loss": 1.0917, + "mean_token_accuracy": 0.667980432510376, + "num_tokens": 106534135.0, + "step": 4121 + }, + { + "epoch": 0.452668570173512, + "grad_norm": 2.318159341812134, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7167156934738159, + "num_tokens": 106558166.0, + "step": 4122 + }, + { + "epoch": 0.4527783878761256, + "grad_norm": 2.143317222595215, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6898223757743835, + "num_tokens": 106585004.0, + "step": 4123 + }, + { + "epoch": 0.45288820557873927, + "grad_norm": 2.1541523933410645, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.728495180606842, + "num_tokens": 106609383.0, + "step": 4124 + }, + { + "epoch": 0.45299802328135297, + "grad_norm": 2.033804416656494, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6979497075080872, + "num_tokens": 106641061.0, + "step": 4125 + }, + { + "epoch": 0.4531078409839666, + "grad_norm": 1.9586421251296997, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7098644375801086, + "num_tokens": 106672179.0, + "step": 4126 + }, + { + "epoch": 0.45321765868658026, + "grad_norm": 2.3748793601989746, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6931797862052917, + "num_tokens": 106695512.0, + "step": 4127 + }, + { + "epoch": 0.45332747638919396, + "grad_norm": 1.9864294528961182, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7282640933990479, + "num_tokens": 106723126.0, + "step": 4128 + }, + { + "epoch": 0.4534372940918076, + "grad_norm": 2.3110907077789307, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7060877084732056, + "num_tokens": 106747737.0, + "step": 4129 + }, + { + "epoch": 0.45354711179442125, + "grad_norm": 1.9809163808822632, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6927604079246521, + "num_tokens": 106778830.0, + "step": 4130 + }, + { + "epoch": 0.4536569294970349, + "grad_norm": 2.1967294216156006, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6927380561828613, + "num_tokens": 106803825.0, + "step": 4131 + }, + { + "epoch": 0.4537667471996486, + "grad_norm": 1.9156509637832642, + "learning_rate": 1e-06, + "loss": 1.0942, + "mean_token_accuracy": 0.6770486831665039, + "num_tokens": 106836896.0, + "step": 4132 + }, + { + "epoch": 0.45387656490226225, + "grad_norm": 2.1785264015197754, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6765973567962646, + "num_tokens": 106862131.0, + "step": 4133 + }, + { + "epoch": 0.4539863826048759, + "grad_norm": 2.2003884315490723, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.701601505279541, + "num_tokens": 106888132.0, + "step": 4134 + }, + { + "epoch": 0.4540962003074896, + "grad_norm": 2.0904979705810547, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7123557925224304, + "num_tokens": 106916637.0, + "step": 4135 + }, + { + "epoch": 0.45420601801010324, + "grad_norm": 2.155559539794922, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.7006170749664307, + "num_tokens": 106943004.0, + "step": 4136 + }, + { + "epoch": 0.4543158357127169, + "grad_norm": 2.0784404277801514, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6971448063850403, + "num_tokens": 106972634.0, + "step": 4137 + }, + { + "epoch": 0.45442565341533053, + "grad_norm": 2.3851189613342285, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6886868476867676, + "num_tokens": 106995022.0, + "step": 4138 + }, + { + "epoch": 0.45453547111794423, + "grad_norm": 2.534252405166626, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7197542786598206, + "num_tokens": 107014998.0, + "step": 4139 + }, + { + "epoch": 0.4546452888205579, + "grad_norm": 2.027371644973755, + "learning_rate": 1e-06, + "loss": 1.065, + "mean_token_accuracy": 0.6769608855247498, + "num_tokens": 107047636.0, + "step": 4140 + }, + { + "epoch": 0.4547551065231715, + "grad_norm": 2.384570837020874, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7070214748382568, + "num_tokens": 107073075.0, + "step": 4141 + }, + { + "epoch": 0.45486492422578517, + "grad_norm": 2.4245219230651855, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7089656591415405, + "num_tokens": 107095074.0, + "step": 4142 + }, + { + "epoch": 0.4549747419283989, + "grad_norm": 2.225576400756836, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6978071331977844, + "num_tokens": 107120478.0, + "step": 4143 + }, + { + "epoch": 0.4550845596310125, + "grad_norm": 2.097456693649292, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.7011269927024841, + "num_tokens": 107149935.0, + "step": 4144 + }, + { + "epoch": 0.45519437733362617, + "grad_norm": 2.2821555137634277, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7023090124130249, + "num_tokens": 107176438.0, + "step": 4145 + }, + { + "epoch": 0.45530419503623987, + "grad_norm": 2.0712292194366455, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.707111120223999, + "num_tokens": 107202811.0, + "step": 4146 + }, + { + "epoch": 0.4554140127388535, + "grad_norm": 2.0724539756774902, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6935033798217773, + "num_tokens": 107233423.0, + "step": 4147 + }, + { + "epoch": 0.45552383044146716, + "grad_norm": 2.1491031646728516, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.7019989490509033, + "num_tokens": 107259979.0, + "step": 4148 + }, + { + "epoch": 0.4556336481440808, + "grad_norm": 2.1199259757995605, + "learning_rate": 1e-06, + "loss": 1.0557, + "mean_token_accuracy": 0.6865756511688232, + "num_tokens": 107289683.0, + "step": 4149 + }, + { + "epoch": 0.4557434658466945, + "grad_norm": 2.2898929119110107, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7063950300216675, + "num_tokens": 107315025.0, + "step": 4150 + }, + { + "epoch": 0.45585328354930815, + "grad_norm": 2.442669630050659, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.702406644821167, + "num_tokens": 107337420.0, + "step": 4151 + }, + { + "epoch": 0.4559631012519218, + "grad_norm": 2.460538625717163, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6974883079528809, + "num_tokens": 107360034.0, + "step": 4152 + }, + { + "epoch": 0.45607291895453544, + "grad_norm": 2.0655558109283447, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6869783401489258, + "num_tokens": 107389164.0, + "step": 4153 + }, + { + "epoch": 0.45618273665714915, + "grad_norm": 2.2123539447784424, + "learning_rate": 1e-06, + "loss": 1.1085, + "mean_token_accuracy": 0.674179196357727, + "num_tokens": 107417239.0, + "step": 4154 + }, + { + "epoch": 0.4562925543597628, + "grad_norm": 2.3043601512908936, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6949566602706909, + "num_tokens": 107442241.0, + "step": 4155 + }, + { + "epoch": 0.45640237206237644, + "grad_norm": 2.2815232276916504, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7032914161682129, + "num_tokens": 107464525.0, + "step": 4156 + }, + { + "epoch": 0.45651218976499014, + "grad_norm": 2.415367603302002, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7178769707679749, + "num_tokens": 107487863.0, + "step": 4157 + }, + { + "epoch": 0.4566220074676038, + "grad_norm": 2.123551607131958, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6967889070510864, + "num_tokens": 107516463.0, + "step": 4158 + }, + { + "epoch": 0.45673182517021743, + "grad_norm": 2.055006504058838, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6831961870193481, + "num_tokens": 107547372.0, + "step": 4159 + }, + { + "epoch": 0.4568416428728311, + "grad_norm": 2.3786375522613525, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7077146172523499, + "num_tokens": 107570039.0, + "step": 4160 + }, + { + "epoch": 0.4569514605754448, + "grad_norm": 2.227942705154419, + "learning_rate": 1e-06, + "loss": 1.0335, + "mean_token_accuracy": 0.6912425756454468, + "num_tokens": 107595999.0, + "step": 4161 + }, + { + "epoch": 0.4570612782780584, + "grad_norm": 2.0757572650909424, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6986191272735596, + "num_tokens": 107624303.0, + "step": 4162 + }, + { + "epoch": 0.45717109598067207, + "grad_norm": 2.135498285293579, + "learning_rate": 1e-06, + "loss": 1.0639, + "mean_token_accuracy": 0.6864875555038452, + "num_tokens": 107654015.0, + "step": 4163 + }, + { + "epoch": 0.4572809136832858, + "grad_norm": 2.0428831577301025, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7322714924812317, + "num_tokens": 107681066.0, + "step": 4164 + }, + { + "epoch": 0.4573907313858994, + "grad_norm": 2.3809032440185547, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.717022180557251, + "num_tokens": 107704738.0, + "step": 4165 + }, + { + "epoch": 0.45750054908851306, + "grad_norm": 2.0433456897735596, + "learning_rate": 1e-06, + "loss": 1.0705, + "mean_token_accuracy": 0.6771947145462036, + "num_tokens": 107738308.0, + "step": 4166 + }, + { + "epoch": 0.4576103667911267, + "grad_norm": 2.014106273651123, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6844768524169922, + "num_tokens": 107770284.0, + "step": 4167 + }, + { + "epoch": 0.4577201844937404, + "grad_norm": 2.0784194469451904, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6895250082015991, + "num_tokens": 107800283.0, + "step": 4168 + }, + { + "epoch": 0.45783000219635406, + "grad_norm": 2.2820956707000732, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7195945978164673, + "num_tokens": 107824988.0, + "step": 4169 + }, + { + "epoch": 0.4579398198989677, + "grad_norm": 2.3270609378814697, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7233754396438599, + "num_tokens": 107847507.0, + "step": 4170 + }, + { + "epoch": 0.45804963760158135, + "grad_norm": 2.004073143005371, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6952494382858276, + "num_tokens": 107878874.0, + "step": 4171 + }, + { + "epoch": 0.45815945530419505, + "grad_norm": 2.0447375774383545, + "learning_rate": 1e-06, + "loss": 1.0638, + "mean_token_accuracy": 0.6916460990905762, + "num_tokens": 107907987.0, + "step": 4172 + }, + { + "epoch": 0.4582692730068087, + "grad_norm": 1.9737805128097534, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.692907452583313, + "num_tokens": 107935921.0, + "step": 4173 + }, + { + "epoch": 0.45837909070942234, + "grad_norm": 1.9740183353424072, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7004748582839966, + "num_tokens": 107969167.0, + "step": 4174 + }, + { + "epoch": 0.45848890841203604, + "grad_norm": 2.238994598388672, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7136960625648499, + "num_tokens": 107993529.0, + "step": 4175 + }, + { + "epoch": 0.4585987261146497, + "grad_norm": 2.2977170944213867, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.684600830078125, + "num_tokens": 108019805.0, + "step": 4176 + }, + { + "epoch": 0.45870854381726334, + "grad_norm": 2.0699214935302734, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6915384531021118, + "num_tokens": 108048592.0, + "step": 4177 + }, + { + "epoch": 0.458818361519877, + "grad_norm": 1.9768245220184326, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7169408202171326, + "num_tokens": 108079410.0, + "step": 4178 + }, + { + "epoch": 0.4589281792224907, + "grad_norm": 2.1926465034484863, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7131099104881287, + "num_tokens": 108106842.0, + "step": 4179 + }, + { + "epoch": 0.45903799692510433, + "grad_norm": 2.1241304874420166, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.6997743248939514, + "num_tokens": 108134819.0, + "step": 4180 + }, + { + "epoch": 0.459147814627718, + "grad_norm": 2.222325086593628, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7120131850242615, + "num_tokens": 108160211.0, + "step": 4181 + }, + { + "epoch": 0.4592576323303316, + "grad_norm": 2.3034605979919434, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7062099575996399, + "num_tokens": 108184231.0, + "step": 4182 + }, + { + "epoch": 0.4593674500329453, + "grad_norm": 2.3397419452667236, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.700213611125946, + "num_tokens": 108206916.0, + "step": 4183 + }, + { + "epoch": 0.45947726773555897, + "grad_norm": 2.3497846126556396, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7149938344955444, + "num_tokens": 108227851.0, + "step": 4184 + }, + { + "epoch": 0.4595870854381726, + "grad_norm": 2.191439151763916, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.695428192615509, + "num_tokens": 108254622.0, + "step": 4185 + }, + { + "epoch": 0.4596969031407863, + "grad_norm": 2.1341989040374756, + "learning_rate": 1e-06, + "loss": 1.0391, + "mean_token_accuracy": 0.7003077864646912, + "num_tokens": 108282366.0, + "step": 4186 + }, + { + "epoch": 0.45980672084339996, + "grad_norm": 1.997254490852356, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6905682682991028, + "num_tokens": 108314359.0, + "step": 4187 + }, + { + "epoch": 0.4599165385460136, + "grad_norm": 2.020632266998291, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7004120349884033, + "num_tokens": 108342960.0, + "step": 4188 + }, + { + "epoch": 0.46002635624862726, + "grad_norm": 2.115043878555298, + "learning_rate": 1e-06, + "loss": 1.0322, + "mean_token_accuracy": 0.6888014674186707, + "num_tokens": 108369224.0, + "step": 4189 + }, + { + "epoch": 0.46013617395124096, + "grad_norm": 2.133394956588745, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6928966045379639, + "num_tokens": 108397077.0, + "step": 4190 + }, + { + "epoch": 0.4602459916538546, + "grad_norm": 2.2160768508911133, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6958644390106201, + "num_tokens": 108421690.0, + "step": 4191 + }, + { + "epoch": 0.46035580935646825, + "grad_norm": 2.1421070098876953, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6799038648605347, + "num_tokens": 108449768.0, + "step": 4192 + }, + { + "epoch": 0.46046562705908195, + "grad_norm": 2.9831032752990723, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7076562643051147, + "num_tokens": 108466772.0, + "step": 4193 + }, + { + "epoch": 0.4605754447616956, + "grad_norm": 2.427156686782837, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.722245991230011, + "num_tokens": 108488184.0, + "step": 4194 + }, + { + "epoch": 0.46068526246430924, + "grad_norm": 2.1638171672821045, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.699458122253418, + "num_tokens": 108514984.0, + "step": 4195 + }, + { + "epoch": 0.4607950801669229, + "grad_norm": 2.174269437789917, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7125489711761475, + "num_tokens": 108540998.0, + "step": 4196 + }, + { + "epoch": 0.4609048978695366, + "grad_norm": 2.001164674758911, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7074611186981201, + "num_tokens": 108570211.0, + "step": 4197 + }, + { + "epoch": 0.46101471557215024, + "grad_norm": 2.029905319213867, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6789653301239014, + "num_tokens": 108599725.0, + "step": 4198 + }, + { + "epoch": 0.4611245332747639, + "grad_norm": 2.309504270553589, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6927366256713867, + "num_tokens": 108624771.0, + "step": 4199 + }, + { + "epoch": 0.4612343509773775, + "grad_norm": 2.0887465476989746, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6903594732284546, + "num_tokens": 108655965.0, + "step": 4200 + }, + { + "epoch": 0.46134416867999123, + "grad_norm": 2.352670907974243, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7028348445892334, + "num_tokens": 108678648.0, + "step": 4201 + }, + { + "epoch": 0.4614539863826049, + "grad_norm": 2.1164119243621826, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.692204475402832, + "num_tokens": 108708088.0, + "step": 4202 + }, + { + "epoch": 0.4615638040852185, + "grad_norm": 1.955316185951233, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.712074875831604, + "num_tokens": 108737497.0, + "step": 4203 + }, + { + "epoch": 0.4616736217878322, + "grad_norm": 2.180989980697632, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7087349891662598, + "num_tokens": 108764609.0, + "step": 4204 + }, + { + "epoch": 0.46178343949044587, + "grad_norm": 2.0542917251586914, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7027729153633118, + "num_tokens": 108790774.0, + "step": 4205 + }, + { + "epoch": 0.4618932571930595, + "grad_norm": 2.0205078125, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7172549962997437, + "num_tokens": 108820188.0, + "step": 4206 + }, + { + "epoch": 0.46200307489567316, + "grad_norm": 2.455186367034912, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7245705127716064, + "num_tokens": 108841518.0, + "step": 4207 + }, + { + "epoch": 0.46211289259828686, + "grad_norm": 2.4013478755950928, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6931608319282532, + "num_tokens": 108865351.0, + "step": 4208 + }, + { + "epoch": 0.4622227103009005, + "grad_norm": 2.5493054389953613, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6899391412734985, + "num_tokens": 108887902.0, + "step": 4209 + }, + { + "epoch": 0.46233252800351415, + "grad_norm": 1.9970061779022217, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7056321501731873, + "num_tokens": 108917553.0, + "step": 4210 + }, + { + "epoch": 0.46244234570612786, + "grad_norm": 1.9690089225769043, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6980037689208984, + "num_tokens": 108949999.0, + "step": 4211 + }, + { + "epoch": 0.4625521634087415, + "grad_norm": 2.231372356414795, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.7051732540130615, + "num_tokens": 108976054.0, + "step": 4212 + }, + { + "epoch": 0.46266198111135515, + "grad_norm": 2.208378791809082, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7161284685134888, + "num_tokens": 109001853.0, + "step": 4213 + }, + { + "epoch": 0.4627717988139688, + "grad_norm": 2.7438063621520996, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7330218553543091, + "num_tokens": 109018515.0, + "step": 4214 + }, + { + "epoch": 0.4628816165165825, + "grad_norm": 2.342144012451172, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7078506350517273, + "num_tokens": 109043524.0, + "step": 4215 + }, + { + "epoch": 0.46299143421919614, + "grad_norm": 2.368791341781616, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6939669251441956, + "num_tokens": 109065870.0, + "step": 4216 + }, + { + "epoch": 0.4631012519218098, + "grad_norm": 2.1249544620513916, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6904288530349731, + "num_tokens": 109092819.0, + "step": 4217 + }, + { + "epoch": 0.46321106962442343, + "grad_norm": 2.1425113677978516, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.717383861541748, + "num_tokens": 109119366.0, + "step": 4218 + }, + { + "epoch": 0.46332088732703713, + "grad_norm": 2.505753755569458, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6982284784317017, + "num_tokens": 109142027.0, + "step": 4219 + }, + { + "epoch": 0.4634307050296508, + "grad_norm": 2.4072017669677734, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7042376399040222, + "num_tokens": 109164566.0, + "step": 4220 + }, + { + "epoch": 0.4635405227322644, + "grad_norm": 2.5165441036224365, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7253552079200745, + "num_tokens": 109185471.0, + "step": 4221 + }, + { + "epoch": 0.4636503404348781, + "grad_norm": 2.3161303997039795, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7012354135513306, + "num_tokens": 109209593.0, + "step": 4222 + }, + { + "epoch": 0.4637601581374918, + "grad_norm": 2.342660427093506, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.6868381500244141, + "num_tokens": 109233085.0, + "step": 4223 + }, + { + "epoch": 0.4638699758401054, + "grad_norm": 2.3495078086853027, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7111427783966064, + "num_tokens": 109255439.0, + "step": 4224 + }, + { + "epoch": 0.46397979354271907, + "grad_norm": 2.140800952911377, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6930592060089111, + "num_tokens": 109285240.0, + "step": 4225 + }, + { + "epoch": 0.46408961124533277, + "grad_norm": 2.52949595451355, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7185254693031311, + "num_tokens": 109306534.0, + "step": 4226 + }, + { + "epoch": 0.4641994289479464, + "grad_norm": 2.1633896827697754, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.693337619304657, + "num_tokens": 109335144.0, + "step": 4227 + }, + { + "epoch": 0.46430924665056006, + "grad_norm": 2.0289764404296875, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6947184801101685, + "num_tokens": 109363223.0, + "step": 4228 + }, + { + "epoch": 0.4644190643531737, + "grad_norm": 2.38240122795105, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7109823822975159, + "num_tokens": 109385976.0, + "step": 4229 + }, + { + "epoch": 0.4645288820557874, + "grad_norm": 2.1959872245788574, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6929758787155151, + "num_tokens": 109413991.0, + "step": 4230 + }, + { + "epoch": 0.46463869975840105, + "grad_norm": 2.388045072555542, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.72199547290802, + "num_tokens": 109434389.0, + "step": 4231 + }, + { + "epoch": 0.4647485174610147, + "grad_norm": 1.9413197040557861, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.690652072429657, + "num_tokens": 109466391.0, + "step": 4232 + }, + { + "epoch": 0.4648583351636284, + "grad_norm": 2.349409818649292, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.71153724193573, + "num_tokens": 109489623.0, + "step": 4233 + }, + { + "epoch": 0.46496815286624205, + "grad_norm": 2.213719129562378, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6901469826698303, + "num_tokens": 109515780.0, + "step": 4234 + }, + { + "epoch": 0.4650779705688557, + "grad_norm": 2.28401255607605, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7066293358802795, + "num_tokens": 109539212.0, + "step": 4235 + }, + { + "epoch": 0.46518778827146934, + "grad_norm": 2.5047647953033447, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6891745328903198, + "num_tokens": 109560287.0, + "step": 4236 + }, + { + "epoch": 0.46529760597408304, + "grad_norm": 2.0792596340179443, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7053713202476501, + "num_tokens": 109588456.0, + "step": 4237 + }, + { + "epoch": 0.4654074236766967, + "grad_norm": 2.1485581398010254, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6813114881515503, + "num_tokens": 109616581.0, + "step": 4238 + }, + { + "epoch": 0.46551724137931033, + "grad_norm": 2.3148915767669678, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7192122340202332, + "num_tokens": 109641331.0, + "step": 4239 + }, + { + "epoch": 0.46562705908192403, + "grad_norm": 2.037303924560547, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6906690001487732, + "num_tokens": 109670886.0, + "step": 4240 + }, + { + "epoch": 0.4657368767845377, + "grad_norm": 2.154867649078369, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6939176321029663, + "num_tokens": 109697801.0, + "step": 4241 + }, + { + "epoch": 0.4658466944871513, + "grad_norm": 2.154327154159546, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7038354873657227, + "num_tokens": 109724159.0, + "step": 4242 + }, + { + "epoch": 0.46595651218976497, + "grad_norm": 2.2856290340423584, + "learning_rate": 1e-06, + "loss": 1.0649, + "mean_token_accuracy": 0.6723470687866211, + "num_tokens": 109749383.0, + "step": 4243 + }, + { + "epoch": 0.4660663298923787, + "grad_norm": 2.2244210243225098, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7088513374328613, + "num_tokens": 109776098.0, + "step": 4244 + }, + { + "epoch": 0.4661761475949923, + "grad_norm": 2.350054979324341, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7111560106277466, + "num_tokens": 109797559.0, + "step": 4245 + }, + { + "epoch": 0.46628596529760596, + "grad_norm": 1.9436713457107544, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.729828953742981, + "num_tokens": 109828678.0, + "step": 4246 + }, + { + "epoch": 0.4663957830002196, + "grad_norm": 2.0924713611602783, + "learning_rate": 1e-06, + "loss": 1.0499, + "mean_token_accuracy": 0.6833061575889587, + "num_tokens": 109857652.0, + "step": 4247 + }, + { + "epoch": 0.4665056007028333, + "grad_norm": 2.150496006011963, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.7002488970756531, + "num_tokens": 109884554.0, + "step": 4248 + }, + { + "epoch": 0.46661541840544696, + "grad_norm": 2.104466199874878, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7169616222381592, + "num_tokens": 109911682.0, + "step": 4249 + }, + { + "epoch": 0.4667252361080606, + "grad_norm": 2.3186123371124268, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.685797929763794, + "num_tokens": 109936157.0, + "step": 4250 + }, + { + "epoch": 0.4668350538106743, + "grad_norm": 2.055894136428833, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.6956648230552673, + "num_tokens": 109963461.0, + "step": 4251 + }, + { + "epoch": 0.46694487151328795, + "grad_norm": 2.3983938694000244, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7273188233375549, + "num_tokens": 109985239.0, + "step": 4252 + }, + { + "epoch": 0.4670546892159016, + "grad_norm": 2.224813461303711, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6901869177818298, + "num_tokens": 110009774.0, + "step": 4253 + }, + { + "epoch": 0.46716450691851524, + "grad_norm": 2.488267660140991, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.7026270627975464, + "num_tokens": 110030532.0, + "step": 4254 + }, + { + "epoch": 0.46727432462112894, + "grad_norm": 1.9263508319854736, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6875590085983276, + "num_tokens": 110064065.0, + "step": 4255 + }, + { + "epoch": 0.4673841423237426, + "grad_norm": 2.1700172424316406, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7026984691619873, + "num_tokens": 110089461.0, + "step": 4256 + }, + { + "epoch": 0.46749396002635624, + "grad_norm": 1.895315408706665, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6966694593429565, + "num_tokens": 110124072.0, + "step": 4257 + }, + { + "epoch": 0.4676037777289699, + "grad_norm": 2.488525867462158, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7061303853988647, + "num_tokens": 110145663.0, + "step": 4258 + }, + { + "epoch": 0.4677135954315836, + "grad_norm": 2.0772705078125, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6883862018585205, + "num_tokens": 110175157.0, + "step": 4259 + }, + { + "epoch": 0.46782341313419723, + "grad_norm": 2.4139323234558105, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7063859105110168, + "num_tokens": 110198393.0, + "step": 4260 + }, + { + "epoch": 0.4679332308368109, + "grad_norm": 1.9238872528076172, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6962853670120239, + "num_tokens": 110231770.0, + "step": 4261 + }, + { + "epoch": 0.4680430485394246, + "grad_norm": 2.1131811141967773, + "learning_rate": 1e-06, + "loss": 1.0647, + "mean_token_accuracy": 0.6847758293151855, + "num_tokens": 110261755.0, + "step": 4262 + }, + { + "epoch": 0.4681528662420382, + "grad_norm": 2.545071840286255, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6881436109542847, + "num_tokens": 110281687.0, + "step": 4263 + }, + { + "epoch": 0.46826268394465187, + "grad_norm": 2.365931987762451, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7151162624359131, + "num_tokens": 110305069.0, + "step": 4264 + }, + { + "epoch": 0.4683725016472655, + "grad_norm": 2.4072511196136475, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.737663745880127, + "num_tokens": 110324935.0, + "step": 4265 + }, + { + "epoch": 0.4684823193498792, + "grad_norm": 2.429795265197754, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6953721046447754, + "num_tokens": 110346955.0, + "step": 4266 + }, + { + "epoch": 0.46859213705249286, + "grad_norm": 2.0818068981170654, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7154278755187988, + "num_tokens": 110373544.0, + "step": 4267 + }, + { + "epoch": 0.4687019547551065, + "grad_norm": 2.050440549850464, + "learning_rate": 1e-06, + "loss": 1.0431, + "mean_token_accuracy": 0.6842323541641235, + "num_tokens": 110405117.0, + "step": 4268 + }, + { + "epoch": 0.4688117724577202, + "grad_norm": 2.4085958003997803, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6966323256492615, + "num_tokens": 110426767.0, + "step": 4269 + }, + { + "epoch": 0.46892159016033386, + "grad_norm": 2.32776141166687, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6892709732055664, + "num_tokens": 110450394.0, + "step": 4270 + }, + { + "epoch": 0.4690314078629475, + "grad_norm": 2.077195405960083, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6853848099708557, + "num_tokens": 110479240.0, + "step": 4271 + }, + { + "epoch": 0.46914122556556115, + "grad_norm": 2.2389614582061768, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7093523740768433, + "num_tokens": 110503201.0, + "step": 4272 + }, + { + "epoch": 0.46925104326817485, + "grad_norm": 2.100825309753418, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6875368356704712, + "num_tokens": 110533873.0, + "step": 4273 + }, + { + "epoch": 0.4693608609707885, + "grad_norm": 2.576810836791992, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.713318407535553, + "num_tokens": 110553795.0, + "step": 4274 + }, + { + "epoch": 0.46947067867340214, + "grad_norm": 2.1409850120544434, + "learning_rate": 1e-06, + "loss": 1.0513, + "mean_token_accuracy": 0.6839667558670044, + "num_tokens": 110582884.0, + "step": 4275 + }, + { + "epoch": 0.4695804963760158, + "grad_norm": 2.146998882293701, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7068586349487305, + "num_tokens": 110608619.0, + "step": 4276 + }, + { + "epoch": 0.4696903140786295, + "grad_norm": 2.2682485580444336, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7085797786712646, + "num_tokens": 110633119.0, + "step": 4277 + }, + { + "epoch": 0.46980013178124314, + "grad_norm": 2.207385778427124, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6889121532440186, + "num_tokens": 110658497.0, + "step": 4278 + }, + { + "epoch": 0.4699099494838568, + "grad_norm": 2.2576000690460205, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7123234868049622, + "num_tokens": 110682326.0, + "step": 4279 + }, + { + "epoch": 0.4700197671864705, + "grad_norm": 2.109484910964966, + "learning_rate": 1e-06, + "loss": 1.06, + "mean_token_accuracy": 0.6891385912895203, + "num_tokens": 110712977.0, + "step": 4280 + }, + { + "epoch": 0.47012958488908413, + "grad_norm": 2.4042868614196777, + "learning_rate": 1e-06, + "loss": 1.0293, + "mean_token_accuracy": 0.6936219930648804, + "num_tokens": 110735079.0, + "step": 4281 + }, + { + "epoch": 0.4702394025916978, + "grad_norm": 2.1140990257263184, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6892350912094116, + "num_tokens": 110763270.0, + "step": 4282 + }, + { + "epoch": 0.4703492202943114, + "grad_norm": 2.258796215057373, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6936012506484985, + "num_tokens": 110789102.0, + "step": 4283 + }, + { + "epoch": 0.4704590379969251, + "grad_norm": 1.9954060316085815, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7182656526565552, + "num_tokens": 110820768.0, + "step": 4284 + }, + { + "epoch": 0.47056885569953877, + "grad_norm": 2.028247356414795, + "learning_rate": 1e-06, + "loss": 1.0859, + "mean_token_accuracy": 0.6745721101760864, + "num_tokens": 110852042.0, + "step": 4285 + }, + { + "epoch": 0.4706786734021524, + "grad_norm": 2.2664525508880615, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7171111702919006, + "num_tokens": 110877114.0, + "step": 4286 + }, + { + "epoch": 0.4707884911047661, + "grad_norm": 1.9493367671966553, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7414325475692749, + "num_tokens": 110905762.0, + "step": 4287 + }, + { + "epoch": 0.47089830880737976, + "grad_norm": 2.4236721992492676, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.6985921859741211, + "num_tokens": 110926649.0, + "step": 4288 + }, + { + "epoch": 0.4710081265099934, + "grad_norm": 2.3136937618255615, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6932845711708069, + "num_tokens": 110952555.0, + "step": 4289 + }, + { + "epoch": 0.47111794421260705, + "grad_norm": 2.2561452388763428, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.690888524055481, + "num_tokens": 110978068.0, + "step": 4290 + }, + { + "epoch": 0.47122776191522076, + "grad_norm": 2.4920711517333984, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.7116915583610535, + "num_tokens": 111000492.0, + "step": 4291 + }, + { + "epoch": 0.4713375796178344, + "grad_norm": 2.079317569732666, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6878593564033508, + "num_tokens": 111031966.0, + "step": 4292 + }, + { + "epoch": 0.47144739732044805, + "grad_norm": 2.172821283340454, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6840237379074097, + "num_tokens": 111060055.0, + "step": 4293 + }, + { + "epoch": 0.4715572150230617, + "grad_norm": 2.0763182640075684, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6912773847579956, + "num_tokens": 111088646.0, + "step": 4294 + }, + { + "epoch": 0.4716670327256754, + "grad_norm": 2.286163091659546, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7065454721450806, + "num_tokens": 111113189.0, + "step": 4295 + }, + { + "epoch": 0.47177685042828904, + "grad_norm": 2.3195369243621826, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7045892477035522, + "num_tokens": 111136436.0, + "step": 4296 + }, + { + "epoch": 0.4718866681309027, + "grad_norm": 2.192101001739502, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7115106582641602, + "num_tokens": 111162183.0, + "step": 4297 + }, + { + "epoch": 0.4719964858335164, + "grad_norm": 2.1166908740997314, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6878601312637329, + "num_tokens": 111191417.0, + "step": 4298 + }, + { + "epoch": 0.47210630353613003, + "grad_norm": 2.067546844482422, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7034275531768799, + "num_tokens": 111219309.0, + "step": 4299 + }, + { + "epoch": 0.4722161212387437, + "grad_norm": 2.111052989959717, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7170824408531189, + "num_tokens": 111247213.0, + "step": 4300 + }, + { + "epoch": 0.4723259389413573, + "grad_norm": 2.4113502502441406, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6889406442642212, + "num_tokens": 111270004.0, + "step": 4301 + }, + { + "epoch": 0.472435756643971, + "grad_norm": 2.3089656829833984, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7112225890159607, + "num_tokens": 111294602.0, + "step": 4302 + }, + { + "epoch": 0.4725455743465847, + "grad_norm": 2.3028151988983154, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6889934539794922, + "num_tokens": 111321839.0, + "step": 4303 + }, + { + "epoch": 0.4726553920491983, + "grad_norm": 2.2648603916168213, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6883281469345093, + "num_tokens": 111348923.0, + "step": 4304 + }, + { + "epoch": 0.47276520975181197, + "grad_norm": 1.956336498260498, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6914317011833191, + "num_tokens": 111378979.0, + "step": 4305 + }, + { + "epoch": 0.47287502745442567, + "grad_norm": 2.4046835899353027, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6990334391593933, + "num_tokens": 111402072.0, + "step": 4306 + }, + { + "epoch": 0.4729848451570393, + "grad_norm": 2.3639798164367676, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7079647183418274, + "num_tokens": 111424366.0, + "step": 4307 + }, + { + "epoch": 0.47309466285965296, + "grad_norm": 1.9018197059631348, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6799275875091553, + "num_tokens": 111457727.0, + "step": 4308 + }, + { + "epoch": 0.47320448056226666, + "grad_norm": 2.3525288105010986, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7135293483734131, + "num_tokens": 111479473.0, + "step": 4309 + }, + { + "epoch": 0.4733142982648803, + "grad_norm": 2.204766273498535, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6869833469390869, + "num_tokens": 111505459.0, + "step": 4310 + }, + { + "epoch": 0.47342411596749395, + "grad_norm": 2.1667230129241943, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6953631043434143, + "num_tokens": 111529820.0, + "step": 4311 + }, + { + "epoch": 0.4735339336701076, + "grad_norm": 2.274761438369751, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6988548040390015, + "num_tokens": 111556109.0, + "step": 4312 + }, + { + "epoch": 0.4736437513727213, + "grad_norm": 2.233388900756836, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7089896202087402, + "num_tokens": 111580721.0, + "step": 4313 + }, + { + "epoch": 0.47375356907533495, + "grad_norm": 2.290809154510498, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7160590887069702, + "num_tokens": 111604459.0, + "step": 4314 + }, + { + "epoch": 0.4738633867779486, + "grad_norm": 2.3220362663269043, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7032572031021118, + "num_tokens": 111629930.0, + "step": 4315 + }, + { + "epoch": 0.4739732044805623, + "grad_norm": 2.098215103149414, + "learning_rate": 1e-06, + "loss": 1.061, + "mean_token_accuracy": 0.6847156882286072, + "num_tokens": 111657613.0, + "step": 4316 + }, + { + "epoch": 0.47408302218317594, + "grad_norm": 2.510481834411621, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7279695868492126, + "num_tokens": 111677344.0, + "step": 4317 + }, + { + "epoch": 0.4741928398857896, + "grad_norm": 2.305807590484619, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7149723768234253, + "num_tokens": 111701017.0, + "step": 4318 + }, + { + "epoch": 0.47430265758840323, + "grad_norm": 2.4929208755493164, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7176487445831299, + "num_tokens": 111721446.0, + "step": 4319 + }, + { + "epoch": 0.47441247529101693, + "grad_norm": 2.1471426486968994, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.6871787309646606, + "num_tokens": 111749768.0, + "step": 4320 + }, + { + "epoch": 0.4745222929936306, + "grad_norm": 2.114550828933716, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6994626522064209, + "num_tokens": 111780595.0, + "step": 4321 + }, + { + "epoch": 0.4746321106962442, + "grad_norm": 2.0472118854522705, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.695945680141449, + "num_tokens": 111808453.0, + "step": 4322 + }, + { + "epoch": 0.47474192839885787, + "grad_norm": 2.0785932540893555, + "learning_rate": 1e-06, + "loss": 1.0661, + "mean_token_accuracy": 0.6790564060211182, + "num_tokens": 111838333.0, + "step": 4323 + }, + { + "epoch": 0.4748517461014716, + "grad_norm": 2.2736499309539795, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7032351493835449, + "num_tokens": 111862836.0, + "step": 4324 + }, + { + "epoch": 0.4749615638040852, + "grad_norm": 2.2597572803497314, + "learning_rate": 1e-06, + "loss": 1.0951, + "mean_token_accuracy": 0.6745543479919434, + "num_tokens": 111889887.0, + "step": 4325 + }, + { + "epoch": 0.47507138150669886, + "grad_norm": 2.285229444503784, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6906217336654663, + "num_tokens": 111916363.0, + "step": 4326 + }, + { + "epoch": 0.47518119920931257, + "grad_norm": 2.009275197982788, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7156619429588318, + "num_tokens": 111947248.0, + "step": 4327 + }, + { + "epoch": 0.4752910169119262, + "grad_norm": 1.996144413948059, + "learning_rate": 1e-06, + "loss": 1.0661, + "mean_token_accuracy": 0.6880390048027039, + "num_tokens": 111979006.0, + "step": 4328 + }, + { + "epoch": 0.47540083461453986, + "grad_norm": 2.264760971069336, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7202064990997314, + "num_tokens": 112001590.0, + "step": 4329 + }, + { + "epoch": 0.4755106523171535, + "grad_norm": 2.5292611122131348, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7103533148765564, + "num_tokens": 112022137.0, + "step": 4330 + }, + { + "epoch": 0.4756204700197672, + "grad_norm": 2.205333709716797, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7279379963874817, + "num_tokens": 112047019.0, + "step": 4331 + }, + { + "epoch": 0.47573028772238085, + "grad_norm": 2.140374183654785, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6929015517234802, + "num_tokens": 112073720.0, + "step": 4332 + }, + { + "epoch": 0.4758401054249945, + "grad_norm": 2.2212774753570557, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.703825831413269, + "num_tokens": 112100905.0, + "step": 4333 + }, + { + "epoch": 0.47594992312760814, + "grad_norm": 2.0704588890075684, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7046926021575928, + "num_tokens": 112127185.0, + "step": 4334 + }, + { + "epoch": 0.47605974083022184, + "grad_norm": 2.3860085010528564, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7169502973556519, + "num_tokens": 112148344.0, + "step": 4335 + }, + { + "epoch": 0.4761695585328355, + "grad_norm": 2.2865285873413086, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6961331367492676, + "num_tokens": 112172753.0, + "step": 4336 + }, + { + "epoch": 0.47627937623544914, + "grad_norm": 2.2661550045013428, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7083896398544312, + "num_tokens": 112195835.0, + "step": 4337 + }, + { + "epoch": 0.47638919393806284, + "grad_norm": 2.0030741691589355, + "learning_rate": 1e-06, + "loss": 1.0756, + "mean_token_accuracy": 0.6768174171447754, + "num_tokens": 112225281.0, + "step": 4338 + }, + { + "epoch": 0.4764990116406765, + "grad_norm": 2.277813196182251, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7032190561294556, + "num_tokens": 112249829.0, + "step": 4339 + }, + { + "epoch": 0.47660882934329013, + "grad_norm": 1.9846585988998413, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6929094791412354, + "num_tokens": 112279445.0, + "step": 4340 + }, + { + "epoch": 0.4767186470459038, + "grad_norm": 2.0645766258239746, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6918944120407104, + "num_tokens": 112309786.0, + "step": 4341 + }, + { + "epoch": 0.4768284647485175, + "grad_norm": 2.1354732513427734, + "learning_rate": 1e-06, + "loss": 1.0555, + "mean_token_accuracy": 0.6845608949661255, + "num_tokens": 112338737.0, + "step": 4342 + }, + { + "epoch": 0.4769382824511311, + "grad_norm": 1.8263108730316162, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6856485605239868, + "num_tokens": 112376012.0, + "step": 4343 + }, + { + "epoch": 0.47704810015374477, + "grad_norm": 2.5177998542785645, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6909803748130798, + "num_tokens": 112398477.0, + "step": 4344 + }, + { + "epoch": 0.47715791785635847, + "grad_norm": 2.3172249794006348, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.703370988368988, + "num_tokens": 112421259.0, + "step": 4345 + }, + { + "epoch": 0.4772677355589721, + "grad_norm": 2.297990322113037, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6883794069290161, + "num_tokens": 112445827.0, + "step": 4346 + }, + { + "epoch": 0.47737755326158576, + "grad_norm": 2.4915525913238525, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6840415596961975, + "num_tokens": 112468017.0, + "step": 4347 + }, + { + "epoch": 0.4774873709641994, + "grad_norm": 2.2363524436950684, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6799676418304443, + "num_tokens": 112493136.0, + "step": 4348 + }, + { + "epoch": 0.4775971886668131, + "grad_norm": 2.136415719985962, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6919879913330078, + "num_tokens": 112517619.0, + "step": 4349 + }, + { + "epoch": 0.47770700636942676, + "grad_norm": 1.9783008098602295, + "learning_rate": 1e-06, + "loss": 1.121, + "mean_token_accuracy": 0.6704065203666687, + "num_tokens": 112553523.0, + "step": 4350 + }, + { + "epoch": 0.4778168240720404, + "grad_norm": 2.214317560195923, + "learning_rate": 1e-06, + "loss": 1.1568, + "mean_token_accuracy": 0.657930314540863, + "num_tokens": 112581580.0, + "step": 4351 + }, + { + "epoch": 0.47792664177465405, + "grad_norm": 2.053274631500244, + "learning_rate": 1e-06, + "loss": 1.066, + "mean_token_accuracy": 0.6845711469650269, + "num_tokens": 112612031.0, + "step": 4352 + }, + { + "epoch": 0.47803645947726775, + "grad_norm": 1.9674804210662842, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.707984983921051, + "num_tokens": 112642814.0, + "step": 4353 + }, + { + "epoch": 0.4781462771798814, + "grad_norm": 2.4629642963409424, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7039221525192261, + "num_tokens": 112665706.0, + "step": 4354 + }, + { + "epoch": 0.47825609488249504, + "grad_norm": 2.276150941848755, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6915817260742188, + "num_tokens": 112690358.0, + "step": 4355 + }, + { + "epoch": 0.47836591258510874, + "grad_norm": 1.9800015687942505, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6962817907333374, + "num_tokens": 112719946.0, + "step": 4356 + }, + { + "epoch": 0.4784757302877224, + "grad_norm": 2.278249502182007, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6944482326507568, + "num_tokens": 112744322.0, + "step": 4357 + }, + { + "epoch": 0.47858554799033604, + "grad_norm": 2.0619442462921143, + "learning_rate": 1e-06, + "loss": 1.0993, + "mean_token_accuracy": 0.6726663112640381, + "num_tokens": 112775838.0, + "step": 4358 + }, + { + "epoch": 0.4786953656929497, + "grad_norm": 2.0798892974853516, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7120602130889893, + "num_tokens": 112803649.0, + "step": 4359 + }, + { + "epoch": 0.4788051833955634, + "grad_norm": 2.163473606109619, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6841856241226196, + "num_tokens": 112831812.0, + "step": 4360 + }, + { + "epoch": 0.47891500109817703, + "grad_norm": 2.1180970668792725, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7247984409332275, + "num_tokens": 112858557.0, + "step": 4361 + }, + { + "epoch": 0.4790248188007907, + "grad_norm": 2.1659584045410156, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7065849304199219, + "num_tokens": 112884800.0, + "step": 4362 + }, + { + "epoch": 0.4791346365034044, + "grad_norm": 2.1056125164031982, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7012110352516174, + "num_tokens": 112913848.0, + "step": 4363 + }, + { + "epoch": 0.479244454206018, + "grad_norm": 2.238842725753784, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7063357830047607, + "num_tokens": 112939866.0, + "step": 4364 + }, + { + "epoch": 0.47935427190863167, + "grad_norm": 2.1782031059265137, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7364871501922607, + "num_tokens": 112965056.0, + "step": 4365 + }, + { + "epoch": 0.4794640896112453, + "grad_norm": 2.077566623687744, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6877542734146118, + "num_tokens": 112993346.0, + "step": 4366 + }, + { + "epoch": 0.479573907313859, + "grad_norm": 2.238356351852417, + "learning_rate": 1e-06, + "loss": 1.077, + "mean_token_accuracy": 0.6837068796157837, + "num_tokens": 113020155.0, + "step": 4367 + }, + { + "epoch": 0.47968372501647266, + "grad_norm": 2.2549006938934326, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6972531080245972, + "num_tokens": 113044250.0, + "step": 4368 + }, + { + "epoch": 0.4797935427190863, + "grad_norm": 2.123765468597412, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6966037154197693, + "num_tokens": 113072237.0, + "step": 4369 + }, + { + "epoch": 0.47990336042169995, + "grad_norm": 2.263857126235962, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7082390785217285, + "num_tokens": 113096600.0, + "step": 4370 + }, + { + "epoch": 0.48001317812431366, + "grad_norm": 1.9126904010772705, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6942144632339478, + "num_tokens": 113131476.0, + "step": 4371 + }, + { + "epoch": 0.4801229958269273, + "grad_norm": 2.0890746116638184, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6881539821624756, + "num_tokens": 113161614.0, + "step": 4372 + }, + { + "epoch": 0.48023281352954095, + "grad_norm": 2.3725764751434326, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7003042101860046, + "num_tokens": 113184723.0, + "step": 4373 + }, + { + "epoch": 0.48034263123215465, + "grad_norm": 2.227562189102173, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.6964746117591858, + "num_tokens": 113207737.0, + "step": 4374 + }, + { + "epoch": 0.4804524489347683, + "grad_norm": 2.1547915935516357, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6993456482887268, + "num_tokens": 113236517.0, + "step": 4375 + }, + { + "epoch": 0.48056226663738194, + "grad_norm": 2.271378517150879, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7143515944480896, + "num_tokens": 113258955.0, + "step": 4376 + }, + { + "epoch": 0.4806720843399956, + "grad_norm": 2.1376936435699463, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.700729250907898, + "num_tokens": 113287178.0, + "step": 4377 + }, + { + "epoch": 0.4807819020426093, + "grad_norm": 2.1638081073760986, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7067384123802185, + "num_tokens": 113314384.0, + "step": 4378 + }, + { + "epoch": 0.48089171974522293, + "grad_norm": 2.479417085647583, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.695508599281311, + "num_tokens": 113335624.0, + "step": 4379 + }, + { + "epoch": 0.4810015374478366, + "grad_norm": 2.1658554077148438, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.6827227473258972, + "num_tokens": 113364171.0, + "step": 4380 + }, + { + "epoch": 0.4811113551504502, + "grad_norm": 2.60658597946167, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7145739197731018, + "num_tokens": 113385080.0, + "step": 4381 + }, + { + "epoch": 0.4812211728530639, + "grad_norm": 2.1427671909332275, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.6987795829772949, + "num_tokens": 113413851.0, + "step": 4382 + }, + { + "epoch": 0.4813309905556776, + "grad_norm": 2.141606569290161, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6825717687606812, + "num_tokens": 113444131.0, + "step": 4383 + }, + { + "epoch": 0.4814408082582912, + "grad_norm": 2.0288898944854736, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.688965380191803, + "num_tokens": 113473839.0, + "step": 4384 + }, + { + "epoch": 0.4815506259609049, + "grad_norm": 2.340705156326294, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6932741403579712, + "num_tokens": 113496652.0, + "step": 4385 + }, + { + "epoch": 0.48166044366351857, + "grad_norm": 2.403151750564575, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7011879086494446, + "num_tokens": 113517813.0, + "step": 4386 + }, + { + "epoch": 0.4817702613661322, + "grad_norm": 2.175903081893921, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.691008448600769, + "num_tokens": 113544589.0, + "step": 4387 + }, + { + "epoch": 0.48188007906874586, + "grad_norm": 2.0300827026367188, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6966861486434937, + "num_tokens": 113575425.0, + "step": 4388 + }, + { + "epoch": 0.48198989677135956, + "grad_norm": 2.141558885574341, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7147006392478943, + "num_tokens": 113600542.0, + "step": 4389 + }, + { + "epoch": 0.4820997144739732, + "grad_norm": 2.1634771823883057, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7085578441619873, + "num_tokens": 113626938.0, + "step": 4390 + }, + { + "epoch": 0.48220953217658685, + "grad_norm": 2.223400831222534, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6949846744537354, + "num_tokens": 113653288.0, + "step": 4391 + }, + { + "epoch": 0.48231934987920055, + "grad_norm": 2.1589174270629883, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7010942697525024, + "num_tokens": 113680097.0, + "step": 4392 + }, + { + "epoch": 0.4824291675818142, + "grad_norm": 2.3999135494232178, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6946367621421814, + "num_tokens": 113703073.0, + "step": 4393 + }, + { + "epoch": 0.48253898528442785, + "grad_norm": 2.3243608474731445, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7033030986785889, + "num_tokens": 113727831.0, + "step": 4394 + }, + { + "epoch": 0.4826488029870415, + "grad_norm": 2.3793997764587402, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.7041659355163574, + "num_tokens": 113750652.0, + "step": 4395 + }, + { + "epoch": 0.4827586206896552, + "grad_norm": 2.173151969909668, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7009804844856262, + "num_tokens": 113777924.0, + "step": 4396 + }, + { + "epoch": 0.48286843839226884, + "grad_norm": 2.325456142425537, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7150207757949829, + "num_tokens": 113801616.0, + "step": 4397 + }, + { + "epoch": 0.4829782560948825, + "grad_norm": 2.223809003829956, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6869929432868958, + "num_tokens": 113827240.0, + "step": 4398 + }, + { + "epoch": 0.48308807379749613, + "grad_norm": 2.450953483581543, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.733593225479126, + "num_tokens": 113847276.0, + "step": 4399 + }, + { + "epoch": 0.48319789150010983, + "grad_norm": 2.029958724975586, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.703027069568634, + "num_tokens": 113876130.0, + "step": 4400 + }, + { + "epoch": 0.4833077092027235, + "grad_norm": 2.314133882522583, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.7068685293197632, + "num_tokens": 113900524.0, + "step": 4401 + }, + { + "epoch": 0.4834175269053371, + "grad_norm": 2.522390604019165, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7155415415763855, + "num_tokens": 113920621.0, + "step": 4402 + }, + { + "epoch": 0.4835273446079508, + "grad_norm": 2.299116373062134, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7233448624610901, + "num_tokens": 113942610.0, + "step": 4403 + }, + { + "epoch": 0.4836371623105645, + "grad_norm": 2.128934621810913, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6892075538635254, + "num_tokens": 113968029.0, + "step": 4404 + }, + { + "epoch": 0.4837469800131781, + "grad_norm": 2.324986696243286, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.712974488735199, + "num_tokens": 113992927.0, + "step": 4405 + }, + { + "epoch": 0.48385679771579176, + "grad_norm": 2.2506182193756104, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6967138051986694, + "num_tokens": 114018573.0, + "step": 4406 + }, + { + "epoch": 0.48396661541840547, + "grad_norm": 2.1100404262542725, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6844161748886108, + "num_tokens": 114048655.0, + "step": 4407 + }, + { + "epoch": 0.4840764331210191, + "grad_norm": 2.2725703716278076, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6968897581100464, + "num_tokens": 114071771.0, + "step": 4408 + }, + { + "epoch": 0.48418625082363276, + "grad_norm": 2.4115819931030273, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.7051980495452881, + "num_tokens": 114094334.0, + "step": 4409 + }, + { + "epoch": 0.4842960685262464, + "grad_norm": 2.3376193046569824, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7112374305725098, + "num_tokens": 114118168.0, + "step": 4410 + }, + { + "epoch": 0.4844058862288601, + "grad_norm": 2.5547268390655518, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6959241032600403, + "num_tokens": 114138611.0, + "step": 4411 + }, + { + "epoch": 0.48451570393147375, + "grad_norm": 2.309593677520752, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7192562818527222, + "num_tokens": 114160678.0, + "step": 4412 + }, + { + "epoch": 0.4846255216340874, + "grad_norm": 2.4768009185791016, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7321508526802063, + "num_tokens": 114180098.0, + "step": 4413 + }, + { + "epoch": 0.4847353393367011, + "grad_norm": 2.333507776260376, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7217538356781006, + "num_tokens": 114203210.0, + "step": 4414 + }, + { + "epoch": 0.48484515703931474, + "grad_norm": 2.004099130630493, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7041372060775757, + "num_tokens": 114232636.0, + "step": 4415 + }, + { + "epoch": 0.4849549747419284, + "grad_norm": 2.285550832748413, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6985521912574768, + "num_tokens": 114256563.0, + "step": 4416 + }, + { + "epoch": 0.48506479244454204, + "grad_norm": 2.440260171890259, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6988366842269897, + "num_tokens": 114282338.0, + "step": 4417 + }, + { + "epoch": 0.48517461014715574, + "grad_norm": 2.25408673286438, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6919301748275757, + "num_tokens": 114306573.0, + "step": 4418 + }, + { + "epoch": 0.4852844278497694, + "grad_norm": 2.1808223724365234, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7148895859718323, + "num_tokens": 114330804.0, + "step": 4419 + }, + { + "epoch": 0.48539424555238303, + "grad_norm": 1.9739141464233398, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.682019829750061, + "num_tokens": 114362071.0, + "step": 4420 + }, + { + "epoch": 0.48550406325499673, + "grad_norm": 2.321180820465088, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.6856716871261597, + "num_tokens": 114386985.0, + "step": 4421 + }, + { + "epoch": 0.4856138809576104, + "grad_norm": 2.3899779319763184, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7136412858963013, + "num_tokens": 114408362.0, + "step": 4422 + }, + { + "epoch": 0.485723698660224, + "grad_norm": 2.0839786529541016, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6957315802574158, + "num_tokens": 114437239.0, + "step": 4423 + }, + { + "epoch": 0.48583351636283767, + "grad_norm": 2.0650570392608643, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6805495023727417, + "num_tokens": 114466495.0, + "step": 4424 + }, + { + "epoch": 0.48594333406545137, + "grad_norm": 2.5854640007019043, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7136520147323608, + "num_tokens": 114486989.0, + "step": 4425 + }, + { + "epoch": 0.486053151768065, + "grad_norm": 2.3615524768829346, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6993118524551392, + "num_tokens": 114509544.0, + "step": 4426 + }, + { + "epoch": 0.48616296947067866, + "grad_norm": 2.347149610519409, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6900277137756348, + "num_tokens": 114533663.0, + "step": 4427 + }, + { + "epoch": 0.4862727871732923, + "grad_norm": 2.0510990619659424, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6816385984420776, + "num_tokens": 114562909.0, + "step": 4428 + }, + { + "epoch": 0.486382604875906, + "grad_norm": 2.346466541290283, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6856143474578857, + "num_tokens": 114585352.0, + "step": 4429 + }, + { + "epoch": 0.48649242257851966, + "grad_norm": 2.432651996612549, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7173563838005066, + "num_tokens": 114607829.0, + "step": 4430 + }, + { + "epoch": 0.4866022402811333, + "grad_norm": 2.0576348304748535, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7212160229682922, + "num_tokens": 114635523.0, + "step": 4431 + }, + { + "epoch": 0.486712057983747, + "grad_norm": 2.114347457885742, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7187221050262451, + "num_tokens": 114660260.0, + "step": 4432 + }, + { + "epoch": 0.48682187568636065, + "grad_norm": 2.1432504653930664, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6836546659469604, + "num_tokens": 114688684.0, + "step": 4433 + }, + { + "epoch": 0.4869316933889743, + "grad_norm": 2.3540618419647217, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7183033227920532, + "num_tokens": 114711677.0, + "step": 4434 + }, + { + "epoch": 0.48704151109158794, + "grad_norm": 2.2994227409362793, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.7001560926437378, + "num_tokens": 114735640.0, + "step": 4435 + }, + { + "epoch": 0.48715132879420164, + "grad_norm": 2.0424952507019043, + "learning_rate": 1e-06, + "loss": 1.0403, + "mean_token_accuracy": 0.688179075717926, + "num_tokens": 114765799.0, + "step": 4436 + }, + { + "epoch": 0.4872611464968153, + "grad_norm": 1.9440985918045044, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.687950611114502, + "num_tokens": 114795705.0, + "step": 4437 + }, + { + "epoch": 0.48737096419942894, + "grad_norm": 2.3368687629699707, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6939618587493896, + "num_tokens": 114820558.0, + "step": 4438 + }, + { + "epoch": 0.48748078190204264, + "grad_norm": 2.1135923862457275, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7002049684524536, + "num_tokens": 114847518.0, + "step": 4439 + }, + { + "epoch": 0.4875905996046563, + "grad_norm": 2.0488386154174805, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.692990779876709, + "num_tokens": 114875666.0, + "step": 4440 + }, + { + "epoch": 0.48770041730726993, + "grad_norm": 2.1204230785369873, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.6988241672515869, + "num_tokens": 114906232.0, + "step": 4441 + }, + { + "epoch": 0.4878102350098836, + "grad_norm": 2.4555318355560303, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7057263255119324, + "num_tokens": 114926526.0, + "step": 4442 + }, + { + "epoch": 0.4879200527124973, + "grad_norm": 2.202164649963379, + "learning_rate": 1e-06, + "loss": 1.0685, + "mean_token_accuracy": 0.6873137950897217, + "num_tokens": 114954475.0, + "step": 4443 + }, + { + "epoch": 0.4880298704151109, + "grad_norm": 2.322120428085327, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6913274526596069, + "num_tokens": 114981638.0, + "step": 4444 + }, + { + "epoch": 0.48813968811772457, + "grad_norm": 2.2220020294189453, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7192338705062866, + "num_tokens": 115006462.0, + "step": 4445 + }, + { + "epoch": 0.4882495058203382, + "grad_norm": 2.2236709594726562, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.716528058052063, + "num_tokens": 115031636.0, + "step": 4446 + }, + { + "epoch": 0.4883593235229519, + "grad_norm": 2.2017035484313965, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7093398571014404, + "num_tokens": 115056335.0, + "step": 4447 + }, + { + "epoch": 0.48846914122556556, + "grad_norm": 2.3600475788116455, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6951100826263428, + "num_tokens": 115078790.0, + "step": 4448 + }, + { + "epoch": 0.4885789589281792, + "grad_norm": 2.3887581825256348, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.717747688293457, + "num_tokens": 115101699.0, + "step": 4449 + }, + { + "epoch": 0.4886887766307929, + "grad_norm": 2.254035234451294, + "learning_rate": 1e-06, + "loss": 1.0845, + "mean_token_accuracy": 0.6747778058052063, + "num_tokens": 115128503.0, + "step": 4450 + }, + { + "epoch": 0.48879859433340656, + "grad_norm": 1.8830138444900513, + "learning_rate": 1e-06, + "loss": 1.0766, + "mean_token_accuracy": 0.6845173835754395, + "num_tokens": 115164634.0, + "step": 4451 + }, + { + "epoch": 0.4889084120360202, + "grad_norm": 2.384597063064575, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6896920204162598, + "num_tokens": 115187246.0, + "step": 4452 + }, + { + "epoch": 0.48901822973863385, + "grad_norm": 2.125462293624878, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7057109475135803, + "num_tokens": 115213586.0, + "step": 4453 + }, + { + "epoch": 0.48912804744124755, + "grad_norm": 2.4153220653533936, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7254765629768372, + "num_tokens": 115235227.0, + "step": 4454 + }, + { + "epoch": 0.4892378651438612, + "grad_norm": 2.023644208908081, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7060418128967285, + "num_tokens": 115263710.0, + "step": 4455 + }, + { + "epoch": 0.48934768284647484, + "grad_norm": 2.2710156440734863, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7131777405738831, + "num_tokens": 115289195.0, + "step": 4456 + }, + { + "epoch": 0.4894575005490885, + "grad_norm": 2.215120315551758, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7014430165290833, + "num_tokens": 115313974.0, + "step": 4457 + }, + { + "epoch": 0.4895673182517022, + "grad_norm": 2.2388434410095215, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6997016072273254, + "num_tokens": 115339114.0, + "step": 4458 + }, + { + "epoch": 0.48967713595431583, + "grad_norm": 2.2253644466400146, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.695270836353302, + "num_tokens": 115365289.0, + "step": 4459 + }, + { + "epoch": 0.4897869536569295, + "grad_norm": 2.331068515777588, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7062544822692871, + "num_tokens": 115387444.0, + "step": 4460 + }, + { + "epoch": 0.4898967713595432, + "grad_norm": 2.325309991836548, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6981235146522522, + "num_tokens": 115409230.0, + "step": 4461 + }, + { + "epoch": 0.4900065890621568, + "grad_norm": 2.455329179763794, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7185531854629517, + "num_tokens": 115430694.0, + "step": 4462 + }, + { + "epoch": 0.4901164067647705, + "grad_norm": 2.227381706237793, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6857244968414307, + "num_tokens": 115456685.0, + "step": 4463 + }, + { + "epoch": 0.4902262244673841, + "grad_norm": 2.1176390647888184, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6917213201522827, + "num_tokens": 115483755.0, + "step": 4464 + }, + { + "epoch": 0.4903360421699978, + "grad_norm": 2.549388885498047, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.723731279373169, + "num_tokens": 115502535.0, + "step": 4465 + }, + { + "epoch": 0.49044585987261147, + "grad_norm": 2.281886100769043, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7013343572616577, + "num_tokens": 115526350.0, + "step": 4466 + }, + { + "epoch": 0.4905556775752251, + "grad_norm": 2.4808943271636963, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7146027088165283, + "num_tokens": 115548180.0, + "step": 4467 + }, + { + "epoch": 0.4906654952778388, + "grad_norm": 2.079336643218994, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6959108710289001, + "num_tokens": 115575251.0, + "step": 4468 + }, + { + "epoch": 0.49077531298045246, + "grad_norm": 2.385861873626709, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7278814315795898, + "num_tokens": 115595870.0, + "step": 4469 + }, + { + "epoch": 0.4908851306830661, + "grad_norm": 2.165107011795044, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6893719434738159, + "num_tokens": 115622500.0, + "step": 4470 + }, + { + "epoch": 0.49099494838567975, + "grad_norm": 2.4635977745056152, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.6904343962669373, + "num_tokens": 115644812.0, + "step": 4471 + }, + { + "epoch": 0.49110476608829345, + "grad_norm": 2.294576406478882, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7121753096580505, + "num_tokens": 115668225.0, + "step": 4472 + }, + { + "epoch": 0.4912145837909071, + "grad_norm": 2.455043077468872, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7243953347206116, + "num_tokens": 115690591.0, + "step": 4473 + }, + { + "epoch": 0.49132440149352075, + "grad_norm": 2.3417904376983643, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.684624433517456, + "num_tokens": 115715069.0, + "step": 4474 + }, + { + "epoch": 0.4914342191961344, + "grad_norm": 2.406963348388672, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7026269435882568, + "num_tokens": 115737035.0, + "step": 4475 + }, + { + "epoch": 0.4915440368987481, + "grad_norm": 2.1525509357452393, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7058336734771729, + "num_tokens": 115762455.0, + "step": 4476 + }, + { + "epoch": 0.49165385460136174, + "grad_norm": 2.2815637588500977, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7072202563285828, + "num_tokens": 115786117.0, + "step": 4477 + }, + { + "epoch": 0.4917636723039754, + "grad_norm": 2.3482978343963623, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7261002063751221, + "num_tokens": 115809341.0, + "step": 4478 + }, + { + "epoch": 0.4918734900065891, + "grad_norm": 2.327813148498535, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7071161270141602, + "num_tokens": 115832269.0, + "step": 4479 + }, + { + "epoch": 0.49198330770920273, + "grad_norm": 2.223614454269409, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7016887664794922, + "num_tokens": 115859033.0, + "step": 4480 + }, + { + "epoch": 0.4920931254118164, + "grad_norm": 2.3517415523529053, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7078531980514526, + "num_tokens": 115881543.0, + "step": 4481 + }, + { + "epoch": 0.49220294311443, + "grad_norm": 2.1812095642089844, + "learning_rate": 1e-06, + "loss": 1.0805, + "mean_token_accuracy": 0.6833807229995728, + "num_tokens": 115908358.0, + "step": 4482 + }, + { + "epoch": 0.4923127608170437, + "grad_norm": 2.1554996967315674, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7046080827713013, + "num_tokens": 115934190.0, + "step": 4483 + }, + { + "epoch": 0.4924225785196574, + "grad_norm": 2.257371664047241, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6824781894683838, + "num_tokens": 115958843.0, + "step": 4484 + }, + { + "epoch": 0.492532396222271, + "grad_norm": 2.150273561477661, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6841171979904175, + "num_tokens": 115984943.0, + "step": 4485 + }, + { + "epoch": 0.49264221392488466, + "grad_norm": 2.0270118713378906, + "learning_rate": 1e-06, + "loss": 1.108, + "mean_token_accuracy": 0.6785356998443604, + "num_tokens": 116015263.0, + "step": 4486 + }, + { + "epoch": 0.49275203162749837, + "grad_norm": 2.1484804153442383, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6902450323104858, + "num_tokens": 116042471.0, + "step": 4487 + }, + { + "epoch": 0.492861849330112, + "grad_norm": 2.304882526397705, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.7043519020080566, + "num_tokens": 116067499.0, + "step": 4488 + }, + { + "epoch": 0.49297166703272566, + "grad_norm": 2.3770532608032227, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7184726595878601, + "num_tokens": 116088440.0, + "step": 4489 + }, + { + "epoch": 0.49308148473533936, + "grad_norm": 2.4077517986297607, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6866725087165833, + "num_tokens": 116111112.0, + "step": 4490 + }, + { + "epoch": 0.493191302437953, + "grad_norm": 2.2636539936065674, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7023283243179321, + "num_tokens": 116135455.0, + "step": 4491 + }, + { + "epoch": 0.49330112014056665, + "grad_norm": 2.283273935317993, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6897271871566772, + "num_tokens": 116160890.0, + "step": 4492 + }, + { + "epoch": 0.4934109378431803, + "grad_norm": 2.140726089477539, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7145403623580933, + "num_tokens": 116187808.0, + "step": 4493 + }, + { + "epoch": 0.493520755545794, + "grad_norm": 2.069391965866089, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6877928972244263, + "num_tokens": 116217838.0, + "step": 4494 + }, + { + "epoch": 0.49363057324840764, + "grad_norm": 2.082942008972168, + "learning_rate": 1e-06, + "loss": 1.0649, + "mean_token_accuracy": 0.6874297857284546, + "num_tokens": 116246981.0, + "step": 4495 + }, + { + "epoch": 0.4937403909510213, + "grad_norm": 2.079662799835205, + "learning_rate": 1e-06, + "loss": 1.0712, + "mean_token_accuracy": 0.6785938739776611, + "num_tokens": 116276532.0, + "step": 4496 + }, + { + "epoch": 0.493850208653635, + "grad_norm": 2.1486918926239014, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.708477258682251, + "num_tokens": 116302544.0, + "step": 4497 + }, + { + "epoch": 0.49396002635624864, + "grad_norm": 2.366929054260254, + "learning_rate": 1e-06, + "loss": 1.1011, + "mean_token_accuracy": 0.6749081611633301, + "num_tokens": 116326255.0, + "step": 4498 + }, + { + "epoch": 0.4940698440588623, + "grad_norm": 2.2206780910491943, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7055613994598389, + "num_tokens": 116351717.0, + "step": 4499 + }, + { + "epoch": 0.49417966176147593, + "grad_norm": 2.0902528762817383, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6895748972892761, + "num_tokens": 116381049.0, + "step": 4500 + }, + { + "epoch": 0.49428947946408963, + "grad_norm": 2.3051912784576416, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7141166925430298, + "num_tokens": 116404558.0, + "step": 4501 + }, + { + "epoch": 0.4943992971667033, + "grad_norm": 2.2583065032958984, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6951645612716675, + "num_tokens": 116430821.0, + "step": 4502 + }, + { + "epoch": 0.4945091148693169, + "grad_norm": 2.184558629989624, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7161504626274109, + "num_tokens": 116456843.0, + "step": 4503 + }, + { + "epoch": 0.49461893257193057, + "grad_norm": 1.999565839767456, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7022198438644409, + "num_tokens": 116488522.0, + "step": 4504 + }, + { + "epoch": 0.49472875027454427, + "grad_norm": 1.9432545900344849, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6904398202896118, + "num_tokens": 116518860.0, + "step": 4505 + }, + { + "epoch": 0.4948385679771579, + "grad_norm": 2.2762043476104736, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7126747369766235, + "num_tokens": 116545374.0, + "step": 4506 + }, + { + "epoch": 0.49494838567977156, + "grad_norm": 2.2272353172302246, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7046313285827637, + "num_tokens": 116572772.0, + "step": 4507 + }, + { + "epoch": 0.49505820338238526, + "grad_norm": 2.033717632293701, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6950317025184631, + "num_tokens": 116601968.0, + "step": 4508 + }, + { + "epoch": 0.4951680210849989, + "grad_norm": 2.4174344539642334, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.699494481086731, + "num_tokens": 116624913.0, + "step": 4509 + }, + { + "epoch": 0.49527783878761256, + "grad_norm": 2.4199070930480957, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6990050673484802, + "num_tokens": 116646611.0, + "step": 4510 + }, + { + "epoch": 0.4953876564902262, + "grad_norm": 2.3808376789093018, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.694622278213501, + "num_tokens": 116669267.0, + "step": 4511 + }, + { + "epoch": 0.4954974741928399, + "grad_norm": 2.2646186351776123, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7091478109359741, + "num_tokens": 116694288.0, + "step": 4512 + }, + { + "epoch": 0.49560729189545355, + "grad_norm": 2.2474162578582764, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7115495204925537, + "num_tokens": 116719128.0, + "step": 4513 + }, + { + "epoch": 0.4957171095980672, + "grad_norm": 2.320979595184326, + "learning_rate": 1e-06, + "loss": 1.0762, + "mean_token_accuracy": 0.6862464547157288, + "num_tokens": 116745096.0, + "step": 4514 + }, + { + "epoch": 0.4958269273006809, + "grad_norm": 2.60321044921875, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7142416834831238, + "num_tokens": 116763826.0, + "step": 4515 + }, + { + "epoch": 0.49593674500329454, + "grad_norm": 2.130825996398926, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7036228179931641, + "num_tokens": 116795641.0, + "step": 4516 + }, + { + "epoch": 0.4960465627059082, + "grad_norm": 2.128006935119629, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6900194883346558, + "num_tokens": 116822911.0, + "step": 4517 + }, + { + "epoch": 0.49615638040852184, + "grad_norm": 2.3180477619171143, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7018988132476807, + "num_tokens": 116846246.0, + "step": 4518 + }, + { + "epoch": 0.49626619811113554, + "grad_norm": 2.0176846981048584, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7007637023925781, + "num_tokens": 116874624.0, + "step": 4519 + }, + { + "epoch": 0.4963760158137492, + "grad_norm": 2.0984532833099365, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7149311304092407, + "num_tokens": 116899231.0, + "step": 4520 + }, + { + "epoch": 0.49648583351636283, + "grad_norm": 2.3584418296813965, + "learning_rate": 1e-06, + "loss": 1.0846, + "mean_token_accuracy": 0.6730368733406067, + "num_tokens": 116924433.0, + "step": 4521 + }, + { + "epoch": 0.4965956512189765, + "grad_norm": 2.1836390495300293, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.6860435009002686, + "num_tokens": 116954796.0, + "step": 4522 + }, + { + "epoch": 0.4967054689215902, + "grad_norm": 1.9528781175613403, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7036216259002686, + "num_tokens": 116984925.0, + "step": 4523 + }, + { + "epoch": 0.4968152866242038, + "grad_norm": 2.093808650970459, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6983901858329773, + "num_tokens": 117012418.0, + "step": 4524 + }, + { + "epoch": 0.49692510432681747, + "grad_norm": 2.0175840854644775, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7151921987533569, + "num_tokens": 117040062.0, + "step": 4525 + }, + { + "epoch": 0.49703492202943117, + "grad_norm": 1.973805546760559, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6870739459991455, + "num_tokens": 117069854.0, + "step": 4526 + }, + { + "epoch": 0.4971447397320448, + "grad_norm": 2.005014657974243, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7003630995750427, + "num_tokens": 117098015.0, + "step": 4527 + }, + { + "epoch": 0.49725455743465846, + "grad_norm": 2.1953136920928955, + "learning_rate": 1e-06, + "loss": 1.0676, + "mean_token_accuracy": 0.6819332838058472, + "num_tokens": 117127797.0, + "step": 4528 + }, + { + "epoch": 0.4973643751372721, + "grad_norm": 2.4098124504089355, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.6962201595306396, + "num_tokens": 117147759.0, + "step": 4529 + }, + { + "epoch": 0.4974741928398858, + "grad_norm": 2.2554614543914795, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6936150193214417, + "num_tokens": 117171512.0, + "step": 4530 + }, + { + "epoch": 0.49758401054249946, + "grad_norm": 2.2816972732543945, + "learning_rate": 1e-06, + "loss": 1.0576, + "mean_token_accuracy": 0.6912941932678223, + "num_tokens": 117196904.0, + "step": 4531 + }, + { + "epoch": 0.4976938282451131, + "grad_norm": 2.4897537231445312, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7063430547714233, + "num_tokens": 117218115.0, + "step": 4532 + }, + { + "epoch": 0.49780364594772675, + "grad_norm": 2.096259593963623, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7105622887611389, + "num_tokens": 117244201.0, + "step": 4533 + }, + { + "epoch": 0.49791346365034045, + "grad_norm": 2.186786413192749, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.735440731048584, + "num_tokens": 117267186.0, + "step": 4534 + }, + { + "epoch": 0.4980232813529541, + "grad_norm": 1.9802043437957764, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6946396827697754, + "num_tokens": 117297522.0, + "step": 4535 + }, + { + "epoch": 0.49813309905556774, + "grad_norm": 2.296254873275757, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6937448382377625, + "num_tokens": 117323164.0, + "step": 4536 + }, + { + "epoch": 0.49824291675818144, + "grad_norm": 2.290318489074707, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.7006917595863342, + "num_tokens": 117347079.0, + "step": 4537 + }, + { + "epoch": 0.4983527344607951, + "grad_norm": 2.266757011413574, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7059733867645264, + "num_tokens": 117373678.0, + "step": 4538 + }, + { + "epoch": 0.49846255216340873, + "grad_norm": 2.1814534664154053, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6916707754135132, + "num_tokens": 117400721.0, + "step": 4539 + }, + { + "epoch": 0.4985723698660224, + "grad_norm": 2.283015251159668, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7122447490692139, + "num_tokens": 117425568.0, + "step": 4540 + }, + { + "epoch": 0.4986821875686361, + "grad_norm": 2.2870051860809326, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7005244493484497, + "num_tokens": 117448642.0, + "step": 4541 + }, + { + "epoch": 0.4987920052712497, + "grad_norm": 2.417983293533325, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7017076015472412, + "num_tokens": 117471867.0, + "step": 4542 + }, + { + "epoch": 0.4989018229738634, + "grad_norm": 1.9220302104949951, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.7006538510322571, + "num_tokens": 117503893.0, + "step": 4543 + }, + { + "epoch": 0.4990116406764771, + "grad_norm": 2.1257035732269287, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6950631141662598, + "num_tokens": 117532300.0, + "step": 4544 + }, + { + "epoch": 0.4991214583790907, + "grad_norm": 2.257415533065796, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7107893228530884, + "num_tokens": 117557057.0, + "step": 4545 + }, + { + "epoch": 0.49923127608170437, + "grad_norm": 2.0627408027648926, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6948195099830627, + "num_tokens": 117586763.0, + "step": 4546 + }, + { + "epoch": 0.499341093784318, + "grad_norm": 2.1890509128570557, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.7067708373069763, + "num_tokens": 117611783.0, + "step": 4547 + }, + { + "epoch": 0.4994509114869317, + "grad_norm": 2.111942768096924, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7275798320770264, + "num_tokens": 117638085.0, + "step": 4548 + }, + { + "epoch": 0.49956072918954536, + "grad_norm": 2.021425247192383, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6964349746704102, + "num_tokens": 117669103.0, + "step": 4549 + }, + { + "epoch": 0.499670546892159, + "grad_norm": 2.327709197998047, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6887370347976685, + "num_tokens": 117693830.0, + "step": 4550 + }, + { + "epoch": 0.49978036459477265, + "grad_norm": 2.497495651245117, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.691548228263855, + "num_tokens": 117714230.0, + "step": 4551 + }, + { + "epoch": 0.49989018229738635, + "grad_norm": 1.9121284484863281, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.6932452917098999, + "num_tokens": 117747833.0, + "step": 4552 + }, + { + "epoch": 0.5, + "grad_norm": 2.3007848262786865, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.684825599193573, + "num_tokens": 117773052.0, + "step": 4553 + }, + { + "epoch": 0.5001098177026136, + "grad_norm": 2.1334996223449707, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7068653106689453, + "num_tokens": 117798234.0, + "step": 4554 + }, + { + "epoch": 0.5002196354052273, + "grad_norm": 2.506024122238159, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7108460664749146, + "num_tokens": 117820361.0, + "step": 4555 + }, + { + "epoch": 0.5003294531078409, + "grad_norm": 2.245668649673462, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.687017560005188, + "num_tokens": 117846772.0, + "step": 4556 + }, + { + "epoch": 0.5004392708104547, + "grad_norm": 2.011032819747925, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.691055178642273, + "num_tokens": 117877864.0, + "step": 4557 + }, + { + "epoch": 0.5005490885130683, + "grad_norm": 2.176126480102539, + "learning_rate": 1e-06, + "loss": 1.0514, + "mean_token_accuracy": 0.6810296773910522, + "num_tokens": 117905908.0, + "step": 4558 + }, + { + "epoch": 0.500658906215682, + "grad_norm": 2.3303205966949463, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6995985507965088, + "num_tokens": 117931453.0, + "step": 4559 + }, + { + "epoch": 0.5007687239182956, + "grad_norm": 2.4366259574890137, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.707539975643158, + "num_tokens": 117953672.0, + "step": 4560 + }, + { + "epoch": 0.5008785416209093, + "grad_norm": 2.195266008377075, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6876538395881653, + "num_tokens": 117979254.0, + "step": 4561 + }, + { + "epoch": 0.5009883593235229, + "grad_norm": 2.0928707122802734, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.704609215259552, + "num_tokens": 118007249.0, + "step": 4562 + }, + { + "epoch": 0.5010981770261366, + "grad_norm": 2.0943431854248047, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.6824921369552612, + "num_tokens": 118038228.0, + "step": 4563 + }, + { + "epoch": 0.5012079947287503, + "grad_norm": 2.1326704025268555, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.7037478685379028, + "num_tokens": 118064982.0, + "step": 4564 + }, + { + "epoch": 0.501317812431364, + "grad_norm": 2.228989839553833, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7044933438301086, + "num_tokens": 118088105.0, + "step": 4565 + }, + { + "epoch": 0.5014276301339776, + "grad_norm": 1.888451099395752, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6956332921981812, + "num_tokens": 118123176.0, + "step": 4566 + }, + { + "epoch": 0.5015374478365913, + "grad_norm": 2.131950616836548, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7023516297340393, + "num_tokens": 118150578.0, + "step": 4567 + }, + { + "epoch": 0.5016472655392049, + "grad_norm": 2.109530210494995, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7232646942138672, + "num_tokens": 118177358.0, + "step": 4568 + }, + { + "epoch": 0.5017570832418186, + "grad_norm": 2.1467032432556152, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.700469970703125, + "num_tokens": 118207315.0, + "step": 4569 + }, + { + "epoch": 0.5018669009444322, + "grad_norm": 2.221923351287842, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7011791467666626, + "num_tokens": 118231938.0, + "step": 4570 + }, + { + "epoch": 0.5019767186470458, + "grad_norm": 2.0205624103546143, + "learning_rate": 1e-06, + "loss": 1.098, + "mean_token_accuracy": 0.6697096228599548, + "num_tokens": 118264767.0, + "step": 4571 + }, + { + "epoch": 0.5020865363496596, + "grad_norm": 2.3255343437194824, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6880275011062622, + "num_tokens": 118287546.0, + "step": 4572 + }, + { + "epoch": 0.5021963540522733, + "grad_norm": 2.2950892448425293, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7109057307243347, + "num_tokens": 118310519.0, + "step": 4573 + }, + { + "epoch": 0.5023061717548869, + "grad_norm": 2.1505231857299805, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7023793458938599, + "num_tokens": 118335584.0, + "step": 4574 + }, + { + "epoch": 0.5024159894575005, + "grad_norm": 2.418919563293457, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7166111469268799, + "num_tokens": 118355882.0, + "step": 4575 + }, + { + "epoch": 0.5025258071601142, + "grad_norm": 2.2781903743743896, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6988682746887207, + "num_tokens": 118379758.0, + "step": 4576 + }, + { + "epoch": 0.5026356248627278, + "grad_norm": 2.086514472961426, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6938419342041016, + "num_tokens": 118409998.0, + "step": 4577 + }, + { + "epoch": 0.5027454425653415, + "grad_norm": 2.064201593399048, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7032725811004639, + "num_tokens": 118437895.0, + "step": 4578 + }, + { + "epoch": 0.5028552602679552, + "grad_norm": 1.9911032915115356, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6953251361846924, + "num_tokens": 118466254.0, + "step": 4579 + }, + { + "epoch": 0.5029650779705689, + "grad_norm": 2.298449754714966, + "learning_rate": 1e-06, + "loss": 1.0649, + "mean_token_accuracy": 0.6878041625022888, + "num_tokens": 118491266.0, + "step": 4580 + }, + { + "epoch": 0.5030748956731825, + "grad_norm": 2.433431625366211, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.6876669526100159, + "num_tokens": 118513453.0, + "step": 4581 + }, + { + "epoch": 0.5031847133757962, + "grad_norm": 2.373302459716797, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6879304051399231, + "num_tokens": 118536240.0, + "step": 4582 + }, + { + "epoch": 0.5032945310784098, + "grad_norm": 2.0519275665283203, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6963467001914978, + "num_tokens": 118566445.0, + "step": 4583 + }, + { + "epoch": 0.5034043487810235, + "grad_norm": 2.2540385723114014, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6941207647323608, + "num_tokens": 118591213.0, + "step": 4584 + }, + { + "epoch": 0.5035141664836371, + "grad_norm": 2.1448471546173096, + "learning_rate": 1e-06, + "loss": 1.1238, + "mean_token_accuracy": 0.6707041263580322, + "num_tokens": 118621747.0, + "step": 4585 + }, + { + "epoch": 0.5036239841862509, + "grad_norm": 1.76329505443573, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6959381103515625, + "num_tokens": 118660496.0, + "step": 4586 + }, + { + "epoch": 0.5037338018888645, + "grad_norm": 2.3959555625915527, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.714589536190033, + "num_tokens": 118682445.0, + "step": 4587 + }, + { + "epoch": 0.5038436195914782, + "grad_norm": 2.1337060928344727, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.7021665573120117, + "num_tokens": 118709958.0, + "step": 4588 + }, + { + "epoch": 0.5039534372940918, + "grad_norm": 2.198692560195923, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7069158554077148, + "num_tokens": 118736291.0, + "step": 4589 + }, + { + "epoch": 0.5040632549967055, + "grad_norm": 2.0969974994659424, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6888085603713989, + "num_tokens": 118763755.0, + "step": 4590 + }, + { + "epoch": 0.5041730726993191, + "grad_norm": 2.3503425121307373, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7110925912857056, + "num_tokens": 118785592.0, + "step": 4591 + }, + { + "epoch": 0.5042828904019327, + "grad_norm": 1.9320701360702515, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6992034912109375, + "num_tokens": 118815809.0, + "step": 4592 + }, + { + "epoch": 0.5043927081045465, + "grad_norm": 1.9739536046981812, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7020223140716553, + "num_tokens": 118844118.0, + "step": 4593 + }, + { + "epoch": 0.5045025258071602, + "grad_norm": 1.8657500743865967, + "learning_rate": 1e-06, + "loss": 1.1341, + "mean_token_accuracy": 0.6599940657615662, + "num_tokens": 118878278.0, + "step": 4594 + }, + { + "epoch": 0.5046123435097738, + "grad_norm": 2.1088576316833496, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6939961910247803, + "num_tokens": 118904888.0, + "step": 4595 + }, + { + "epoch": 0.5047221612123874, + "grad_norm": 1.988187551498413, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.700215220451355, + "num_tokens": 118936516.0, + "step": 4596 + }, + { + "epoch": 0.5048319789150011, + "grad_norm": 1.975420355796814, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6875752210617065, + "num_tokens": 118965982.0, + "step": 4597 + }, + { + "epoch": 0.5049417966176147, + "grad_norm": 2.225273609161377, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7144040465354919, + "num_tokens": 118989660.0, + "step": 4598 + }, + { + "epoch": 0.5050516143202284, + "grad_norm": 2.2915029525756836, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7123355269432068, + "num_tokens": 119014512.0, + "step": 4599 + }, + { + "epoch": 0.505161432022842, + "grad_norm": 1.958397388458252, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6800810098648071, + "num_tokens": 119046114.0, + "step": 4600 + }, + { + "epoch": 0.5052712497254558, + "grad_norm": 2.4902331829071045, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7001990675926208, + "num_tokens": 119069967.0, + "step": 4601 + }, + { + "epoch": 0.5053810674280694, + "grad_norm": 2.056643486022949, + "learning_rate": 1e-06, + "loss": 1.0758, + "mean_token_accuracy": 0.6805078387260437, + "num_tokens": 119102712.0, + "step": 4602 + }, + { + "epoch": 0.5054908851306831, + "grad_norm": 2.1445841789245605, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.699700117111206, + "num_tokens": 119131270.0, + "step": 4603 + }, + { + "epoch": 0.5056007028332967, + "grad_norm": 2.9387295246124268, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7159692645072937, + "num_tokens": 119147580.0, + "step": 4604 + }, + { + "epoch": 0.5057105205359104, + "grad_norm": 2.0890872478485107, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6803457140922546, + "num_tokens": 119176404.0, + "step": 4605 + }, + { + "epoch": 0.505820338238524, + "grad_norm": 2.1852571964263916, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6989436745643616, + "num_tokens": 119203393.0, + "step": 4606 + }, + { + "epoch": 0.5059301559411377, + "grad_norm": 2.1129891872406006, + "learning_rate": 1e-06, + "loss": 1.0904, + "mean_token_accuracy": 0.674360990524292, + "num_tokens": 119231833.0, + "step": 4607 + }, + { + "epoch": 0.5060399736437514, + "grad_norm": 2.3522446155548096, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7009961605072021, + "num_tokens": 119256386.0, + "step": 4608 + }, + { + "epoch": 0.5061497913463651, + "grad_norm": 2.067883014678955, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6830322742462158, + "num_tokens": 119285024.0, + "step": 4609 + }, + { + "epoch": 0.5062596090489787, + "grad_norm": 2.28262996673584, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7217714786529541, + "num_tokens": 119308733.0, + "step": 4610 + }, + { + "epoch": 0.5063694267515924, + "grad_norm": 2.312537431716919, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6946385502815247, + "num_tokens": 119332389.0, + "step": 4611 + }, + { + "epoch": 0.506479244454206, + "grad_norm": 2.1395022869110107, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6937799453735352, + "num_tokens": 119360140.0, + "step": 4612 + }, + { + "epoch": 0.5065890621568196, + "grad_norm": 2.472364664077759, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.718761682510376, + "num_tokens": 119381098.0, + "step": 4613 + }, + { + "epoch": 0.5066988798594333, + "grad_norm": 2.4623894691467285, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7288844585418701, + "num_tokens": 119406309.0, + "step": 4614 + }, + { + "epoch": 0.506808697562047, + "grad_norm": 2.3188397884368896, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7247058153152466, + "num_tokens": 119429642.0, + "step": 4615 + }, + { + "epoch": 0.5069185152646607, + "grad_norm": 1.9316614866256714, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6924373507499695, + "num_tokens": 119461475.0, + "step": 4616 + }, + { + "epoch": 0.5070283329672743, + "grad_norm": 2.281477451324463, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.711689829826355, + "num_tokens": 119485600.0, + "step": 4617 + }, + { + "epoch": 0.507138150669888, + "grad_norm": 2.2047626972198486, + "learning_rate": 1e-06, + "loss": 1.0812, + "mean_token_accuracy": 0.68398118019104, + "num_tokens": 119511824.0, + "step": 4618 + }, + { + "epoch": 0.5072479683725016, + "grad_norm": 2.4529173374176025, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7100197076797485, + "num_tokens": 119533133.0, + "step": 4619 + }, + { + "epoch": 0.5073577860751153, + "grad_norm": 2.267587661743164, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7010245323181152, + "num_tokens": 119557080.0, + "step": 4620 + }, + { + "epoch": 0.5074676037777289, + "grad_norm": 2.471935987472534, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6911236047744751, + "num_tokens": 119577569.0, + "step": 4621 + }, + { + "epoch": 0.5075774214803427, + "grad_norm": 2.2563602924346924, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6924718618392944, + "num_tokens": 119605645.0, + "step": 4622 + }, + { + "epoch": 0.5076872391829563, + "grad_norm": 2.3947904109954834, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7072919011116028, + "num_tokens": 119626742.0, + "step": 4623 + }, + { + "epoch": 0.50779705688557, + "grad_norm": 2.390291690826416, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7009230852127075, + "num_tokens": 119648331.0, + "step": 4624 + }, + { + "epoch": 0.5079068745881836, + "grad_norm": 2.3154115676879883, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7084300518035889, + "num_tokens": 119671073.0, + "step": 4625 + }, + { + "epoch": 0.5080166922907973, + "grad_norm": 2.190861225128174, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7064846158027649, + "num_tokens": 119697145.0, + "step": 4626 + }, + { + "epoch": 0.5081265099934109, + "grad_norm": 2.4276976585388184, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6920651197433472, + "num_tokens": 119719279.0, + "step": 4627 + }, + { + "epoch": 0.5082363276960246, + "grad_norm": 2.0524344444274902, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7069610357284546, + "num_tokens": 119746924.0, + "step": 4628 + }, + { + "epoch": 0.5083461453986382, + "grad_norm": 2.4692227840423584, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6822605133056641, + "num_tokens": 119769037.0, + "step": 4629 + }, + { + "epoch": 0.508455963101252, + "grad_norm": 2.0118255615234375, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.687263011932373, + "num_tokens": 119797640.0, + "step": 4630 + }, + { + "epoch": 0.5085657808038656, + "grad_norm": 2.156262159347534, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7148748636245728, + "num_tokens": 119821460.0, + "step": 4631 + }, + { + "epoch": 0.5086755985064793, + "grad_norm": 2.088839054107666, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6846227645874023, + "num_tokens": 119850542.0, + "step": 4632 + }, + { + "epoch": 0.5087854162090929, + "grad_norm": 2.0397696495056152, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7070634365081787, + "num_tokens": 119877816.0, + "step": 4633 + }, + { + "epoch": 0.5088952339117065, + "grad_norm": 2.0339040756225586, + "learning_rate": 1e-06, + "loss": 1.0466, + "mean_token_accuracy": 0.6834989786148071, + "num_tokens": 119907607.0, + "step": 4634 + }, + { + "epoch": 0.5090050516143202, + "grad_norm": 2.1664648056030273, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6976402997970581, + "num_tokens": 119933810.0, + "step": 4635 + }, + { + "epoch": 0.5091148693169338, + "grad_norm": 2.2463736534118652, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6789859533309937, + "num_tokens": 119960256.0, + "step": 4636 + }, + { + "epoch": 0.5092246870195476, + "grad_norm": 2.031614303588867, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7022758722305298, + "num_tokens": 119988437.0, + "step": 4637 + }, + { + "epoch": 0.5093345047221612, + "grad_norm": 2.248342275619507, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7029738426208496, + "num_tokens": 120012553.0, + "step": 4638 + }, + { + "epoch": 0.5094443224247749, + "grad_norm": 2.19939923286438, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7001867294311523, + "num_tokens": 120037367.0, + "step": 4639 + }, + { + "epoch": 0.5095541401273885, + "grad_norm": 2.5386226177215576, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7205380797386169, + "num_tokens": 120056816.0, + "step": 4640 + }, + { + "epoch": 0.5096639578300022, + "grad_norm": 1.921915054321289, + "learning_rate": 1e-06, + "loss": 1.0581, + "mean_token_accuracy": 0.6829391717910767, + "num_tokens": 120090690.0, + "step": 4641 + }, + { + "epoch": 0.5097737755326158, + "grad_norm": 2.2665798664093018, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7053459286689758, + "num_tokens": 120114819.0, + "step": 4642 + }, + { + "epoch": 0.5098835932352295, + "grad_norm": 2.033942222595215, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7068951725959778, + "num_tokens": 120144814.0, + "step": 4643 + }, + { + "epoch": 0.5099934109378432, + "grad_norm": 2.288060188293457, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7206450700759888, + "num_tokens": 120168876.0, + "step": 4644 + }, + { + "epoch": 0.5101032286404569, + "grad_norm": 2.215024709701538, + "learning_rate": 1e-06, + "loss": 1.0576, + "mean_token_accuracy": 0.6814022064208984, + "num_tokens": 120196920.0, + "step": 4645 + }, + { + "epoch": 0.5102130463430705, + "grad_norm": 2.323235273361206, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.6977741718292236, + "num_tokens": 120222102.0, + "step": 4646 + }, + { + "epoch": 0.5103228640456842, + "grad_norm": 2.19632887840271, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7030943036079407, + "num_tokens": 120247712.0, + "step": 4647 + }, + { + "epoch": 0.5104326817482978, + "grad_norm": 1.917293906211853, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.6897921562194824, + "num_tokens": 120281315.0, + "step": 4648 + }, + { + "epoch": 0.5105424994509115, + "grad_norm": 2.1113016605377197, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.7033357620239258, + "num_tokens": 120309377.0, + "step": 4649 + }, + { + "epoch": 0.5106523171535251, + "grad_norm": 2.0823612213134766, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6893436908721924, + "num_tokens": 120337949.0, + "step": 4650 + }, + { + "epoch": 0.5107621348561389, + "grad_norm": 2.1736767292022705, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7071244120597839, + "num_tokens": 120362736.0, + "step": 4651 + }, + { + "epoch": 0.5108719525587525, + "grad_norm": 2.1173861026763916, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6911419630050659, + "num_tokens": 120394003.0, + "step": 4652 + }, + { + "epoch": 0.5109817702613662, + "grad_norm": 2.384885311126709, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6933771967887878, + "num_tokens": 120417078.0, + "step": 4653 + }, + { + "epoch": 0.5110915879639798, + "grad_norm": 2.244732141494751, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7232117652893066, + "num_tokens": 120440605.0, + "step": 4654 + }, + { + "epoch": 0.5112014056665934, + "grad_norm": 2.371760845184326, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7057453393936157, + "num_tokens": 120463428.0, + "step": 4655 + }, + { + "epoch": 0.5113112233692071, + "grad_norm": 2.089763879776001, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7024168968200684, + "num_tokens": 120493073.0, + "step": 4656 + }, + { + "epoch": 0.5114210410718207, + "grad_norm": 2.201301097869873, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7013100385665894, + "num_tokens": 120518143.0, + "step": 4657 + }, + { + "epoch": 0.5115308587744345, + "grad_norm": 2.286278247833252, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7052783370018005, + "num_tokens": 120542359.0, + "step": 4658 + }, + { + "epoch": 0.5116406764770481, + "grad_norm": 2.145759105682373, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7043961882591248, + "num_tokens": 120571395.0, + "step": 4659 + }, + { + "epoch": 0.5117504941796618, + "grad_norm": 1.850968360900879, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6825979351997375, + "num_tokens": 120607912.0, + "step": 4660 + }, + { + "epoch": 0.5118603118822754, + "grad_norm": 1.9978346824645996, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7007318139076233, + "num_tokens": 120639223.0, + "step": 4661 + }, + { + "epoch": 0.5119701295848891, + "grad_norm": 2.5711746215820312, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7078834176063538, + "num_tokens": 120659595.0, + "step": 4662 + }, + { + "epoch": 0.5120799472875027, + "grad_norm": 1.875617504119873, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.6870065927505493, + "num_tokens": 120694465.0, + "step": 4663 + }, + { + "epoch": 0.5121897649901164, + "grad_norm": 2.1527676582336426, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6980470418930054, + "num_tokens": 120722417.0, + "step": 4664 + }, + { + "epoch": 0.51229958269273, + "grad_norm": 2.1144087314605713, + "learning_rate": 1e-06, + "loss": 1.0724, + "mean_token_accuracy": 0.6824828386306763, + "num_tokens": 120750601.0, + "step": 4665 + }, + { + "epoch": 0.5124094003953438, + "grad_norm": 2.009631633758545, + "learning_rate": 1e-06, + "loss": 1.0637, + "mean_token_accuracy": 0.6768641471862793, + "num_tokens": 120781378.0, + "step": 4666 + }, + { + "epoch": 0.5125192180979574, + "grad_norm": 2.204714059829712, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6932306289672852, + "num_tokens": 120807616.0, + "step": 4667 + }, + { + "epoch": 0.5126290358005711, + "grad_norm": 2.2784039974212646, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6943204402923584, + "num_tokens": 120832422.0, + "step": 4668 + }, + { + "epoch": 0.5127388535031847, + "grad_norm": 2.1512112617492676, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.6756174564361572, + "num_tokens": 120859714.0, + "step": 4669 + }, + { + "epoch": 0.5128486712057984, + "grad_norm": 2.3308322429656982, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7138220071792603, + "num_tokens": 120883835.0, + "step": 4670 + }, + { + "epoch": 0.512958488908412, + "grad_norm": 2.235722541809082, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.695742130279541, + "num_tokens": 120907148.0, + "step": 4671 + }, + { + "epoch": 0.5130683066110256, + "grad_norm": 2.0213942527770996, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7020962238311768, + "num_tokens": 120937197.0, + "step": 4672 + }, + { + "epoch": 0.5131781243136394, + "grad_norm": 2.290842056274414, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7124735116958618, + "num_tokens": 120960227.0, + "step": 4673 + }, + { + "epoch": 0.513287942016253, + "grad_norm": 2.076298713684082, + "learning_rate": 1e-06, + "loss": 1.0638, + "mean_token_accuracy": 0.6961957216262817, + "num_tokens": 120987671.0, + "step": 4674 + }, + { + "epoch": 0.5133977597188667, + "grad_norm": 2.1844372749328613, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7173192501068115, + "num_tokens": 121014518.0, + "step": 4675 + }, + { + "epoch": 0.5135075774214803, + "grad_norm": 2.0891103744506836, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6904329657554626, + "num_tokens": 121046518.0, + "step": 4676 + }, + { + "epoch": 0.513617395124094, + "grad_norm": 2.3357996940612793, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7077348232269287, + "num_tokens": 121070381.0, + "step": 4677 + }, + { + "epoch": 0.5137272128267076, + "grad_norm": 2.2398436069488525, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7068860530853271, + "num_tokens": 121093702.0, + "step": 4678 + }, + { + "epoch": 0.5138370305293213, + "grad_norm": 2.645686149597168, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6959914565086365, + "num_tokens": 121114919.0, + "step": 4679 + }, + { + "epoch": 0.513946848231935, + "grad_norm": 2.269012212753296, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6987594366073608, + "num_tokens": 121137886.0, + "step": 4680 + }, + { + "epoch": 0.5140566659345487, + "grad_norm": 2.1513452529907227, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6998737454414368, + "num_tokens": 121164767.0, + "step": 4681 + }, + { + "epoch": 0.5141664836371623, + "grad_norm": 2.0337483882904053, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6917000412940979, + "num_tokens": 121194080.0, + "step": 4682 + }, + { + "epoch": 0.514276301339776, + "grad_norm": 1.9805814027786255, + "learning_rate": 1e-06, + "loss": 1.1084, + "mean_token_accuracy": 0.6687383055686951, + "num_tokens": 121226531.0, + "step": 4683 + }, + { + "epoch": 0.5143861190423896, + "grad_norm": 2.244870662689209, + "learning_rate": 1e-06, + "loss": 1.0966, + "mean_token_accuracy": 0.6759359836578369, + "num_tokens": 121254072.0, + "step": 4684 + }, + { + "epoch": 0.5144959367450033, + "grad_norm": 2.271450996398926, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6996619701385498, + "num_tokens": 121279731.0, + "step": 4685 + }, + { + "epoch": 0.5146057544476169, + "grad_norm": 1.8617384433746338, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7183982133865356, + "num_tokens": 121312266.0, + "step": 4686 + }, + { + "epoch": 0.5147155721502307, + "grad_norm": 2.204211711883545, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6979238986968994, + "num_tokens": 121337329.0, + "step": 4687 + }, + { + "epoch": 0.5148253898528443, + "grad_norm": 1.9502570629119873, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6939487457275391, + "num_tokens": 121369123.0, + "step": 4688 + }, + { + "epoch": 0.514935207555458, + "grad_norm": 2.325751781463623, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7066580653190613, + "num_tokens": 121391602.0, + "step": 4689 + }, + { + "epoch": 0.5150450252580716, + "grad_norm": 2.117866039276123, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6877729296684265, + "num_tokens": 121418720.0, + "step": 4690 + }, + { + "epoch": 0.5151548429606853, + "grad_norm": 2.115196466445923, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7196043133735657, + "num_tokens": 121444116.0, + "step": 4691 + }, + { + "epoch": 0.5152646606632989, + "grad_norm": 1.9733097553253174, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7001831531524658, + "num_tokens": 121474478.0, + "step": 4692 + }, + { + "epoch": 0.5153744783659125, + "grad_norm": 2.072627305984497, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6972314715385437, + "num_tokens": 121502744.0, + "step": 4693 + }, + { + "epoch": 0.5154842960685262, + "grad_norm": 2.0827929973602295, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7226809859275818, + "num_tokens": 121530615.0, + "step": 4694 + }, + { + "epoch": 0.51559411377114, + "grad_norm": 2.044624090194702, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.6993367075920105, + "num_tokens": 121559108.0, + "step": 4695 + }, + { + "epoch": 0.5157039314737536, + "grad_norm": 2.0864012241363525, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.7087034583091736, + "num_tokens": 121588392.0, + "step": 4696 + }, + { + "epoch": 0.5158137491763672, + "grad_norm": 2.029958724975586, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7090501189231873, + "num_tokens": 121617083.0, + "step": 4697 + }, + { + "epoch": 0.5159235668789809, + "grad_norm": 2.2671046257019043, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7137482166290283, + "num_tokens": 121641301.0, + "step": 4698 + }, + { + "epoch": 0.5160333845815945, + "grad_norm": 2.016726016998291, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7055872082710266, + "num_tokens": 121671011.0, + "step": 4699 + }, + { + "epoch": 0.5161432022842082, + "grad_norm": 2.144371509552002, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7143256664276123, + "num_tokens": 121697278.0, + "step": 4700 + }, + { + "epoch": 0.5162530199868218, + "grad_norm": 2.128410577774048, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7030085921287537, + "num_tokens": 121724693.0, + "step": 4701 + }, + { + "epoch": 0.5163628376894356, + "grad_norm": 2.645624876022339, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6979550719261169, + "num_tokens": 121753616.0, + "step": 4702 + }, + { + "epoch": 0.5164726553920492, + "grad_norm": 2.0650835037231445, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7071229219436646, + "num_tokens": 121781118.0, + "step": 4703 + }, + { + "epoch": 0.5165824730946629, + "grad_norm": 2.0969431400299072, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7206946611404419, + "num_tokens": 121807666.0, + "step": 4704 + }, + { + "epoch": 0.5166922907972765, + "grad_norm": 2.5084691047668457, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7324776649475098, + "num_tokens": 121826186.0, + "step": 4705 + }, + { + "epoch": 0.5168021084998902, + "grad_norm": 2.16542387008667, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.709949791431427, + "num_tokens": 121853749.0, + "step": 4706 + }, + { + "epoch": 0.5169119262025038, + "grad_norm": 2.2480552196502686, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6975225210189819, + "num_tokens": 121879214.0, + "step": 4707 + }, + { + "epoch": 0.5170217439051175, + "grad_norm": 2.1767749786376953, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.7037160396575928, + "num_tokens": 121905782.0, + "step": 4708 + }, + { + "epoch": 0.5171315616077312, + "grad_norm": 2.0987963676452637, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6905098557472229, + "num_tokens": 121935314.0, + "step": 4709 + }, + { + "epoch": 0.5172413793103449, + "grad_norm": 1.9146395921707153, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7146986722946167, + "num_tokens": 121965963.0, + "step": 4710 + }, + { + "epoch": 0.5173511970129585, + "grad_norm": 1.9909838438034058, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7074195146560669, + "num_tokens": 121994548.0, + "step": 4711 + }, + { + "epoch": 0.5174610147155722, + "grad_norm": 2.1157689094543457, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7174031734466553, + "num_tokens": 122021964.0, + "step": 4712 + }, + { + "epoch": 0.5175708324181858, + "grad_norm": 2.5345346927642822, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.714497447013855, + "num_tokens": 122043758.0, + "step": 4713 + }, + { + "epoch": 0.5176806501207994, + "grad_norm": 2.236292839050293, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.715118408203125, + "num_tokens": 122068302.0, + "step": 4714 + }, + { + "epoch": 0.5177904678234131, + "grad_norm": 2.227492570877075, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7045391798019409, + "num_tokens": 122095220.0, + "step": 4715 + }, + { + "epoch": 0.5179002855260268, + "grad_norm": 2.3669190406799316, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7020889520645142, + "num_tokens": 122117264.0, + "step": 4716 + }, + { + "epoch": 0.5180101032286405, + "grad_norm": 2.464003562927246, + "learning_rate": 1e-06, + "loss": 1.1105, + "mean_token_accuracy": 0.6617193818092346, + "num_tokens": 122140503.0, + "step": 4717 + }, + { + "epoch": 0.5181199209312541, + "grad_norm": 2.2910611629486084, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.698582112789154, + "num_tokens": 122162514.0, + "step": 4718 + }, + { + "epoch": 0.5182297386338678, + "grad_norm": 2.3608012199401855, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6956315040588379, + "num_tokens": 122184999.0, + "step": 4719 + }, + { + "epoch": 0.5183395563364814, + "grad_norm": 2.151630163192749, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7116858959197998, + "num_tokens": 122213918.0, + "step": 4720 + }, + { + "epoch": 0.5184493740390951, + "grad_norm": 2.359726667404175, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7063272595405579, + "num_tokens": 122237734.0, + "step": 4721 + }, + { + "epoch": 0.5185591917417087, + "grad_norm": 2.304807662963867, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7012776136398315, + "num_tokens": 122260160.0, + "step": 4722 + }, + { + "epoch": 0.5186690094443224, + "grad_norm": 2.107978582382202, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7199749946594238, + "num_tokens": 122286833.0, + "step": 4723 + }, + { + "epoch": 0.5187788271469361, + "grad_norm": 2.2070343494415283, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7357337474822998, + "num_tokens": 122310433.0, + "step": 4724 + }, + { + "epoch": 0.5188886448495498, + "grad_norm": 2.26088547706604, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.6987020969390869, + "num_tokens": 122334006.0, + "step": 4725 + }, + { + "epoch": 0.5189984625521634, + "grad_norm": 2.1482067108154297, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7060890197753906, + "num_tokens": 122361228.0, + "step": 4726 + }, + { + "epoch": 0.5191082802547771, + "grad_norm": 2.2289905548095703, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7216169834136963, + "num_tokens": 122386834.0, + "step": 4727 + }, + { + "epoch": 0.5192180979573907, + "grad_norm": 2.30582857131958, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6996098756790161, + "num_tokens": 122411805.0, + "step": 4728 + }, + { + "epoch": 0.5193279156600044, + "grad_norm": 1.9587243795394897, + "learning_rate": 1e-06, + "loss": 1.1036, + "mean_token_accuracy": 0.6779139637947083, + "num_tokens": 122446482.0, + "step": 4729 + }, + { + "epoch": 0.519437733362618, + "grad_norm": 2.1845693588256836, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7084189653396606, + "num_tokens": 122473380.0, + "step": 4730 + }, + { + "epoch": 0.5195475510652318, + "grad_norm": 2.3489956855773926, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7212701439857483, + "num_tokens": 122497542.0, + "step": 4731 + }, + { + "epoch": 0.5196573687678454, + "grad_norm": 2.038757801055908, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6919909715652466, + "num_tokens": 122526094.0, + "step": 4732 + }, + { + "epoch": 0.519767186470459, + "grad_norm": 2.051713705062866, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6908972859382629, + "num_tokens": 122555978.0, + "step": 4733 + }, + { + "epoch": 0.5198770041730727, + "grad_norm": 2.0803823471069336, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7058608531951904, + "num_tokens": 122584360.0, + "step": 4734 + }, + { + "epoch": 0.5199868218756863, + "grad_norm": 2.3520939350128174, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6955751180648804, + "num_tokens": 122606846.0, + "step": 4735 + }, + { + "epoch": 0.5200966395783, + "grad_norm": 2.588782548904419, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7176234722137451, + "num_tokens": 122626112.0, + "step": 4736 + }, + { + "epoch": 0.5202064572809136, + "grad_norm": 2.301412582397461, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6924156546592712, + "num_tokens": 122650732.0, + "step": 4737 + }, + { + "epoch": 0.5203162749835274, + "grad_norm": 2.27955961227417, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.711986243724823, + "num_tokens": 122674126.0, + "step": 4738 + }, + { + "epoch": 0.520426092686141, + "grad_norm": 2.412432909011841, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7100757360458374, + "num_tokens": 122697488.0, + "step": 4739 + }, + { + "epoch": 0.5205359103887547, + "grad_norm": 2.354954481124878, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7012811899185181, + "num_tokens": 122722311.0, + "step": 4740 + }, + { + "epoch": 0.5206457280913683, + "grad_norm": 2.2212178707122803, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7083175182342529, + "num_tokens": 122748210.0, + "step": 4741 + }, + { + "epoch": 0.520755545793982, + "grad_norm": 2.3124542236328125, + "learning_rate": 1e-06, + "loss": 1.0611, + "mean_token_accuracy": 0.6884700059890747, + "num_tokens": 122772245.0, + "step": 4742 + }, + { + "epoch": 0.5208653634965956, + "grad_norm": 2.14357328414917, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7007904052734375, + "num_tokens": 122798903.0, + "step": 4743 + }, + { + "epoch": 0.5209751811992093, + "grad_norm": 2.7323203086853027, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6997947692871094, + "num_tokens": 122818671.0, + "step": 4744 + }, + { + "epoch": 0.521084998901823, + "grad_norm": 2.087639093399048, + "learning_rate": 1e-06, + "loss": 1.1347, + "mean_token_accuracy": 0.6681228280067444, + "num_tokens": 122849140.0, + "step": 4745 + }, + { + "epoch": 0.5211948166044367, + "grad_norm": 2.0701563358306885, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.694132387638092, + "num_tokens": 122879171.0, + "step": 4746 + }, + { + "epoch": 0.5213046343070503, + "grad_norm": 2.1459858417510986, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.6967957019805908, + "num_tokens": 122908443.0, + "step": 4747 + }, + { + "epoch": 0.521414452009664, + "grad_norm": 2.3988120555877686, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7169339656829834, + "num_tokens": 122928803.0, + "step": 4748 + }, + { + "epoch": 0.5215242697122776, + "grad_norm": 2.21224308013916, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7014182209968567, + "num_tokens": 122954480.0, + "step": 4749 + }, + { + "epoch": 0.5216340874148913, + "grad_norm": 2.1155543327331543, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7105028629302979, + "num_tokens": 122982025.0, + "step": 4750 + }, + { + "epoch": 0.5217439051175049, + "grad_norm": 2.268742322921753, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7062113285064697, + "num_tokens": 123009122.0, + "step": 4751 + }, + { + "epoch": 0.5218537228201185, + "grad_norm": 2.2358224391937256, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7242642045021057, + "num_tokens": 123034418.0, + "step": 4752 + }, + { + "epoch": 0.5219635405227323, + "grad_norm": 2.4069647789001465, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7053825855255127, + "num_tokens": 123055576.0, + "step": 4753 + }, + { + "epoch": 0.522073358225346, + "grad_norm": 2.652107000350952, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.704372763633728, + "num_tokens": 123074540.0, + "step": 4754 + }, + { + "epoch": 0.5221831759279596, + "grad_norm": 2.436537265777588, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7073556184768677, + "num_tokens": 123097449.0, + "step": 4755 + }, + { + "epoch": 0.5222929936305732, + "grad_norm": 2.265578031539917, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6916489601135254, + "num_tokens": 123125290.0, + "step": 4756 + }, + { + "epoch": 0.5224028113331869, + "grad_norm": 1.977240800857544, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.6986652612686157, + "num_tokens": 123155920.0, + "step": 4757 + }, + { + "epoch": 0.5225126290358005, + "grad_norm": 2.350710868835449, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7175132632255554, + "num_tokens": 123176959.0, + "step": 4758 + }, + { + "epoch": 0.5226224467384142, + "grad_norm": 2.195472002029419, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6908040642738342, + "num_tokens": 123202287.0, + "step": 4759 + }, + { + "epoch": 0.5227322644410279, + "grad_norm": 2.350327253341675, + "learning_rate": 1e-06, + "loss": 1.0665, + "mean_token_accuracy": 0.6885855793952942, + "num_tokens": 123225282.0, + "step": 4760 + }, + { + "epoch": 0.5228420821436416, + "grad_norm": 2.3034543991088867, + "learning_rate": 1e-06, + "loss": 1.0638, + "mean_token_accuracy": 0.681605339050293, + "num_tokens": 123251470.0, + "step": 4761 + }, + { + "epoch": 0.5229518998462552, + "grad_norm": 2.0150532722473145, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6871960759162903, + "num_tokens": 123280914.0, + "step": 4762 + }, + { + "epoch": 0.5230617175488689, + "grad_norm": 2.173884153366089, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7110849618911743, + "num_tokens": 123307373.0, + "step": 4763 + }, + { + "epoch": 0.5231715352514825, + "grad_norm": 1.9426705837249756, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6815805435180664, + "num_tokens": 123341593.0, + "step": 4764 + }, + { + "epoch": 0.5232813529540962, + "grad_norm": 2.2162444591522217, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7122302055358887, + "num_tokens": 123367583.0, + "step": 4765 + }, + { + "epoch": 0.5233911706567098, + "grad_norm": 1.8395023345947266, + "learning_rate": 1e-06, + "loss": 1.0499, + "mean_token_accuracy": 0.6878299713134766, + "num_tokens": 123404988.0, + "step": 4766 + }, + { + "epoch": 0.5235009883593236, + "grad_norm": 2.132904052734375, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7143001556396484, + "num_tokens": 123429995.0, + "step": 4767 + }, + { + "epoch": 0.5236108060619372, + "grad_norm": 2.102752447128296, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6940587759017944, + "num_tokens": 123458296.0, + "step": 4768 + }, + { + "epoch": 0.5237206237645509, + "grad_norm": 2.1200172901153564, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.7012196779251099, + "num_tokens": 123484696.0, + "step": 4769 + }, + { + "epoch": 0.5238304414671645, + "grad_norm": 2.2120656967163086, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.699380099773407, + "num_tokens": 123511954.0, + "step": 4770 + }, + { + "epoch": 0.5239402591697782, + "grad_norm": 2.502350091934204, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7170779705047607, + "num_tokens": 123532302.0, + "step": 4771 + }, + { + "epoch": 0.5240500768723918, + "grad_norm": 2.3221116065979004, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7085622549057007, + "num_tokens": 123557394.0, + "step": 4772 + }, + { + "epoch": 0.5241598945750054, + "grad_norm": 2.3889777660369873, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.691352367401123, + "num_tokens": 123581035.0, + "step": 4773 + }, + { + "epoch": 0.5242697122776192, + "grad_norm": 2.104372262954712, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.70203697681427, + "num_tokens": 123609927.0, + "step": 4774 + }, + { + "epoch": 0.5243795299802329, + "grad_norm": 2.2761383056640625, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.684646725654602, + "num_tokens": 123636801.0, + "step": 4775 + }, + { + "epoch": 0.5244893476828465, + "grad_norm": 2.133296489715576, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7236238718032837, + "num_tokens": 123662412.0, + "step": 4776 + }, + { + "epoch": 0.5245991653854601, + "grad_norm": 2.2204947471618652, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6897223591804504, + "num_tokens": 123688200.0, + "step": 4777 + }, + { + "epoch": 0.5247089830880738, + "grad_norm": 2.635758638381958, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7024513483047485, + "num_tokens": 123707255.0, + "step": 4778 + }, + { + "epoch": 0.5248188007906874, + "grad_norm": 2.368706226348877, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7110570073127747, + "num_tokens": 123730727.0, + "step": 4779 + }, + { + "epoch": 0.5249286184933011, + "grad_norm": 2.397913694381714, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.744174599647522, + "num_tokens": 123750769.0, + "step": 4780 + }, + { + "epoch": 0.5250384361959147, + "grad_norm": 2.198495626449585, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7191442251205444, + "num_tokens": 123774443.0, + "step": 4781 + }, + { + "epoch": 0.5251482538985285, + "grad_norm": 2.4475016593933105, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.7022503614425659, + "num_tokens": 123797174.0, + "step": 4782 + }, + { + "epoch": 0.5252580716011421, + "grad_norm": 2.227501392364502, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6964905261993408, + "num_tokens": 123823711.0, + "step": 4783 + }, + { + "epoch": 0.5253678893037558, + "grad_norm": 2.376544237136841, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6913824081420898, + "num_tokens": 123847343.0, + "step": 4784 + }, + { + "epoch": 0.5254777070063694, + "grad_norm": 2.2393410205841064, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6855170130729675, + "num_tokens": 123872690.0, + "step": 4785 + }, + { + "epoch": 0.5255875247089831, + "grad_norm": 2.2807891368865967, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7034401893615723, + "num_tokens": 123897410.0, + "step": 4786 + }, + { + "epoch": 0.5256973424115967, + "grad_norm": 2.339376926422119, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6937440633773804, + "num_tokens": 123921194.0, + "step": 4787 + }, + { + "epoch": 0.5258071601142104, + "grad_norm": 2.1505300998687744, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7096041440963745, + "num_tokens": 123946528.0, + "step": 4788 + }, + { + "epoch": 0.5259169778168241, + "grad_norm": 2.091665744781494, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7049789428710938, + "num_tokens": 123974449.0, + "step": 4789 + }, + { + "epoch": 0.5260267955194378, + "grad_norm": 2.4256012439727783, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7191815376281738, + "num_tokens": 123994838.0, + "step": 4790 + }, + { + "epoch": 0.5261366132220514, + "grad_norm": 2.140660285949707, + "learning_rate": 1e-06, + "loss": 1.119, + "mean_token_accuracy": 0.6649973392486572, + "num_tokens": 124025635.0, + "step": 4791 + }, + { + "epoch": 0.526246430924665, + "grad_norm": 2.217031240463257, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.7020083665847778, + "num_tokens": 124052257.0, + "step": 4792 + }, + { + "epoch": 0.5263562486272787, + "grad_norm": 2.301955223083496, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7056295871734619, + "num_tokens": 124076722.0, + "step": 4793 + }, + { + "epoch": 0.5264660663298923, + "grad_norm": 2.692626953125, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6905345320701599, + "num_tokens": 124095514.0, + "step": 4794 + }, + { + "epoch": 0.526575884032506, + "grad_norm": 2.1707570552825928, + "learning_rate": 1e-06, + "loss": 1.0577, + "mean_token_accuracy": 0.675177276134491, + "num_tokens": 124123063.0, + "step": 4795 + }, + { + "epoch": 0.5266857017351197, + "grad_norm": 2.317262887954712, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6892533898353577, + "num_tokens": 124146795.0, + "step": 4796 + }, + { + "epoch": 0.5267955194377334, + "grad_norm": 2.2661635875701904, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7086657881736755, + "num_tokens": 124170669.0, + "step": 4797 + }, + { + "epoch": 0.526905337140347, + "grad_norm": 2.170870780944824, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6890336275100708, + "num_tokens": 124195844.0, + "step": 4798 + }, + { + "epoch": 0.5270151548429607, + "grad_norm": 2.23028302192688, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6861010789871216, + "num_tokens": 124223053.0, + "step": 4799 + }, + { + "epoch": 0.5271249725455743, + "grad_norm": 2.1652820110321045, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6982871294021606, + "num_tokens": 124248218.0, + "step": 4800 + }, + { + "epoch": 0.527234790248188, + "grad_norm": 2.1572654247283936, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.718483567237854, + "num_tokens": 124273859.0, + "step": 4801 + }, + { + "epoch": 0.5273446079508016, + "grad_norm": 2.5193710327148438, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7169408798217773, + "num_tokens": 124294213.0, + "step": 4802 + }, + { + "epoch": 0.5274544256534154, + "grad_norm": 2.060840606689453, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.699066162109375, + "num_tokens": 124319793.0, + "step": 4803 + }, + { + "epoch": 0.527564243356029, + "grad_norm": 2.127622604370117, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7075961232185364, + "num_tokens": 124346866.0, + "step": 4804 + }, + { + "epoch": 0.5276740610586427, + "grad_norm": 2.462721824645996, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7066447138786316, + "num_tokens": 124368693.0, + "step": 4805 + }, + { + "epoch": 0.5277838787612563, + "grad_norm": 2.5168209075927734, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.686883807182312, + "num_tokens": 124388582.0, + "step": 4806 + }, + { + "epoch": 0.52789369646387, + "grad_norm": 2.1695749759674072, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7140613794326782, + "num_tokens": 124413481.0, + "step": 4807 + }, + { + "epoch": 0.5280035141664836, + "grad_norm": 2.2809572219848633, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7109401226043701, + "num_tokens": 124436784.0, + "step": 4808 + }, + { + "epoch": 0.5281133318690973, + "grad_norm": 2.267007350921631, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6893234252929688, + "num_tokens": 124460433.0, + "step": 4809 + }, + { + "epoch": 0.528223149571711, + "grad_norm": 2.303057909011841, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6894623041152954, + "num_tokens": 124484380.0, + "step": 4810 + }, + { + "epoch": 0.5283329672743247, + "grad_norm": 2.2074105739593506, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.713450014591217, + "num_tokens": 124509622.0, + "step": 4811 + }, + { + "epoch": 0.5284427849769383, + "grad_norm": 1.9612706899642944, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6900162696838379, + "num_tokens": 124540590.0, + "step": 4812 + }, + { + "epoch": 0.528552602679552, + "grad_norm": 2.5900449752807617, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7086994647979736, + "num_tokens": 124562283.0, + "step": 4813 + }, + { + "epoch": 0.5286624203821656, + "grad_norm": 2.0770742893218994, + "learning_rate": 1e-06, + "loss": 1.0649, + "mean_token_accuracy": 0.6832547187805176, + "num_tokens": 124594488.0, + "step": 4814 + }, + { + "epoch": 0.5287722380847792, + "grad_norm": 2.203843355178833, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7130767107009888, + "num_tokens": 124621489.0, + "step": 4815 + }, + { + "epoch": 0.5288820557873929, + "grad_norm": 1.9689229726791382, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7402154207229614, + "num_tokens": 124649436.0, + "step": 4816 + }, + { + "epoch": 0.5289918734900065, + "grad_norm": 2.3964014053344727, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7124195694923401, + "num_tokens": 124671012.0, + "step": 4817 + }, + { + "epoch": 0.5291016911926203, + "grad_norm": 2.2570700645446777, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6959421634674072, + "num_tokens": 124694958.0, + "step": 4818 + }, + { + "epoch": 0.5292115088952339, + "grad_norm": 2.4729971885681152, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7058364152908325, + "num_tokens": 124715344.0, + "step": 4819 + }, + { + "epoch": 0.5293213265978476, + "grad_norm": 2.1424224376678467, + "learning_rate": 1e-06, + "loss": 1.1327, + "mean_token_accuracy": 0.666124701499939, + "num_tokens": 124743751.0, + "step": 4820 + }, + { + "epoch": 0.5294311443004612, + "grad_norm": 2.2726776599884033, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7249089479446411, + "num_tokens": 124769264.0, + "step": 4821 + }, + { + "epoch": 0.5295409620030749, + "grad_norm": 2.2122230529785156, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.694608211517334, + "num_tokens": 124796623.0, + "step": 4822 + }, + { + "epoch": 0.5296507797056885, + "grad_norm": 2.1019539833068848, + "learning_rate": 1e-06, + "loss": 1.0907, + "mean_token_accuracy": 0.6724964380264282, + "num_tokens": 124825236.0, + "step": 4823 + }, + { + "epoch": 0.5297605974083022, + "grad_norm": 2.260007619857788, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.695531964302063, + "num_tokens": 124849570.0, + "step": 4824 + }, + { + "epoch": 0.5298704151109159, + "grad_norm": 2.1570935249328613, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6876499652862549, + "num_tokens": 124876140.0, + "step": 4825 + }, + { + "epoch": 0.5299802328135296, + "grad_norm": 2.0922584533691406, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6912973523139954, + "num_tokens": 124902546.0, + "step": 4826 + }, + { + "epoch": 0.5300900505161432, + "grad_norm": 2.278726816177368, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7046939134597778, + "num_tokens": 124926641.0, + "step": 4827 + }, + { + "epoch": 0.5301998682187569, + "grad_norm": 2.0496835708618164, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7006021738052368, + "num_tokens": 124955451.0, + "step": 4828 + }, + { + "epoch": 0.5303096859213705, + "grad_norm": 2.0472571849823, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7148644328117371, + "num_tokens": 124983402.0, + "step": 4829 + }, + { + "epoch": 0.5304195036239842, + "grad_norm": 2.238508939743042, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.7019808292388916, + "num_tokens": 125008433.0, + "step": 4830 + }, + { + "epoch": 0.5305293213265978, + "grad_norm": 2.043246269226074, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.699701189994812, + "num_tokens": 125036151.0, + "step": 4831 + }, + { + "epoch": 0.5306391390292116, + "grad_norm": 2.4054434299468994, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7213056087493896, + "num_tokens": 125058766.0, + "step": 4832 + }, + { + "epoch": 0.5307489567318252, + "grad_norm": 2.1331186294555664, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7140833139419556, + "num_tokens": 125084381.0, + "step": 4833 + }, + { + "epoch": 0.5308587744344389, + "grad_norm": 2.1758270263671875, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6875346899032593, + "num_tokens": 125108921.0, + "step": 4834 + }, + { + "epoch": 0.5309685921370525, + "grad_norm": 1.9918179512023926, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7023518085479736, + "num_tokens": 125140054.0, + "step": 4835 + }, + { + "epoch": 0.5310784098396661, + "grad_norm": 2.110781669616699, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.686740517616272, + "num_tokens": 125167042.0, + "step": 4836 + }, + { + "epoch": 0.5311882275422798, + "grad_norm": 2.0108189582824707, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7157902717590332, + "num_tokens": 125195924.0, + "step": 4837 + }, + { + "epoch": 0.5312980452448934, + "grad_norm": 2.2245733737945557, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6985657215118408, + "num_tokens": 125221340.0, + "step": 4838 + }, + { + "epoch": 0.5314078629475072, + "grad_norm": 2.3123698234558105, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.6985107660293579, + "num_tokens": 125243894.0, + "step": 4839 + }, + { + "epoch": 0.5315176806501208, + "grad_norm": 2.26422381401062, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.712421178817749, + "num_tokens": 125268549.0, + "step": 4840 + }, + { + "epoch": 0.5316274983527345, + "grad_norm": 2.15022611618042, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.7003003358840942, + "num_tokens": 125295550.0, + "step": 4841 + }, + { + "epoch": 0.5317373160553481, + "grad_norm": 2.1978094577789307, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7011486887931824, + "num_tokens": 125320744.0, + "step": 4842 + }, + { + "epoch": 0.5318471337579618, + "grad_norm": 2.165128707885742, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6864751577377319, + "num_tokens": 125348066.0, + "step": 4843 + }, + { + "epoch": 0.5319569514605754, + "grad_norm": 1.91519296169281, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6971882581710815, + "num_tokens": 125378429.0, + "step": 4844 + }, + { + "epoch": 0.5320667691631891, + "grad_norm": 2.1664881706237793, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6952964663505554, + "num_tokens": 125403980.0, + "step": 4845 + }, + { + "epoch": 0.5321765868658027, + "grad_norm": 2.054013967514038, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6925657987594604, + "num_tokens": 125434953.0, + "step": 4846 + }, + { + "epoch": 0.5322864045684165, + "grad_norm": 2.213670492172241, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7036712169647217, + "num_tokens": 125461614.0, + "step": 4847 + }, + { + "epoch": 0.5323962222710301, + "grad_norm": 1.9734294414520264, + "learning_rate": 1e-06, + "loss": 1.0818, + "mean_token_accuracy": 0.678712010383606, + "num_tokens": 125494269.0, + "step": 4848 + }, + { + "epoch": 0.5325060399736438, + "grad_norm": 2.1678178310394287, + "learning_rate": 1e-06, + "loss": 1.0814, + "mean_token_accuracy": 0.6723219156265259, + "num_tokens": 125523086.0, + "step": 4849 + }, + { + "epoch": 0.5326158576762574, + "grad_norm": 2.332017183303833, + "learning_rate": 1e-06, + "loss": 1.0415, + "mean_token_accuracy": 0.692459225654602, + "num_tokens": 125548116.0, + "step": 4850 + }, + { + "epoch": 0.532725675378871, + "grad_norm": 2.1180267333984375, + "learning_rate": 1e-06, + "loss": 1.0796, + "mean_token_accuracy": 0.6739193201065063, + "num_tokens": 125576092.0, + "step": 4851 + }, + { + "epoch": 0.5328354930814847, + "grad_norm": 2.1046462059020996, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7044482231140137, + "num_tokens": 125606678.0, + "step": 4852 + }, + { + "epoch": 0.5329453107840983, + "grad_norm": 2.34727144241333, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7062970399856567, + "num_tokens": 125628470.0, + "step": 4853 + }, + { + "epoch": 0.5330551284867121, + "grad_norm": 2.5004589557647705, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6935728788375854, + "num_tokens": 125649194.0, + "step": 4854 + }, + { + "epoch": 0.5331649461893258, + "grad_norm": 2.0799477100372314, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.6884414553642273, + "num_tokens": 125681786.0, + "step": 4855 + }, + { + "epoch": 0.5332747638919394, + "grad_norm": 2.1203134059906006, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7061285972595215, + "num_tokens": 125708871.0, + "step": 4856 + }, + { + "epoch": 0.533384581594553, + "grad_norm": 2.10744309425354, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7074625492095947, + "num_tokens": 125736407.0, + "step": 4857 + }, + { + "epoch": 0.5334943992971667, + "grad_norm": 2.072453022003174, + "learning_rate": 1e-06, + "loss": 1.0496, + "mean_token_accuracy": 0.6967409253120422, + "num_tokens": 125764625.0, + "step": 4858 + }, + { + "epoch": 0.5336042169997803, + "grad_norm": 2.309936285018921, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6999026536941528, + "num_tokens": 125788811.0, + "step": 4859 + }, + { + "epoch": 0.533714034702394, + "grad_norm": 2.0577046871185303, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6967572569847107, + "num_tokens": 125816450.0, + "step": 4860 + }, + { + "epoch": 0.5338238524050077, + "grad_norm": 2.2185328006744385, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.7032657861709595, + "num_tokens": 125843742.0, + "step": 4861 + }, + { + "epoch": 0.5339336701076214, + "grad_norm": 2.4533822536468506, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6865171194076538, + "num_tokens": 125867806.0, + "step": 4862 + }, + { + "epoch": 0.534043487810235, + "grad_norm": 2.040130138397217, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6929061412811279, + "num_tokens": 125898897.0, + "step": 4863 + }, + { + "epoch": 0.5341533055128487, + "grad_norm": 2.1284968852996826, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7152112126350403, + "num_tokens": 125926333.0, + "step": 4864 + }, + { + "epoch": 0.5342631232154623, + "grad_norm": 2.1743667125701904, + "learning_rate": 1e-06, + "loss": 1.0493, + "mean_token_accuracy": 0.6813240051269531, + "num_tokens": 125955239.0, + "step": 4865 + }, + { + "epoch": 0.534372940918076, + "grad_norm": 2.0195999145507812, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7098960876464844, + "num_tokens": 125984817.0, + "step": 4866 + }, + { + "epoch": 0.5344827586206896, + "grad_norm": 2.453446626663208, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7076585292816162, + "num_tokens": 126007118.0, + "step": 4867 + }, + { + "epoch": 0.5345925763233034, + "grad_norm": 2.167912721633911, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7038756608963013, + "num_tokens": 126033833.0, + "step": 4868 + }, + { + "epoch": 0.534702394025917, + "grad_norm": 2.396196126937866, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6970891356468201, + "num_tokens": 126058969.0, + "step": 4869 + }, + { + "epoch": 0.5348122117285307, + "grad_norm": 2.11568284034729, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6827220916748047, + "num_tokens": 126087601.0, + "step": 4870 + }, + { + "epoch": 0.5349220294311443, + "grad_norm": 2.5097572803497314, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7011531591415405, + "num_tokens": 126108863.0, + "step": 4871 + }, + { + "epoch": 0.535031847133758, + "grad_norm": 2.090517997741699, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7145417332649231, + "num_tokens": 126135520.0, + "step": 4872 + }, + { + "epoch": 0.5351416648363716, + "grad_norm": 2.21712589263916, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7008832693099976, + "num_tokens": 126159142.0, + "step": 4873 + }, + { + "epoch": 0.5352514825389852, + "grad_norm": 2.0281028747558594, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7121443748474121, + "num_tokens": 126185981.0, + "step": 4874 + }, + { + "epoch": 0.5353613002415989, + "grad_norm": 2.2553598880767822, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6963249444961548, + "num_tokens": 126211810.0, + "step": 4875 + }, + { + "epoch": 0.5354711179442126, + "grad_norm": 2.208857297897339, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7047755122184753, + "num_tokens": 126236974.0, + "step": 4876 + }, + { + "epoch": 0.5355809356468263, + "grad_norm": 2.313117027282715, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7033370733261108, + "num_tokens": 126261310.0, + "step": 4877 + }, + { + "epoch": 0.5356907533494399, + "grad_norm": 2.635122776031494, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6929854154586792, + "num_tokens": 126281643.0, + "step": 4878 + }, + { + "epoch": 0.5358005710520536, + "grad_norm": 2.104217767715454, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.689018726348877, + "num_tokens": 126312906.0, + "step": 4879 + }, + { + "epoch": 0.5359103887546672, + "grad_norm": 1.9501612186431885, + "learning_rate": 1e-06, + "loss": 1.1087, + "mean_token_accuracy": 0.6683170795440674, + "num_tokens": 126345375.0, + "step": 4880 + }, + { + "epoch": 0.5360202064572809, + "grad_norm": 2.018005847930908, + "learning_rate": 1e-06, + "loss": 1.0577, + "mean_token_accuracy": 0.6808667182922363, + "num_tokens": 126376856.0, + "step": 4881 + }, + { + "epoch": 0.5361300241598945, + "grad_norm": 2.236194133758545, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6985238194465637, + "num_tokens": 126401481.0, + "step": 4882 + }, + { + "epoch": 0.5362398418625083, + "grad_norm": 2.6547625064849854, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.6989023685455322, + "num_tokens": 126421653.0, + "step": 4883 + }, + { + "epoch": 0.5363496595651219, + "grad_norm": 2.282742738723755, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6905415654182434, + "num_tokens": 126446537.0, + "step": 4884 + }, + { + "epoch": 0.5364594772677356, + "grad_norm": 2.207115650177002, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.729629397392273, + "num_tokens": 126472324.0, + "step": 4885 + }, + { + "epoch": 0.5365692949703492, + "grad_norm": 2.1506383419036865, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.728058934211731, + "num_tokens": 126498138.0, + "step": 4886 + }, + { + "epoch": 0.5366791126729629, + "grad_norm": 2.3048315048217773, + "learning_rate": 1e-06, + "loss": 1.0606, + "mean_token_accuracy": 0.676274836063385, + "num_tokens": 126524152.0, + "step": 4887 + }, + { + "epoch": 0.5367889303755765, + "grad_norm": 2.1473641395568848, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.702453076839447, + "num_tokens": 126551253.0, + "step": 4888 + }, + { + "epoch": 0.5368987480781902, + "grad_norm": 1.9107805490493774, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6902774572372437, + "num_tokens": 126583556.0, + "step": 4889 + }, + { + "epoch": 0.5370085657808039, + "grad_norm": 2.5550880432128906, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7168401479721069, + "num_tokens": 126602628.0, + "step": 4890 + }, + { + "epoch": 0.5371183834834176, + "grad_norm": 2.0307390689849854, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7096315622329712, + "num_tokens": 126630945.0, + "step": 4891 + }, + { + "epoch": 0.5372282011860312, + "grad_norm": 1.9987339973449707, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7034046649932861, + "num_tokens": 126660562.0, + "step": 4892 + }, + { + "epoch": 0.5373380188886449, + "grad_norm": 1.9641060829162598, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7038766741752625, + "num_tokens": 126691431.0, + "step": 4893 + }, + { + "epoch": 0.5374478365912585, + "grad_norm": 2.0899410247802734, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7042602896690369, + "num_tokens": 126718942.0, + "step": 4894 + }, + { + "epoch": 0.5375576542938721, + "grad_norm": 2.782235860824585, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7297953367233276, + "num_tokens": 126737774.0, + "step": 4895 + }, + { + "epoch": 0.5376674719964858, + "grad_norm": 2.262190341949463, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6972336173057556, + "num_tokens": 126764676.0, + "step": 4896 + }, + { + "epoch": 0.5377772896990995, + "grad_norm": 2.2931880950927734, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6987563371658325, + "num_tokens": 126790046.0, + "step": 4897 + }, + { + "epoch": 0.5378871074017132, + "grad_norm": 2.5958333015441895, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7366606593132019, + "num_tokens": 126808339.0, + "step": 4898 + }, + { + "epoch": 0.5379969251043268, + "grad_norm": 2.1814074516296387, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7259872555732727, + "num_tokens": 126832548.0, + "step": 4899 + }, + { + "epoch": 0.5381067428069405, + "grad_norm": 1.958619475364685, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6953945159912109, + "num_tokens": 126864910.0, + "step": 4900 + }, + { + "epoch": 0.5382165605095541, + "grad_norm": 2.127265214920044, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6934921145439148, + "num_tokens": 126893270.0, + "step": 4901 + }, + { + "epoch": 0.5383263782121678, + "grad_norm": 2.0913307666778564, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6981773376464844, + "num_tokens": 126921753.0, + "step": 4902 + }, + { + "epoch": 0.5384361959147814, + "grad_norm": 1.9487149715423584, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6853757500648499, + "num_tokens": 126955918.0, + "step": 4903 + }, + { + "epoch": 0.5385460136173951, + "grad_norm": 2.3538637161254883, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7110027074813843, + "num_tokens": 126978163.0, + "step": 4904 + }, + { + "epoch": 0.5386558313200088, + "grad_norm": 2.3855481147766113, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7057284116744995, + "num_tokens": 127001935.0, + "step": 4905 + }, + { + "epoch": 0.5387656490226225, + "grad_norm": 2.1758620738983154, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7058840394020081, + "num_tokens": 127026364.0, + "step": 4906 + }, + { + "epoch": 0.5388754667252361, + "grad_norm": 2.323793649673462, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6862894296646118, + "num_tokens": 127051022.0, + "step": 4907 + }, + { + "epoch": 0.5389852844278498, + "grad_norm": 2.280683994293213, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.697097897529602, + "num_tokens": 127077302.0, + "step": 4908 + }, + { + "epoch": 0.5390951021304634, + "grad_norm": 1.9875239133834839, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6966677904129028, + "num_tokens": 127107594.0, + "step": 4909 + }, + { + "epoch": 0.5392049198330771, + "grad_norm": 2.415323495864868, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7054073214530945, + "num_tokens": 127129814.0, + "step": 4910 + }, + { + "epoch": 0.5393147375356907, + "grad_norm": 2.4925572872161865, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7153220176696777, + "num_tokens": 127148691.0, + "step": 4911 + }, + { + "epoch": 0.5394245552383045, + "grad_norm": 2.328803062438965, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7359364628791809, + "num_tokens": 127168865.0, + "step": 4912 + }, + { + "epoch": 0.5395343729409181, + "grad_norm": 2.0111334323883057, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6953232288360596, + "num_tokens": 127199139.0, + "step": 4913 + }, + { + "epoch": 0.5396441906435318, + "grad_norm": 2.6125295162200928, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.710040807723999, + "num_tokens": 127219551.0, + "step": 4914 + }, + { + "epoch": 0.5397540083461454, + "grad_norm": 2.1646265983581543, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7077234983444214, + "num_tokens": 127247441.0, + "step": 4915 + }, + { + "epoch": 0.539863826048759, + "grad_norm": 1.8995214700698853, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7086318135261536, + "num_tokens": 127280922.0, + "step": 4916 + }, + { + "epoch": 0.5399736437513727, + "grad_norm": 2.17234468460083, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7240504026412964, + "num_tokens": 127303976.0, + "step": 4917 + }, + { + "epoch": 0.5400834614539863, + "grad_norm": 2.137455463409424, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7242692708969116, + "num_tokens": 127329285.0, + "step": 4918 + }, + { + "epoch": 0.5401932791566001, + "grad_norm": 2.358403444290161, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7194992303848267, + "num_tokens": 127352261.0, + "step": 4919 + }, + { + "epoch": 0.5403030968592137, + "grad_norm": 2.1295077800750732, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7023585438728333, + "num_tokens": 127380086.0, + "step": 4920 + }, + { + "epoch": 0.5404129145618274, + "grad_norm": 2.5481879711151123, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7173779606819153, + "num_tokens": 127402010.0, + "step": 4921 + }, + { + "epoch": 0.540522732264441, + "grad_norm": 2.225156545639038, + "learning_rate": 1e-06, + "loss": 1.0335, + "mean_token_accuracy": 0.6911228895187378, + "num_tokens": 127429070.0, + "step": 4922 + }, + { + "epoch": 0.5406325499670547, + "grad_norm": 2.0400733947753906, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7109662890434265, + "num_tokens": 127458094.0, + "step": 4923 + }, + { + "epoch": 0.5407423676696683, + "grad_norm": 2.074044942855835, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7022624015808105, + "num_tokens": 127487961.0, + "step": 4924 + }, + { + "epoch": 0.540852185372282, + "grad_norm": 2.202562093734741, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.702156662940979, + "num_tokens": 127514020.0, + "step": 4925 + }, + { + "epoch": 0.5409620030748957, + "grad_norm": 2.05126953125, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6909000873565674, + "num_tokens": 127542668.0, + "step": 4926 + }, + { + "epoch": 0.5410718207775094, + "grad_norm": 2.529684066772461, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7066807150840759, + "num_tokens": 127562630.0, + "step": 4927 + }, + { + "epoch": 0.541181638480123, + "grad_norm": 2.0981781482696533, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7053806185722351, + "num_tokens": 127592931.0, + "step": 4928 + }, + { + "epoch": 0.5412914561827367, + "grad_norm": 2.0921566486358643, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6923192739486694, + "num_tokens": 127621487.0, + "step": 4929 + }, + { + "epoch": 0.5414012738853503, + "grad_norm": 1.9717894792556763, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7299869060516357, + "num_tokens": 127650196.0, + "step": 4930 + }, + { + "epoch": 0.541511091587964, + "grad_norm": 2.346715211868286, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7180224657058716, + "num_tokens": 127673970.0, + "step": 4931 + }, + { + "epoch": 0.5416209092905776, + "grad_norm": 2.0763769149780273, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6868102550506592, + "num_tokens": 127703486.0, + "step": 4932 + }, + { + "epoch": 0.5417307269931912, + "grad_norm": 2.0632386207580566, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.705559492111206, + "num_tokens": 127731693.0, + "step": 4933 + }, + { + "epoch": 0.541840544695805, + "grad_norm": 2.2595231533050537, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7249186038970947, + "num_tokens": 127754534.0, + "step": 4934 + }, + { + "epoch": 0.5419503623984187, + "grad_norm": 1.9671900272369385, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6971086263656616, + "num_tokens": 127785708.0, + "step": 4935 + }, + { + "epoch": 0.5420601801010323, + "grad_norm": 2.4020493030548096, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7252998948097229, + "num_tokens": 127807870.0, + "step": 4936 + }, + { + "epoch": 0.5421699978036459, + "grad_norm": 2.348104476928711, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6954267024993896, + "num_tokens": 127830538.0, + "step": 4937 + }, + { + "epoch": 0.5422798155062596, + "grad_norm": 2.0072288513183594, + "learning_rate": 1e-06, + "loss": 1.069, + "mean_token_accuracy": 0.6773338913917542, + "num_tokens": 127861733.0, + "step": 4938 + }, + { + "epoch": 0.5423896332088732, + "grad_norm": 2.259899616241455, + "learning_rate": 1e-06, + "loss": 1.0431, + "mean_token_accuracy": 0.6883604526519775, + "num_tokens": 127886190.0, + "step": 4939 + }, + { + "epoch": 0.5424994509114869, + "grad_norm": 2.1630897521972656, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6969084739685059, + "num_tokens": 127914203.0, + "step": 4940 + }, + { + "epoch": 0.5426092686141006, + "grad_norm": 2.0970969200134277, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6833668351173401, + "num_tokens": 127942783.0, + "step": 4941 + }, + { + "epoch": 0.5427190863167143, + "grad_norm": 2.267348051071167, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6972933411598206, + "num_tokens": 127967032.0, + "step": 4942 + }, + { + "epoch": 0.5428289040193279, + "grad_norm": 2.4798221588134766, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.716478705406189, + "num_tokens": 127988188.0, + "step": 4943 + }, + { + "epoch": 0.5429387217219416, + "grad_norm": 2.397541046142578, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6900884509086609, + "num_tokens": 128012088.0, + "step": 4944 + }, + { + "epoch": 0.5430485394245552, + "grad_norm": 2.238319158554077, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6983004808425903, + "num_tokens": 128037208.0, + "step": 4945 + }, + { + "epoch": 0.5431583571271689, + "grad_norm": 2.1768910884857178, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7324616312980652, + "num_tokens": 128062799.0, + "step": 4946 + }, + { + "epoch": 0.5432681748297825, + "grad_norm": 2.0027127265930176, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7160485982894897, + "num_tokens": 128093529.0, + "step": 4947 + }, + { + "epoch": 0.5433779925323963, + "grad_norm": 1.98752760887146, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6798588633537292, + "num_tokens": 128123690.0, + "step": 4948 + }, + { + "epoch": 0.5434878102350099, + "grad_norm": 2.270883321762085, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6935415267944336, + "num_tokens": 128148433.0, + "step": 4949 + }, + { + "epoch": 0.5435976279376236, + "grad_norm": 2.2252306938171387, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6872416734695435, + "num_tokens": 128173520.0, + "step": 4950 + }, + { + "epoch": 0.5437074456402372, + "grad_norm": 1.977192997932434, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.7021963596343994, + "num_tokens": 128205469.0, + "step": 4951 + }, + { + "epoch": 0.5438172633428509, + "grad_norm": 2.2652840614318848, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7292460203170776, + "num_tokens": 128228251.0, + "step": 4952 + }, + { + "epoch": 0.5439270810454645, + "grad_norm": 2.03670334815979, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7121778726577759, + "num_tokens": 128255630.0, + "step": 4953 + }, + { + "epoch": 0.5440368987480781, + "grad_norm": 2.347684621810913, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7075255513191223, + "num_tokens": 128278625.0, + "step": 4954 + }, + { + "epoch": 0.5441467164506919, + "grad_norm": 2.3767542839050293, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7162994146347046, + "num_tokens": 128299970.0, + "step": 4955 + }, + { + "epoch": 0.5442565341533055, + "grad_norm": 2.177063465118408, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7114729285240173, + "num_tokens": 128325057.0, + "step": 4956 + }, + { + "epoch": 0.5443663518559192, + "grad_norm": 2.3841071128845215, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7001264691352844, + "num_tokens": 128348579.0, + "step": 4957 + }, + { + "epoch": 0.5444761695585328, + "grad_norm": 2.1438286304473877, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7135515213012695, + "num_tokens": 128374047.0, + "step": 4958 + }, + { + "epoch": 0.5445859872611465, + "grad_norm": 2.6189966201782227, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7071105241775513, + "num_tokens": 128394609.0, + "step": 4959 + }, + { + "epoch": 0.5446958049637601, + "grad_norm": 2.166975259780884, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7003484964370728, + "num_tokens": 128419846.0, + "step": 4960 + }, + { + "epoch": 0.5448056226663738, + "grad_norm": 1.8673043251037598, + "learning_rate": 1e-06, + "loss": 1.0765, + "mean_token_accuracy": 0.6777397394180298, + "num_tokens": 128453276.0, + "step": 4961 + }, + { + "epoch": 0.5449154403689875, + "grad_norm": 2.250488519668579, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6976763010025024, + "num_tokens": 128476588.0, + "step": 4962 + }, + { + "epoch": 0.5450252580716012, + "grad_norm": 2.135164976119995, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7010340690612793, + "num_tokens": 128503964.0, + "step": 4963 + }, + { + "epoch": 0.5451350757742148, + "grad_norm": 2.0630452632904053, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.717455267906189, + "num_tokens": 128531796.0, + "step": 4964 + }, + { + "epoch": 0.5452448934768285, + "grad_norm": 2.2626850605010986, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7080960869789124, + "num_tokens": 128554987.0, + "step": 4965 + }, + { + "epoch": 0.5453547111794421, + "grad_norm": 2.4177968502044678, + "learning_rate": 1e-06, + "loss": 1.0409, + "mean_token_accuracy": 0.6889945268630981, + "num_tokens": 128580111.0, + "step": 4966 + }, + { + "epoch": 0.5454645288820558, + "grad_norm": 1.9967106580734253, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6772472262382507, + "num_tokens": 128612332.0, + "step": 4967 + }, + { + "epoch": 0.5455743465846694, + "grad_norm": 2.3316662311553955, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6885075569152832, + "num_tokens": 128636853.0, + "step": 4968 + }, + { + "epoch": 0.5456841642872831, + "grad_norm": 2.084049701690674, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.72409588098526, + "num_tokens": 128664436.0, + "step": 4969 + }, + { + "epoch": 0.5457939819898968, + "grad_norm": 2.070343494415283, + "learning_rate": 1e-06, + "loss": 1.0596, + "mean_token_accuracy": 0.6850601434707642, + "num_tokens": 128694240.0, + "step": 4970 + }, + { + "epoch": 0.5459037996925105, + "grad_norm": 2.2579479217529297, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.69678795337677, + "num_tokens": 128719182.0, + "step": 4971 + }, + { + "epoch": 0.5460136173951241, + "grad_norm": 2.07523250579834, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7006678581237793, + "num_tokens": 128746729.0, + "step": 4972 + }, + { + "epoch": 0.5461234350977378, + "grad_norm": 2.2687313556671143, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6944547295570374, + "num_tokens": 128772278.0, + "step": 4973 + }, + { + "epoch": 0.5462332528003514, + "grad_norm": 2.2219650745391846, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.690399169921875, + "num_tokens": 128799048.0, + "step": 4974 + }, + { + "epoch": 0.546343070502965, + "grad_norm": 2.1257004737854004, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.703665554523468, + "num_tokens": 128825845.0, + "step": 4975 + }, + { + "epoch": 0.5464528882055787, + "grad_norm": 2.195887565612793, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7067739963531494, + "num_tokens": 128854929.0, + "step": 4976 + }, + { + "epoch": 0.5465627059081924, + "grad_norm": 2.338230848312378, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7060889005661011, + "num_tokens": 128878255.0, + "step": 4977 + }, + { + "epoch": 0.5466725236108061, + "grad_norm": 1.980915904045105, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.691249430179596, + "num_tokens": 128910671.0, + "step": 4978 + }, + { + "epoch": 0.5467823413134197, + "grad_norm": 2.127246856689453, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6931071281433105, + "num_tokens": 128941001.0, + "step": 4979 + }, + { + "epoch": 0.5468921590160334, + "grad_norm": 2.2266271114349365, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7195907235145569, + "num_tokens": 128965677.0, + "step": 4980 + }, + { + "epoch": 0.547001976718647, + "grad_norm": 2.318479299545288, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.6799936294555664, + "num_tokens": 128992259.0, + "step": 4981 + }, + { + "epoch": 0.5471117944212607, + "grad_norm": 1.860216736793518, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7144946455955505, + "num_tokens": 129026500.0, + "step": 4982 + }, + { + "epoch": 0.5472216121238743, + "grad_norm": 2.135113000869751, + "learning_rate": 1e-06, + "loss": 1.0804, + "mean_token_accuracy": 0.6774274706840515, + "num_tokens": 129055142.0, + "step": 4983 + }, + { + "epoch": 0.5473314298264881, + "grad_norm": 1.9324373006820679, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7062702775001526, + "num_tokens": 129087650.0, + "step": 4984 + }, + { + "epoch": 0.5474412475291017, + "grad_norm": 1.9237140417099, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7001343965530396, + "num_tokens": 129117894.0, + "step": 4985 + }, + { + "epoch": 0.5475510652317154, + "grad_norm": 2.1612634658813477, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6972209811210632, + "num_tokens": 129144524.0, + "step": 4986 + }, + { + "epoch": 0.547660882934329, + "grad_norm": 2.4170985221862793, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6919581294059753, + "num_tokens": 129167161.0, + "step": 4987 + }, + { + "epoch": 0.5477707006369427, + "grad_norm": 2.2829554080963135, + "learning_rate": 1e-06, + "loss": 1.0895, + "mean_token_accuracy": 0.676893413066864, + "num_tokens": 129192187.0, + "step": 4988 + }, + { + "epoch": 0.5478805183395563, + "grad_norm": 2.071789503097534, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.7014261484146118, + "num_tokens": 129220666.0, + "step": 4989 + }, + { + "epoch": 0.54799033604217, + "grad_norm": 2.3575992584228516, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7093930840492249, + "num_tokens": 129242823.0, + "step": 4990 + }, + { + "epoch": 0.5481001537447837, + "grad_norm": 2.3071939945220947, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6887969970703125, + "num_tokens": 129267928.0, + "step": 4991 + }, + { + "epoch": 0.5482099714473974, + "grad_norm": 2.344785690307617, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6889498829841614, + "num_tokens": 129292838.0, + "step": 4992 + }, + { + "epoch": 0.548319789150011, + "grad_norm": 2.3790035247802734, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7115367650985718, + "num_tokens": 129314953.0, + "step": 4993 + }, + { + "epoch": 0.5484296068526247, + "grad_norm": 2.2185721397399902, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.720158576965332, + "num_tokens": 129339660.0, + "step": 4994 + }, + { + "epoch": 0.5485394245552383, + "grad_norm": 2.3898472785949707, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7153769731521606, + "num_tokens": 129363111.0, + "step": 4995 + }, + { + "epoch": 0.5486492422578519, + "grad_norm": 2.2438411712646484, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.6881138682365417, + "num_tokens": 129390318.0, + "step": 4996 + }, + { + "epoch": 0.5487590599604656, + "grad_norm": 2.24002742767334, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6942818760871887, + "num_tokens": 129416078.0, + "step": 4997 + }, + { + "epoch": 0.5488688776630792, + "grad_norm": 2.6348323822021484, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.690394401550293, + "num_tokens": 129437106.0, + "step": 4998 + }, + { + "epoch": 0.548978695365693, + "grad_norm": 1.9544681310653687, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7047282457351685, + "num_tokens": 129469652.0, + "step": 4999 + }, + { + "epoch": 0.5490885130683066, + "grad_norm": 2.0914885997772217, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7105337381362915, + "num_tokens": 129496588.0, + "step": 5000 + }, + { + "epoch": 0.5491983307709203, + "grad_norm": 2.2220754623413086, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7057023644447327, + "num_tokens": 129521502.0, + "step": 5001 + }, + { + "epoch": 0.5493081484735339, + "grad_norm": 2.3115410804748535, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7213608026504517, + "num_tokens": 129544748.0, + "step": 5002 + }, + { + "epoch": 0.5494179661761476, + "grad_norm": 2.023266315460205, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7021294236183167, + "num_tokens": 129574117.0, + "step": 5003 + }, + { + "epoch": 0.5495277838787612, + "grad_norm": 2.7816643714904785, + "learning_rate": 1e-06, + "loss": 1.1058, + "mean_token_accuracy": 0.6849749088287354, + "num_tokens": 129595211.0, + "step": 5004 + }, + { + "epoch": 0.5496376015813749, + "grad_norm": 2.325056791305542, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6997728943824768, + "num_tokens": 129619967.0, + "step": 5005 + }, + { + "epoch": 0.5497474192839886, + "grad_norm": 2.1674561500549316, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7127481698989868, + "num_tokens": 129645294.0, + "step": 5006 + }, + { + "epoch": 0.5498572369866023, + "grad_norm": 2.529590129852295, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7042001485824585, + "num_tokens": 129664580.0, + "step": 5007 + }, + { + "epoch": 0.5499670546892159, + "grad_norm": 2.1171281337738037, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6941084861755371, + "num_tokens": 129691874.0, + "step": 5008 + }, + { + "epoch": 0.5500768723918296, + "grad_norm": 2.1832611560821533, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.685122549533844, + "num_tokens": 129719829.0, + "step": 5009 + }, + { + "epoch": 0.5501866900944432, + "grad_norm": 2.202017068862915, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7286814451217651, + "num_tokens": 129743967.0, + "step": 5010 + }, + { + "epoch": 0.5502965077970569, + "grad_norm": 2.318239688873291, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7160879969596863, + "num_tokens": 129767756.0, + "step": 5011 + }, + { + "epoch": 0.5504063254996705, + "grad_norm": 2.3116135597229004, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.708455502986908, + "num_tokens": 129790423.0, + "step": 5012 + }, + { + "epoch": 0.5505161432022843, + "grad_norm": 2.2853198051452637, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6930650472640991, + "num_tokens": 129815734.0, + "step": 5013 + }, + { + "epoch": 0.5506259609048979, + "grad_norm": 2.0767998695373535, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7140387892723083, + "num_tokens": 129842341.0, + "step": 5014 + }, + { + "epoch": 0.5507357786075116, + "grad_norm": 1.8930383920669556, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7026610374450684, + "num_tokens": 129873370.0, + "step": 5015 + }, + { + "epoch": 0.5508455963101252, + "grad_norm": 2.3351473808288574, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7025495767593384, + "num_tokens": 129897220.0, + "step": 5016 + }, + { + "epoch": 0.5509554140127388, + "grad_norm": 2.1913845539093018, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7001227140426636, + "num_tokens": 129922265.0, + "step": 5017 + }, + { + "epoch": 0.5510652317153525, + "grad_norm": 2.344360828399658, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.70325767993927, + "num_tokens": 129946103.0, + "step": 5018 + }, + { + "epoch": 0.5511750494179661, + "grad_norm": 2.15221905708313, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6990891695022583, + "num_tokens": 129973655.0, + "step": 5019 + }, + { + "epoch": 0.5512848671205799, + "grad_norm": 2.190211296081543, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7273644804954529, + "num_tokens": 129998339.0, + "step": 5020 + }, + { + "epoch": 0.5513946848231935, + "grad_norm": 2.0333776473999023, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6907249689102173, + "num_tokens": 130031600.0, + "step": 5021 + }, + { + "epoch": 0.5515045025258072, + "grad_norm": 2.2535781860351562, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6962301731109619, + "num_tokens": 130055600.0, + "step": 5022 + }, + { + "epoch": 0.5516143202284208, + "grad_norm": 2.3269846439361572, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7186995148658752, + "num_tokens": 130077860.0, + "step": 5023 + }, + { + "epoch": 0.5517241379310345, + "grad_norm": 2.2550811767578125, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7125959396362305, + "num_tokens": 130102686.0, + "step": 5024 + }, + { + "epoch": 0.5518339556336481, + "grad_norm": 2.3956198692321777, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6915607452392578, + "num_tokens": 130125007.0, + "step": 5025 + }, + { + "epoch": 0.5519437733362618, + "grad_norm": 1.954215407371521, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.7023102045059204, + "num_tokens": 130154385.0, + "step": 5026 + }, + { + "epoch": 0.5520535910388754, + "grad_norm": 2.249586343765259, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7185028195381165, + "num_tokens": 130179220.0, + "step": 5027 + }, + { + "epoch": 0.5521634087414892, + "grad_norm": 2.270956039428711, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7029545903205872, + "num_tokens": 130204245.0, + "step": 5028 + }, + { + "epoch": 0.5522732264441028, + "grad_norm": 1.9129506349563599, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7209919095039368, + "num_tokens": 130237468.0, + "step": 5029 + }, + { + "epoch": 0.5523830441467165, + "grad_norm": 2.2562873363494873, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6967129111289978, + "num_tokens": 130261337.0, + "step": 5030 + }, + { + "epoch": 0.5524928618493301, + "grad_norm": 1.9848074913024902, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6908338069915771, + "num_tokens": 130292001.0, + "step": 5031 + }, + { + "epoch": 0.5526026795519438, + "grad_norm": 2.3986899852752686, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6995232105255127, + "num_tokens": 130315471.0, + "step": 5032 + }, + { + "epoch": 0.5527124972545574, + "grad_norm": 2.172741174697876, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6968348026275635, + "num_tokens": 130343018.0, + "step": 5033 + }, + { + "epoch": 0.552822314957171, + "grad_norm": 2.215930223464966, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6926840543746948, + "num_tokens": 130368875.0, + "step": 5034 + }, + { + "epoch": 0.5529321326597848, + "grad_norm": 2.05285906791687, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6993587017059326, + "num_tokens": 130396092.0, + "step": 5035 + }, + { + "epoch": 0.5530419503623984, + "grad_norm": 2.3355934619903564, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6934813261032104, + "num_tokens": 130420468.0, + "step": 5036 + }, + { + "epoch": 0.5531517680650121, + "grad_norm": 2.518352746963501, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7062805891036987, + "num_tokens": 130440923.0, + "step": 5037 + }, + { + "epoch": 0.5532615857676257, + "grad_norm": 2.1724538803100586, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7297930717468262, + "num_tokens": 130465065.0, + "step": 5038 + }, + { + "epoch": 0.5533714034702394, + "grad_norm": 2.5437772274017334, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6869249939918518, + "num_tokens": 130487004.0, + "step": 5039 + }, + { + "epoch": 0.553481221172853, + "grad_norm": 2.312173843383789, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6780953407287598, + "num_tokens": 130514627.0, + "step": 5040 + }, + { + "epoch": 0.5535910388754667, + "grad_norm": 2.454838991165161, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6883915066719055, + "num_tokens": 130536856.0, + "step": 5041 + }, + { + "epoch": 0.5537008565780804, + "grad_norm": 2.1362738609313965, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7067080736160278, + "num_tokens": 130562912.0, + "step": 5042 + }, + { + "epoch": 0.5538106742806941, + "grad_norm": 2.1217455863952637, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7055665254592896, + "num_tokens": 130589178.0, + "step": 5043 + }, + { + "epoch": 0.5539204919833077, + "grad_norm": 2.1000289916992188, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.702639639377594, + "num_tokens": 130617557.0, + "step": 5044 + }, + { + "epoch": 0.5540303096859214, + "grad_norm": 2.208315849304199, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.683211624622345, + "num_tokens": 130645531.0, + "step": 5045 + }, + { + "epoch": 0.554140127388535, + "grad_norm": 2.5432417392730713, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7056567668914795, + "num_tokens": 130666322.0, + "step": 5046 + }, + { + "epoch": 0.5542499450911487, + "grad_norm": 2.3668856620788574, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.716541588306427, + "num_tokens": 130689313.0, + "step": 5047 + }, + { + "epoch": 0.5543597627937623, + "grad_norm": 2.0510544776916504, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7067160606384277, + "num_tokens": 130717803.0, + "step": 5048 + }, + { + "epoch": 0.5544695804963761, + "grad_norm": 1.9730933904647827, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6935070753097534, + "num_tokens": 130747161.0, + "step": 5049 + }, + { + "epoch": 0.5545793981989897, + "grad_norm": 2.41463041305542, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6995410919189453, + "num_tokens": 130768925.0, + "step": 5050 + }, + { + "epoch": 0.5546892159016034, + "grad_norm": 2.321659803390503, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7049688696861267, + "num_tokens": 130792964.0, + "step": 5051 + }, + { + "epoch": 0.554799033604217, + "grad_norm": 2.5194077491760254, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7190686464309692, + "num_tokens": 130813028.0, + "step": 5052 + }, + { + "epoch": 0.5549088513068307, + "grad_norm": 2.1839871406555176, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7019897699356079, + "num_tokens": 130838883.0, + "step": 5053 + }, + { + "epoch": 0.5550186690094443, + "grad_norm": 2.1431379318237305, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.696711540222168, + "num_tokens": 130866937.0, + "step": 5054 + }, + { + "epoch": 0.555128486712058, + "grad_norm": 2.2397046089172363, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.69008868932724, + "num_tokens": 130891268.0, + "step": 5055 + }, + { + "epoch": 0.5552383044146716, + "grad_norm": 2.163478374481201, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.698196530342102, + "num_tokens": 130920607.0, + "step": 5056 + }, + { + "epoch": 0.5553481221172853, + "grad_norm": 2.1839752197265625, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.704173743724823, + "num_tokens": 130947053.0, + "step": 5057 + }, + { + "epoch": 0.555457939819899, + "grad_norm": 2.598341226577759, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6939524412155151, + "num_tokens": 130968392.0, + "step": 5058 + }, + { + "epoch": 0.5555677575225126, + "grad_norm": 2.381960868835449, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7291592955589294, + "num_tokens": 130990351.0, + "step": 5059 + }, + { + "epoch": 0.5556775752251263, + "grad_norm": 2.1564834117889404, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7090595960617065, + "num_tokens": 131016161.0, + "step": 5060 + }, + { + "epoch": 0.5557873929277399, + "grad_norm": 2.447951555252075, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7106683254241943, + "num_tokens": 131037117.0, + "step": 5061 + }, + { + "epoch": 0.5558972106303536, + "grad_norm": 2.3530800342559814, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6984710693359375, + "num_tokens": 131060013.0, + "step": 5062 + }, + { + "epoch": 0.5560070283329672, + "grad_norm": 2.3883254528045654, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6954779624938965, + "num_tokens": 131084337.0, + "step": 5063 + }, + { + "epoch": 0.556116846035581, + "grad_norm": 2.0454978942871094, + "learning_rate": 1e-06, + "loss": 1.1051, + "mean_token_accuracy": 0.6710935831069946, + "num_tokens": 131116657.0, + "step": 5064 + }, + { + "epoch": 0.5562266637381946, + "grad_norm": 2.275789499282837, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7065001726150513, + "num_tokens": 131140951.0, + "step": 5065 + }, + { + "epoch": 0.5563364814408083, + "grad_norm": 2.2025022506713867, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7015721201896667, + "num_tokens": 131166105.0, + "step": 5066 + }, + { + "epoch": 0.5564462991434219, + "grad_norm": 2.062978982925415, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6931185722351074, + "num_tokens": 131197081.0, + "step": 5067 + }, + { + "epoch": 0.5565561168460356, + "grad_norm": 2.4446775913238525, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7276444435119629, + "num_tokens": 131218222.0, + "step": 5068 + }, + { + "epoch": 0.5566659345486492, + "grad_norm": 2.2288663387298584, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6943995952606201, + "num_tokens": 131242354.0, + "step": 5069 + }, + { + "epoch": 0.5567757522512629, + "grad_norm": 2.1973040103912354, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7075045108795166, + "num_tokens": 131267390.0, + "step": 5070 + }, + { + "epoch": 0.5568855699538766, + "grad_norm": 2.0751445293426514, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7043598890304565, + "num_tokens": 131295486.0, + "step": 5071 + }, + { + "epoch": 0.5569953876564903, + "grad_norm": 2.35758900642395, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7058905959129333, + "num_tokens": 131317985.0, + "step": 5072 + }, + { + "epoch": 0.5571052053591039, + "grad_norm": 2.665555238723755, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7208064794540405, + "num_tokens": 131336640.0, + "step": 5073 + }, + { + "epoch": 0.5572150230617176, + "grad_norm": 2.326270341873169, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6895627379417419, + "num_tokens": 131361817.0, + "step": 5074 + }, + { + "epoch": 0.5573248407643312, + "grad_norm": 2.0757315158843994, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.72971510887146, + "num_tokens": 131388140.0, + "step": 5075 + }, + { + "epoch": 0.5574346584669448, + "grad_norm": 2.1796741485595703, + "learning_rate": 1e-06, + "loss": 1.0655, + "mean_token_accuracy": 0.6844686269760132, + "num_tokens": 131414579.0, + "step": 5076 + }, + { + "epoch": 0.5575444761695585, + "grad_norm": 1.9415624141693115, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.718223512172699, + "num_tokens": 131445760.0, + "step": 5077 + }, + { + "epoch": 0.5576542938721722, + "grad_norm": 2.1179635524749756, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7002695798873901, + "num_tokens": 131472746.0, + "step": 5078 + }, + { + "epoch": 0.5577641115747859, + "grad_norm": 2.3093485832214355, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7139117121696472, + "num_tokens": 131494842.0, + "step": 5079 + }, + { + "epoch": 0.5578739292773995, + "grad_norm": 2.397984027862549, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7014853954315186, + "num_tokens": 131517296.0, + "step": 5080 + }, + { + "epoch": 0.5579837469800132, + "grad_norm": 2.0649616718292236, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6807259321212769, + "num_tokens": 131546387.0, + "step": 5081 + }, + { + "epoch": 0.5580935646826268, + "grad_norm": 2.665891647338867, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7126126885414124, + "num_tokens": 131564509.0, + "step": 5082 + }, + { + "epoch": 0.5582033823852405, + "grad_norm": 2.377023935317993, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6996133327484131, + "num_tokens": 131588213.0, + "step": 5083 + }, + { + "epoch": 0.5583132000878541, + "grad_norm": 2.1512880325317383, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6929755806922913, + "num_tokens": 131616361.0, + "step": 5084 + }, + { + "epoch": 0.5584230177904678, + "grad_norm": 2.0532350540161133, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6903685927391052, + "num_tokens": 131646562.0, + "step": 5085 + }, + { + "epoch": 0.5585328354930815, + "grad_norm": 2.224036931991577, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7109030485153198, + "num_tokens": 131670099.0, + "step": 5086 + }, + { + "epoch": 0.5586426531956952, + "grad_norm": 2.105301856994629, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.692555844783783, + "num_tokens": 131697822.0, + "step": 5087 + }, + { + "epoch": 0.5587524708983088, + "grad_norm": 2.282766580581665, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7115498781204224, + "num_tokens": 131721692.0, + "step": 5088 + }, + { + "epoch": 0.5588622886009225, + "grad_norm": 2.4780285358428955, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7055724263191223, + "num_tokens": 131743548.0, + "step": 5089 + }, + { + "epoch": 0.5589721063035361, + "grad_norm": 2.240535020828247, + "learning_rate": 1e-06, + "loss": 1.1346, + "mean_token_accuracy": 0.6678884625434875, + "num_tokens": 131769790.0, + "step": 5090 + }, + { + "epoch": 0.5590819240061498, + "grad_norm": 2.0353152751922607, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6962846517562866, + "num_tokens": 131799241.0, + "step": 5091 + }, + { + "epoch": 0.5591917417087634, + "grad_norm": 1.7450125217437744, + "learning_rate": 1e-06, + "loss": 1.0416, + "mean_token_accuracy": 0.6870077848434448, + "num_tokens": 131839024.0, + "step": 5092 + }, + { + "epoch": 0.5593015594113772, + "grad_norm": 2.1108925342559814, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6835628747940063, + "num_tokens": 131866780.0, + "step": 5093 + }, + { + "epoch": 0.5594113771139908, + "grad_norm": 2.148191213607788, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.6967610120773315, + "num_tokens": 131894181.0, + "step": 5094 + }, + { + "epoch": 0.5595211948166045, + "grad_norm": 2.6842637062072754, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6951196789741516, + "num_tokens": 131914061.0, + "step": 5095 + }, + { + "epoch": 0.5596310125192181, + "grad_norm": 2.0750205516815186, + "learning_rate": 1e-06, + "loss": 1.0923, + "mean_token_accuracy": 0.6755515336990356, + "num_tokens": 131944439.0, + "step": 5096 + }, + { + "epoch": 0.5597408302218317, + "grad_norm": 2.1498725414276123, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7246381640434265, + "num_tokens": 131969212.0, + "step": 5097 + }, + { + "epoch": 0.5598506479244454, + "grad_norm": 2.1915695667266846, + "learning_rate": 1e-06, + "loss": 1.1416, + "mean_token_accuracy": 0.6598319411277771, + "num_tokens": 131998108.0, + "step": 5098 + }, + { + "epoch": 0.559960465627059, + "grad_norm": 1.9076842069625854, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6784988641738892, + "num_tokens": 132031655.0, + "step": 5099 + }, + { + "epoch": 0.5600702833296728, + "grad_norm": 2.4833223819732666, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7150943279266357, + "num_tokens": 132051282.0, + "step": 5100 + }, + { + "epoch": 0.5601801010322864, + "grad_norm": 2.1410298347473145, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6969347596168518, + "num_tokens": 132075903.0, + "step": 5101 + }, + { + "epoch": 0.5602899187349001, + "grad_norm": 2.1774914264678955, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6981106996536255, + "num_tokens": 132101575.0, + "step": 5102 + }, + { + "epoch": 0.5603997364375137, + "grad_norm": 2.1987268924713135, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7014723420143127, + "num_tokens": 132127711.0, + "step": 5103 + }, + { + "epoch": 0.5605095541401274, + "grad_norm": 2.3426358699798584, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7170695066452026, + "num_tokens": 132150607.0, + "step": 5104 + }, + { + "epoch": 0.560619371842741, + "grad_norm": 2.2292869091033936, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.7018610239028931, + "num_tokens": 132174942.0, + "step": 5105 + }, + { + "epoch": 0.5607291895453547, + "grad_norm": 2.302833080291748, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.719458281993866, + "num_tokens": 132197226.0, + "step": 5106 + }, + { + "epoch": 0.5608390072479684, + "grad_norm": 2.1382575035095215, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7179639339447021, + "num_tokens": 132224700.0, + "step": 5107 + }, + { + "epoch": 0.5609488249505821, + "grad_norm": 2.2567555904388428, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.690218985080719, + "num_tokens": 132249511.0, + "step": 5108 + }, + { + "epoch": 0.5610586426531957, + "grad_norm": 2.2097649574279785, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6856364011764526, + "num_tokens": 132276127.0, + "step": 5109 + }, + { + "epoch": 0.5611684603558094, + "grad_norm": 2.020707130432129, + "learning_rate": 1e-06, + "loss": 1.0694, + "mean_token_accuracy": 0.6770834922790527, + "num_tokens": 132306672.0, + "step": 5110 + }, + { + "epoch": 0.561278278058423, + "grad_norm": 2.143737316131592, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.7041504383087158, + "num_tokens": 132333910.0, + "step": 5111 + }, + { + "epoch": 0.5613880957610367, + "grad_norm": 2.121711015701294, + "learning_rate": 1e-06, + "loss": 1.0678, + "mean_token_accuracy": 0.6868805885314941, + "num_tokens": 132363479.0, + "step": 5112 + }, + { + "epoch": 0.5614979134636503, + "grad_norm": 2.119394540786743, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7053414583206177, + "num_tokens": 132390880.0, + "step": 5113 + }, + { + "epoch": 0.5616077311662641, + "grad_norm": 2.2294418811798096, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7041422128677368, + "num_tokens": 132416496.0, + "step": 5114 + }, + { + "epoch": 0.5617175488688777, + "grad_norm": 2.227151870727539, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7101269960403442, + "num_tokens": 132441545.0, + "step": 5115 + }, + { + "epoch": 0.5618273665714913, + "grad_norm": 2.135716676712036, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7198569774627686, + "num_tokens": 132466047.0, + "step": 5116 + }, + { + "epoch": 0.561937184274105, + "grad_norm": 2.2829554080963135, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7223063707351685, + "num_tokens": 132487881.0, + "step": 5117 + }, + { + "epoch": 0.5620470019767186, + "grad_norm": 2.1815528869628906, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7197096347808838, + "num_tokens": 132515071.0, + "step": 5118 + }, + { + "epoch": 0.5621568196793323, + "grad_norm": 2.3246757984161377, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6825681924819946, + "num_tokens": 132539234.0, + "step": 5119 + }, + { + "epoch": 0.5622666373819459, + "grad_norm": 2.116164445877075, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6886533498764038, + "num_tokens": 132569368.0, + "step": 5120 + }, + { + "epoch": 0.5623764550845596, + "grad_norm": 2.098771095275879, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7041245102882385, + "num_tokens": 132595696.0, + "step": 5121 + }, + { + "epoch": 0.5624862727871733, + "grad_norm": 1.9686920642852783, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6917855143547058, + "num_tokens": 132627634.0, + "step": 5122 + }, + { + "epoch": 0.562596090489787, + "grad_norm": 2.2178544998168945, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6993106603622437, + "num_tokens": 132652802.0, + "step": 5123 + }, + { + "epoch": 0.5627059081924006, + "grad_norm": 2.1193079948425293, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7255342602729797, + "num_tokens": 132677330.0, + "step": 5124 + }, + { + "epoch": 0.5628157258950143, + "grad_norm": 2.1104366779327393, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7033600807189941, + "num_tokens": 132703704.0, + "step": 5125 + }, + { + "epoch": 0.5629255435976279, + "grad_norm": 2.115232467651367, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7138278484344482, + "num_tokens": 132731686.0, + "step": 5126 + }, + { + "epoch": 0.5630353613002416, + "grad_norm": 2.009219169616699, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7125261425971985, + "num_tokens": 132761771.0, + "step": 5127 + }, + { + "epoch": 0.5631451790028552, + "grad_norm": 2.22526216506958, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6924135684967041, + "num_tokens": 132786439.0, + "step": 5128 + }, + { + "epoch": 0.563254996705469, + "grad_norm": 2.167752504348755, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7256355285644531, + "num_tokens": 132811005.0, + "step": 5129 + }, + { + "epoch": 0.5633648144080826, + "grad_norm": 2.1680216789245605, + "learning_rate": 1e-06, + "loss": 1.1077, + "mean_token_accuracy": 0.6722511053085327, + "num_tokens": 132839399.0, + "step": 5130 + }, + { + "epoch": 0.5634746321106963, + "grad_norm": 2.019228458404541, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6887514591217041, + "num_tokens": 132867618.0, + "step": 5131 + }, + { + "epoch": 0.5635844498133099, + "grad_norm": 1.9280840158462524, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6883459091186523, + "num_tokens": 132901567.0, + "step": 5132 + }, + { + "epoch": 0.5636942675159236, + "grad_norm": 2.270012855529785, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7157682180404663, + "num_tokens": 132925044.0, + "step": 5133 + }, + { + "epoch": 0.5638040852185372, + "grad_norm": 2.527777671813965, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7229003310203552, + "num_tokens": 132944209.0, + "step": 5134 + }, + { + "epoch": 0.5639139029211508, + "grad_norm": 2.039525032043457, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.7021133899688721, + "num_tokens": 132973354.0, + "step": 5135 + }, + { + "epoch": 0.5640237206237646, + "grad_norm": 2.5195987224578857, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.732588529586792, + "num_tokens": 132992898.0, + "step": 5136 + }, + { + "epoch": 0.5641335383263782, + "grad_norm": 2.5124449729919434, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.7011037468910217, + "num_tokens": 133013935.0, + "step": 5137 + }, + { + "epoch": 0.5642433560289919, + "grad_norm": 2.7572851181030273, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7258949279785156, + "num_tokens": 133033719.0, + "step": 5138 + }, + { + "epoch": 0.5643531737316055, + "grad_norm": 2.09747314453125, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6986193656921387, + "num_tokens": 133062200.0, + "step": 5139 + }, + { + "epoch": 0.5644629914342192, + "grad_norm": 2.131671905517578, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.698065996170044, + "num_tokens": 133090360.0, + "step": 5140 + }, + { + "epoch": 0.5645728091368328, + "grad_norm": 2.203906297683716, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6935322284698486, + "num_tokens": 133117850.0, + "step": 5141 + }, + { + "epoch": 0.5646826268394465, + "grad_norm": 2.225322961807251, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7155240178108215, + "num_tokens": 133142324.0, + "step": 5142 + }, + { + "epoch": 0.5647924445420602, + "grad_norm": 2.429887056350708, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7413037419319153, + "num_tokens": 133163820.0, + "step": 5143 + }, + { + "epoch": 0.5649022622446739, + "grad_norm": 2.1363964080810547, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7148709893226624, + "num_tokens": 133190171.0, + "step": 5144 + }, + { + "epoch": 0.5650120799472875, + "grad_norm": 2.3880817890167236, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7140947580337524, + "num_tokens": 133211488.0, + "step": 5145 + }, + { + "epoch": 0.5651218976499012, + "grad_norm": 2.4741194248199463, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7070137858390808, + "num_tokens": 133232643.0, + "step": 5146 + }, + { + "epoch": 0.5652317153525148, + "grad_norm": 2.6717851161956787, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7123832702636719, + "num_tokens": 133250041.0, + "step": 5147 + }, + { + "epoch": 0.5653415330551285, + "grad_norm": 2.2469019889831543, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.692769467830658, + "num_tokens": 133274638.0, + "step": 5148 + }, + { + "epoch": 0.5654513507577421, + "grad_norm": 2.158555269241333, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6916025876998901, + "num_tokens": 133307670.0, + "step": 5149 + }, + { + "epoch": 0.5655611684603558, + "grad_norm": 2.1533331871032715, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7044932842254639, + "num_tokens": 133332989.0, + "step": 5150 + }, + { + "epoch": 0.5656709861629695, + "grad_norm": 2.0832643508911133, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7080777287483215, + "num_tokens": 133360939.0, + "step": 5151 + }, + { + "epoch": 0.5657808038655832, + "grad_norm": 2.118051290512085, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.712218165397644, + "num_tokens": 133388730.0, + "step": 5152 + }, + { + "epoch": 0.5658906215681968, + "grad_norm": 2.0642642974853516, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7018067836761475, + "num_tokens": 133416971.0, + "step": 5153 + }, + { + "epoch": 0.5660004392708105, + "grad_norm": 2.1070773601531982, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.688027024269104, + "num_tokens": 133444888.0, + "step": 5154 + }, + { + "epoch": 0.5661102569734241, + "grad_norm": 2.1579508781433105, + "learning_rate": 1e-06, + "loss": 1.0723, + "mean_token_accuracy": 0.6830668449401855, + "num_tokens": 133471556.0, + "step": 5155 + }, + { + "epoch": 0.5662200746760377, + "grad_norm": 2.000821113586426, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6942676305770874, + "num_tokens": 133504244.0, + "step": 5156 + }, + { + "epoch": 0.5663298923786514, + "grad_norm": 2.0053927898406982, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6857420206069946, + "num_tokens": 133533015.0, + "step": 5157 + }, + { + "epoch": 0.5664397100812651, + "grad_norm": 2.550198554992676, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6940630078315735, + "num_tokens": 133553724.0, + "step": 5158 + }, + { + "epoch": 0.5665495277838788, + "grad_norm": 1.9046868085861206, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6859350800514221, + "num_tokens": 133586088.0, + "step": 5159 + }, + { + "epoch": 0.5666593454864924, + "grad_norm": 2.225597620010376, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6984379291534424, + "num_tokens": 133610842.0, + "step": 5160 + }, + { + "epoch": 0.5667691631891061, + "grad_norm": 2.385605812072754, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7007635831832886, + "num_tokens": 133632718.0, + "step": 5161 + }, + { + "epoch": 0.5668789808917197, + "grad_norm": 2.134770631790161, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6955329179763794, + "num_tokens": 133657697.0, + "step": 5162 + }, + { + "epoch": 0.5669887985943334, + "grad_norm": 2.2243216037750244, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7027018070220947, + "num_tokens": 133680641.0, + "step": 5163 + }, + { + "epoch": 0.567098616296947, + "grad_norm": 2.131420612335205, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7224342823028564, + "num_tokens": 133707898.0, + "step": 5164 + }, + { + "epoch": 0.5672084339995608, + "grad_norm": 2.412414073944092, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.7043637633323669, + "num_tokens": 133729956.0, + "step": 5165 + }, + { + "epoch": 0.5673182517021744, + "grad_norm": 2.0301151275634766, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7327108979225159, + "num_tokens": 133756836.0, + "step": 5166 + }, + { + "epoch": 0.5674280694047881, + "grad_norm": 2.3626468181610107, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7192658185958862, + "num_tokens": 133778886.0, + "step": 5167 + }, + { + "epoch": 0.5675378871074017, + "grad_norm": 2.1379306316375732, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6835513114929199, + "num_tokens": 133809059.0, + "step": 5168 + }, + { + "epoch": 0.5676477048100154, + "grad_norm": 1.918657660484314, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7051417231559753, + "num_tokens": 133839123.0, + "step": 5169 + }, + { + "epoch": 0.567757522512629, + "grad_norm": 1.981795310974121, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7350956201553345, + "num_tokens": 133866533.0, + "step": 5170 + }, + { + "epoch": 0.5678673402152427, + "grad_norm": 2.043107032775879, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6889566779136658, + "num_tokens": 133897919.0, + "step": 5171 + }, + { + "epoch": 0.5679771579178564, + "grad_norm": 1.9748560190200806, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6956558227539062, + "num_tokens": 133928244.0, + "step": 5172 + }, + { + "epoch": 0.5680869756204701, + "grad_norm": 2.118741273880005, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6924530863761902, + "num_tokens": 133956029.0, + "step": 5173 + }, + { + "epoch": 0.5681967933230837, + "grad_norm": 2.0845961570739746, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.6834107637405396, + "num_tokens": 133986082.0, + "step": 5174 + }, + { + "epoch": 0.5683066110256974, + "grad_norm": 2.2061092853546143, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6980119943618774, + "num_tokens": 134011899.0, + "step": 5175 + }, + { + "epoch": 0.568416428728311, + "grad_norm": 2.43453049659729, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7068452835083008, + "num_tokens": 134032637.0, + "step": 5176 + }, + { + "epoch": 0.5685262464309246, + "grad_norm": 2.303318977355957, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7056436538696289, + "num_tokens": 134056987.0, + "step": 5177 + }, + { + "epoch": 0.5686360641335383, + "grad_norm": 2.4026589393615723, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6907936334609985, + "num_tokens": 134081097.0, + "step": 5178 + }, + { + "epoch": 0.5687458818361519, + "grad_norm": 2.3727610111236572, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7093051075935364, + "num_tokens": 134104706.0, + "step": 5179 + }, + { + "epoch": 0.5688556995387657, + "grad_norm": 2.178583860397339, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6867231130599976, + "num_tokens": 134131050.0, + "step": 5180 + }, + { + "epoch": 0.5689655172413793, + "grad_norm": 2.2461886405944824, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6935734748840332, + "num_tokens": 134157262.0, + "step": 5181 + }, + { + "epoch": 0.569075334943993, + "grad_norm": 2.007473945617676, + "learning_rate": 1e-06, + "loss": 1.1317, + "mean_token_accuracy": 0.6682538390159607, + "num_tokens": 134188743.0, + "step": 5182 + }, + { + "epoch": 0.5691851526466066, + "grad_norm": 2.135640859603882, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7039291858673096, + "num_tokens": 134215418.0, + "step": 5183 + }, + { + "epoch": 0.5692949703492203, + "grad_norm": 1.832672119140625, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7084829211235046, + "num_tokens": 134248000.0, + "step": 5184 + }, + { + "epoch": 0.5694047880518339, + "grad_norm": 2.3895061016082764, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7114427089691162, + "num_tokens": 134271086.0, + "step": 5185 + }, + { + "epoch": 0.5695146057544476, + "grad_norm": 2.4437479972839355, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7155894041061401, + "num_tokens": 134293514.0, + "step": 5186 + }, + { + "epoch": 0.5696244234570613, + "grad_norm": 2.1034297943115234, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.6853611469268799, + "num_tokens": 134324497.0, + "step": 5187 + }, + { + "epoch": 0.569734241159675, + "grad_norm": 2.303335666656494, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7088978290557861, + "num_tokens": 134348726.0, + "step": 5188 + }, + { + "epoch": 0.5698440588622886, + "grad_norm": 2.1318626403808594, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7162429690361023, + "num_tokens": 134375115.0, + "step": 5189 + }, + { + "epoch": 0.5699538765649023, + "grad_norm": 2.39623761177063, + "learning_rate": 1e-06, + "loss": 1.0683, + "mean_token_accuracy": 0.6837732195854187, + "num_tokens": 134397922.0, + "step": 5190 + }, + { + "epoch": 0.5700636942675159, + "grad_norm": 2.1219680309295654, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7204705476760864, + "num_tokens": 134425011.0, + "step": 5191 + }, + { + "epoch": 0.5701735119701296, + "grad_norm": 2.310077428817749, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.693925142288208, + "num_tokens": 134452007.0, + "step": 5192 + }, + { + "epoch": 0.5702833296727432, + "grad_norm": 2.0875518321990967, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.700188398361206, + "num_tokens": 134480460.0, + "step": 5193 + }, + { + "epoch": 0.570393147375357, + "grad_norm": 2.2314443588256836, + "learning_rate": 1e-06, + "loss": 1.0876, + "mean_token_accuracy": 0.6760143637657166, + "num_tokens": 134507478.0, + "step": 5194 + }, + { + "epoch": 0.5705029650779706, + "grad_norm": 2.313993215560913, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6885005235671997, + "num_tokens": 134531741.0, + "step": 5195 + }, + { + "epoch": 0.5706127827805842, + "grad_norm": 2.1335928440093994, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7084974050521851, + "num_tokens": 134557636.0, + "step": 5196 + }, + { + "epoch": 0.5707226004831979, + "grad_norm": 2.3282628059387207, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7025783061981201, + "num_tokens": 134579402.0, + "step": 5197 + }, + { + "epoch": 0.5708324181858115, + "grad_norm": 2.131937265396118, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7034813165664673, + "num_tokens": 134606547.0, + "step": 5198 + }, + { + "epoch": 0.5709422358884252, + "grad_norm": 2.307488203048706, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.718506932258606, + "num_tokens": 134630304.0, + "step": 5199 + }, + { + "epoch": 0.5710520535910388, + "grad_norm": 2.3142852783203125, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.700707197189331, + "num_tokens": 134654483.0, + "step": 5200 + }, + { + "epoch": 0.5711618712936526, + "grad_norm": 1.9750173091888428, + "learning_rate": 1e-06, + "loss": 1.0655, + "mean_token_accuracy": 0.6804803609848022, + "num_tokens": 134687064.0, + "step": 5201 + }, + { + "epoch": 0.5712716889962662, + "grad_norm": 2.2304093837738037, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7078059911727905, + "num_tokens": 134713343.0, + "step": 5202 + }, + { + "epoch": 0.5713815066988799, + "grad_norm": 2.419278144836426, + "learning_rate": 1e-06, + "loss": 1.0469, + "mean_token_accuracy": 0.6878218054771423, + "num_tokens": 134736151.0, + "step": 5203 + }, + { + "epoch": 0.5714913244014935, + "grad_norm": 2.2836432456970215, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7185097932815552, + "num_tokens": 134760650.0, + "step": 5204 + }, + { + "epoch": 0.5716011421041072, + "grad_norm": 2.535261631011963, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7091077566146851, + "num_tokens": 134780341.0, + "step": 5205 + }, + { + "epoch": 0.5717109598067208, + "grad_norm": 2.1993000507354736, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.700769305229187, + "num_tokens": 134804164.0, + "step": 5206 + }, + { + "epoch": 0.5718207775093345, + "grad_norm": 2.16365122795105, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7013516426086426, + "num_tokens": 134830883.0, + "step": 5207 + }, + { + "epoch": 0.5719305952119481, + "grad_norm": 2.2158195972442627, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6844915151596069, + "num_tokens": 134857646.0, + "step": 5208 + }, + { + "epoch": 0.5720404129145619, + "grad_norm": 2.0507431030273438, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7066657543182373, + "num_tokens": 134886570.0, + "step": 5209 + }, + { + "epoch": 0.5721502306171755, + "grad_norm": 2.3386969566345215, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.719114363193512, + "num_tokens": 134909165.0, + "step": 5210 + }, + { + "epoch": 0.5722600483197892, + "grad_norm": 2.167137861251831, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.7037736177444458, + "num_tokens": 134935723.0, + "step": 5211 + }, + { + "epoch": 0.5723698660224028, + "grad_norm": 2.658475875854492, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7116413116455078, + "num_tokens": 134952760.0, + "step": 5212 + }, + { + "epoch": 0.5724796837250165, + "grad_norm": 2.159682035446167, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7040549516677856, + "num_tokens": 134978712.0, + "step": 5213 + }, + { + "epoch": 0.5725895014276301, + "grad_norm": 2.1241304874420166, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6840865612030029, + "num_tokens": 135007948.0, + "step": 5214 + }, + { + "epoch": 0.5726993191302437, + "grad_norm": 1.9518052339553833, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6878556609153748, + "num_tokens": 135039118.0, + "step": 5215 + }, + { + "epoch": 0.5728091368328575, + "grad_norm": 1.8734582662582397, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7040092945098877, + "num_tokens": 135076922.0, + "step": 5216 + }, + { + "epoch": 0.5729189545354711, + "grad_norm": 2.0160374641418457, + "learning_rate": 1e-06, + "loss": 1.0618, + "mean_token_accuracy": 0.6860118508338928, + "num_tokens": 135106676.0, + "step": 5217 + }, + { + "epoch": 0.5730287722380848, + "grad_norm": 2.348693609237671, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7173609733581543, + "num_tokens": 135127502.0, + "step": 5218 + }, + { + "epoch": 0.5731385899406984, + "grad_norm": 2.2252674102783203, + "learning_rate": 1e-06, + "loss": 1.0601, + "mean_token_accuracy": 0.6831198930740356, + "num_tokens": 135152895.0, + "step": 5219 + }, + { + "epoch": 0.5732484076433121, + "grad_norm": 1.8962152004241943, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7079579830169678, + "num_tokens": 135185791.0, + "step": 5220 + }, + { + "epoch": 0.5733582253459257, + "grad_norm": 1.9547505378723145, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6897783875465393, + "num_tokens": 135216237.0, + "step": 5221 + }, + { + "epoch": 0.5734680430485394, + "grad_norm": 2.1689107418060303, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6814504861831665, + "num_tokens": 135241407.0, + "step": 5222 + }, + { + "epoch": 0.5735778607511531, + "grad_norm": 2.445028305053711, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7193430662155151, + "num_tokens": 135262741.0, + "step": 5223 + }, + { + "epoch": 0.5736876784537668, + "grad_norm": 2.1509792804718018, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7322655320167542, + "num_tokens": 135285964.0, + "step": 5224 + }, + { + "epoch": 0.5737974961563804, + "grad_norm": 2.0905959606170654, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.692103385925293, + "num_tokens": 135314741.0, + "step": 5225 + }, + { + "epoch": 0.5739073138589941, + "grad_norm": 2.3752222061157227, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.701214075088501, + "num_tokens": 135336909.0, + "step": 5226 + }, + { + "epoch": 0.5740171315616077, + "grad_norm": 1.9886665344238281, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7263548374176025, + "num_tokens": 135364698.0, + "step": 5227 + }, + { + "epoch": 0.5741269492642214, + "grad_norm": 2.2567408084869385, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7027508616447449, + "num_tokens": 135389786.0, + "step": 5228 + }, + { + "epoch": 0.574236766966835, + "grad_norm": 2.270205020904541, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7178357243537903, + "num_tokens": 135413067.0, + "step": 5229 + }, + { + "epoch": 0.5743465846694488, + "grad_norm": 2.43711519241333, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.704669177532196, + "num_tokens": 135436186.0, + "step": 5230 + }, + { + "epoch": 0.5744564023720624, + "grad_norm": 2.3634755611419678, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7303042411804199, + "num_tokens": 135457201.0, + "step": 5231 + }, + { + "epoch": 0.5745662200746761, + "grad_norm": 2.276261806488037, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7103738784790039, + "num_tokens": 135481982.0, + "step": 5232 + }, + { + "epoch": 0.5746760377772897, + "grad_norm": 2.409449338912964, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6922650933265686, + "num_tokens": 135505097.0, + "step": 5233 + }, + { + "epoch": 0.5747858554799034, + "grad_norm": 2.105264663696289, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6962432861328125, + "num_tokens": 135533256.0, + "step": 5234 + }, + { + "epoch": 0.574895673182517, + "grad_norm": 2.0508275032043457, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7193868160247803, + "num_tokens": 135559313.0, + "step": 5235 + }, + { + "epoch": 0.5750054908851306, + "grad_norm": 2.231294631958008, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7101031541824341, + "num_tokens": 135583174.0, + "step": 5236 + }, + { + "epoch": 0.5751153085877443, + "grad_norm": 2.089505434036255, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.703335702419281, + "num_tokens": 135610717.0, + "step": 5237 + }, + { + "epoch": 0.575225126290358, + "grad_norm": 2.269958019256592, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6962077021598816, + "num_tokens": 135635496.0, + "step": 5238 + }, + { + "epoch": 0.5753349439929717, + "grad_norm": 2.0375077724456787, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7194098234176636, + "num_tokens": 135662797.0, + "step": 5239 + }, + { + "epoch": 0.5754447616955853, + "grad_norm": 2.68814754486084, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7391891479492188, + "num_tokens": 135681463.0, + "step": 5240 + }, + { + "epoch": 0.575554579398199, + "grad_norm": 1.9876923561096191, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6965574026107788, + "num_tokens": 135712559.0, + "step": 5241 + }, + { + "epoch": 0.5756643971008126, + "grad_norm": 2.3061466217041016, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7135539650917053, + "num_tokens": 135735649.0, + "step": 5242 + }, + { + "epoch": 0.5757742148034263, + "grad_norm": 2.2283120155334473, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6967087388038635, + "num_tokens": 135763058.0, + "step": 5243 + }, + { + "epoch": 0.5758840325060399, + "grad_norm": 2.303964853286743, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7156988382339478, + "num_tokens": 135785873.0, + "step": 5244 + }, + { + "epoch": 0.5759938502086537, + "grad_norm": 2.2927896976470947, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7188670635223389, + "num_tokens": 135809402.0, + "step": 5245 + }, + { + "epoch": 0.5761036679112673, + "grad_norm": 2.4307100772857666, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6876019239425659, + "num_tokens": 135832842.0, + "step": 5246 + }, + { + "epoch": 0.576213485613881, + "grad_norm": 2.220358371734619, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6969698667526245, + "num_tokens": 135859345.0, + "step": 5247 + }, + { + "epoch": 0.5763233033164946, + "grad_norm": 2.208162784576416, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7044858932495117, + "num_tokens": 135885536.0, + "step": 5248 + }, + { + "epoch": 0.5764331210191083, + "grad_norm": 2.2130866050720215, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.6812914609909058, + "num_tokens": 135912140.0, + "step": 5249 + }, + { + "epoch": 0.5765429387217219, + "grad_norm": 2.1314525604248047, + "learning_rate": 1e-06, + "loss": 1.0996, + "mean_token_accuracy": 0.6795930862426758, + "num_tokens": 135942276.0, + "step": 5250 + }, + { + "epoch": 0.5766527564243356, + "grad_norm": 2.0565879344940186, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7081430554389954, + "num_tokens": 135969380.0, + "step": 5251 + }, + { + "epoch": 0.5767625741269493, + "grad_norm": 1.9741981029510498, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7245699167251587, + "num_tokens": 135997935.0, + "step": 5252 + }, + { + "epoch": 0.576872391829563, + "grad_norm": 2.1842923164367676, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6956741809844971, + "num_tokens": 136022158.0, + "step": 5253 + }, + { + "epoch": 0.5769822095321766, + "grad_norm": 2.0963833332061768, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.708547830581665, + "num_tokens": 136048515.0, + "step": 5254 + }, + { + "epoch": 0.5770920272347903, + "grad_norm": 2.3349788188934326, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7087159752845764, + "num_tokens": 136072347.0, + "step": 5255 + }, + { + "epoch": 0.5772018449374039, + "grad_norm": 2.317368984222412, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6821893453598022, + "num_tokens": 136098672.0, + "step": 5256 + }, + { + "epoch": 0.5773116626400175, + "grad_norm": 2.2234489917755127, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6990754008293152, + "num_tokens": 136123095.0, + "step": 5257 + }, + { + "epoch": 0.5774214803426312, + "grad_norm": 2.1476330757141113, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.6819332838058472, + "num_tokens": 136150673.0, + "step": 5258 + }, + { + "epoch": 0.577531298045245, + "grad_norm": 1.9034262895584106, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6878838539123535, + "num_tokens": 136185183.0, + "step": 5259 + }, + { + "epoch": 0.5776411157478586, + "grad_norm": 1.8743993043899536, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7096129059791565, + "num_tokens": 136217660.0, + "step": 5260 + }, + { + "epoch": 0.5777509334504722, + "grad_norm": 2.30937123298645, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7135690450668335, + "num_tokens": 136240480.0, + "step": 5261 + }, + { + "epoch": 0.5778607511530859, + "grad_norm": 2.349940061569214, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6878050565719604, + "num_tokens": 136265258.0, + "step": 5262 + }, + { + "epoch": 0.5779705688556995, + "grad_norm": 2.175374746322632, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.706372857093811, + "num_tokens": 136291840.0, + "step": 5263 + }, + { + "epoch": 0.5780803865583132, + "grad_norm": 2.0900864601135254, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7128094434738159, + "num_tokens": 136320057.0, + "step": 5264 + }, + { + "epoch": 0.5781902042609268, + "grad_norm": 2.0415432453155518, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6831445693969727, + "num_tokens": 136350046.0, + "step": 5265 + }, + { + "epoch": 0.5783000219635406, + "grad_norm": 2.120877265930176, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.712037205696106, + "num_tokens": 136378683.0, + "step": 5266 + }, + { + "epoch": 0.5784098396661542, + "grad_norm": 2.0518088340759277, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.6811327338218689, + "num_tokens": 136408297.0, + "step": 5267 + }, + { + "epoch": 0.5785196573687679, + "grad_norm": 2.0349273681640625, + "learning_rate": 1e-06, + "loss": 1.0554, + "mean_token_accuracy": 0.6788793802261353, + "num_tokens": 136438358.0, + "step": 5268 + }, + { + "epoch": 0.5786294750713815, + "grad_norm": 2.1224422454833984, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.7041116952896118, + "num_tokens": 136464614.0, + "step": 5269 + }, + { + "epoch": 0.5787392927739952, + "grad_norm": 2.190777540206909, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7041881680488586, + "num_tokens": 136487921.0, + "step": 5270 + }, + { + "epoch": 0.5788491104766088, + "grad_norm": 2.2709619998931885, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7004870772361755, + "num_tokens": 136512898.0, + "step": 5271 + }, + { + "epoch": 0.5789589281792225, + "grad_norm": 2.178926706314087, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7264530658721924, + "num_tokens": 136539048.0, + "step": 5272 + }, + { + "epoch": 0.5790687458818361, + "grad_norm": 2.1622416973114014, + "learning_rate": 1e-06, + "loss": 1.0679, + "mean_token_accuracy": 0.6760867834091187, + "num_tokens": 136567431.0, + "step": 5273 + }, + { + "epoch": 0.5791785635844499, + "grad_norm": 2.1597533226013184, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7136610746383667, + "num_tokens": 136593366.0, + "step": 5274 + }, + { + "epoch": 0.5792883812870635, + "grad_norm": 2.504403829574585, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6920022368431091, + "num_tokens": 136614886.0, + "step": 5275 + }, + { + "epoch": 0.5793981989896771, + "grad_norm": 2.0925207138061523, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6869527101516724, + "num_tokens": 136644948.0, + "step": 5276 + }, + { + "epoch": 0.5795080166922908, + "grad_norm": 2.452733039855957, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7049557566642761, + "num_tokens": 136667190.0, + "step": 5277 + }, + { + "epoch": 0.5796178343949044, + "grad_norm": 2.169219970703125, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7195017337799072, + "num_tokens": 136691470.0, + "step": 5278 + }, + { + "epoch": 0.5797276520975181, + "grad_norm": 2.011945962905884, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.7004324197769165, + "num_tokens": 136720589.0, + "step": 5279 + }, + { + "epoch": 0.5798374698001317, + "grad_norm": 1.9299763441085815, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6866239309310913, + "num_tokens": 136751274.0, + "step": 5280 + }, + { + "epoch": 0.5799472875027455, + "grad_norm": 2.6745753288269043, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6959271430969238, + "num_tokens": 136772880.0, + "step": 5281 + }, + { + "epoch": 0.5800571052053591, + "grad_norm": 2.1404316425323486, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6981350183486938, + "num_tokens": 136800349.0, + "step": 5282 + }, + { + "epoch": 0.5801669229079728, + "grad_norm": 2.5862109661102295, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7223353385925293, + "num_tokens": 136819943.0, + "step": 5283 + }, + { + "epoch": 0.5802767406105864, + "grad_norm": 2.020395278930664, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.7017036080360413, + "num_tokens": 136849882.0, + "step": 5284 + }, + { + "epoch": 0.5803865583132001, + "grad_norm": 2.0708343982696533, + "learning_rate": 1e-06, + "loss": 1.0656, + "mean_token_accuracy": 0.6743288040161133, + "num_tokens": 136878293.0, + "step": 5285 + }, + { + "epoch": 0.5804963760158137, + "grad_norm": 2.171419382095337, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6947187185287476, + "num_tokens": 136906377.0, + "step": 5286 + }, + { + "epoch": 0.5806061937184274, + "grad_norm": 2.0834858417510986, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6930621862411499, + "num_tokens": 136937276.0, + "step": 5287 + }, + { + "epoch": 0.5807160114210411, + "grad_norm": 2.1937084197998047, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7017862796783447, + "num_tokens": 136962371.0, + "step": 5288 + }, + { + "epoch": 0.5808258291236548, + "grad_norm": 2.0404906272888184, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.717131495475769, + "num_tokens": 136989852.0, + "step": 5289 + }, + { + "epoch": 0.5809356468262684, + "grad_norm": 2.5543107986450195, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7109126448631287, + "num_tokens": 137008127.0, + "step": 5290 + }, + { + "epoch": 0.5810454645288821, + "grad_norm": 2.1682071685791016, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7432224750518799, + "num_tokens": 137031663.0, + "step": 5291 + }, + { + "epoch": 0.5811552822314957, + "grad_norm": 2.223677635192871, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6962062120437622, + "num_tokens": 137059073.0, + "step": 5292 + }, + { + "epoch": 0.5812650999341094, + "grad_norm": 2.3979504108428955, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7279735207557678, + "num_tokens": 137079263.0, + "step": 5293 + }, + { + "epoch": 0.581374917636723, + "grad_norm": 1.9953173398971558, + "learning_rate": 1e-06, + "loss": 1.0598, + "mean_token_accuracy": 0.6833559274673462, + "num_tokens": 137111579.0, + "step": 5294 + }, + { + "epoch": 0.5814847353393368, + "grad_norm": 2.103609323501587, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7213839292526245, + "num_tokens": 137136444.0, + "step": 5295 + }, + { + "epoch": 0.5815945530419504, + "grad_norm": 2.139718770980835, + "learning_rate": 1e-06, + "loss": 1.1282, + "mean_token_accuracy": 0.6608518958091736, + "num_tokens": 137165499.0, + "step": 5296 + }, + { + "epoch": 0.581704370744564, + "grad_norm": 2.0047030448913574, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6950418949127197, + "num_tokens": 137194506.0, + "step": 5297 + }, + { + "epoch": 0.5818141884471777, + "grad_norm": 2.1244313716888428, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7017650604248047, + "num_tokens": 137222222.0, + "step": 5298 + }, + { + "epoch": 0.5819240061497913, + "grad_norm": 2.2179815769195557, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6897766590118408, + "num_tokens": 137248256.0, + "step": 5299 + }, + { + "epoch": 0.582033823852405, + "grad_norm": 2.150033950805664, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7355262041091919, + "num_tokens": 137272022.0, + "step": 5300 + }, + { + "epoch": 0.5821436415550186, + "grad_norm": 2.334279775619507, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7030592560768127, + "num_tokens": 137294545.0, + "step": 5301 + }, + { + "epoch": 0.5822534592576323, + "grad_norm": 2.4747657775878906, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7097513675689697, + "num_tokens": 137314838.0, + "step": 5302 + }, + { + "epoch": 0.582363276960246, + "grad_norm": 2.1272356510162354, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6998379826545715, + "num_tokens": 137340879.0, + "step": 5303 + }, + { + "epoch": 0.5824730946628597, + "grad_norm": 2.1658713817596436, + "learning_rate": 1e-06, + "loss": 1.0738, + "mean_token_accuracy": 0.6937469244003296, + "num_tokens": 137368376.0, + "step": 5304 + }, + { + "epoch": 0.5825829123654733, + "grad_norm": 2.0965447425842285, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6975631713867188, + "num_tokens": 137396813.0, + "step": 5305 + }, + { + "epoch": 0.582692730068087, + "grad_norm": 2.2429912090301514, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7059509754180908, + "num_tokens": 137421814.0, + "step": 5306 + }, + { + "epoch": 0.5828025477707006, + "grad_norm": 2.098607301712036, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7055429220199585, + "num_tokens": 137448987.0, + "step": 5307 + }, + { + "epoch": 0.5829123654733143, + "grad_norm": 1.9130324125289917, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7024241089820862, + "num_tokens": 137478806.0, + "step": 5308 + }, + { + "epoch": 0.5830221831759279, + "grad_norm": 2.2079689502716064, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6846995949745178, + "num_tokens": 137504325.0, + "step": 5309 + }, + { + "epoch": 0.5831320008785417, + "grad_norm": 2.4960520267486572, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.704595685005188, + "num_tokens": 137525204.0, + "step": 5310 + }, + { + "epoch": 0.5832418185811553, + "grad_norm": 2.1977972984313965, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7095699906349182, + "num_tokens": 137551618.0, + "step": 5311 + }, + { + "epoch": 0.583351636283769, + "grad_norm": 2.2069902420043945, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7060266733169556, + "num_tokens": 137577095.0, + "step": 5312 + }, + { + "epoch": 0.5834614539863826, + "grad_norm": 2.0128955841064453, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7168379426002502, + "num_tokens": 137606454.0, + "step": 5313 + }, + { + "epoch": 0.5835712716889963, + "grad_norm": 2.0815627574920654, + "learning_rate": 1e-06, + "loss": 1.077, + "mean_token_accuracy": 0.6720321178436279, + "num_tokens": 137634449.0, + "step": 5314 + }, + { + "epoch": 0.5836810893916099, + "grad_norm": 2.1920549869537354, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.7021588087081909, + "num_tokens": 137658097.0, + "step": 5315 + }, + { + "epoch": 0.5837909070942235, + "grad_norm": 2.331937551498413, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6870723962783813, + "num_tokens": 137683043.0, + "step": 5316 + }, + { + "epoch": 0.5839007247968373, + "grad_norm": 2.4404919147491455, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.719937801361084, + "num_tokens": 137703335.0, + "step": 5317 + }, + { + "epoch": 0.584010542499451, + "grad_norm": 2.5218546390533447, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7085201144218445, + "num_tokens": 137722898.0, + "step": 5318 + }, + { + "epoch": 0.5841203602020646, + "grad_norm": 1.8661584854125977, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7071462869644165, + "num_tokens": 137755357.0, + "step": 5319 + }, + { + "epoch": 0.5842301779046782, + "grad_norm": 2.019819498062134, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7022433280944824, + "num_tokens": 137785996.0, + "step": 5320 + }, + { + "epoch": 0.5843399956072919, + "grad_norm": 2.179857015609741, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.7044992446899414, + "num_tokens": 137812226.0, + "step": 5321 + }, + { + "epoch": 0.5844498133099055, + "grad_norm": 1.932637333869934, + "learning_rate": 1e-06, + "loss": 1.1306, + "mean_token_accuracy": 0.665429413318634, + "num_tokens": 137845770.0, + "step": 5322 + }, + { + "epoch": 0.5845596310125192, + "grad_norm": 2.10201096534729, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6973324418067932, + "num_tokens": 137874955.0, + "step": 5323 + }, + { + "epoch": 0.5846694487151329, + "grad_norm": 1.9255915880203247, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7096912860870361, + "num_tokens": 137908104.0, + "step": 5324 + }, + { + "epoch": 0.5847792664177466, + "grad_norm": 2.2491533756256104, + "learning_rate": 1e-06, + "loss": 1.0632, + "mean_token_accuracy": 0.6832302808761597, + "num_tokens": 137933143.0, + "step": 5325 + }, + { + "epoch": 0.5848890841203602, + "grad_norm": 1.9915869235992432, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6824511289596558, + "num_tokens": 137963416.0, + "step": 5326 + }, + { + "epoch": 0.5849989018229739, + "grad_norm": 2.187582492828369, + "learning_rate": 1e-06, + "loss": 1.084, + "mean_token_accuracy": 0.6708192825317383, + "num_tokens": 137990308.0, + "step": 5327 + }, + { + "epoch": 0.5851087195255875, + "grad_norm": 2.276592254638672, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.725670337677002, + "num_tokens": 138012509.0, + "step": 5328 + }, + { + "epoch": 0.5852185372282012, + "grad_norm": 2.5060276985168457, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.720605194568634, + "num_tokens": 138032504.0, + "step": 5329 + }, + { + "epoch": 0.5853283549308148, + "grad_norm": 2.122466564178467, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.726070761680603, + "num_tokens": 138057651.0, + "step": 5330 + }, + { + "epoch": 0.5854381726334285, + "grad_norm": 2.097602367401123, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7178325653076172, + "num_tokens": 138086199.0, + "step": 5331 + }, + { + "epoch": 0.5855479903360422, + "grad_norm": 2.0957937240600586, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7157090306282043, + "num_tokens": 138113881.0, + "step": 5332 + }, + { + "epoch": 0.5856578080386559, + "grad_norm": 2.4405477046966553, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7206730842590332, + "num_tokens": 138135806.0, + "step": 5333 + }, + { + "epoch": 0.5857676257412695, + "grad_norm": 2.297988176345825, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7110812664031982, + "num_tokens": 138159020.0, + "step": 5334 + }, + { + "epoch": 0.5858774434438832, + "grad_norm": 2.111707925796509, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7031238079071045, + "num_tokens": 138184601.0, + "step": 5335 + }, + { + "epoch": 0.5859872611464968, + "grad_norm": 2.1409904956817627, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6852489709854126, + "num_tokens": 138212102.0, + "step": 5336 + }, + { + "epoch": 0.5860970788491104, + "grad_norm": 2.24798846244812, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6853411197662354, + "num_tokens": 138238813.0, + "step": 5337 + }, + { + "epoch": 0.5862068965517241, + "grad_norm": 2.4192938804626465, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7003908157348633, + "num_tokens": 138260141.0, + "step": 5338 + }, + { + "epoch": 0.5863167142543378, + "grad_norm": 2.1944079399108887, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7266751527786255, + "num_tokens": 138283651.0, + "step": 5339 + }, + { + "epoch": 0.5864265319569515, + "grad_norm": 1.880815863609314, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6971907019615173, + "num_tokens": 138315435.0, + "step": 5340 + }, + { + "epoch": 0.5865363496595651, + "grad_norm": 2.0591342449188232, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7104400396347046, + "num_tokens": 138342803.0, + "step": 5341 + }, + { + "epoch": 0.5866461673621788, + "grad_norm": 2.3901796340942383, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6967065334320068, + "num_tokens": 138365491.0, + "step": 5342 + }, + { + "epoch": 0.5867559850647924, + "grad_norm": 1.9380919933319092, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7201968431472778, + "num_tokens": 138395065.0, + "step": 5343 + }, + { + "epoch": 0.5868658027674061, + "grad_norm": 2.312195062637329, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.6825602054595947, + "num_tokens": 138420604.0, + "step": 5344 + }, + { + "epoch": 0.5869756204700197, + "grad_norm": 2.1482667922973633, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7035057544708252, + "num_tokens": 138448634.0, + "step": 5345 + }, + { + "epoch": 0.5870854381726335, + "grad_norm": 2.0586166381835938, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6981082558631897, + "num_tokens": 138477350.0, + "step": 5346 + }, + { + "epoch": 0.5871952558752471, + "grad_norm": 2.514019250869751, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7082058191299438, + "num_tokens": 138499062.0, + "step": 5347 + }, + { + "epoch": 0.5873050735778608, + "grad_norm": 2.38657808303833, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.702358067035675, + "num_tokens": 138520145.0, + "step": 5348 + }, + { + "epoch": 0.5874148912804744, + "grad_norm": 2.433502435684204, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7010746002197266, + "num_tokens": 138541908.0, + "step": 5349 + }, + { + "epoch": 0.5875247089830881, + "grad_norm": 2.0663719177246094, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7393777370452881, + "num_tokens": 138566648.0, + "step": 5350 + }, + { + "epoch": 0.5876345266857017, + "grad_norm": 2.048264265060425, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7161364555358887, + "num_tokens": 138593706.0, + "step": 5351 + }, + { + "epoch": 0.5877443443883154, + "grad_norm": 2.116424083709717, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7061420679092407, + "num_tokens": 138622965.0, + "step": 5352 + }, + { + "epoch": 0.5878541620909291, + "grad_norm": 2.2228355407714844, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6917035579681396, + "num_tokens": 138649693.0, + "step": 5353 + }, + { + "epoch": 0.5879639797935428, + "grad_norm": 1.823933720588684, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6953825950622559, + "num_tokens": 138681641.0, + "step": 5354 + }, + { + "epoch": 0.5880737974961564, + "grad_norm": 2.242365598678589, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6969847083091736, + "num_tokens": 138705096.0, + "step": 5355 + }, + { + "epoch": 0.58818361519877, + "grad_norm": 2.4883601665496826, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7121566534042358, + "num_tokens": 138725318.0, + "step": 5356 + }, + { + "epoch": 0.5882934329013837, + "grad_norm": 2.135641098022461, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7101776003837585, + "num_tokens": 138752583.0, + "step": 5357 + }, + { + "epoch": 0.5884032506039973, + "grad_norm": 2.2962300777435303, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7122516632080078, + "num_tokens": 138775165.0, + "step": 5358 + }, + { + "epoch": 0.588513068306611, + "grad_norm": 2.0437960624694824, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7112183570861816, + "num_tokens": 138803464.0, + "step": 5359 + }, + { + "epoch": 0.5886228860092246, + "grad_norm": 2.191505193710327, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6948472857475281, + "num_tokens": 138830517.0, + "step": 5360 + }, + { + "epoch": 0.5887327037118384, + "grad_norm": 2.4082765579223633, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.7019352912902832, + "num_tokens": 138852743.0, + "step": 5361 + }, + { + "epoch": 0.588842521414452, + "grad_norm": 2.231421709060669, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7076334953308105, + "num_tokens": 138877770.0, + "step": 5362 + }, + { + "epoch": 0.5889523391170657, + "grad_norm": 2.0020601749420166, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6965579390525818, + "num_tokens": 138907114.0, + "step": 5363 + }, + { + "epoch": 0.5890621568196793, + "grad_norm": 2.0952811241149902, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7136119604110718, + "num_tokens": 138934234.0, + "step": 5364 + }, + { + "epoch": 0.589171974522293, + "grad_norm": 2.6244800090789795, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7088490128517151, + "num_tokens": 138953750.0, + "step": 5365 + }, + { + "epoch": 0.5892817922249066, + "grad_norm": 2.028968334197998, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.70383620262146, + "num_tokens": 138985148.0, + "step": 5366 + }, + { + "epoch": 0.5893916099275203, + "grad_norm": 2.187256336212158, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7204477787017822, + "num_tokens": 139009650.0, + "step": 5367 + }, + { + "epoch": 0.589501427630134, + "grad_norm": 2.3082330226898193, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7260854840278625, + "num_tokens": 139030423.0, + "step": 5368 + }, + { + "epoch": 0.5896112453327477, + "grad_norm": 2.2632479667663574, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.702901303768158, + "num_tokens": 139056646.0, + "step": 5369 + }, + { + "epoch": 0.5897210630353613, + "grad_norm": 2.4952592849731445, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7181228399276733, + "num_tokens": 139076987.0, + "step": 5370 + }, + { + "epoch": 0.589830880737975, + "grad_norm": 2.1131322383880615, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7252021431922913, + "num_tokens": 139101805.0, + "step": 5371 + }, + { + "epoch": 0.5899406984405886, + "grad_norm": 1.9656286239624023, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6980769634246826, + "num_tokens": 139129225.0, + "step": 5372 + }, + { + "epoch": 0.5900505161432023, + "grad_norm": 2.11555814743042, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.712225615978241, + "num_tokens": 139155600.0, + "step": 5373 + }, + { + "epoch": 0.5901603338458159, + "grad_norm": 2.1979596614837646, + "learning_rate": 1e-06, + "loss": 1.0492, + "mean_token_accuracy": 0.6898609399795532, + "num_tokens": 139183045.0, + "step": 5374 + }, + { + "epoch": 0.5902701515484297, + "grad_norm": 1.8559321165084839, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7018440961837769, + "num_tokens": 139216042.0, + "step": 5375 + }, + { + "epoch": 0.5903799692510433, + "grad_norm": 2.255760669708252, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7066216468811035, + "num_tokens": 139240011.0, + "step": 5376 + }, + { + "epoch": 0.590489786953657, + "grad_norm": 2.165536642074585, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7138845920562744, + "num_tokens": 139268069.0, + "step": 5377 + }, + { + "epoch": 0.5905996046562706, + "grad_norm": 2.4021785259246826, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7238013744354248, + "num_tokens": 139289186.0, + "step": 5378 + }, + { + "epoch": 0.5907094223588842, + "grad_norm": 2.0937535762786865, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7066327333450317, + "num_tokens": 139318862.0, + "step": 5379 + }, + { + "epoch": 0.5908192400614979, + "grad_norm": 2.3702807426452637, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7038545608520508, + "num_tokens": 139342120.0, + "step": 5380 + }, + { + "epoch": 0.5909290577641115, + "grad_norm": 2.0335426330566406, + "learning_rate": 1e-06, + "loss": 1.1154, + "mean_token_accuracy": 0.6660165786743164, + "num_tokens": 139374511.0, + "step": 5381 + }, + { + "epoch": 0.5910388754667253, + "grad_norm": 2.4589545726776123, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7266777157783508, + "num_tokens": 139395596.0, + "step": 5382 + }, + { + "epoch": 0.5911486931693389, + "grad_norm": 2.2083852291107178, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6951849460601807, + "num_tokens": 139420950.0, + "step": 5383 + }, + { + "epoch": 0.5912585108719526, + "grad_norm": 2.0918619632720947, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7133089303970337, + "num_tokens": 139449011.0, + "step": 5384 + }, + { + "epoch": 0.5913683285745662, + "grad_norm": 1.8957011699676514, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.6905843019485474, + "num_tokens": 139480828.0, + "step": 5385 + }, + { + "epoch": 0.5914781462771799, + "grad_norm": 2.179755210876465, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7017635107040405, + "num_tokens": 139507020.0, + "step": 5386 + }, + { + "epoch": 0.5915879639797935, + "grad_norm": 2.09714937210083, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7015564441680908, + "num_tokens": 139534335.0, + "step": 5387 + }, + { + "epoch": 0.5916977816824072, + "grad_norm": 2.084296941757202, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.708154022693634, + "num_tokens": 139561396.0, + "step": 5388 + }, + { + "epoch": 0.5918075993850208, + "grad_norm": 2.1216351985931396, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6893526315689087, + "num_tokens": 139588667.0, + "step": 5389 + }, + { + "epoch": 0.5919174170876346, + "grad_norm": 1.9751019477844238, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6871165633201599, + "num_tokens": 139621085.0, + "step": 5390 + }, + { + "epoch": 0.5920272347902482, + "grad_norm": 2.201327323913574, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7242255210876465, + "num_tokens": 139644535.0, + "step": 5391 + }, + { + "epoch": 0.5921370524928619, + "grad_norm": 1.9175078868865967, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7050440311431885, + "num_tokens": 139675145.0, + "step": 5392 + }, + { + "epoch": 0.5922468701954755, + "grad_norm": 2.284827470779419, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7164040803909302, + "num_tokens": 139698271.0, + "step": 5393 + }, + { + "epoch": 0.5923566878980892, + "grad_norm": 2.0129120349884033, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7074024677276611, + "num_tokens": 139728259.0, + "step": 5394 + }, + { + "epoch": 0.5924665056007028, + "grad_norm": 2.130251407623291, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7162869572639465, + "num_tokens": 139751081.0, + "step": 5395 + }, + { + "epoch": 0.5925763233033164, + "grad_norm": 1.9478226900100708, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6972182989120483, + "num_tokens": 139783392.0, + "step": 5396 + }, + { + "epoch": 0.5926861410059302, + "grad_norm": 2.155670642852783, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7067058682441711, + "num_tokens": 139809224.0, + "step": 5397 + }, + { + "epoch": 0.5927959587085438, + "grad_norm": 2.1414294242858887, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7315493226051331, + "num_tokens": 139835349.0, + "step": 5398 + }, + { + "epoch": 0.5929057764111575, + "grad_norm": 2.2592878341674805, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7223945260047913, + "num_tokens": 139858710.0, + "step": 5399 + }, + { + "epoch": 0.5930155941137711, + "grad_norm": 2.180548667907715, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6924698352813721, + "num_tokens": 139883838.0, + "step": 5400 + }, + { + "epoch": 0.5931254118163848, + "grad_norm": 2.3268520832061768, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6980432271957397, + "num_tokens": 139908596.0, + "step": 5401 + }, + { + "epoch": 0.5932352295189984, + "grad_norm": 2.116551637649536, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.706525444984436, + "num_tokens": 139937133.0, + "step": 5402 + }, + { + "epoch": 0.5933450472216121, + "grad_norm": 2.207975387573242, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6898589134216309, + "num_tokens": 139962948.0, + "step": 5403 + }, + { + "epoch": 0.5934548649242258, + "grad_norm": 2.031891345977783, + "learning_rate": 1e-06, + "loss": 1.1098, + "mean_token_accuracy": 0.6636438369750977, + "num_tokens": 139993617.0, + "step": 5404 + }, + { + "epoch": 0.5935646826268395, + "grad_norm": 2.082883834838867, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.696471095085144, + "num_tokens": 140021652.0, + "step": 5405 + }, + { + "epoch": 0.5936745003294531, + "grad_norm": 2.151604175567627, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6950074434280396, + "num_tokens": 140049633.0, + "step": 5406 + }, + { + "epoch": 0.5937843180320668, + "grad_norm": 2.453648805618286, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7378714680671692, + "num_tokens": 140070854.0, + "step": 5407 + }, + { + "epoch": 0.5938941357346804, + "grad_norm": 2.2670376300811768, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7086470723152161, + "num_tokens": 140094943.0, + "step": 5408 + }, + { + "epoch": 0.5940039534372941, + "grad_norm": 2.4687929153442383, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7035109400749207, + "num_tokens": 140114169.0, + "step": 5409 + }, + { + "epoch": 0.5941137711399077, + "grad_norm": 2.2052245140075684, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.7019284963607788, + "num_tokens": 140139128.0, + "step": 5410 + }, + { + "epoch": 0.5942235888425215, + "grad_norm": 2.1572153568267822, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7175604104995728, + "num_tokens": 140165992.0, + "step": 5411 + }, + { + "epoch": 0.5943334065451351, + "grad_norm": 2.3432137966156006, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6898775100708008, + "num_tokens": 140190420.0, + "step": 5412 + }, + { + "epoch": 0.5944432242477488, + "grad_norm": 2.0590524673461914, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.713468074798584, + "num_tokens": 140218466.0, + "step": 5413 + }, + { + "epoch": 0.5945530419503624, + "grad_norm": 2.251711130142212, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.721280574798584, + "num_tokens": 140241243.0, + "step": 5414 + }, + { + "epoch": 0.594662859652976, + "grad_norm": 2.1204426288604736, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7100082039833069, + "num_tokens": 140269198.0, + "step": 5415 + }, + { + "epoch": 0.5947726773555897, + "grad_norm": 2.235691785812378, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7276394367218018, + "num_tokens": 140291919.0, + "step": 5416 + }, + { + "epoch": 0.5948824950582033, + "grad_norm": 2.5305964946746826, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.718304455280304, + "num_tokens": 140314477.0, + "step": 5417 + }, + { + "epoch": 0.5949923127608171, + "grad_norm": 2.3852202892303467, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7082000374794006, + "num_tokens": 140337521.0, + "step": 5418 + }, + { + "epoch": 0.5951021304634307, + "grad_norm": 2.284090042114258, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.7001948952674866, + "num_tokens": 140360715.0, + "step": 5419 + }, + { + "epoch": 0.5952119481660444, + "grad_norm": 2.2943148612976074, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7059929370880127, + "num_tokens": 140384016.0, + "step": 5420 + }, + { + "epoch": 0.595321765868658, + "grad_norm": 2.265514612197876, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7271000146865845, + "num_tokens": 140408011.0, + "step": 5421 + }, + { + "epoch": 0.5954315835712717, + "grad_norm": 1.8808729648590088, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7036412358283997, + "num_tokens": 140439623.0, + "step": 5422 + }, + { + "epoch": 0.5955414012738853, + "grad_norm": 2.236149787902832, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6875039339065552, + "num_tokens": 140463762.0, + "step": 5423 + }, + { + "epoch": 0.595651218976499, + "grad_norm": 2.165356159210205, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6869767904281616, + "num_tokens": 140489649.0, + "step": 5424 + }, + { + "epoch": 0.5957610366791126, + "grad_norm": 2.44364857673645, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7071332335472107, + "num_tokens": 140509941.0, + "step": 5425 + }, + { + "epoch": 0.5958708543817264, + "grad_norm": 2.2195026874542236, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6967626810073853, + "num_tokens": 140535871.0, + "step": 5426 + }, + { + "epoch": 0.59598067208434, + "grad_norm": 2.158005475997925, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7093916535377502, + "num_tokens": 140562930.0, + "step": 5427 + }, + { + "epoch": 0.5960904897869537, + "grad_norm": 2.028334379196167, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7137638330459595, + "num_tokens": 140590643.0, + "step": 5428 + }, + { + "epoch": 0.5962003074895673, + "grad_norm": 2.288844585418701, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6935656666755676, + "num_tokens": 140615200.0, + "step": 5429 + }, + { + "epoch": 0.596310125192181, + "grad_norm": 1.9557822942733765, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7187215089797974, + "num_tokens": 140645253.0, + "step": 5430 + }, + { + "epoch": 0.5964199428947946, + "grad_norm": 2.22918963432312, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.722103476524353, + "num_tokens": 140667899.0, + "step": 5431 + }, + { + "epoch": 0.5965297605974083, + "grad_norm": 2.3257226943969727, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6919150352478027, + "num_tokens": 140691109.0, + "step": 5432 + }, + { + "epoch": 0.596639578300022, + "grad_norm": 2.1368539333343506, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6954419612884521, + "num_tokens": 140717798.0, + "step": 5433 + }, + { + "epoch": 0.5967493960026357, + "grad_norm": 1.9418845176696777, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6920603513717651, + "num_tokens": 140749119.0, + "step": 5434 + }, + { + "epoch": 0.5968592137052493, + "grad_norm": 2.167327880859375, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7133535146713257, + "num_tokens": 140774407.0, + "step": 5435 + }, + { + "epoch": 0.596969031407863, + "grad_norm": 2.193946361541748, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6977212429046631, + "num_tokens": 140800384.0, + "step": 5436 + }, + { + "epoch": 0.5970788491104766, + "grad_norm": 2.2804057598114014, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7182118892669678, + "num_tokens": 140823824.0, + "step": 5437 + }, + { + "epoch": 0.5971886668130902, + "grad_norm": 2.3224995136260986, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7033776044845581, + "num_tokens": 140845377.0, + "step": 5438 + }, + { + "epoch": 0.5972984845157039, + "grad_norm": 2.1025807857513428, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7159590721130371, + "num_tokens": 140869853.0, + "step": 5439 + }, + { + "epoch": 0.5974083022183176, + "grad_norm": 2.243952989578247, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7216756343841553, + "num_tokens": 140893376.0, + "step": 5440 + }, + { + "epoch": 0.5975181199209313, + "grad_norm": 1.894638180732727, + "learning_rate": 1e-06, + "loss": 1.0703, + "mean_token_accuracy": 0.6735633611679077, + "num_tokens": 140930049.0, + "step": 5441 + }, + { + "epoch": 0.5976279376235449, + "grad_norm": 1.9943263530731201, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.679206371307373, + "num_tokens": 140960774.0, + "step": 5442 + }, + { + "epoch": 0.5977377553261586, + "grad_norm": 1.8609404563903809, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6976557970046997, + "num_tokens": 140991688.0, + "step": 5443 + }, + { + "epoch": 0.5978475730287722, + "grad_norm": 2.1362252235412598, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7124304175376892, + "num_tokens": 141018534.0, + "step": 5444 + }, + { + "epoch": 0.5979573907313859, + "grad_norm": 2.4838926792144775, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.6947304010391235, + "num_tokens": 141040847.0, + "step": 5445 + }, + { + "epoch": 0.5980672084339995, + "grad_norm": 2.1354498863220215, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.69379723072052, + "num_tokens": 141068941.0, + "step": 5446 + }, + { + "epoch": 0.5981770261366133, + "grad_norm": 2.5084073543548584, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7291210293769836, + "num_tokens": 141088818.0, + "step": 5447 + }, + { + "epoch": 0.5982868438392269, + "grad_norm": 2.1076560020446777, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7411031723022461, + "num_tokens": 141111050.0, + "step": 5448 + }, + { + "epoch": 0.5983966615418406, + "grad_norm": 2.1691911220550537, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7241030931472778, + "num_tokens": 141134434.0, + "step": 5449 + }, + { + "epoch": 0.5985064792444542, + "grad_norm": 2.3740317821502686, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7168430685997009, + "num_tokens": 141156074.0, + "step": 5450 + }, + { + "epoch": 0.5986162969470679, + "grad_norm": 2.103592872619629, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6991490125656128, + "num_tokens": 141184073.0, + "step": 5451 + }, + { + "epoch": 0.5987261146496815, + "grad_norm": 2.238145589828491, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7111735939979553, + "num_tokens": 141207261.0, + "step": 5452 + }, + { + "epoch": 0.5988359323522952, + "grad_norm": 2.4718945026397705, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.718854546546936, + "num_tokens": 141227943.0, + "step": 5453 + }, + { + "epoch": 0.5989457500549088, + "grad_norm": 1.938764214515686, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6984052062034607, + "num_tokens": 141259938.0, + "step": 5454 + }, + { + "epoch": 0.5990555677575226, + "grad_norm": 2.406817674636841, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7046123147010803, + "num_tokens": 141281372.0, + "step": 5455 + }, + { + "epoch": 0.5991653854601362, + "grad_norm": 2.400334358215332, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7205692529678345, + "num_tokens": 141301792.0, + "step": 5456 + }, + { + "epoch": 0.5992752031627498, + "grad_norm": 2.138237953186035, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7022035121917725, + "num_tokens": 141330241.0, + "step": 5457 + }, + { + "epoch": 0.5993850208653635, + "grad_norm": 2.1050586700439453, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6937634348869324, + "num_tokens": 141360496.0, + "step": 5458 + }, + { + "epoch": 0.5994948385679771, + "grad_norm": 2.1584486961364746, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6904689073562622, + "num_tokens": 141386431.0, + "step": 5459 + }, + { + "epoch": 0.5996046562705908, + "grad_norm": 2.098926067352295, + "learning_rate": 1e-06, + "loss": 1.101, + "mean_token_accuracy": 0.667749285697937, + "num_tokens": 141415944.0, + "step": 5460 + }, + { + "epoch": 0.5997144739732044, + "grad_norm": 2.073596239089966, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6998327970504761, + "num_tokens": 141444005.0, + "step": 5461 + }, + { + "epoch": 0.5998242916758182, + "grad_norm": 2.156856060028076, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.69106525182724, + "num_tokens": 141470558.0, + "step": 5462 + }, + { + "epoch": 0.5999341093784318, + "grad_norm": 2.068336009979248, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6886833906173706, + "num_tokens": 141499619.0, + "step": 5463 + }, + { + "epoch": 0.6000439270810455, + "grad_norm": 2.4224658012390137, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.72148597240448, + "num_tokens": 141519971.0, + "step": 5464 + }, + { + "epoch": 0.6001537447836591, + "grad_norm": 2.114933967590332, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7229450941085815, + "num_tokens": 141547413.0, + "step": 5465 + }, + { + "epoch": 0.6002635624862728, + "grad_norm": 1.9068132638931274, + "learning_rate": 1e-06, + "loss": 1.067, + "mean_token_accuracy": 0.6784652471542358, + "num_tokens": 141582883.0, + "step": 5466 + }, + { + "epoch": 0.6003733801888864, + "grad_norm": 2.1897082328796387, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7046009302139282, + "num_tokens": 141609111.0, + "step": 5467 + }, + { + "epoch": 0.6004831978915001, + "grad_norm": 2.139437437057495, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7229098081588745, + "num_tokens": 141632586.0, + "step": 5468 + }, + { + "epoch": 0.6005930155941138, + "grad_norm": 2.457831621170044, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7139695882797241, + "num_tokens": 141653030.0, + "step": 5469 + }, + { + "epoch": 0.6007028332967275, + "grad_norm": 2.0526020526885986, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6929317712783813, + "num_tokens": 141680762.0, + "step": 5470 + }, + { + "epoch": 0.6008126509993411, + "grad_norm": 2.336203098297119, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6898940801620483, + "num_tokens": 141705139.0, + "step": 5471 + }, + { + "epoch": 0.6009224687019548, + "grad_norm": 2.189739942550659, + "learning_rate": 1e-06, + "loss": 1.0637, + "mean_token_accuracy": 0.6803421974182129, + "num_tokens": 141732248.0, + "step": 5472 + }, + { + "epoch": 0.6010322864045684, + "grad_norm": 2.373094081878662, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6872372627258301, + "num_tokens": 141756347.0, + "step": 5473 + }, + { + "epoch": 0.601142104107182, + "grad_norm": 1.8565373420715332, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6895425319671631, + "num_tokens": 141790860.0, + "step": 5474 + }, + { + "epoch": 0.6012519218097957, + "grad_norm": 2.1079368591308594, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7045887112617493, + "num_tokens": 141818768.0, + "step": 5475 + }, + { + "epoch": 0.6013617395124095, + "grad_norm": 2.0063695907592773, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7178510427474976, + "num_tokens": 141846167.0, + "step": 5476 + }, + { + "epoch": 0.6014715572150231, + "grad_norm": 2.4407553672790527, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.707926869392395, + "num_tokens": 141867109.0, + "step": 5477 + }, + { + "epoch": 0.6015813749176367, + "grad_norm": 2.1285808086395264, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6892446279525757, + "num_tokens": 141894244.0, + "step": 5478 + }, + { + "epoch": 0.6016911926202504, + "grad_norm": 2.2753472328186035, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7137017250061035, + "num_tokens": 141918529.0, + "step": 5479 + }, + { + "epoch": 0.601801010322864, + "grad_norm": 2.230879545211792, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7122543454170227, + "num_tokens": 141941287.0, + "step": 5480 + }, + { + "epoch": 0.6019108280254777, + "grad_norm": 2.3312575817108154, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6852315068244934, + "num_tokens": 141965353.0, + "step": 5481 + }, + { + "epoch": 0.6020206457280913, + "grad_norm": 2.2270429134368896, + "learning_rate": 1e-06, + "loss": 1.1, + "mean_token_accuracy": 0.674549400806427, + "num_tokens": 141992824.0, + "step": 5482 + }, + { + "epoch": 0.602130463430705, + "grad_norm": 2.34563946723938, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.69089674949646, + "num_tokens": 142016379.0, + "step": 5483 + }, + { + "epoch": 0.6022402811333187, + "grad_norm": 1.912555456161499, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6822036504745483, + "num_tokens": 142051236.0, + "step": 5484 + }, + { + "epoch": 0.6023500988359324, + "grad_norm": 2.2538132667541504, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6931911706924438, + "num_tokens": 142076418.0, + "step": 5485 + }, + { + "epoch": 0.602459916538546, + "grad_norm": 2.557030439376831, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7203962206840515, + "num_tokens": 142096091.0, + "step": 5486 + }, + { + "epoch": 0.6025697342411597, + "grad_norm": 2.0774002075195312, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7064661383628845, + "num_tokens": 142124345.0, + "step": 5487 + }, + { + "epoch": 0.6026795519437733, + "grad_norm": 2.193725347518921, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6888392567634583, + "num_tokens": 142154203.0, + "step": 5488 + }, + { + "epoch": 0.602789369646387, + "grad_norm": 2.2248973846435547, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.6990292072296143, + "num_tokens": 142178440.0, + "step": 5489 + }, + { + "epoch": 0.6028991873490006, + "grad_norm": 2.4278717041015625, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7079253196716309, + "num_tokens": 142203260.0, + "step": 5490 + }, + { + "epoch": 0.6030090050516144, + "grad_norm": 2.6059443950653076, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7129102945327759, + "num_tokens": 142222642.0, + "step": 5491 + }, + { + "epoch": 0.603118822754228, + "grad_norm": 2.137594699859619, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7120063304901123, + "num_tokens": 142248504.0, + "step": 5492 + }, + { + "epoch": 0.6032286404568417, + "grad_norm": 2.1937811374664307, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7078679800033569, + "num_tokens": 142275888.0, + "step": 5493 + }, + { + "epoch": 0.6033384581594553, + "grad_norm": 2.225703239440918, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6962084770202637, + "num_tokens": 142302175.0, + "step": 5494 + }, + { + "epoch": 0.603448275862069, + "grad_norm": 2.3582546710968018, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7090774774551392, + "num_tokens": 142326863.0, + "step": 5495 + }, + { + "epoch": 0.6035580935646826, + "grad_norm": 2.032132148742676, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7049691081047058, + "num_tokens": 142355356.0, + "step": 5496 + }, + { + "epoch": 0.6036679112672962, + "grad_norm": 2.2833034992218018, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7074638605117798, + "num_tokens": 142379442.0, + "step": 5497 + }, + { + "epoch": 0.60377772896991, + "grad_norm": 2.3111953735351562, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7034686207771301, + "num_tokens": 142402486.0, + "step": 5498 + }, + { + "epoch": 0.6038875466725236, + "grad_norm": 2.23114013671875, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6933033466339111, + "num_tokens": 142427756.0, + "step": 5499 + }, + { + "epoch": 0.6039973643751373, + "grad_norm": 2.0962510108947754, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7050055265426636, + "num_tokens": 142456504.0, + "step": 5500 + }, + { + "epoch": 0.6041071820777509, + "grad_norm": 2.0678107738494873, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7181878089904785, + "num_tokens": 142481999.0, + "step": 5501 + }, + { + "epoch": 0.6042169997803646, + "grad_norm": 2.107149839401245, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7150954008102417, + "num_tokens": 142508085.0, + "step": 5502 + }, + { + "epoch": 0.6043268174829782, + "grad_norm": 2.330204963684082, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6830639243125916, + "num_tokens": 142533502.0, + "step": 5503 + }, + { + "epoch": 0.6044366351855919, + "grad_norm": 2.116013288497925, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7083165645599365, + "num_tokens": 142562232.0, + "step": 5504 + }, + { + "epoch": 0.6045464528882056, + "grad_norm": 2.282862901687622, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7196090221405029, + "num_tokens": 142584535.0, + "step": 5505 + }, + { + "epoch": 0.6046562705908193, + "grad_norm": 2.0581541061401367, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6975421905517578, + "num_tokens": 142613548.0, + "step": 5506 + }, + { + "epoch": 0.6047660882934329, + "grad_norm": 2.400233507156372, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7035965919494629, + "num_tokens": 142634501.0, + "step": 5507 + }, + { + "epoch": 0.6048759059960466, + "grad_norm": 2.139889717102051, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7119966745376587, + "num_tokens": 142660053.0, + "step": 5508 + }, + { + "epoch": 0.6049857236986602, + "grad_norm": 2.075511932373047, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.701697826385498, + "num_tokens": 142686587.0, + "step": 5509 + }, + { + "epoch": 0.6050955414012739, + "grad_norm": 2.2298636436462402, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6993080377578735, + "num_tokens": 142710842.0, + "step": 5510 + }, + { + "epoch": 0.6052053591038875, + "grad_norm": 1.9127002954483032, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6990878582000732, + "num_tokens": 142744140.0, + "step": 5511 + }, + { + "epoch": 0.6053151768065012, + "grad_norm": 2.1852364540100098, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7059672474861145, + "num_tokens": 142769131.0, + "step": 5512 + }, + { + "epoch": 0.6054249945091149, + "grad_norm": 2.1510939598083496, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7101848125457764, + "num_tokens": 142798422.0, + "step": 5513 + }, + { + "epoch": 0.6055348122117286, + "grad_norm": 2.3427681922912598, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7126946449279785, + "num_tokens": 142820626.0, + "step": 5514 + }, + { + "epoch": 0.6056446299143422, + "grad_norm": 2.2067196369171143, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7147190570831299, + "num_tokens": 142842981.0, + "step": 5515 + }, + { + "epoch": 0.6057544476169558, + "grad_norm": 2.495626211166382, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7096554040908813, + "num_tokens": 142862928.0, + "step": 5516 + }, + { + "epoch": 0.6058642653195695, + "grad_norm": 2.1391401290893555, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.694233775138855, + "num_tokens": 142891872.0, + "step": 5517 + }, + { + "epoch": 0.6059740830221831, + "grad_norm": 2.0528550148010254, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7005118131637573, + "num_tokens": 142919028.0, + "step": 5518 + }, + { + "epoch": 0.6060839007247968, + "grad_norm": 1.9620641469955444, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7164263725280762, + "num_tokens": 142947374.0, + "step": 5519 + }, + { + "epoch": 0.6061937184274105, + "grad_norm": 2.1932480335235596, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6878708600997925, + "num_tokens": 142974095.0, + "step": 5520 + }, + { + "epoch": 0.6063035361300242, + "grad_norm": 2.4743175506591797, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6887128353118896, + "num_tokens": 142995967.0, + "step": 5521 + }, + { + "epoch": 0.6064133538326378, + "grad_norm": 2.0849828720092773, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.6785334348678589, + "num_tokens": 143025221.0, + "step": 5522 + }, + { + "epoch": 0.6065231715352515, + "grad_norm": 2.101750135421753, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7044557332992554, + "num_tokens": 143051527.0, + "step": 5523 + }, + { + "epoch": 0.6066329892378651, + "grad_norm": 2.364797353744507, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6973389983177185, + "num_tokens": 143075544.0, + "step": 5524 + }, + { + "epoch": 0.6067428069404788, + "grad_norm": 2.1531739234924316, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7210615873336792, + "num_tokens": 143099855.0, + "step": 5525 + }, + { + "epoch": 0.6068526246430924, + "grad_norm": 1.9762215614318848, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7138372659683228, + "num_tokens": 143129817.0, + "step": 5526 + }, + { + "epoch": 0.6069624423457062, + "grad_norm": 2.466001033782959, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.697861909866333, + "num_tokens": 143150698.0, + "step": 5527 + }, + { + "epoch": 0.6070722600483198, + "grad_norm": 2.4577484130859375, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.705851674079895, + "num_tokens": 143171043.0, + "step": 5528 + }, + { + "epoch": 0.6071820777509335, + "grad_norm": 2.3014538288116455, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7144280672073364, + "num_tokens": 143193800.0, + "step": 5529 + }, + { + "epoch": 0.6072918954535471, + "grad_norm": 2.1089413166046143, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6976382732391357, + "num_tokens": 143222251.0, + "step": 5530 + }, + { + "epoch": 0.6074017131561608, + "grad_norm": 2.398667573928833, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7102243900299072, + "num_tokens": 143245623.0, + "step": 5531 + }, + { + "epoch": 0.6075115308587744, + "grad_norm": 2.188044548034668, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7198851108551025, + "num_tokens": 143271328.0, + "step": 5532 + }, + { + "epoch": 0.607621348561388, + "grad_norm": 2.5389626026153564, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7184287905693054, + "num_tokens": 143290653.0, + "step": 5533 + }, + { + "epoch": 0.6077311662640018, + "grad_norm": 2.1239352226257324, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7236453294754028, + "num_tokens": 143316964.0, + "step": 5534 + }, + { + "epoch": 0.6078409839666155, + "grad_norm": 2.5335021018981934, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.723866879940033, + "num_tokens": 143339087.0, + "step": 5535 + }, + { + "epoch": 0.6079508016692291, + "grad_norm": 1.8863211870193481, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7003909945487976, + "num_tokens": 143372999.0, + "step": 5536 + }, + { + "epoch": 0.6080606193718427, + "grad_norm": 2.229418992996216, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7047126293182373, + "num_tokens": 143398034.0, + "step": 5537 + }, + { + "epoch": 0.6081704370744564, + "grad_norm": 2.342162847518921, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7198002338409424, + "num_tokens": 143419511.0, + "step": 5538 + }, + { + "epoch": 0.60828025477707, + "grad_norm": 2.05545973777771, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6900640726089478, + "num_tokens": 143447620.0, + "step": 5539 + }, + { + "epoch": 0.6083900724796837, + "grad_norm": 2.3722963333129883, + "learning_rate": 1e-06, + "loss": 1.0423, + "mean_token_accuracy": 0.6951721906661987, + "num_tokens": 143470927.0, + "step": 5540 + }, + { + "epoch": 0.6084998901822973, + "grad_norm": 2.1802480220794678, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.703900933265686, + "num_tokens": 143495362.0, + "step": 5541 + }, + { + "epoch": 0.6086097078849111, + "grad_norm": 2.121394157409668, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7142316102981567, + "num_tokens": 143522298.0, + "step": 5542 + }, + { + "epoch": 0.6087195255875247, + "grad_norm": 2.6238338947296143, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7198550701141357, + "num_tokens": 143540762.0, + "step": 5543 + }, + { + "epoch": 0.6088293432901384, + "grad_norm": 2.4084842205047607, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7293328046798706, + "num_tokens": 143562935.0, + "step": 5544 + }, + { + "epoch": 0.608939160992752, + "grad_norm": 2.369811773300171, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6978217363357544, + "num_tokens": 143586717.0, + "step": 5545 + }, + { + "epoch": 0.6090489786953657, + "grad_norm": 2.128328323364258, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6905848383903503, + "num_tokens": 143613936.0, + "step": 5546 + }, + { + "epoch": 0.6091587963979793, + "grad_norm": 2.0998265743255615, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7093971967697144, + "num_tokens": 143641181.0, + "step": 5547 + }, + { + "epoch": 0.609268614100593, + "grad_norm": 2.158226251602173, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7044333815574646, + "num_tokens": 143666141.0, + "step": 5548 + }, + { + "epoch": 0.6093784318032067, + "grad_norm": 2.0918500423431396, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.6998271942138672, + "num_tokens": 143694096.0, + "step": 5549 + }, + { + "epoch": 0.6094882495058204, + "grad_norm": 2.018695592880249, + "learning_rate": 1e-06, + "loss": 1.0322, + "mean_token_accuracy": 0.6972661018371582, + "num_tokens": 143724483.0, + "step": 5550 + }, + { + "epoch": 0.609598067208434, + "grad_norm": 2.17217755317688, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7107325196266174, + "num_tokens": 143748850.0, + "step": 5551 + }, + { + "epoch": 0.6097078849110477, + "grad_norm": 2.108339786529541, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6984375715255737, + "num_tokens": 143776175.0, + "step": 5552 + }, + { + "epoch": 0.6098177026136613, + "grad_norm": 2.188810110092163, + "learning_rate": 1e-06, + "loss": 1.0637, + "mean_token_accuracy": 0.6886357069015503, + "num_tokens": 143804042.0, + "step": 5553 + }, + { + "epoch": 0.609927520316275, + "grad_norm": 2.1610634326934814, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7064220905303955, + "num_tokens": 143831258.0, + "step": 5554 + }, + { + "epoch": 0.6100373380188886, + "grad_norm": 1.9899073839187622, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6932946443557739, + "num_tokens": 143861686.0, + "step": 5555 + }, + { + "epoch": 0.6101471557215024, + "grad_norm": 2.416238784790039, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7001742124557495, + "num_tokens": 143883454.0, + "step": 5556 + }, + { + "epoch": 0.610256973424116, + "grad_norm": 2.2283384799957275, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.70649653673172, + "num_tokens": 143907384.0, + "step": 5557 + }, + { + "epoch": 0.6103667911267296, + "grad_norm": 2.1821060180664062, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6945653557777405, + "num_tokens": 143934663.0, + "step": 5558 + }, + { + "epoch": 0.6104766088293433, + "grad_norm": 2.06097674369812, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7111476063728333, + "num_tokens": 143962719.0, + "step": 5559 + }, + { + "epoch": 0.6105864265319569, + "grad_norm": 2.512378692626953, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7125368118286133, + "num_tokens": 143983202.0, + "step": 5560 + }, + { + "epoch": 0.6106962442345706, + "grad_norm": 2.435763120651245, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7135807871818542, + "num_tokens": 144004714.0, + "step": 5561 + }, + { + "epoch": 0.6108060619371842, + "grad_norm": 2.5403456687927246, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6981058120727539, + "num_tokens": 144025861.0, + "step": 5562 + }, + { + "epoch": 0.610915879639798, + "grad_norm": 2.328735589981079, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7143424153327942, + "num_tokens": 144048018.0, + "step": 5563 + }, + { + "epoch": 0.6110256973424116, + "grad_norm": 2.104738473892212, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7257353067398071, + "num_tokens": 144073098.0, + "step": 5564 + }, + { + "epoch": 0.6111355150450253, + "grad_norm": 1.9928175210952759, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6910380125045776, + "num_tokens": 144103692.0, + "step": 5565 + }, + { + "epoch": 0.6112453327476389, + "grad_norm": 2.177032232284546, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6987553834915161, + "num_tokens": 144130071.0, + "step": 5566 + }, + { + "epoch": 0.6113551504502526, + "grad_norm": 2.0173144340515137, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7134656310081482, + "num_tokens": 144160736.0, + "step": 5567 + }, + { + "epoch": 0.6114649681528662, + "grad_norm": 2.1442785263061523, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7157796621322632, + "num_tokens": 144187160.0, + "step": 5568 + }, + { + "epoch": 0.6115747858554799, + "grad_norm": 2.175513982772827, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6924634575843811, + "num_tokens": 144215373.0, + "step": 5569 + }, + { + "epoch": 0.6116846035580935, + "grad_norm": 1.9283720254898071, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7233695983886719, + "num_tokens": 144246010.0, + "step": 5570 + }, + { + "epoch": 0.6117944212607073, + "grad_norm": 2.277313232421875, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6895620822906494, + "num_tokens": 144271534.0, + "step": 5571 + }, + { + "epoch": 0.6119042389633209, + "grad_norm": 2.0689477920532227, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6979559659957886, + "num_tokens": 144296962.0, + "step": 5572 + }, + { + "epoch": 0.6120140566659346, + "grad_norm": 2.3841054439544678, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7174094915390015, + "num_tokens": 144319660.0, + "step": 5573 + }, + { + "epoch": 0.6121238743685482, + "grad_norm": 2.045079231262207, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7064068913459778, + "num_tokens": 144345996.0, + "step": 5574 + }, + { + "epoch": 0.6122336920711619, + "grad_norm": 2.188814163208008, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.718617856502533, + "num_tokens": 144370199.0, + "step": 5575 + }, + { + "epoch": 0.6123435097737755, + "grad_norm": 2.145601749420166, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6824643015861511, + "num_tokens": 144398358.0, + "step": 5576 + }, + { + "epoch": 0.6124533274763891, + "grad_norm": 2.286926507949829, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7108955383300781, + "num_tokens": 144422666.0, + "step": 5577 + }, + { + "epoch": 0.6125631451790029, + "grad_norm": 1.9975285530090332, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6907420754432678, + "num_tokens": 144453770.0, + "step": 5578 + }, + { + "epoch": 0.6126729628816165, + "grad_norm": 1.9211113452911377, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.698230504989624, + "num_tokens": 144487826.0, + "step": 5579 + }, + { + "epoch": 0.6127827805842302, + "grad_norm": 2.3604931831359863, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7208913564682007, + "num_tokens": 144510583.0, + "step": 5580 + }, + { + "epoch": 0.6128925982868438, + "grad_norm": 2.143433094024658, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7069859504699707, + "num_tokens": 144537090.0, + "step": 5581 + }, + { + "epoch": 0.6130024159894575, + "grad_norm": 1.81163489818573, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7073168754577637, + "num_tokens": 144573721.0, + "step": 5582 + }, + { + "epoch": 0.6131122336920711, + "grad_norm": 2.104207992553711, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.7016758322715759, + "num_tokens": 144600228.0, + "step": 5583 + }, + { + "epoch": 0.6132220513946848, + "grad_norm": 2.1847095489501953, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6989122629165649, + "num_tokens": 144624825.0, + "step": 5584 + }, + { + "epoch": 0.6133318690972985, + "grad_norm": 2.0441858768463135, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7071560621261597, + "num_tokens": 144653870.0, + "step": 5585 + }, + { + "epoch": 0.6134416867999122, + "grad_norm": 2.027632474899292, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7009754180908203, + "num_tokens": 144682126.0, + "step": 5586 + }, + { + "epoch": 0.6135515045025258, + "grad_norm": 2.083425998687744, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.6770721077919006, + "num_tokens": 144711108.0, + "step": 5587 + }, + { + "epoch": 0.6136613222051395, + "grad_norm": 2.3145389556884766, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7371256351470947, + "num_tokens": 144732059.0, + "step": 5588 + }, + { + "epoch": 0.6137711399077531, + "grad_norm": 2.227402925491333, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7198445200920105, + "num_tokens": 144756402.0, + "step": 5589 + }, + { + "epoch": 0.6138809576103668, + "grad_norm": 2.468892812728882, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.699223518371582, + "num_tokens": 144778478.0, + "step": 5590 + }, + { + "epoch": 0.6139907753129804, + "grad_norm": 2.156608819961548, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6971807479858398, + "num_tokens": 144803171.0, + "step": 5591 + }, + { + "epoch": 0.6141005930155942, + "grad_norm": 2.2212936878204346, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.695704996585846, + "num_tokens": 144829122.0, + "step": 5592 + }, + { + "epoch": 0.6142104107182078, + "grad_norm": 2.286728858947754, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.704746425151825, + "num_tokens": 144853840.0, + "step": 5593 + }, + { + "epoch": 0.6143202284208215, + "grad_norm": 2.1379010677337646, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.728564977645874, + "num_tokens": 144879799.0, + "step": 5594 + }, + { + "epoch": 0.6144300461234351, + "grad_norm": 2.1367833614349365, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7309216856956482, + "num_tokens": 144903876.0, + "step": 5595 + }, + { + "epoch": 0.6145398638260487, + "grad_norm": 2.1161985397338867, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6896553039550781, + "num_tokens": 144932095.0, + "step": 5596 + }, + { + "epoch": 0.6146496815286624, + "grad_norm": 2.281531810760498, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7157007455825806, + "num_tokens": 144955621.0, + "step": 5597 + }, + { + "epoch": 0.614759499231276, + "grad_norm": 2.080249786376953, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.702806830406189, + "num_tokens": 144985577.0, + "step": 5598 + }, + { + "epoch": 0.6148693169338898, + "grad_norm": 2.4225621223449707, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.730448305606842, + "num_tokens": 145005958.0, + "step": 5599 + }, + { + "epoch": 0.6149791346365034, + "grad_norm": 2.465019464492798, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7172061800956726, + "num_tokens": 145025793.0, + "step": 5600 + }, + { + "epoch": 0.6150889523391171, + "grad_norm": 2.011383533477783, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.6997356414794922, + "num_tokens": 145056561.0, + "step": 5601 + }, + { + "epoch": 0.6151987700417307, + "grad_norm": 2.039313316345215, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7206096649169922, + "num_tokens": 145084877.0, + "step": 5602 + }, + { + "epoch": 0.6153085877443444, + "grad_norm": 2.2723119258880615, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7279918193817139, + "num_tokens": 145108660.0, + "step": 5603 + }, + { + "epoch": 0.615418405446958, + "grad_norm": 2.2674412727355957, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7079096436500549, + "num_tokens": 145132828.0, + "step": 5604 + }, + { + "epoch": 0.6155282231495717, + "grad_norm": 2.2888917922973633, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7017834186553955, + "num_tokens": 145154962.0, + "step": 5605 + }, + { + "epoch": 0.6156380408521853, + "grad_norm": 2.3854143619537354, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6831321120262146, + "num_tokens": 145182406.0, + "step": 5606 + }, + { + "epoch": 0.6157478585547991, + "grad_norm": 2.1601619720458984, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6872061491012573, + "num_tokens": 145208430.0, + "step": 5607 + }, + { + "epoch": 0.6158576762574127, + "grad_norm": 2.2491679191589355, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7101610898971558, + "num_tokens": 145236623.0, + "step": 5608 + }, + { + "epoch": 0.6159674939600264, + "grad_norm": 2.1933791637420654, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7080532312393188, + "num_tokens": 145261618.0, + "step": 5609 + }, + { + "epoch": 0.61607731166264, + "grad_norm": 2.135885000228882, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6966392993927002, + "num_tokens": 145291071.0, + "step": 5610 + }, + { + "epoch": 0.6161871293652537, + "grad_norm": 1.8524725437164307, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7354111671447754, + "num_tokens": 145323857.0, + "step": 5611 + }, + { + "epoch": 0.6162969470678673, + "grad_norm": 2.121828556060791, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6913472414016724, + "num_tokens": 145353905.0, + "step": 5612 + }, + { + "epoch": 0.616406764770481, + "grad_norm": 1.9504296779632568, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7217036485671997, + "num_tokens": 145384056.0, + "step": 5613 + }, + { + "epoch": 0.6165165824730947, + "grad_norm": 2.1373953819274902, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7282848358154297, + "num_tokens": 145410449.0, + "step": 5614 + }, + { + "epoch": 0.6166264001757084, + "grad_norm": 2.2690958976745605, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7267812490463257, + "num_tokens": 145433886.0, + "step": 5615 + }, + { + "epoch": 0.616736217878322, + "grad_norm": 2.4415245056152344, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6886080503463745, + "num_tokens": 145455812.0, + "step": 5616 + }, + { + "epoch": 0.6168460355809356, + "grad_norm": 2.1062633991241455, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6884695291519165, + "num_tokens": 145483198.0, + "step": 5617 + }, + { + "epoch": 0.6169558532835493, + "grad_norm": 2.2080795764923096, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7079482674598694, + "num_tokens": 145508062.0, + "step": 5618 + }, + { + "epoch": 0.6170656709861629, + "grad_norm": 2.1891517639160156, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6955373883247375, + "num_tokens": 145533803.0, + "step": 5619 + }, + { + "epoch": 0.6171754886887766, + "grad_norm": 2.0985774993896484, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6904934644699097, + "num_tokens": 145561163.0, + "step": 5620 + }, + { + "epoch": 0.6172853063913903, + "grad_norm": 2.071169137954712, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7055656909942627, + "num_tokens": 145587924.0, + "step": 5621 + }, + { + "epoch": 0.617395124094004, + "grad_norm": 2.075448751449585, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7132249474525452, + "num_tokens": 145614186.0, + "step": 5622 + }, + { + "epoch": 0.6175049417966176, + "grad_norm": 2.218721628189087, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7152934670448303, + "num_tokens": 145636444.0, + "step": 5623 + }, + { + "epoch": 0.6176147594992313, + "grad_norm": 2.039247751235962, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7032723426818848, + "num_tokens": 145665371.0, + "step": 5624 + }, + { + "epoch": 0.6177245772018449, + "grad_norm": 2.1848855018615723, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7035441994667053, + "num_tokens": 145689331.0, + "step": 5625 + }, + { + "epoch": 0.6178343949044586, + "grad_norm": 2.4492263793945312, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7013795375823975, + "num_tokens": 145709752.0, + "step": 5626 + }, + { + "epoch": 0.6179442126070722, + "grad_norm": 2.322176933288574, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7152035236358643, + "num_tokens": 145732554.0, + "step": 5627 + }, + { + "epoch": 0.618054030309686, + "grad_norm": 1.829425573348999, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6842403411865234, + "num_tokens": 145768213.0, + "step": 5628 + }, + { + "epoch": 0.6181638480122996, + "grad_norm": 2.4368433952331543, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7155883312225342, + "num_tokens": 145788714.0, + "step": 5629 + }, + { + "epoch": 0.6182736657149133, + "grad_norm": 2.3892292976379395, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.719773530960083, + "num_tokens": 145811217.0, + "step": 5630 + }, + { + "epoch": 0.6183834834175269, + "grad_norm": 2.2750885486602783, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6971187591552734, + "num_tokens": 145834633.0, + "step": 5631 + }, + { + "epoch": 0.6184933011201406, + "grad_norm": 2.457362174987793, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7060106992721558, + "num_tokens": 145855250.0, + "step": 5632 + }, + { + "epoch": 0.6186031188227542, + "grad_norm": 1.9310071468353271, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6944655776023865, + "num_tokens": 145888420.0, + "step": 5633 + }, + { + "epoch": 0.6187129365253679, + "grad_norm": 2.270449161529541, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7108709812164307, + "num_tokens": 145911814.0, + "step": 5634 + }, + { + "epoch": 0.6188227542279815, + "grad_norm": 2.1583199501037598, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.744134247303009, + "num_tokens": 145937059.0, + "step": 5635 + }, + { + "epoch": 0.6189325719305953, + "grad_norm": 2.189441204071045, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.7032957673072815, + "num_tokens": 145964732.0, + "step": 5636 + }, + { + "epoch": 0.6190423896332089, + "grad_norm": 2.188068151473999, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.700441837310791, + "num_tokens": 145990592.0, + "step": 5637 + }, + { + "epoch": 0.6191522073358225, + "grad_norm": 1.9753025770187378, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6925936341285706, + "num_tokens": 146020581.0, + "step": 5638 + }, + { + "epoch": 0.6192620250384362, + "grad_norm": 2.2469468116760254, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6927227973937988, + "num_tokens": 146045852.0, + "step": 5639 + }, + { + "epoch": 0.6193718427410498, + "grad_norm": 2.2206764221191406, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7441877126693726, + "num_tokens": 146069988.0, + "step": 5640 + }, + { + "epoch": 0.6194816604436635, + "grad_norm": 2.49461030960083, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.706777811050415, + "num_tokens": 146090334.0, + "step": 5641 + }, + { + "epoch": 0.6195914781462771, + "grad_norm": 2.2155368328094482, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7122312188148499, + "num_tokens": 146113724.0, + "step": 5642 + }, + { + "epoch": 0.6197012958488909, + "grad_norm": 2.213653326034546, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7136307954788208, + "num_tokens": 146138973.0, + "step": 5643 + }, + { + "epoch": 0.6198111135515045, + "grad_norm": 1.9139987230300903, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7314763069152832, + "num_tokens": 146168630.0, + "step": 5644 + }, + { + "epoch": 0.6199209312541182, + "grad_norm": 1.9970474243164062, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6906282901763916, + "num_tokens": 146198815.0, + "step": 5645 + }, + { + "epoch": 0.6200307489567318, + "grad_norm": 2.483299732208252, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7256057262420654, + "num_tokens": 146219566.0, + "step": 5646 + }, + { + "epoch": 0.6201405666593455, + "grad_norm": 2.287019729614258, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.6826378107070923, + "num_tokens": 146243718.0, + "step": 5647 + }, + { + "epoch": 0.6202503843619591, + "grad_norm": 2.2117857933044434, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6964523792266846, + "num_tokens": 146268636.0, + "step": 5648 + }, + { + "epoch": 0.6203602020645728, + "grad_norm": 2.301342010498047, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7154253721237183, + "num_tokens": 146291335.0, + "step": 5649 + }, + { + "epoch": 0.6204700197671865, + "grad_norm": 2.5877745151519775, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7170390486717224, + "num_tokens": 146311255.0, + "step": 5650 + }, + { + "epoch": 0.6205798374698002, + "grad_norm": 2.2412219047546387, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6992377638816833, + "num_tokens": 146335164.0, + "step": 5651 + }, + { + "epoch": 0.6206896551724138, + "grad_norm": 2.5560109615325928, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.727946937084198, + "num_tokens": 146354287.0, + "step": 5652 + }, + { + "epoch": 0.6207994728750275, + "grad_norm": 2.1619160175323486, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7211101651191711, + "num_tokens": 146380335.0, + "step": 5653 + }, + { + "epoch": 0.6209092905776411, + "grad_norm": 2.3665518760681152, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6990599632263184, + "num_tokens": 146401453.0, + "step": 5654 + }, + { + "epoch": 0.6210191082802548, + "grad_norm": 2.0040152072906494, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6994339227676392, + "num_tokens": 146431847.0, + "step": 5655 + }, + { + "epoch": 0.6211289259828684, + "grad_norm": 2.262040853500366, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.7038977742195129, + "num_tokens": 146456537.0, + "step": 5656 + }, + { + "epoch": 0.6212387436854822, + "grad_norm": 2.2503445148468018, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7069483399391174, + "num_tokens": 146479809.0, + "step": 5657 + }, + { + "epoch": 0.6213485613880958, + "grad_norm": 2.108177900314331, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6961756348609924, + "num_tokens": 146505965.0, + "step": 5658 + }, + { + "epoch": 0.6214583790907094, + "grad_norm": 2.222119092941284, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7101671099662781, + "num_tokens": 146531551.0, + "step": 5659 + }, + { + "epoch": 0.6215681967933231, + "grad_norm": 1.9278090000152588, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6811628341674805, + "num_tokens": 146563551.0, + "step": 5660 + }, + { + "epoch": 0.6216780144959367, + "grad_norm": 2.1382641792297363, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6931309700012207, + "num_tokens": 146591120.0, + "step": 5661 + }, + { + "epoch": 0.6217878321985504, + "grad_norm": 2.425905227661133, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7207435369491577, + "num_tokens": 146611965.0, + "step": 5662 + }, + { + "epoch": 0.621897649901164, + "grad_norm": 1.8077961206436157, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7051271796226501, + "num_tokens": 146648461.0, + "step": 5663 + }, + { + "epoch": 0.6220074676037777, + "grad_norm": 2.141225576400757, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6996364593505859, + "num_tokens": 146676427.0, + "step": 5664 + }, + { + "epoch": 0.6221172853063914, + "grad_norm": 2.1763148307800293, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7187975645065308, + "num_tokens": 146700165.0, + "step": 5665 + }, + { + "epoch": 0.6222271030090051, + "grad_norm": 2.1861891746520996, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7017307281494141, + "num_tokens": 146724996.0, + "step": 5666 + }, + { + "epoch": 0.6223369207116187, + "grad_norm": 2.264732837677002, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7120472192764282, + "num_tokens": 146750905.0, + "step": 5667 + }, + { + "epoch": 0.6224467384142324, + "grad_norm": 2.1534078121185303, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7053108215332031, + "num_tokens": 146776383.0, + "step": 5668 + }, + { + "epoch": 0.622556556116846, + "grad_norm": 2.1705007553100586, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7332043647766113, + "num_tokens": 146800241.0, + "step": 5669 + }, + { + "epoch": 0.6226663738194597, + "grad_norm": 1.9579083919525146, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7093775272369385, + "num_tokens": 146831318.0, + "step": 5670 + }, + { + "epoch": 0.6227761915220733, + "grad_norm": 2.3604798316955566, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7012282013893127, + "num_tokens": 146853834.0, + "step": 5671 + }, + { + "epoch": 0.6228860092246871, + "grad_norm": 2.5363759994506836, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7339908480644226, + "num_tokens": 146872499.0, + "step": 5672 + }, + { + "epoch": 0.6229958269273007, + "grad_norm": 2.2346723079681396, + "learning_rate": 1e-06, + "loss": 1.0651, + "mean_token_accuracy": 0.6825248599052429, + "num_tokens": 146897374.0, + "step": 5673 + }, + { + "epoch": 0.6231056446299144, + "grad_norm": 1.9782135486602783, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6945154666900635, + "num_tokens": 146927243.0, + "step": 5674 + }, + { + "epoch": 0.623215462332528, + "grad_norm": 2.277789831161499, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7161687612533569, + "num_tokens": 146949952.0, + "step": 5675 + }, + { + "epoch": 0.6233252800351416, + "grad_norm": 2.143852472305298, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.689407229423523, + "num_tokens": 146979182.0, + "step": 5676 + }, + { + "epoch": 0.6234350977377553, + "grad_norm": 2.0142767429351807, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6972026228904724, + "num_tokens": 147009131.0, + "step": 5677 + }, + { + "epoch": 0.6235449154403689, + "grad_norm": 2.105454444885254, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6975149512290955, + "num_tokens": 147038561.0, + "step": 5678 + }, + { + "epoch": 0.6236547331429827, + "grad_norm": 2.7047994136810303, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.731865644454956, + "num_tokens": 147055735.0, + "step": 5679 + }, + { + "epoch": 0.6237645508455963, + "grad_norm": 2.350121021270752, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7171192765235901, + "num_tokens": 147076421.0, + "step": 5680 + }, + { + "epoch": 0.62387436854821, + "grad_norm": 2.2603211402893066, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7084095478057861, + "num_tokens": 147099457.0, + "step": 5681 + }, + { + "epoch": 0.6239841862508236, + "grad_norm": 2.238236427307129, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.695030927658081, + "num_tokens": 147125823.0, + "step": 5682 + }, + { + "epoch": 0.6240940039534373, + "grad_norm": 2.0513455867767334, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7072945237159729, + "num_tokens": 147154551.0, + "step": 5683 + }, + { + "epoch": 0.6242038216560509, + "grad_norm": 2.3449959754943848, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7029239535331726, + "num_tokens": 147177596.0, + "step": 5684 + }, + { + "epoch": 0.6243136393586646, + "grad_norm": 2.3705408573150635, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6974287629127502, + "num_tokens": 147199678.0, + "step": 5685 + }, + { + "epoch": 0.6244234570612783, + "grad_norm": 2.1551992893218994, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7293590903282166, + "num_tokens": 147223851.0, + "step": 5686 + }, + { + "epoch": 0.624533274763892, + "grad_norm": 2.0744597911834717, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7187347412109375, + "num_tokens": 147251227.0, + "step": 5687 + }, + { + "epoch": 0.6246430924665056, + "grad_norm": 2.170379638671875, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6883785724639893, + "num_tokens": 147277361.0, + "step": 5688 + }, + { + "epoch": 0.6247529101691193, + "grad_norm": 2.330385684967041, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7048486471176147, + "num_tokens": 147297440.0, + "step": 5689 + }, + { + "epoch": 0.6248627278717329, + "grad_norm": 2.1117422580718994, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7087460160255432, + "num_tokens": 147324500.0, + "step": 5690 + }, + { + "epoch": 0.6249725455743466, + "grad_norm": 2.1180455684661865, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6948227882385254, + "num_tokens": 147352353.0, + "step": 5691 + }, + { + "epoch": 0.6250823632769602, + "grad_norm": 2.5502264499664307, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7139000296592712, + "num_tokens": 147371328.0, + "step": 5692 + }, + { + "epoch": 0.6251921809795739, + "grad_norm": 2.4128148555755615, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7196229696273804, + "num_tokens": 147390884.0, + "step": 5693 + }, + { + "epoch": 0.6253019986821876, + "grad_norm": 2.5602293014526367, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7206172943115234, + "num_tokens": 147410305.0, + "step": 5694 + }, + { + "epoch": 0.6254118163848013, + "grad_norm": 2.2348155975341797, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7063384056091309, + "num_tokens": 147434813.0, + "step": 5695 + }, + { + "epoch": 0.6255216340874149, + "grad_norm": 1.958522081375122, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7016158699989319, + "num_tokens": 147466456.0, + "step": 5696 + }, + { + "epoch": 0.6256314517900285, + "grad_norm": 2.6322686672210693, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7142599821090698, + "num_tokens": 147485784.0, + "step": 5697 + }, + { + "epoch": 0.6257412694926422, + "grad_norm": 2.185237407684326, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7255648374557495, + "num_tokens": 147509484.0, + "step": 5698 + }, + { + "epoch": 0.6258510871952558, + "grad_norm": 2.4300365447998047, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7061210870742798, + "num_tokens": 147530235.0, + "step": 5699 + }, + { + "epoch": 0.6259609048978695, + "grad_norm": 2.231318235397339, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6864290833473206, + "num_tokens": 147556317.0, + "step": 5700 + }, + { + "epoch": 0.6260707226004832, + "grad_norm": 2.212496519088745, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7083716988563538, + "num_tokens": 147580805.0, + "step": 5701 + }, + { + "epoch": 0.6261805403030969, + "grad_norm": 2.236558675765991, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7145301699638367, + "num_tokens": 147605713.0, + "step": 5702 + }, + { + "epoch": 0.6262903580057105, + "grad_norm": 2.154622793197632, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6955262422561646, + "num_tokens": 147632335.0, + "step": 5703 + }, + { + "epoch": 0.6264001757083242, + "grad_norm": 2.0485291481018066, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6998835802078247, + "num_tokens": 147658970.0, + "step": 5704 + }, + { + "epoch": 0.6265099934109378, + "grad_norm": 2.147463798522949, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7002884149551392, + "num_tokens": 147686015.0, + "step": 5705 + }, + { + "epoch": 0.6266198111135515, + "grad_norm": 2.4280788898468018, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7075713872909546, + "num_tokens": 147708507.0, + "step": 5706 + }, + { + "epoch": 0.6267296288161651, + "grad_norm": 2.1610915660858154, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6819219589233398, + "num_tokens": 147736217.0, + "step": 5707 + }, + { + "epoch": 0.6268394465187789, + "grad_norm": 2.2754006385803223, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7041453719139099, + "num_tokens": 147760731.0, + "step": 5708 + }, + { + "epoch": 0.6269492642213925, + "grad_norm": 2.1524105072021484, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6893302798271179, + "num_tokens": 147786980.0, + "step": 5709 + }, + { + "epoch": 0.6270590819240062, + "grad_norm": 2.296623468399048, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7220557928085327, + "num_tokens": 147807755.0, + "step": 5710 + }, + { + "epoch": 0.6271688996266198, + "grad_norm": 2.1554038524627686, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6975756883621216, + "num_tokens": 147834274.0, + "step": 5711 + }, + { + "epoch": 0.6272787173292335, + "grad_norm": 2.1122663021087646, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7194480895996094, + "num_tokens": 147860854.0, + "step": 5712 + }, + { + "epoch": 0.6273885350318471, + "grad_norm": 2.2420144081115723, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6816693544387817, + "num_tokens": 147887369.0, + "step": 5713 + }, + { + "epoch": 0.6274983527344608, + "grad_norm": 2.0342724323272705, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.722893238067627, + "num_tokens": 147915582.0, + "step": 5714 + }, + { + "epoch": 0.6276081704370745, + "grad_norm": 2.0628507137298584, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6929803490638733, + "num_tokens": 147943778.0, + "step": 5715 + }, + { + "epoch": 0.6277179881396882, + "grad_norm": 2.240788698196411, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6950973272323608, + "num_tokens": 147967687.0, + "step": 5716 + }, + { + "epoch": 0.6278278058423018, + "grad_norm": 2.0743236541748047, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.698957085609436, + "num_tokens": 147996893.0, + "step": 5717 + }, + { + "epoch": 0.6279376235449154, + "grad_norm": 1.9488790035247803, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.6808820962905884, + "num_tokens": 148030636.0, + "step": 5718 + }, + { + "epoch": 0.6280474412475291, + "grad_norm": 2.236396312713623, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7131460905075073, + "num_tokens": 148054012.0, + "step": 5719 + }, + { + "epoch": 0.6281572589501427, + "grad_norm": 1.9931033849716187, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6788216829299927, + "num_tokens": 148085309.0, + "step": 5720 + }, + { + "epoch": 0.6282670766527564, + "grad_norm": 2.2836380004882812, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6951111555099487, + "num_tokens": 148109119.0, + "step": 5721 + }, + { + "epoch": 0.62837689435537, + "grad_norm": 2.3787031173706055, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6989899277687073, + "num_tokens": 148132385.0, + "step": 5722 + }, + { + "epoch": 0.6284867120579838, + "grad_norm": 1.8444149494171143, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7034738063812256, + "num_tokens": 148165336.0, + "step": 5723 + }, + { + "epoch": 0.6285965297605974, + "grad_norm": 2.014571189880371, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6990910768508911, + "num_tokens": 148195345.0, + "step": 5724 + }, + { + "epoch": 0.6287063474632111, + "grad_norm": 2.3113174438476562, + "learning_rate": 1e-06, + "loss": 1.0632, + "mean_token_accuracy": 0.6860684156417847, + "num_tokens": 148218406.0, + "step": 5725 + }, + { + "epoch": 0.6288161651658247, + "grad_norm": 1.977109432220459, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7068419456481934, + "num_tokens": 148249170.0, + "step": 5726 + }, + { + "epoch": 0.6289259828684384, + "grad_norm": 2.4481446743011475, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7216625213623047, + "num_tokens": 148269680.0, + "step": 5727 + }, + { + "epoch": 0.629035800571052, + "grad_norm": 2.365377426147461, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7250678539276123, + "num_tokens": 148291251.0, + "step": 5728 + }, + { + "epoch": 0.6291456182736657, + "grad_norm": 2.198131799697876, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7142485976219177, + "num_tokens": 148318381.0, + "step": 5729 + }, + { + "epoch": 0.6292554359762794, + "grad_norm": 2.0647759437561035, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7122970819473267, + "num_tokens": 148345808.0, + "step": 5730 + }, + { + "epoch": 0.6293652536788931, + "grad_norm": 2.3891146183013916, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7236607074737549, + "num_tokens": 148367241.0, + "step": 5731 + }, + { + "epoch": 0.6294750713815067, + "grad_norm": 2.213221788406372, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6865714192390442, + "num_tokens": 148394134.0, + "step": 5732 + }, + { + "epoch": 0.6295848890841204, + "grad_norm": 2.2098801136016846, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7137409448623657, + "num_tokens": 148418898.0, + "step": 5733 + }, + { + "epoch": 0.629694706786734, + "grad_norm": 2.1309077739715576, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7030593156814575, + "num_tokens": 148445447.0, + "step": 5734 + }, + { + "epoch": 0.6298045244893477, + "grad_norm": 2.308997631072998, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6997794508934021, + "num_tokens": 148468281.0, + "step": 5735 + }, + { + "epoch": 0.6299143421919613, + "grad_norm": 2.1658966541290283, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7081084847450256, + "num_tokens": 148494389.0, + "step": 5736 + }, + { + "epoch": 0.630024159894575, + "grad_norm": 2.2159581184387207, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7053650617599487, + "num_tokens": 148521662.0, + "step": 5737 + }, + { + "epoch": 0.6301339775971887, + "grad_norm": 2.535740375518799, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7165261507034302, + "num_tokens": 148541457.0, + "step": 5738 + }, + { + "epoch": 0.6302437952998023, + "grad_norm": 2.14316987991333, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.687903881072998, + "num_tokens": 148569634.0, + "step": 5739 + }, + { + "epoch": 0.630353613002416, + "grad_norm": 1.9678843021392822, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6997332572937012, + "num_tokens": 148600795.0, + "step": 5740 + }, + { + "epoch": 0.6304634307050296, + "grad_norm": 1.995300531387329, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.713268518447876, + "num_tokens": 148629434.0, + "step": 5741 + }, + { + "epoch": 0.6305732484076433, + "grad_norm": 2.054328680038452, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7060421109199524, + "num_tokens": 148658548.0, + "step": 5742 + }, + { + "epoch": 0.6306830661102569, + "grad_norm": 2.120119094848633, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7151997089385986, + "num_tokens": 148685449.0, + "step": 5743 + }, + { + "epoch": 0.6307928838128707, + "grad_norm": 2.2278759479522705, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6988397836685181, + "num_tokens": 148709970.0, + "step": 5744 + }, + { + "epoch": 0.6309027015154843, + "grad_norm": 2.1500632762908936, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7043071985244751, + "num_tokens": 148737062.0, + "step": 5745 + }, + { + "epoch": 0.631012519218098, + "grad_norm": 1.9776837825775146, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7059240341186523, + "num_tokens": 148767671.0, + "step": 5746 + }, + { + "epoch": 0.6311223369207116, + "grad_norm": 2.043750286102295, + "learning_rate": 1e-06, + "loss": 1.0694, + "mean_token_accuracy": 0.6765499114990234, + "num_tokens": 148798121.0, + "step": 5747 + }, + { + "epoch": 0.6312321546233253, + "grad_norm": 2.1826181411743164, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7425971031188965, + "num_tokens": 148822303.0, + "step": 5748 + }, + { + "epoch": 0.6313419723259389, + "grad_norm": 2.2435598373413086, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7074213027954102, + "num_tokens": 148846091.0, + "step": 5749 + }, + { + "epoch": 0.6314517900285526, + "grad_norm": 2.1898088455200195, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7052000761032104, + "num_tokens": 148875284.0, + "step": 5750 + }, + { + "epoch": 0.6315616077311663, + "grad_norm": 2.2378737926483154, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6930146217346191, + "num_tokens": 148903453.0, + "step": 5751 + }, + { + "epoch": 0.63167142543378, + "grad_norm": 2.2124900817871094, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7032768130302429, + "num_tokens": 148930804.0, + "step": 5752 + }, + { + "epoch": 0.6317812431363936, + "grad_norm": 2.0986521244049072, + "learning_rate": 1e-06, + "loss": 1.0599, + "mean_token_accuracy": 0.6828368902206421, + "num_tokens": 148961324.0, + "step": 5753 + }, + { + "epoch": 0.6318910608390073, + "grad_norm": 2.0593745708465576, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7082014083862305, + "num_tokens": 148989833.0, + "step": 5754 + }, + { + "epoch": 0.6320008785416209, + "grad_norm": 1.9336788654327393, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.700747013092041, + "num_tokens": 149021071.0, + "step": 5755 + }, + { + "epoch": 0.6321106962442345, + "grad_norm": 2.1546761989593506, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7167290449142456, + "num_tokens": 149044826.0, + "step": 5756 + }, + { + "epoch": 0.6322205139468482, + "grad_norm": 2.0359442234039307, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.700302004814148, + "num_tokens": 149071338.0, + "step": 5757 + }, + { + "epoch": 0.6323303316494618, + "grad_norm": 2.032238483428955, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7088029384613037, + "num_tokens": 149100176.0, + "step": 5758 + }, + { + "epoch": 0.6324401493520756, + "grad_norm": 2.2357711791992188, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6915621757507324, + "num_tokens": 149126871.0, + "step": 5759 + }, + { + "epoch": 0.6325499670546892, + "grad_norm": 2.427844524383545, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7042529582977295, + "num_tokens": 149149003.0, + "step": 5760 + }, + { + "epoch": 0.6326597847573029, + "grad_norm": 2.5105719566345215, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6812916994094849, + "num_tokens": 149172255.0, + "step": 5761 + }, + { + "epoch": 0.6327696024599165, + "grad_norm": 2.336412191390991, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6942099332809448, + "num_tokens": 149194487.0, + "step": 5762 + }, + { + "epoch": 0.6328794201625302, + "grad_norm": 2.167234420776367, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7023420333862305, + "num_tokens": 149219638.0, + "step": 5763 + }, + { + "epoch": 0.6329892378651438, + "grad_norm": 2.6920061111450195, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.698559582233429, + "num_tokens": 149238091.0, + "step": 5764 + }, + { + "epoch": 0.6330990555677575, + "grad_norm": 2.2700018882751465, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7069287300109863, + "num_tokens": 149261916.0, + "step": 5765 + }, + { + "epoch": 0.6332088732703712, + "grad_norm": 2.2988438606262207, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7258998155593872, + "num_tokens": 149283449.0, + "step": 5766 + }, + { + "epoch": 0.6333186909729849, + "grad_norm": 2.1218514442443848, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7109392285346985, + "num_tokens": 149309221.0, + "step": 5767 + }, + { + "epoch": 0.6334285086755985, + "grad_norm": 2.009580135345459, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6913684606552124, + "num_tokens": 149338800.0, + "step": 5768 + }, + { + "epoch": 0.6335383263782122, + "grad_norm": 2.103242874145508, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7044473886489868, + "num_tokens": 149367427.0, + "step": 5769 + }, + { + "epoch": 0.6336481440808258, + "grad_norm": 2.317091941833496, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7068470120429993, + "num_tokens": 149391851.0, + "step": 5770 + }, + { + "epoch": 0.6337579617834395, + "grad_norm": 2.31276273727417, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7133008241653442, + "num_tokens": 149415251.0, + "step": 5771 + }, + { + "epoch": 0.6338677794860531, + "grad_norm": 2.055454730987549, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.684654176235199, + "num_tokens": 149444650.0, + "step": 5772 + }, + { + "epoch": 0.6339775971886669, + "grad_norm": 1.9447752237319946, + "learning_rate": 1e-06, + "loss": 1.0766, + "mean_token_accuracy": 0.6731460094451904, + "num_tokens": 149475849.0, + "step": 5773 + }, + { + "epoch": 0.6340874148912805, + "grad_norm": 2.0852975845336914, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6926884651184082, + "num_tokens": 149503813.0, + "step": 5774 + }, + { + "epoch": 0.6341972325938942, + "grad_norm": 2.2777397632598877, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.697725772857666, + "num_tokens": 149527726.0, + "step": 5775 + }, + { + "epoch": 0.6343070502965078, + "grad_norm": 1.9403706789016724, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6930415034294128, + "num_tokens": 149560705.0, + "step": 5776 + }, + { + "epoch": 0.6344168679991214, + "grad_norm": 2.038177490234375, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7035927772521973, + "num_tokens": 149587557.0, + "step": 5777 + }, + { + "epoch": 0.6345266857017351, + "grad_norm": 2.135768175125122, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6962834596633911, + "num_tokens": 149613308.0, + "step": 5778 + }, + { + "epoch": 0.6346365034043487, + "grad_norm": 2.253462314605713, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.675136923789978, + "num_tokens": 149639459.0, + "step": 5779 + }, + { + "epoch": 0.6347463211069625, + "grad_norm": 2.043926239013672, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7022760510444641, + "num_tokens": 149669146.0, + "step": 5780 + }, + { + "epoch": 0.6348561388095761, + "grad_norm": 2.1195566654205322, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.709907591342926, + "num_tokens": 149695518.0, + "step": 5781 + }, + { + "epoch": 0.6349659565121898, + "grad_norm": 2.4380624294281006, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7171430587768555, + "num_tokens": 149718383.0, + "step": 5782 + }, + { + "epoch": 0.6350757742148034, + "grad_norm": 2.34374737739563, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.706039309501648, + "num_tokens": 149741179.0, + "step": 5783 + }, + { + "epoch": 0.6351855919174171, + "grad_norm": 2.380920886993408, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6964014172554016, + "num_tokens": 149763981.0, + "step": 5784 + }, + { + "epoch": 0.6352954096200307, + "grad_norm": 1.9834731817245483, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.698000431060791, + "num_tokens": 149794146.0, + "step": 5785 + }, + { + "epoch": 0.6354052273226444, + "grad_norm": 2.246487855911255, + "learning_rate": 1e-06, + "loss": 1.0534, + "mean_token_accuracy": 0.6928284168243408, + "num_tokens": 149819811.0, + "step": 5786 + }, + { + "epoch": 0.635515045025258, + "grad_norm": 2.0633063316345215, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.695244312286377, + "num_tokens": 149848238.0, + "step": 5787 + }, + { + "epoch": 0.6356248627278718, + "grad_norm": 2.2095448970794678, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7067573666572571, + "num_tokens": 149874801.0, + "step": 5788 + }, + { + "epoch": 0.6357346804304854, + "grad_norm": 2.4139130115509033, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7098773717880249, + "num_tokens": 149897048.0, + "step": 5789 + }, + { + "epoch": 0.6358444981330991, + "grad_norm": 2.391388416290283, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.6912097930908203, + "num_tokens": 149919727.0, + "step": 5790 + }, + { + "epoch": 0.6359543158357127, + "grad_norm": 2.1433634757995605, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7047020792961121, + "num_tokens": 149945061.0, + "step": 5791 + }, + { + "epoch": 0.6360641335383264, + "grad_norm": 2.4436216354370117, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7183716297149658, + "num_tokens": 149966169.0, + "step": 5792 + }, + { + "epoch": 0.63617395124094, + "grad_norm": 2.096736192703247, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7217611074447632, + "num_tokens": 149992607.0, + "step": 5793 + }, + { + "epoch": 0.6362837689435537, + "grad_norm": 2.1854641437530518, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6849589347839355, + "num_tokens": 150019636.0, + "step": 5794 + }, + { + "epoch": 0.6363935866461674, + "grad_norm": 2.1004347801208496, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.6998494863510132, + "num_tokens": 150046268.0, + "step": 5795 + }, + { + "epoch": 0.636503404348781, + "grad_norm": 2.1439263820648193, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6898418664932251, + "num_tokens": 150072283.0, + "step": 5796 + }, + { + "epoch": 0.6366132220513947, + "grad_norm": 2.1181747913360596, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7027601003646851, + "num_tokens": 150099010.0, + "step": 5797 + }, + { + "epoch": 0.6367230397540083, + "grad_norm": 2.10294508934021, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7095082402229309, + "num_tokens": 150124807.0, + "step": 5798 + }, + { + "epoch": 0.636832857456622, + "grad_norm": 2.094397783279419, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7104927897453308, + "num_tokens": 150153282.0, + "step": 5799 + }, + { + "epoch": 0.6369426751592356, + "grad_norm": 2.379589557647705, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6957467794418335, + "num_tokens": 150175704.0, + "step": 5800 + }, + { + "epoch": 0.6370524928618493, + "grad_norm": 2.2002909183502197, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.715523898601532, + "num_tokens": 150201266.0, + "step": 5801 + }, + { + "epoch": 0.637162310564463, + "grad_norm": 2.2335121631622314, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7144898772239685, + "num_tokens": 150223486.0, + "step": 5802 + }, + { + "epoch": 0.6372721282670767, + "grad_norm": 2.1852872371673584, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7045165300369263, + "num_tokens": 150249127.0, + "step": 5803 + }, + { + "epoch": 0.6373819459696903, + "grad_norm": 2.3074424266815186, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7213463187217712, + "num_tokens": 150272652.0, + "step": 5804 + }, + { + "epoch": 0.637491763672304, + "grad_norm": 2.2966818809509277, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7169569730758667, + "num_tokens": 150296163.0, + "step": 5805 + }, + { + "epoch": 0.6376015813749176, + "grad_norm": 2.073070526123047, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6921068429946899, + "num_tokens": 150326248.0, + "step": 5806 + }, + { + "epoch": 0.6377113990775313, + "grad_norm": 2.1544125080108643, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7147791385650635, + "num_tokens": 150350743.0, + "step": 5807 + }, + { + "epoch": 0.6378212167801449, + "grad_norm": 2.2408287525177, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7240804433822632, + "num_tokens": 150374351.0, + "step": 5808 + }, + { + "epoch": 0.6379310344827587, + "grad_norm": 1.9946813583374023, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.684260904788971, + "num_tokens": 150403427.0, + "step": 5809 + }, + { + "epoch": 0.6380408521853723, + "grad_norm": 2.313535690307617, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7185168266296387, + "num_tokens": 150425915.0, + "step": 5810 + }, + { + "epoch": 0.638150669887986, + "grad_norm": 2.3221960067749023, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7130058407783508, + "num_tokens": 150448372.0, + "step": 5811 + }, + { + "epoch": 0.6382604875905996, + "grad_norm": 2.3128013610839844, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7224769592285156, + "num_tokens": 150470554.0, + "step": 5812 + }, + { + "epoch": 0.6383703052932133, + "grad_norm": 2.368457555770874, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6850783228874207, + "num_tokens": 150493751.0, + "step": 5813 + }, + { + "epoch": 0.6384801229958269, + "grad_norm": 2.2635045051574707, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6931006908416748, + "num_tokens": 150518023.0, + "step": 5814 + }, + { + "epoch": 0.6385899406984406, + "grad_norm": 1.9148355722427368, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7215452790260315, + "num_tokens": 150549241.0, + "step": 5815 + }, + { + "epoch": 0.6386997584010542, + "grad_norm": 2.0350472927093506, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7051149010658264, + "num_tokens": 150580556.0, + "step": 5816 + }, + { + "epoch": 0.638809576103668, + "grad_norm": 2.0410373210906982, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6910523176193237, + "num_tokens": 150611486.0, + "step": 5817 + }, + { + "epoch": 0.6389193938062816, + "grad_norm": 2.2480432987213135, + "learning_rate": 1e-06, + "loss": 1.0599, + "mean_token_accuracy": 0.6766794323921204, + "num_tokens": 150636022.0, + "step": 5818 + }, + { + "epoch": 0.6390292115088952, + "grad_norm": 2.3066015243530273, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6987864375114441, + "num_tokens": 150659070.0, + "step": 5819 + }, + { + "epoch": 0.6391390292115089, + "grad_norm": 2.127903938293457, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.73052978515625, + "num_tokens": 150685946.0, + "step": 5820 + }, + { + "epoch": 0.6392488469141225, + "grad_norm": 2.2393739223480225, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7146890759468079, + "num_tokens": 150710552.0, + "step": 5821 + }, + { + "epoch": 0.6393586646167362, + "grad_norm": 2.1426899433135986, + "learning_rate": 1e-06, + "loss": 1.0727, + "mean_token_accuracy": 0.6873778104782104, + "num_tokens": 150739809.0, + "step": 5822 + }, + { + "epoch": 0.6394684823193498, + "grad_norm": 2.4709527492523193, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7094258666038513, + "num_tokens": 150761127.0, + "step": 5823 + }, + { + "epoch": 0.6395783000219636, + "grad_norm": 2.0811824798583984, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7586156129837036, + "num_tokens": 150786097.0, + "step": 5824 + }, + { + "epoch": 0.6396881177245772, + "grad_norm": 2.4211983680725098, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6944432258605957, + "num_tokens": 150807894.0, + "step": 5825 + }, + { + "epoch": 0.6397979354271909, + "grad_norm": 2.2810611724853516, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7081078886985779, + "num_tokens": 150831323.0, + "step": 5826 + }, + { + "epoch": 0.6399077531298045, + "grad_norm": 2.1284923553466797, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6980681419372559, + "num_tokens": 150858194.0, + "step": 5827 + }, + { + "epoch": 0.6400175708324182, + "grad_norm": 2.2273690700531006, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7087425589561462, + "num_tokens": 150882564.0, + "step": 5828 + }, + { + "epoch": 0.6401273885350318, + "grad_norm": 2.1760239601135254, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.724872350692749, + "num_tokens": 150907187.0, + "step": 5829 + }, + { + "epoch": 0.6402372062376455, + "grad_norm": 2.147123336791992, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6879364252090454, + "num_tokens": 150934895.0, + "step": 5830 + }, + { + "epoch": 0.6403470239402592, + "grad_norm": 2.285809278488159, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6972313523292542, + "num_tokens": 150958210.0, + "step": 5831 + }, + { + "epoch": 0.6404568416428729, + "grad_norm": 2.2708559036254883, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7362422347068787, + "num_tokens": 150980567.0, + "step": 5832 + }, + { + "epoch": 0.6405666593454865, + "grad_norm": 2.0250601768493652, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6939119100570679, + "num_tokens": 151009372.0, + "step": 5833 + }, + { + "epoch": 0.6406764770481002, + "grad_norm": 2.2021706104278564, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6970616579055786, + "num_tokens": 151035712.0, + "step": 5834 + }, + { + "epoch": 0.6407862947507138, + "grad_norm": 2.1296215057373047, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7238212823867798, + "num_tokens": 151061318.0, + "step": 5835 + }, + { + "epoch": 0.6408961124533274, + "grad_norm": 2.1233487129211426, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7033513784408569, + "num_tokens": 151087793.0, + "step": 5836 + }, + { + "epoch": 0.6410059301559411, + "grad_norm": 2.2996323108673096, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6953166127204895, + "num_tokens": 151111689.0, + "step": 5837 + }, + { + "epoch": 0.6411157478585549, + "grad_norm": 1.9703258275985718, + "learning_rate": 1e-06, + "loss": 1.0514, + "mean_token_accuracy": 0.6876081228256226, + "num_tokens": 151144752.0, + "step": 5838 + }, + { + "epoch": 0.6412255655611685, + "grad_norm": 2.0224053859710693, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7109001278877258, + "num_tokens": 151175374.0, + "step": 5839 + }, + { + "epoch": 0.6413353832637821, + "grad_norm": 1.8790429830551147, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6974283456802368, + "num_tokens": 151207581.0, + "step": 5840 + }, + { + "epoch": 0.6414452009663958, + "grad_norm": 2.15559458732605, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7233908772468567, + "num_tokens": 151231272.0, + "step": 5841 + }, + { + "epoch": 0.6415550186690094, + "grad_norm": 2.0174732208251953, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6970255970954895, + "num_tokens": 151259848.0, + "step": 5842 + }, + { + "epoch": 0.6416648363716231, + "grad_norm": 2.4306914806365967, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7165223360061646, + "num_tokens": 151283535.0, + "step": 5843 + }, + { + "epoch": 0.6417746540742367, + "grad_norm": 2.3335928916931152, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7177116870880127, + "num_tokens": 151307785.0, + "step": 5844 + }, + { + "epoch": 0.6418844717768504, + "grad_norm": 2.179067850112915, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6895024180412292, + "num_tokens": 151334314.0, + "step": 5845 + }, + { + "epoch": 0.6419942894794641, + "grad_norm": 2.5756664276123047, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7191183567047119, + "num_tokens": 151355616.0, + "step": 5846 + }, + { + "epoch": 0.6421041071820778, + "grad_norm": 2.2630650997161865, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7282472252845764, + "num_tokens": 151377054.0, + "step": 5847 + }, + { + "epoch": 0.6422139248846914, + "grad_norm": 2.180250644683838, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7112319469451904, + "num_tokens": 151402658.0, + "step": 5848 + }, + { + "epoch": 0.6423237425873051, + "grad_norm": 2.1614997386932373, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7057418823242188, + "num_tokens": 151429171.0, + "step": 5849 + }, + { + "epoch": 0.6424335602899187, + "grad_norm": 2.0495269298553467, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7034245133399963, + "num_tokens": 151458553.0, + "step": 5850 + }, + { + "epoch": 0.6425433779925324, + "grad_norm": 2.282608985900879, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7178642749786377, + "num_tokens": 151481152.0, + "step": 5851 + }, + { + "epoch": 0.642653195695146, + "grad_norm": 1.929197907447815, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7075923681259155, + "num_tokens": 151511318.0, + "step": 5852 + }, + { + "epoch": 0.6427630133977598, + "grad_norm": 2.233304977416992, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7120104432106018, + "num_tokens": 151534627.0, + "step": 5853 + }, + { + "epoch": 0.6428728311003734, + "grad_norm": 2.2188467979431152, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6997689008712769, + "num_tokens": 151560458.0, + "step": 5854 + }, + { + "epoch": 0.6429826488029871, + "grad_norm": 2.381009578704834, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7027333974838257, + "num_tokens": 151582610.0, + "step": 5855 + }, + { + "epoch": 0.6430924665056007, + "grad_norm": 2.225616931915283, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7202335596084595, + "num_tokens": 151608983.0, + "step": 5856 + }, + { + "epoch": 0.6432022842082143, + "grad_norm": 2.5559680461883545, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7250227928161621, + "num_tokens": 151627493.0, + "step": 5857 + }, + { + "epoch": 0.643312101910828, + "grad_norm": 1.7713408470153809, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7415968775749207, + "num_tokens": 151661357.0, + "step": 5858 + }, + { + "epoch": 0.6434219196134416, + "grad_norm": 1.88136887550354, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6977864503860474, + "num_tokens": 151696267.0, + "step": 5859 + }, + { + "epoch": 0.6435317373160554, + "grad_norm": 2.0254595279693604, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7027865052223206, + "num_tokens": 151724986.0, + "step": 5860 + }, + { + "epoch": 0.643641555018669, + "grad_norm": 2.0344879627227783, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7216744422912598, + "num_tokens": 151752918.0, + "step": 5861 + }, + { + "epoch": 0.6437513727212827, + "grad_norm": 2.1659610271453857, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7234300374984741, + "num_tokens": 151777243.0, + "step": 5862 + }, + { + "epoch": 0.6438611904238963, + "grad_norm": 2.1705586910247803, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6988701820373535, + "num_tokens": 151805063.0, + "step": 5863 + }, + { + "epoch": 0.64397100812651, + "grad_norm": 2.2141973972320557, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7114611864089966, + "num_tokens": 151831224.0, + "step": 5864 + }, + { + "epoch": 0.6440808258291236, + "grad_norm": 1.9985610246658325, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6799764633178711, + "num_tokens": 151862421.0, + "step": 5865 + }, + { + "epoch": 0.6441906435317373, + "grad_norm": 2.2771899700164795, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6984940767288208, + "num_tokens": 151888055.0, + "step": 5866 + }, + { + "epoch": 0.644300461234351, + "grad_norm": 2.4413161277770996, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7290791273117065, + "num_tokens": 151907627.0, + "step": 5867 + }, + { + "epoch": 0.6444102789369647, + "grad_norm": 2.2418620586395264, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7057268619537354, + "num_tokens": 151931472.0, + "step": 5868 + }, + { + "epoch": 0.6445200966395783, + "grad_norm": 2.192744493484497, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7166825532913208, + "num_tokens": 151958960.0, + "step": 5869 + }, + { + "epoch": 0.644629914342192, + "grad_norm": 2.027620792388916, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7091673612594604, + "num_tokens": 151986428.0, + "step": 5870 + }, + { + "epoch": 0.6447397320448056, + "grad_norm": 2.3187484741210938, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7071748971939087, + "num_tokens": 152009759.0, + "step": 5871 + }, + { + "epoch": 0.6448495497474193, + "grad_norm": 2.039537191390991, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6865209937095642, + "num_tokens": 152039419.0, + "step": 5872 + }, + { + "epoch": 0.6449593674500329, + "grad_norm": 2.307201385498047, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7123232483863831, + "num_tokens": 152061413.0, + "step": 5873 + }, + { + "epoch": 0.6450691851526466, + "grad_norm": 2.2327306270599365, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6881687641143799, + "num_tokens": 152086909.0, + "step": 5874 + }, + { + "epoch": 0.6451790028552603, + "grad_norm": 2.3001303672790527, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7122450470924377, + "num_tokens": 152110852.0, + "step": 5875 + }, + { + "epoch": 0.645288820557874, + "grad_norm": 2.1139214038848877, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6905344724655151, + "num_tokens": 152139170.0, + "step": 5876 + }, + { + "epoch": 0.6453986382604876, + "grad_norm": 2.3433125019073486, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6969361305236816, + "num_tokens": 152162709.0, + "step": 5877 + }, + { + "epoch": 0.6455084559631012, + "grad_norm": 2.194544792175293, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6951799392700195, + "num_tokens": 152190478.0, + "step": 5878 + }, + { + "epoch": 0.6456182736657149, + "grad_norm": 2.103485107421875, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7015864849090576, + "num_tokens": 152218556.0, + "step": 5879 + }, + { + "epoch": 0.6457280913683285, + "grad_norm": 2.262813091278076, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7111035585403442, + "num_tokens": 152242928.0, + "step": 5880 + }, + { + "epoch": 0.6458379090709422, + "grad_norm": 2.2977540493011475, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7106429934501648, + "num_tokens": 152266912.0, + "step": 5881 + }, + { + "epoch": 0.6459477267735559, + "grad_norm": 1.9735376834869385, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.6987888813018799, + "num_tokens": 152297859.0, + "step": 5882 + }, + { + "epoch": 0.6460575444761696, + "grad_norm": 2.3695907592773438, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7064558267593384, + "num_tokens": 152320490.0, + "step": 5883 + }, + { + "epoch": 0.6461673621787832, + "grad_norm": 2.133737087249756, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6904276609420776, + "num_tokens": 152346999.0, + "step": 5884 + }, + { + "epoch": 0.6462771798813969, + "grad_norm": 2.0646820068359375, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7059310674667358, + "num_tokens": 152377770.0, + "step": 5885 + }, + { + "epoch": 0.6463869975840105, + "grad_norm": 2.064934730529785, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6856513023376465, + "num_tokens": 152408079.0, + "step": 5886 + }, + { + "epoch": 0.6464968152866242, + "grad_norm": 2.3278136253356934, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6975828409194946, + "num_tokens": 152430651.0, + "step": 5887 + }, + { + "epoch": 0.6466066329892378, + "grad_norm": 2.09814453125, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6943674087524414, + "num_tokens": 152459557.0, + "step": 5888 + }, + { + "epoch": 0.6467164506918516, + "grad_norm": 2.6015827655792236, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7314215898513794, + "num_tokens": 152478085.0, + "step": 5889 + }, + { + "epoch": 0.6468262683944652, + "grad_norm": 2.128998279571533, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7256383895874023, + "num_tokens": 152506522.0, + "step": 5890 + }, + { + "epoch": 0.6469360860970789, + "grad_norm": 2.075773239135742, + "learning_rate": 1e-06, + "loss": 1.0629, + "mean_token_accuracy": 0.6820801496505737, + "num_tokens": 152535998.0, + "step": 5891 + }, + { + "epoch": 0.6470459037996925, + "grad_norm": 2.178011417388916, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6928662061691284, + "num_tokens": 152562365.0, + "step": 5892 + }, + { + "epoch": 0.6471557215023062, + "grad_norm": 1.9805454015731812, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.687085747718811, + "num_tokens": 152591858.0, + "step": 5893 + }, + { + "epoch": 0.6472655392049198, + "grad_norm": 2.23266339302063, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7292912006378174, + "num_tokens": 152616172.0, + "step": 5894 + }, + { + "epoch": 0.6473753569075335, + "grad_norm": 2.270073175430298, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6997050642967224, + "num_tokens": 152639601.0, + "step": 5895 + }, + { + "epoch": 0.6474851746101472, + "grad_norm": 2.163255453109741, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6749756336212158, + "num_tokens": 152667239.0, + "step": 5896 + }, + { + "epoch": 0.6475949923127609, + "grad_norm": 2.002814292907715, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6878511905670166, + "num_tokens": 152697231.0, + "step": 5897 + }, + { + "epoch": 0.6477048100153745, + "grad_norm": 2.0456430912017822, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6956467628479004, + "num_tokens": 152728080.0, + "step": 5898 + }, + { + "epoch": 0.6478146277179881, + "grad_norm": 1.855770230293274, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.6858265399932861, + "num_tokens": 152762345.0, + "step": 5899 + }, + { + "epoch": 0.6479244454206018, + "grad_norm": 2.0826358795166016, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6953545212745667, + "num_tokens": 152789650.0, + "step": 5900 + }, + { + "epoch": 0.6480342631232154, + "grad_norm": 1.9909344911575317, + "learning_rate": 1e-06, + "loss": 1.0846, + "mean_token_accuracy": 0.6802046895027161, + "num_tokens": 152821126.0, + "step": 5901 + }, + { + "epoch": 0.6481440808258291, + "grad_norm": 2.2632904052734375, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6997241377830505, + "num_tokens": 152846609.0, + "step": 5902 + }, + { + "epoch": 0.6482538985284428, + "grad_norm": 2.1110169887542725, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6999356746673584, + "num_tokens": 152872815.0, + "step": 5903 + }, + { + "epoch": 0.6483637162310565, + "grad_norm": 2.0893824100494385, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6906708478927612, + "num_tokens": 152900870.0, + "step": 5904 + }, + { + "epoch": 0.6484735339336701, + "grad_norm": 2.1449429988861084, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7207913994789124, + "num_tokens": 152926475.0, + "step": 5905 + }, + { + "epoch": 0.6485833516362838, + "grad_norm": 1.9270871877670288, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6923996210098267, + "num_tokens": 152957867.0, + "step": 5906 + }, + { + "epoch": 0.6486931693388974, + "grad_norm": 2.159109592437744, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7141057252883911, + "num_tokens": 152982578.0, + "step": 5907 + }, + { + "epoch": 0.6488029870415111, + "grad_norm": 2.2711968421936035, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7176067233085632, + "num_tokens": 153006120.0, + "step": 5908 + }, + { + "epoch": 0.6489128047441247, + "grad_norm": 2.121624708175659, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7285661101341248, + "num_tokens": 153032777.0, + "step": 5909 + }, + { + "epoch": 0.6490226224467384, + "grad_norm": 2.200005531311035, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.687212347984314, + "num_tokens": 153061135.0, + "step": 5910 + }, + { + "epoch": 0.6491324401493521, + "grad_norm": 1.9921519756317139, + "learning_rate": 1e-06, + "loss": 1.0969, + "mean_token_accuracy": 0.6609451770782471, + "num_tokens": 153094626.0, + "step": 5911 + }, + { + "epoch": 0.6492422578519658, + "grad_norm": 2.3137402534484863, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7021867036819458, + "num_tokens": 153120192.0, + "step": 5912 + }, + { + "epoch": 0.6493520755545794, + "grad_norm": 2.1506237983703613, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7069530487060547, + "num_tokens": 153146784.0, + "step": 5913 + }, + { + "epoch": 0.6494618932571931, + "grad_norm": 2.2972662448883057, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7224059700965881, + "num_tokens": 153170056.0, + "step": 5914 + }, + { + "epoch": 0.6495717109598067, + "grad_norm": 2.155336856842041, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7214453220367432, + "num_tokens": 153195030.0, + "step": 5915 + }, + { + "epoch": 0.6496815286624203, + "grad_norm": 2.1300852298736572, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7159770727157593, + "num_tokens": 153221898.0, + "step": 5916 + }, + { + "epoch": 0.649791346365034, + "grad_norm": 1.9101589918136597, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.700849711894989, + "num_tokens": 153255269.0, + "step": 5917 + }, + { + "epoch": 0.6499011640676478, + "grad_norm": 2.0941481590270996, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7227259874343872, + "num_tokens": 153281596.0, + "step": 5918 + }, + { + "epoch": 0.6500109817702614, + "grad_norm": 2.5788767337799072, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7187415361404419, + "num_tokens": 153299996.0, + "step": 5919 + }, + { + "epoch": 0.650120799472875, + "grad_norm": 2.2257742881774902, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.707175612449646, + "num_tokens": 153324129.0, + "step": 5920 + }, + { + "epoch": 0.6502306171754887, + "grad_norm": 2.2578656673431396, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6999763250350952, + "num_tokens": 153348145.0, + "step": 5921 + }, + { + "epoch": 0.6503404348781023, + "grad_norm": 2.2693092823028564, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7125451564788818, + "num_tokens": 153371737.0, + "step": 5922 + }, + { + "epoch": 0.650450252580716, + "grad_norm": 2.021108865737915, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7011659145355225, + "num_tokens": 153402899.0, + "step": 5923 + }, + { + "epoch": 0.6505600702833296, + "grad_norm": 2.0526795387268066, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7326356768608093, + "num_tokens": 153428385.0, + "step": 5924 + }, + { + "epoch": 0.6506698879859434, + "grad_norm": 1.9089816808700562, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6860636472702026, + "num_tokens": 153461679.0, + "step": 5925 + }, + { + "epoch": 0.650779705688557, + "grad_norm": 2.0990121364593506, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6989859342575073, + "num_tokens": 153491875.0, + "step": 5926 + }, + { + "epoch": 0.6508895233911707, + "grad_norm": 2.466618776321411, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6992129683494568, + "num_tokens": 153512525.0, + "step": 5927 + }, + { + "epoch": 0.6509993410937843, + "grad_norm": 2.2360191345214844, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7057640552520752, + "num_tokens": 153537574.0, + "step": 5928 + }, + { + "epoch": 0.651109158796398, + "grad_norm": 2.2627480030059814, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7133152484893799, + "num_tokens": 153561400.0, + "step": 5929 + }, + { + "epoch": 0.6512189764990116, + "grad_norm": 2.208120107650757, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6931681632995605, + "num_tokens": 153588151.0, + "step": 5930 + }, + { + "epoch": 0.6513287942016253, + "grad_norm": 2.137010097503662, + "learning_rate": 1e-06, + "loss": 1.1083, + "mean_token_accuracy": 0.667682945728302, + "num_tokens": 153619776.0, + "step": 5931 + }, + { + "epoch": 0.651438611904239, + "grad_norm": 2.5879361629486084, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7039231061935425, + "num_tokens": 153639527.0, + "step": 5932 + }, + { + "epoch": 0.6515484296068527, + "grad_norm": 1.9827122688293457, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7254396677017212, + "num_tokens": 153667505.0, + "step": 5933 + }, + { + "epoch": 0.6516582473094663, + "grad_norm": 2.0932118892669678, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6979014873504639, + "num_tokens": 153695390.0, + "step": 5934 + }, + { + "epoch": 0.65176806501208, + "grad_norm": 2.4297616481781006, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7155669331550598, + "num_tokens": 153717638.0, + "step": 5935 + }, + { + "epoch": 0.6518778827146936, + "grad_norm": 2.3572843074798584, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7142621874809265, + "num_tokens": 153738948.0, + "step": 5936 + }, + { + "epoch": 0.6519877004173072, + "grad_norm": 2.7675764560699463, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7294234037399292, + "num_tokens": 153755757.0, + "step": 5937 + }, + { + "epoch": 0.6520975181199209, + "grad_norm": 2.114234685897827, + "learning_rate": 1e-06, + "loss": 1.0444, + "mean_token_accuracy": 0.6862674951553345, + "num_tokens": 153782883.0, + "step": 5938 + }, + { + "epoch": 0.6522073358225345, + "grad_norm": 2.114438056945801, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7179653644561768, + "num_tokens": 153809530.0, + "step": 5939 + }, + { + "epoch": 0.6523171535251483, + "grad_norm": 2.2954328060150146, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7351362705230713, + "num_tokens": 153833445.0, + "step": 5940 + }, + { + "epoch": 0.6524269712277619, + "grad_norm": 2.039414167404175, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6945683360099792, + "num_tokens": 153863494.0, + "step": 5941 + }, + { + "epoch": 0.6525367889303756, + "grad_norm": 2.1839187145233154, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7025208473205566, + "num_tokens": 153889125.0, + "step": 5942 + }, + { + "epoch": 0.6526466066329892, + "grad_norm": 2.3916001319885254, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7032122015953064, + "num_tokens": 153911542.0, + "step": 5943 + }, + { + "epoch": 0.6527564243356029, + "grad_norm": 1.946409821510315, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.6877784729003906, + "num_tokens": 153942533.0, + "step": 5944 + }, + { + "epoch": 0.6528662420382165, + "grad_norm": 2.2011280059814453, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6959173679351807, + "num_tokens": 153967534.0, + "step": 5945 + }, + { + "epoch": 0.6529760597408302, + "grad_norm": 2.0168588161468506, + "learning_rate": 1e-06, + "loss": 1.083, + "mean_token_accuracy": 0.6755622029304504, + "num_tokens": 153998781.0, + "step": 5946 + }, + { + "epoch": 0.6530858774434439, + "grad_norm": 2.587509870529175, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.6983546018600464, + "num_tokens": 154019262.0, + "step": 5947 + }, + { + "epoch": 0.6531956951460576, + "grad_norm": 2.083855628967285, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.699851393699646, + "num_tokens": 154047477.0, + "step": 5948 + }, + { + "epoch": 0.6533055128486712, + "grad_norm": 2.266857385635376, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7217541933059692, + "num_tokens": 154070596.0, + "step": 5949 + }, + { + "epoch": 0.6534153305512849, + "grad_norm": 2.900010347366333, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7146667242050171, + "num_tokens": 154088080.0, + "step": 5950 + }, + { + "epoch": 0.6535251482538985, + "grad_norm": 2.3092687129974365, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7191526889801025, + "num_tokens": 154111183.0, + "step": 5951 + }, + { + "epoch": 0.6536349659565122, + "grad_norm": 2.2762668132781982, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7057468891143799, + "num_tokens": 154133988.0, + "step": 5952 + }, + { + "epoch": 0.6537447836591258, + "grad_norm": 2.0014495849609375, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7073830366134644, + "num_tokens": 154162949.0, + "step": 5953 + }, + { + "epoch": 0.6538546013617396, + "grad_norm": 2.048929452896118, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7042447924613953, + "num_tokens": 154191471.0, + "step": 5954 + }, + { + "epoch": 0.6539644190643532, + "grad_norm": 2.2327582836151123, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7084435224533081, + "num_tokens": 154214934.0, + "step": 5955 + }, + { + "epoch": 0.6540742367669669, + "grad_norm": 2.0553812980651855, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7018048763275146, + "num_tokens": 154242775.0, + "step": 5956 + }, + { + "epoch": 0.6541840544695805, + "grad_norm": 2.1129887104034424, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6845973134040833, + "num_tokens": 154271659.0, + "step": 5957 + }, + { + "epoch": 0.6542938721721941, + "grad_norm": 2.508500337600708, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7365235686302185, + "num_tokens": 154291248.0, + "step": 5958 + }, + { + "epoch": 0.6544036898748078, + "grad_norm": 2.036428451538086, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7019880414009094, + "num_tokens": 154319496.0, + "step": 5959 + }, + { + "epoch": 0.6545135075774214, + "grad_norm": 2.0052363872528076, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.707997739315033, + "num_tokens": 154348439.0, + "step": 5960 + }, + { + "epoch": 0.6546233252800352, + "grad_norm": 2.2237582206726074, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.727950930595398, + "num_tokens": 154372375.0, + "step": 5961 + }, + { + "epoch": 0.6547331429826488, + "grad_norm": 2.106705665588379, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7187386751174927, + "num_tokens": 154397081.0, + "step": 5962 + }, + { + "epoch": 0.6548429606852625, + "grad_norm": 2.1624791622161865, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.6998645067214966, + "num_tokens": 154425629.0, + "step": 5963 + }, + { + "epoch": 0.6549527783878761, + "grad_norm": 2.4307844638824463, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7284847497940063, + "num_tokens": 154447369.0, + "step": 5964 + }, + { + "epoch": 0.6550625960904898, + "grad_norm": 2.472342014312744, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6996528506278992, + "num_tokens": 154469352.0, + "step": 5965 + }, + { + "epoch": 0.6551724137931034, + "grad_norm": 2.150345802307129, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6918818950653076, + "num_tokens": 154498264.0, + "step": 5966 + }, + { + "epoch": 0.6552822314957171, + "grad_norm": 2.4084157943725586, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7320806980133057, + "num_tokens": 154518816.0, + "step": 5967 + }, + { + "epoch": 0.6553920491983307, + "grad_norm": 1.9166414737701416, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6835247874259949, + "num_tokens": 154553216.0, + "step": 5968 + }, + { + "epoch": 0.6555018669009445, + "grad_norm": 2.354153871536255, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7370086312294006, + "num_tokens": 154573981.0, + "step": 5969 + }, + { + "epoch": 0.6556116846035581, + "grad_norm": 2.054677963256836, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.701751708984375, + "num_tokens": 154602589.0, + "step": 5970 + }, + { + "epoch": 0.6557215023061718, + "grad_norm": 2.6160173416137695, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.705959677696228, + "num_tokens": 154621052.0, + "step": 5971 + }, + { + "epoch": 0.6558313200087854, + "grad_norm": 2.266380548477173, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6878149509429932, + "num_tokens": 154647699.0, + "step": 5972 + }, + { + "epoch": 0.6559411377113991, + "grad_norm": 2.3092336654663086, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7197739481925964, + "num_tokens": 154669577.0, + "step": 5973 + }, + { + "epoch": 0.6560509554140127, + "grad_norm": 2.2385404109954834, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7126693725585938, + "num_tokens": 154695662.0, + "step": 5974 + }, + { + "epoch": 0.6561607731166264, + "grad_norm": 2.1245291233062744, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6929975748062134, + "num_tokens": 154723545.0, + "step": 5975 + }, + { + "epoch": 0.6562705908192401, + "grad_norm": 1.9974257946014404, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6879948377609253, + "num_tokens": 154755534.0, + "step": 5976 + }, + { + "epoch": 0.6563804085218538, + "grad_norm": 2.1761255264282227, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6976797580718994, + "num_tokens": 154781348.0, + "step": 5977 + }, + { + "epoch": 0.6564902262244674, + "grad_norm": 2.0634663105010986, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6967449188232422, + "num_tokens": 154810297.0, + "step": 5978 + }, + { + "epoch": 0.656600043927081, + "grad_norm": 2.409722328186035, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7129645347595215, + "num_tokens": 154831098.0, + "step": 5979 + }, + { + "epoch": 0.6567098616296947, + "grad_norm": 2.216740369796753, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6819798350334167, + "num_tokens": 154857877.0, + "step": 5980 + }, + { + "epoch": 0.6568196793323083, + "grad_norm": 2.194051504135132, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7127381563186646, + "num_tokens": 154883350.0, + "step": 5981 + }, + { + "epoch": 0.656929497034922, + "grad_norm": 2.1262881755828857, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6942802667617798, + "num_tokens": 154909297.0, + "step": 5982 + }, + { + "epoch": 0.6570393147375357, + "grad_norm": 2.0101583003997803, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7107269167900085, + "num_tokens": 154936292.0, + "step": 5983 + }, + { + "epoch": 0.6571491324401494, + "grad_norm": 1.9836045503616333, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6843628883361816, + "num_tokens": 154970211.0, + "step": 5984 + }, + { + "epoch": 0.657258950142763, + "grad_norm": 2.5038435459136963, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7199253439903259, + "num_tokens": 154989482.0, + "step": 5985 + }, + { + "epoch": 0.6573687678453767, + "grad_norm": 2.117177724838257, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7068979740142822, + "num_tokens": 155017921.0, + "step": 5986 + }, + { + "epoch": 0.6574785855479903, + "grad_norm": 2.2993783950805664, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6916966438293457, + "num_tokens": 155042883.0, + "step": 5987 + }, + { + "epoch": 0.657588403250604, + "grad_norm": 2.1178221702575684, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6934360265731812, + "num_tokens": 155071816.0, + "step": 5988 + }, + { + "epoch": 0.6576982209532176, + "grad_norm": 2.237309217453003, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.695514440536499, + "num_tokens": 155098747.0, + "step": 5989 + }, + { + "epoch": 0.6578080386558314, + "grad_norm": 2.506037950515747, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7204287052154541, + "num_tokens": 155119942.0, + "step": 5990 + }, + { + "epoch": 0.657917856358445, + "grad_norm": 2.106158494949341, + "learning_rate": 1e-06, + "loss": 1.0503, + "mean_token_accuracy": 0.6839486956596375, + "num_tokens": 155146995.0, + "step": 5991 + }, + { + "epoch": 0.6580276740610587, + "grad_norm": 1.9833853244781494, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6916443705558777, + "num_tokens": 155179232.0, + "step": 5992 + }, + { + "epoch": 0.6581374917636723, + "grad_norm": 2.308086633682251, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.700892984867096, + "num_tokens": 155202130.0, + "step": 5993 + }, + { + "epoch": 0.658247309466286, + "grad_norm": 2.268376588821411, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6895506978034973, + "num_tokens": 155227196.0, + "step": 5994 + }, + { + "epoch": 0.6583571271688996, + "grad_norm": 2.227912187576294, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6893442869186401, + "num_tokens": 155254350.0, + "step": 5995 + }, + { + "epoch": 0.6584669448715132, + "grad_norm": 2.4457037448883057, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7076511979103088, + "num_tokens": 155274055.0, + "step": 5996 + }, + { + "epoch": 0.6585767625741269, + "grad_norm": 2.143280267715454, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.705441951751709, + "num_tokens": 155299575.0, + "step": 5997 + }, + { + "epoch": 0.6586865802767407, + "grad_norm": 2.458576202392578, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7059223651885986, + "num_tokens": 155322020.0, + "step": 5998 + }, + { + "epoch": 0.6587963979793543, + "grad_norm": 2.025155544281006, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6940646171569824, + "num_tokens": 155351256.0, + "step": 5999 + }, + { + "epoch": 0.6589062156819679, + "grad_norm": 2.091977596282959, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.689171314239502, + "num_tokens": 155380192.0, + "step": 6000 + }, + { + "epoch": 0.6590160333845816, + "grad_norm": 2.1927266120910645, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7002047300338745, + "num_tokens": 155404549.0, + "step": 6001 + }, + { + "epoch": 0.6591258510871952, + "grad_norm": 2.081838369369507, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7010776400566101, + "num_tokens": 155432136.0, + "step": 6002 + }, + { + "epoch": 0.6592356687898089, + "grad_norm": 2.0334904193878174, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7003078460693359, + "num_tokens": 155462102.0, + "step": 6003 + }, + { + "epoch": 0.6593454864924225, + "grad_norm": 2.212014675140381, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7045574188232422, + "num_tokens": 155486045.0, + "step": 6004 + }, + { + "epoch": 0.6594553041950363, + "grad_norm": 2.2944788932800293, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.7001059055328369, + "num_tokens": 155510272.0, + "step": 6005 + }, + { + "epoch": 0.6595651218976499, + "grad_norm": 1.9637371301651, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7240476012229919, + "num_tokens": 155541765.0, + "step": 6006 + }, + { + "epoch": 0.6596749396002636, + "grad_norm": 1.901166558265686, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7027299404144287, + "num_tokens": 155574579.0, + "step": 6007 + }, + { + "epoch": 0.6597847573028772, + "grad_norm": 2.0151679515838623, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6931016445159912, + "num_tokens": 155603237.0, + "step": 6008 + }, + { + "epoch": 0.6598945750054909, + "grad_norm": 2.4907772541046143, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7081156969070435, + "num_tokens": 155623349.0, + "step": 6009 + }, + { + "epoch": 0.6600043927081045, + "grad_norm": 2.181365966796875, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6851093769073486, + "num_tokens": 155651660.0, + "step": 6010 + }, + { + "epoch": 0.6601142104107182, + "grad_norm": 2.1956489086151123, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.703514575958252, + "num_tokens": 155676112.0, + "step": 6011 + }, + { + "epoch": 0.6602240281133319, + "grad_norm": 1.8773316144943237, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7214705348014832, + "num_tokens": 155706650.0, + "step": 6012 + }, + { + "epoch": 0.6603338458159456, + "grad_norm": 1.8187675476074219, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7064954042434692, + "num_tokens": 155742884.0, + "step": 6013 + }, + { + "epoch": 0.6604436635185592, + "grad_norm": 2.0385968685150146, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7285712957382202, + "num_tokens": 155769598.0, + "step": 6014 + }, + { + "epoch": 0.6605534812211729, + "grad_norm": 2.260870933532715, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6970879435539246, + "num_tokens": 155792342.0, + "step": 6015 + }, + { + "epoch": 0.6606632989237865, + "grad_norm": 2.0412440299987793, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7045369148254395, + "num_tokens": 155820917.0, + "step": 6016 + }, + { + "epoch": 0.6607731166264001, + "grad_norm": 2.1298134326934814, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7160966396331787, + "num_tokens": 155846192.0, + "step": 6017 + }, + { + "epoch": 0.6608829343290138, + "grad_norm": 2.2962582111358643, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6926001310348511, + "num_tokens": 155869072.0, + "step": 6018 + }, + { + "epoch": 0.6609927520316276, + "grad_norm": 2.141719102859497, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6928011775016785, + "num_tokens": 155898142.0, + "step": 6019 + }, + { + "epoch": 0.6611025697342412, + "grad_norm": 2.0936903953552246, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7206122279167175, + "num_tokens": 155925699.0, + "step": 6020 + }, + { + "epoch": 0.6612123874368548, + "grad_norm": 2.2680535316467285, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6930941939353943, + "num_tokens": 155950313.0, + "step": 6021 + }, + { + "epoch": 0.6613222051394685, + "grad_norm": 2.0484087467193604, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7180569767951965, + "num_tokens": 155977689.0, + "step": 6022 + }, + { + "epoch": 0.6614320228420821, + "grad_norm": 2.1775424480438232, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7061901092529297, + "num_tokens": 156003254.0, + "step": 6023 + }, + { + "epoch": 0.6615418405446958, + "grad_norm": 2.2078006267547607, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7206872701644897, + "num_tokens": 156027004.0, + "step": 6024 + }, + { + "epoch": 0.6616516582473094, + "grad_norm": 2.08172345161438, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6830441355705261, + "num_tokens": 156057555.0, + "step": 6025 + }, + { + "epoch": 0.6617614759499231, + "grad_norm": 2.1715586185455322, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6928046941757202, + "num_tokens": 156083938.0, + "step": 6026 + }, + { + "epoch": 0.6618712936525368, + "grad_norm": 1.9655659198760986, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7071245908737183, + "num_tokens": 156115291.0, + "step": 6027 + }, + { + "epoch": 0.6619811113551505, + "grad_norm": 2.0273990631103516, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7184250354766846, + "num_tokens": 156141710.0, + "step": 6028 + }, + { + "epoch": 0.6620909290577641, + "grad_norm": 2.1195261478424072, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.698223888874054, + "num_tokens": 156167447.0, + "step": 6029 + }, + { + "epoch": 0.6622007467603778, + "grad_norm": 2.0230982303619385, + "learning_rate": 1e-06, + "loss": 1.0655, + "mean_token_accuracy": 0.682522177696228, + "num_tokens": 156197415.0, + "step": 6030 + }, + { + "epoch": 0.6623105644629914, + "grad_norm": 1.740435242652893, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6787147521972656, + "num_tokens": 156235692.0, + "step": 6031 + }, + { + "epoch": 0.6624203821656051, + "grad_norm": 2.0640132427215576, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7160260677337646, + "num_tokens": 156263420.0, + "step": 6032 + }, + { + "epoch": 0.6625301998682187, + "grad_norm": 2.309015989303589, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.723219633102417, + "num_tokens": 156285855.0, + "step": 6033 + }, + { + "epoch": 0.6626400175708325, + "grad_norm": 2.05558705329895, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7026189565658569, + "num_tokens": 156315182.0, + "step": 6034 + }, + { + "epoch": 0.6627498352734461, + "grad_norm": 2.167395830154419, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7163194417953491, + "num_tokens": 156340739.0, + "step": 6035 + }, + { + "epoch": 0.6628596529760598, + "grad_norm": 2.1046664714813232, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6984862089157104, + "num_tokens": 156367369.0, + "step": 6036 + }, + { + "epoch": 0.6629694706786734, + "grad_norm": 2.1972649097442627, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6945859789848328, + "num_tokens": 156393412.0, + "step": 6037 + }, + { + "epoch": 0.663079288381287, + "grad_norm": 2.223299264907837, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7109884023666382, + "num_tokens": 156419102.0, + "step": 6038 + }, + { + "epoch": 0.6631891060839007, + "grad_norm": 2.0887811183929443, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6940226554870605, + "num_tokens": 156446862.0, + "step": 6039 + }, + { + "epoch": 0.6632989237865143, + "grad_norm": 2.2196860313415527, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7077206373214722, + "num_tokens": 156471500.0, + "step": 6040 + }, + { + "epoch": 0.6634087414891281, + "grad_norm": 2.1646652221679688, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7085881233215332, + "num_tokens": 156497443.0, + "step": 6041 + }, + { + "epoch": 0.6635185591917417, + "grad_norm": 1.9989264011383057, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6919850707054138, + "num_tokens": 156527999.0, + "step": 6042 + }, + { + "epoch": 0.6636283768943554, + "grad_norm": 2.2780890464782715, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7001366019248962, + "num_tokens": 156551836.0, + "step": 6043 + }, + { + "epoch": 0.663738194596969, + "grad_norm": 2.1053125858306885, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6862161159515381, + "num_tokens": 156580317.0, + "step": 6044 + }, + { + "epoch": 0.6638480122995827, + "grad_norm": 2.3063442707061768, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7164501547813416, + "num_tokens": 156603323.0, + "step": 6045 + }, + { + "epoch": 0.6639578300021963, + "grad_norm": 2.0868537425994873, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7044579982757568, + "num_tokens": 156630293.0, + "step": 6046 + }, + { + "epoch": 0.66406764770481, + "grad_norm": 2.0666916370391846, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6907994151115417, + "num_tokens": 156658155.0, + "step": 6047 + }, + { + "epoch": 0.6641774654074237, + "grad_norm": 2.112002372741699, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6969586610794067, + "num_tokens": 156685519.0, + "step": 6048 + }, + { + "epoch": 0.6642872831100374, + "grad_norm": 2.0719363689422607, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7150366306304932, + "num_tokens": 156712257.0, + "step": 6049 + }, + { + "epoch": 0.664397100812651, + "grad_norm": 2.2549901008605957, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7028249502182007, + "num_tokens": 156735594.0, + "step": 6050 + }, + { + "epoch": 0.6645069185152647, + "grad_norm": 2.1692471504211426, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.6988157629966736, + "num_tokens": 156760714.0, + "step": 6051 + }, + { + "epoch": 0.6646167362178783, + "grad_norm": 1.992823600769043, + "learning_rate": 1e-06, + "loss": 1.0948, + "mean_token_accuracy": 0.6794935464859009, + "num_tokens": 156791882.0, + "step": 6052 + }, + { + "epoch": 0.664726553920492, + "grad_norm": 2.2625086307525635, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.6978546380996704, + "num_tokens": 156816170.0, + "step": 6053 + }, + { + "epoch": 0.6648363716231056, + "grad_norm": 2.2243402004241943, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.734105110168457, + "num_tokens": 156838617.0, + "step": 6054 + }, + { + "epoch": 0.6649461893257194, + "grad_norm": 2.0283963680267334, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6881231665611267, + "num_tokens": 156870103.0, + "step": 6055 + }, + { + "epoch": 0.665056007028333, + "grad_norm": 2.239306926727295, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7083200812339783, + "num_tokens": 156895180.0, + "step": 6056 + }, + { + "epoch": 0.6651658247309467, + "grad_norm": 2.370950698852539, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7063915729522705, + "num_tokens": 156919471.0, + "step": 6057 + }, + { + "epoch": 0.6652756424335603, + "grad_norm": 2.133589267730713, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.7072689533233643, + "num_tokens": 156947861.0, + "step": 6058 + }, + { + "epoch": 0.665385460136174, + "grad_norm": 2.5137362480163574, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7227991223335266, + "num_tokens": 156966379.0, + "step": 6059 + }, + { + "epoch": 0.6654952778387876, + "grad_norm": 2.2669553756713867, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7401186227798462, + "num_tokens": 156988824.0, + "step": 6060 + }, + { + "epoch": 0.6656050955414012, + "grad_norm": 2.27490496635437, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6951690912246704, + "num_tokens": 157012835.0, + "step": 6061 + }, + { + "epoch": 0.6657149132440149, + "grad_norm": 2.100206136703491, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7139980792999268, + "num_tokens": 157041090.0, + "step": 6062 + }, + { + "epoch": 0.6658247309466286, + "grad_norm": 2.1737983226776123, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6904729008674622, + "num_tokens": 157069696.0, + "step": 6063 + }, + { + "epoch": 0.6659345486492423, + "grad_norm": 2.492698907852173, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7387055158615112, + "num_tokens": 157088566.0, + "step": 6064 + }, + { + "epoch": 0.6660443663518559, + "grad_norm": 2.1120245456695557, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.710691511631012, + "num_tokens": 157114916.0, + "step": 6065 + }, + { + "epoch": 0.6661541840544696, + "grad_norm": 2.0904839038848877, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6797314882278442, + "num_tokens": 157146212.0, + "step": 6066 + }, + { + "epoch": 0.6662640017570832, + "grad_norm": 2.416275978088379, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7024465203285217, + "num_tokens": 157168951.0, + "step": 6067 + }, + { + "epoch": 0.6663738194596969, + "grad_norm": 2.498260021209717, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7118229866027832, + "num_tokens": 157189670.0, + "step": 6068 + }, + { + "epoch": 0.6664836371623105, + "grad_norm": 2.4635066986083984, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7099475264549255, + "num_tokens": 157214174.0, + "step": 6069 + }, + { + "epoch": 0.6665934548649243, + "grad_norm": 2.0863771438598633, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6866126656532288, + "num_tokens": 157242797.0, + "step": 6070 + }, + { + "epoch": 0.6667032725675379, + "grad_norm": 2.5766870975494385, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7064536213874817, + "num_tokens": 157262190.0, + "step": 6071 + }, + { + "epoch": 0.6668130902701516, + "grad_norm": 2.340954303741455, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7162423729896545, + "num_tokens": 157283822.0, + "step": 6072 + }, + { + "epoch": 0.6669229079727652, + "grad_norm": 2.2000179290771484, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6960253119468689, + "num_tokens": 157308337.0, + "step": 6073 + }, + { + "epoch": 0.6670327256753789, + "grad_norm": 2.1107726097106934, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.711701512336731, + "num_tokens": 157332784.0, + "step": 6074 + }, + { + "epoch": 0.6671425433779925, + "grad_norm": 2.1444413661956787, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6890243887901306, + "num_tokens": 157360868.0, + "step": 6075 + }, + { + "epoch": 0.6672523610806061, + "grad_norm": 2.4314544200897217, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6981010437011719, + "num_tokens": 157382569.0, + "step": 6076 + }, + { + "epoch": 0.6673621787832199, + "grad_norm": 2.1060168743133545, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6934108734130859, + "num_tokens": 157409460.0, + "step": 6077 + }, + { + "epoch": 0.6674719964858336, + "grad_norm": 2.01454496383667, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6889716982841492, + "num_tokens": 157440052.0, + "step": 6078 + }, + { + "epoch": 0.6675818141884472, + "grad_norm": 2.2899646759033203, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6959017515182495, + "num_tokens": 157464038.0, + "step": 6079 + }, + { + "epoch": 0.6676916318910608, + "grad_norm": 2.0001869201660156, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6975760459899902, + "num_tokens": 157492927.0, + "step": 6080 + }, + { + "epoch": 0.6678014495936745, + "grad_norm": 1.9284677505493164, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.716902494430542, + "num_tokens": 157522518.0, + "step": 6081 + }, + { + "epoch": 0.6679112672962881, + "grad_norm": 2.230815887451172, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7085398435592651, + "num_tokens": 157547170.0, + "step": 6082 + }, + { + "epoch": 0.6680210849989018, + "grad_norm": 2.146332025527954, + "learning_rate": 1e-06, + "loss": 1.1325, + "mean_token_accuracy": 0.671216607093811, + "num_tokens": 157575044.0, + "step": 6083 + }, + { + "epoch": 0.6681309027015155, + "grad_norm": 1.749819278717041, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6870819330215454, + "num_tokens": 157614962.0, + "step": 6084 + }, + { + "epoch": 0.6682407204041292, + "grad_norm": 2.2664120197296143, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.704241156578064, + "num_tokens": 157638797.0, + "step": 6085 + }, + { + "epoch": 0.6683505381067428, + "grad_norm": 2.172877550125122, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7041630148887634, + "num_tokens": 157663629.0, + "step": 6086 + }, + { + "epoch": 0.6684603558093565, + "grad_norm": 2.317610025405884, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7235407829284668, + "num_tokens": 157686851.0, + "step": 6087 + }, + { + "epoch": 0.6685701735119701, + "grad_norm": 2.451587677001953, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7246384024620056, + "num_tokens": 157706324.0, + "step": 6088 + }, + { + "epoch": 0.6686799912145838, + "grad_norm": 2.3736634254455566, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6956847906112671, + "num_tokens": 157730442.0, + "step": 6089 + }, + { + "epoch": 0.6687898089171974, + "grad_norm": 2.3030846118927, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.720780074596405, + "num_tokens": 157753769.0, + "step": 6090 + }, + { + "epoch": 0.6688996266198111, + "grad_norm": 2.02978777885437, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6961020827293396, + "num_tokens": 157783437.0, + "step": 6091 + }, + { + "epoch": 0.6690094443224248, + "grad_norm": 2.049208879470825, + "learning_rate": 1e-06, + "loss": 1.0713, + "mean_token_accuracy": 0.6736242175102234, + "num_tokens": 157813602.0, + "step": 6092 + }, + { + "epoch": 0.6691192620250385, + "grad_norm": 2.02665114402771, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7122419476509094, + "num_tokens": 157841665.0, + "step": 6093 + }, + { + "epoch": 0.6692290797276521, + "grad_norm": 2.3641164302825928, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7060911655426025, + "num_tokens": 157864011.0, + "step": 6094 + }, + { + "epoch": 0.6693388974302658, + "grad_norm": 1.9998468160629272, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7095059752464294, + "num_tokens": 157893302.0, + "step": 6095 + }, + { + "epoch": 0.6694487151328794, + "grad_norm": 2.24200439453125, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7220728397369385, + "num_tokens": 157917252.0, + "step": 6096 + }, + { + "epoch": 0.669558532835493, + "grad_norm": 1.9969265460968018, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.706918478012085, + "num_tokens": 157946229.0, + "step": 6097 + }, + { + "epoch": 0.6696683505381067, + "grad_norm": 2.2539660930633545, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6906064748764038, + "num_tokens": 157971749.0, + "step": 6098 + }, + { + "epoch": 0.6697781682407205, + "grad_norm": 1.929104208946228, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.682155966758728, + "num_tokens": 158004797.0, + "step": 6099 + }, + { + "epoch": 0.6698879859433341, + "grad_norm": 2.1283771991729736, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7024909853935242, + "num_tokens": 158032561.0, + "step": 6100 + }, + { + "epoch": 0.6699978036459477, + "grad_norm": 2.3254241943359375, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7281280755996704, + "num_tokens": 158054892.0, + "step": 6101 + }, + { + "epoch": 0.6701076213485614, + "grad_norm": 2.1357343196868896, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6886127591133118, + "num_tokens": 158083531.0, + "step": 6102 + }, + { + "epoch": 0.670217439051175, + "grad_norm": 2.4214835166931152, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6942948698997498, + "num_tokens": 158105142.0, + "step": 6103 + }, + { + "epoch": 0.6703272567537887, + "grad_norm": 2.60322642326355, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7075828313827515, + "num_tokens": 158124184.0, + "step": 6104 + }, + { + "epoch": 0.6704370744564023, + "grad_norm": 2.4605095386505127, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.719536304473877, + "num_tokens": 158144656.0, + "step": 6105 + }, + { + "epoch": 0.6705468921590161, + "grad_norm": 2.3371334075927734, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7060428261756897, + "num_tokens": 158168309.0, + "step": 6106 + }, + { + "epoch": 0.6706567098616297, + "grad_norm": 2.383256196975708, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7202350497245789, + "num_tokens": 158189567.0, + "step": 6107 + }, + { + "epoch": 0.6707665275642434, + "grad_norm": 2.380701780319214, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7090585827827454, + "num_tokens": 158210476.0, + "step": 6108 + }, + { + "epoch": 0.670876345266857, + "grad_norm": 2.098006010055542, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7210864424705505, + "num_tokens": 158237155.0, + "step": 6109 + }, + { + "epoch": 0.6709861629694707, + "grad_norm": 2.397061824798584, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6893277764320374, + "num_tokens": 158260792.0, + "step": 6110 + }, + { + "epoch": 0.6710959806720843, + "grad_norm": 2.4800775051116943, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7205101251602173, + "num_tokens": 158280874.0, + "step": 6111 + }, + { + "epoch": 0.671205798374698, + "grad_norm": 2.2612812519073486, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7195701599121094, + "num_tokens": 158304659.0, + "step": 6112 + }, + { + "epoch": 0.6713156160773117, + "grad_norm": 2.070413589477539, + "learning_rate": 1e-06, + "loss": 1.0561, + "mean_token_accuracy": 0.6859481930732727, + "num_tokens": 158334472.0, + "step": 6113 + }, + { + "epoch": 0.6714254337799254, + "grad_norm": 1.995740294456482, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6880897879600525, + "num_tokens": 158367217.0, + "step": 6114 + }, + { + "epoch": 0.671535251482539, + "grad_norm": 1.9878190755844116, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.696425199508667, + "num_tokens": 158396304.0, + "step": 6115 + }, + { + "epoch": 0.6716450691851527, + "grad_norm": 2.1253557205200195, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6926629543304443, + "num_tokens": 158422496.0, + "step": 6116 + }, + { + "epoch": 0.6717548868877663, + "grad_norm": 2.350468158721924, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7056670188903809, + "num_tokens": 158445506.0, + "step": 6117 + }, + { + "epoch": 0.67186470459038, + "grad_norm": 2.0361196994781494, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.71369469165802, + "num_tokens": 158473721.0, + "step": 6118 + }, + { + "epoch": 0.6719745222929936, + "grad_norm": 2.3093154430389404, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7162876129150391, + "num_tokens": 158496941.0, + "step": 6119 + }, + { + "epoch": 0.6720843399956072, + "grad_norm": 2.0477709770202637, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7216858267784119, + "num_tokens": 158525505.0, + "step": 6120 + }, + { + "epoch": 0.672194157698221, + "grad_norm": 2.253774881362915, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7049354314804077, + "num_tokens": 158548342.0, + "step": 6121 + }, + { + "epoch": 0.6723039754008346, + "grad_norm": 2.0230154991149902, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6944847106933594, + "num_tokens": 158577693.0, + "step": 6122 + }, + { + "epoch": 0.6724137931034483, + "grad_norm": 2.1869547367095947, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7174314856529236, + "num_tokens": 158602480.0, + "step": 6123 + }, + { + "epoch": 0.6725236108060619, + "grad_norm": 2.0382790565490723, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7057663798332214, + "num_tokens": 158631339.0, + "step": 6124 + }, + { + "epoch": 0.6726334285086756, + "grad_norm": 2.48095440864563, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7094841003417969, + "num_tokens": 158651126.0, + "step": 6125 + }, + { + "epoch": 0.6727432462112892, + "grad_norm": 2.321298122406006, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.719680666923523, + "num_tokens": 158674262.0, + "step": 6126 + }, + { + "epoch": 0.6728530639139029, + "grad_norm": 2.4351754188537598, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.6965124607086182, + "num_tokens": 158695067.0, + "step": 6127 + }, + { + "epoch": 0.6729628816165166, + "grad_norm": 2.457059621810913, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7308942675590515, + "num_tokens": 158716763.0, + "step": 6128 + }, + { + "epoch": 0.6730726993191303, + "grad_norm": 2.1090383529663086, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7046484351158142, + "num_tokens": 158743441.0, + "step": 6129 + }, + { + "epoch": 0.6731825170217439, + "grad_norm": 2.209030866622925, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7130539417266846, + "num_tokens": 158768551.0, + "step": 6130 + }, + { + "epoch": 0.6732923347243576, + "grad_norm": 2.259026288986206, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7033421993255615, + "num_tokens": 158793558.0, + "step": 6131 + }, + { + "epoch": 0.6734021524269712, + "grad_norm": 2.1215763092041016, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7304109334945679, + "num_tokens": 158818430.0, + "step": 6132 + }, + { + "epoch": 0.6735119701295849, + "grad_norm": 2.291292667388916, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6929007172584534, + "num_tokens": 158843138.0, + "step": 6133 + }, + { + "epoch": 0.6736217878321985, + "grad_norm": 2.075324773788452, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.7050526738166809, + "num_tokens": 158874477.0, + "step": 6134 + }, + { + "epoch": 0.6737316055348123, + "grad_norm": 2.2241175174713135, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7106239795684814, + "num_tokens": 158899004.0, + "step": 6135 + }, + { + "epoch": 0.6738414232374259, + "grad_norm": 2.2003555297851562, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7053805589675903, + "num_tokens": 158924939.0, + "step": 6136 + }, + { + "epoch": 0.6739512409400396, + "grad_norm": 2.1774332523345947, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7179572582244873, + "num_tokens": 158950178.0, + "step": 6137 + }, + { + "epoch": 0.6740610586426532, + "grad_norm": 1.9974669218063354, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7007933855056763, + "num_tokens": 158979755.0, + "step": 6138 + }, + { + "epoch": 0.6741708763452668, + "grad_norm": 2.1188440322875977, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7083884477615356, + "num_tokens": 159007944.0, + "step": 6139 + }, + { + "epoch": 0.6742806940478805, + "grad_norm": 1.975934386253357, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7190084457397461, + "num_tokens": 159037619.0, + "step": 6140 + }, + { + "epoch": 0.6743905117504941, + "grad_norm": 2.128934621810913, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7149314284324646, + "num_tokens": 159065207.0, + "step": 6141 + }, + { + "epoch": 0.6745003294531079, + "grad_norm": 2.169774055480957, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6999799013137817, + "num_tokens": 159091637.0, + "step": 6142 + }, + { + "epoch": 0.6746101471557215, + "grad_norm": 2.3897104263305664, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7089561223983765, + "num_tokens": 159114372.0, + "step": 6143 + }, + { + "epoch": 0.6747199648583352, + "grad_norm": 2.143695592880249, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6944969296455383, + "num_tokens": 159141680.0, + "step": 6144 + }, + { + "epoch": 0.6748297825609488, + "grad_norm": 2.553753137588501, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7158601880073547, + "num_tokens": 159162816.0, + "step": 6145 + }, + { + "epoch": 0.6749396002635625, + "grad_norm": 2.172740936279297, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6974745988845825, + "num_tokens": 159188403.0, + "step": 6146 + }, + { + "epoch": 0.6750494179661761, + "grad_norm": 2.0452115535736084, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6964138746261597, + "num_tokens": 159218711.0, + "step": 6147 + }, + { + "epoch": 0.6751592356687898, + "grad_norm": 2.120863914489746, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.6875483989715576, + "num_tokens": 159246877.0, + "step": 6148 + }, + { + "epoch": 0.6752690533714034, + "grad_norm": 2.0515010356903076, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7106037139892578, + "num_tokens": 159275042.0, + "step": 6149 + }, + { + "epoch": 0.6753788710740172, + "grad_norm": 1.8585638999938965, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7152048349380493, + "num_tokens": 159307958.0, + "step": 6150 + }, + { + "epoch": 0.6754886887766308, + "grad_norm": 2.414862632751465, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7052980661392212, + "num_tokens": 159330280.0, + "step": 6151 + }, + { + "epoch": 0.6755985064792445, + "grad_norm": 2.068603515625, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.7010412812232971, + "num_tokens": 159358692.0, + "step": 6152 + }, + { + "epoch": 0.6757083241818581, + "grad_norm": 1.9679715633392334, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7088724374771118, + "num_tokens": 159387486.0, + "step": 6153 + }, + { + "epoch": 0.6758181418844718, + "grad_norm": 2.250722646713257, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7099170684814453, + "num_tokens": 159411417.0, + "step": 6154 + }, + { + "epoch": 0.6759279595870854, + "grad_norm": 1.9988555908203125, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7112256288528442, + "num_tokens": 159442444.0, + "step": 6155 + }, + { + "epoch": 0.676037777289699, + "grad_norm": 2.2349720001220703, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.698878288269043, + "num_tokens": 159466545.0, + "step": 6156 + }, + { + "epoch": 0.6761475949923128, + "grad_norm": 2.240664005279541, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.69870525598526, + "num_tokens": 159491951.0, + "step": 6157 + }, + { + "epoch": 0.6762574126949265, + "grad_norm": 1.9406802654266357, + "learning_rate": 1e-06, + "loss": 1.1055, + "mean_token_accuracy": 0.6741135120391846, + "num_tokens": 159524107.0, + "step": 6158 + }, + { + "epoch": 0.6763672303975401, + "grad_norm": 1.750864863395691, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7121471166610718, + "num_tokens": 159560268.0, + "step": 6159 + }, + { + "epoch": 0.6764770481001537, + "grad_norm": 2.2687981128692627, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7406693696975708, + "num_tokens": 159582247.0, + "step": 6160 + }, + { + "epoch": 0.6765868658027674, + "grad_norm": 2.2029662132263184, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7142665386199951, + "num_tokens": 159608502.0, + "step": 6161 + }, + { + "epoch": 0.676696683505381, + "grad_norm": 2.4232914447784424, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.722985029220581, + "num_tokens": 159629373.0, + "step": 6162 + }, + { + "epoch": 0.6768065012079947, + "grad_norm": 2.265197277069092, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6927573084831238, + "num_tokens": 159654888.0, + "step": 6163 + }, + { + "epoch": 0.6769163189106084, + "grad_norm": 2.207749843597412, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7138721942901611, + "num_tokens": 159678909.0, + "step": 6164 + }, + { + "epoch": 0.6770261366132221, + "grad_norm": 2.5663259029388428, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7172209620475769, + "num_tokens": 159699296.0, + "step": 6165 + }, + { + "epoch": 0.6771359543158357, + "grad_norm": 2.0639045238494873, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.692730724811554, + "num_tokens": 159728051.0, + "step": 6166 + }, + { + "epoch": 0.6772457720184494, + "grad_norm": 2.2104055881500244, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7176029682159424, + "num_tokens": 159752609.0, + "step": 6167 + }, + { + "epoch": 0.677355589721063, + "grad_norm": 2.4991462230682373, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6980583071708679, + "num_tokens": 159775222.0, + "step": 6168 + }, + { + "epoch": 0.6774654074236767, + "grad_norm": 2.457048177719116, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6974837779998779, + "num_tokens": 159795594.0, + "step": 6169 + }, + { + "epoch": 0.6775752251262903, + "grad_norm": 2.203345537185669, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.7007492780685425, + "num_tokens": 159821063.0, + "step": 6170 + }, + { + "epoch": 0.6776850428289041, + "grad_norm": 1.965315341949463, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7122278213500977, + "num_tokens": 159850148.0, + "step": 6171 + }, + { + "epoch": 0.6777948605315177, + "grad_norm": 2.2581114768981934, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.713881254196167, + "num_tokens": 159875384.0, + "step": 6172 + }, + { + "epoch": 0.6779046782341314, + "grad_norm": 2.214937686920166, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7094515562057495, + "num_tokens": 159898946.0, + "step": 6173 + }, + { + "epoch": 0.678014495936745, + "grad_norm": 2.2540552616119385, + "learning_rate": 1e-06, + "loss": 1.0514, + "mean_token_accuracy": 0.6826120615005493, + "num_tokens": 159925065.0, + "step": 6174 + }, + { + "epoch": 0.6781243136393587, + "grad_norm": 2.177889585494995, + "learning_rate": 1e-06, + "loss": 1.04, + "mean_token_accuracy": 0.6835532188415527, + "num_tokens": 159950968.0, + "step": 6175 + }, + { + "epoch": 0.6782341313419723, + "grad_norm": 2.3875837326049805, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6964660882949829, + "num_tokens": 159972968.0, + "step": 6176 + }, + { + "epoch": 0.678343949044586, + "grad_norm": 2.006702184677124, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.741558849811554, + "num_tokens": 160000824.0, + "step": 6177 + }, + { + "epoch": 0.6784537667471996, + "grad_norm": 2.2842957973480225, + "learning_rate": 1e-06, + "loss": 1.0416, + "mean_token_accuracy": 0.6832380294799805, + "num_tokens": 160026215.0, + "step": 6178 + }, + { + "epoch": 0.6785635844498134, + "grad_norm": 2.274761915206909, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7050339579582214, + "num_tokens": 160048907.0, + "step": 6179 + }, + { + "epoch": 0.678673402152427, + "grad_norm": 2.040318489074707, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.695906937122345, + "num_tokens": 160078943.0, + "step": 6180 + }, + { + "epoch": 0.6787832198550406, + "grad_norm": 2.1840460300445557, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7108674049377441, + "num_tokens": 160105582.0, + "step": 6181 + }, + { + "epoch": 0.6788930375576543, + "grad_norm": 2.2956268787384033, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7078927159309387, + "num_tokens": 160128859.0, + "step": 6182 + }, + { + "epoch": 0.6790028552602679, + "grad_norm": 1.9607009887695312, + "learning_rate": 1e-06, + "loss": 1.1019, + "mean_token_accuracy": 0.6717057228088379, + "num_tokens": 160161771.0, + "step": 6183 + }, + { + "epoch": 0.6791126729628816, + "grad_norm": 2.41664981842041, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7115525007247925, + "num_tokens": 160185315.0, + "step": 6184 + }, + { + "epoch": 0.6792224906654952, + "grad_norm": 2.1042051315307617, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7332003116607666, + "num_tokens": 160210093.0, + "step": 6185 + }, + { + "epoch": 0.679332308368109, + "grad_norm": 2.2551193237304688, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.698715329170227, + "num_tokens": 160235947.0, + "step": 6186 + }, + { + "epoch": 0.6794421260707226, + "grad_norm": 2.087196111679077, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7150168418884277, + "num_tokens": 160262839.0, + "step": 6187 + }, + { + "epoch": 0.6795519437733363, + "grad_norm": 2.104682207107544, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.703089714050293, + "num_tokens": 160291066.0, + "step": 6188 + }, + { + "epoch": 0.6796617614759499, + "grad_norm": 2.2225308418273926, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.697136402130127, + "num_tokens": 160316101.0, + "step": 6189 + }, + { + "epoch": 0.6797715791785636, + "grad_norm": 2.0497255325317383, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7209159731864929, + "num_tokens": 160342531.0, + "step": 6190 + }, + { + "epoch": 0.6798813968811772, + "grad_norm": 2.5533111095428467, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7084559202194214, + "num_tokens": 160361907.0, + "step": 6191 + }, + { + "epoch": 0.6799912145837909, + "grad_norm": 2.2121598720550537, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7293447256088257, + "num_tokens": 160385012.0, + "step": 6192 + }, + { + "epoch": 0.6801010322864046, + "grad_norm": 1.9475194215774536, + "learning_rate": 1e-06, + "loss": 1.0762, + "mean_token_accuracy": 0.6771926879882812, + "num_tokens": 160418417.0, + "step": 6193 + }, + { + "epoch": 0.6802108499890183, + "grad_norm": 2.605059862136841, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7209884524345398, + "num_tokens": 160436585.0, + "step": 6194 + }, + { + "epoch": 0.6803206676916319, + "grad_norm": 2.0396599769592285, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6968023777008057, + "num_tokens": 160465543.0, + "step": 6195 + }, + { + "epoch": 0.6804304853942456, + "grad_norm": 2.1609036922454834, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6902773380279541, + "num_tokens": 160494434.0, + "step": 6196 + }, + { + "epoch": 0.6805403030968592, + "grad_norm": 1.9850512742996216, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6993451118469238, + "num_tokens": 160522554.0, + "step": 6197 + }, + { + "epoch": 0.6806501207994728, + "grad_norm": 2.3463406562805176, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7040600776672363, + "num_tokens": 160545157.0, + "step": 6198 + }, + { + "epoch": 0.6807599385020865, + "grad_norm": 1.9939649105072021, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.696280837059021, + "num_tokens": 160573827.0, + "step": 6199 + }, + { + "epoch": 0.6808697562047002, + "grad_norm": 2.5731911659240723, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7231687903404236, + "num_tokens": 160592697.0, + "step": 6200 + }, + { + "epoch": 0.6809795739073139, + "grad_norm": 2.0346319675445557, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7074611186981201, + "num_tokens": 160621152.0, + "step": 6201 + }, + { + "epoch": 0.6810893916099275, + "grad_norm": 2.0680007934570312, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7066367864608765, + "num_tokens": 160649670.0, + "step": 6202 + }, + { + "epoch": 0.6811992093125412, + "grad_norm": 2.4224064350128174, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7039897441864014, + "num_tokens": 160672004.0, + "step": 6203 + }, + { + "epoch": 0.6813090270151548, + "grad_norm": 2.013836622238159, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6925452947616577, + "num_tokens": 160701671.0, + "step": 6204 + }, + { + "epoch": 0.6814188447177685, + "grad_norm": 2.3621699810028076, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6931471824645996, + "num_tokens": 160724240.0, + "step": 6205 + }, + { + "epoch": 0.6815286624203821, + "grad_norm": 2.176737070083618, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6955841183662415, + "num_tokens": 160750043.0, + "step": 6206 + }, + { + "epoch": 0.6816384801229959, + "grad_norm": 2.282397747039795, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6982535719871521, + "num_tokens": 160775094.0, + "step": 6207 + }, + { + "epoch": 0.6817482978256095, + "grad_norm": 2.024756908416748, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7327154874801636, + "num_tokens": 160804642.0, + "step": 6208 + }, + { + "epoch": 0.6818581155282232, + "grad_norm": 2.278825283050537, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.713699460029602, + "num_tokens": 160828357.0, + "step": 6209 + }, + { + "epoch": 0.6819679332308368, + "grad_norm": 2.275873899459839, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7293566465377808, + "num_tokens": 160852946.0, + "step": 6210 + }, + { + "epoch": 0.6820777509334505, + "grad_norm": 2.1857306957244873, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6884092092514038, + "num_tokens": 160880009.0, + "step": 6211 + }, + { + "epoch": 0.6821875686360641, + "grad_norm": 1.9104450941085815, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7092373967170715, + "num_tokens": 160910735.0, + "step": 6212 + }, + { + "epoch": 0.6822973863386778, + "grad_norm": 2.0764448642730713, + "learning_rate": 1e-06, + "loss": 1.069, + "mean_token_accuracy": 0.6795978546142578, + "num_tokens": 160940187.0, + "step": 6213 + }, + { + "epoch": 0.6824072040412914, + "grad_norm": 2.04187273979187, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7174621224403381, + "num_tokens": 160967733.0, + "step": 6214 + }, + { + "epoch": 0.6825170217439052, + "grad_norm": 2.315183401107788, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7149615287780762, + "num_tokens": 160991485.0, + "step": 6215 + }, + { + "epoch": 0.6826268394465188, + "grad_norm": 2.088275671005249, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6964941024780273, + "num_tokens": 161022484.0, + "step": 6216 + }, + { + "epoch": 0.6827366571491325, + "grad_norm": 2.651778221130371, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7083169221878052, + "num_tokens": 161041677.0, + "step": 6217 + }, + { + "epoch": 0.6828464748517461, + "grad_norm": 1.9078152179718018, + "learning_rate": 1e-06, + "loss": 1.0948, + "mean_token_accuracy": 0.6800632476806641, + "num_tokens": 161078083.0, + "step": 6218 + }, + { + "epoch": 0.6829562925543597, + "grad_norm": 2.07321834564209, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7091259360313416, + "num_tokens": 161105164.0, + "step": 6219 + }, + { + "epoch": 0.6830661102569734, + "grad_norm": 2.0988965034484863, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6997802257537842, + "num_tokens": 161134797.0, + "step": 6220 + }, + { + "epoch": 0.683175927959587, + "grad_norm": 1.9611132144927979, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7133848667144775, + "num_tokens": 161163605.0, + "step": 6221 + }, + { + "epoch": 0.6832857456622008, + "grad_norm": 2.186258554458618, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7067569494247437, + "num_tokens": 161189308.0, + "step": 6222 + }, + { + "epoch": 0.6833955633648144, + "grad_norm": 2.3073465824127197, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7075387239456177, + "num_tokens": 161212337.0, + "step": 6223 + }, + { + "epoch": 0.6835053810674281, + "grad_norm": 2.325934648513794, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7099443674087524, + "num_tokens": 161234858.0, + "step": 6224 + }, + { + "epoch": 0.6836151987700417, + "grad_norm": 2.086784601211548, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7111507058143616, + "num_tokens": 161261948.0, + "step": 6225 + }, + { + "epoch": 0.6837250164726554, + "grad_norm": 2.188239336013794, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7112687230110168, + "num_tokens": 161285655.0, + "step": 6226 + }, + { + "epoch": 0.683834834175269, + "grad_norm": 2.069932460784912, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7075795531272888, + "num_tokens": 161313042.0, + "step": 6227 + }, + { + "epoch": 0.6839446518778827, + "grad_norm": 2.2798473834991455, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6999975442886353, + "num_tokens": 161339421.0, + "step": 6228 + }, + { + "epoch": 0.6840544695804964, + "grad_norm": 1.869829773902893, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.6870059967041016, + "num_tokens": 161374361.0, + "step": 6229 + }, + { + "epoch": 0.6841642872831101, + "grad_norm": 2.1318790912628174, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7066291570663452, + "num_tokens": 161400878.0, + "step": 6230 + }, + { + "epoch": 0.6842741049857237, + "grad_norm": 2.119473695755005, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7061628699302673, + "num_tokens": 161428464.0, + "step": 6231 + }, + { + "epoch": 0.6843839226883374, + "grad_norm": 2.29935884475708, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7319234609603882, + "num_tokens": 161450688.0, + "step": 6232 + }, + { + "epoch": 0.684493740390951, + "grad_norm": 2.3464133739471436, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.706211268901825, + "num_tokens": 161472646.0, + "step": 6233 + }, + { + "epoch": 0.6846035580935647, + "grad_norm": 2.35371470451355, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6970810890197754, + "num_tokens": 161495999.0, + "step": 6234 + }, + { + "epoch": 0.6847133757961783, + "grad_norm": 2.2273380756378174, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7210239768028259, + "num_tokens": 161518854.0, + "step": 6235 + }, + { + "epoch": 0.6848231934987921, + "grad_norm": 2.4001035690307617, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.700988233089447, + "num_tokens": 161541927.0, + "step": 6236 + }, + { + "epoch": 0.6849330112014057, + "grad_norm": 2.1600136756896973, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7070889472961426, + "num_tokens": 161568560.0, + "step": 6237 + }, + { + "epoch": 0.6850428289040194, + "grad_norm": 2.384234666824341, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7055386304855347, + "num_tokens": 161592384.0, + "step": 6238 + }, + { + "epoch": 0.685152646606633, + "grad_norm": 2.107362985610962, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6921418905258179, + "num_tokens": 161620897.0, + "step": 6239 + }, + { + "epoch": 0.6852624643092466, + "grad_norm": 2.2986087799072266, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.716899037361145, + "num_tokens": 161644189.0, + "step": 6240 + }, + { + "epoch": 0.6853722820118603, + "grad_norm": 2.2263312339782715, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6917625665664673, + "num_tokens": 161668350.0, + "step": 6241 + }, + { + "epoch": 0.6854820997144739, + "grad_norm": 2.389866352081299, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7221043109893799, + "num_tokens": 161691163.0, + "step": 6242 + }, + { + "epoch": 0.6855919174170876, + "grad_norm": 2.1775779724121094, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7205508351325989, + "num_tokens": 161715801.0, + "step": 6243 + }, + { + "epoch": 0.6857017351197013, + "grad_norm": 2.199352979660034, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6927879452705383, + "num_tokens": 161743236.0, + "step": 6244 + }, + { + "epoch": 0.685811552822315, + "grad_norm": 2.013303518295288, + "learning_rate": 1e-06, + "loss": 1.0881, + "mean_token_accuracy": 0.6763466596603394, + "num_tokens": 161772576.0, + "step": 6245 + }, + { + "epoch": 0.6859213705249286, + "grad_norm": 2.1627581119537354, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7089403867721558, + "num_tokens": 161798925.0, + "step": 6246 + }, + { + "epoch": 0.6860311882275423, + "grad_norm": 2.4951021671295166, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7124711275100708, + "num_tokens": 161820941.0, + "step": 6247 + }, + { + "epoch": 0.6861410059301559, + "grad_norm": 2.223682403564453, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6972061991691589, + "num_tokens": 161846081.0, + "step": 6248 + }, + { + "epoch": 0.6862508236327696, + "grad_norm": 2.073127031326294, + "learning_rate": 1e-06, + "loss": 1.1004, + "mean_token_accuracy": 0.6713640689849854, + "num_tokens": 161875326.0, + "step": 6249 + }, + { + "epoch": 0.6863606413353832, + "grad_norm": 2.1702706813812256, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7100542187690735, + "num_tokens": 161902576.0, + "step": 6250 + }, + { + "epoch": 0.686470459037997, + "grad_norm": 2.621539354324341, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7185897827148438, + "num_tokens": 161920422.0, + "step": 6251 + }, + { + "epoch": 0.6865802767406106, + "grad_norm": 2.2607483863830566, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7172362208366394, + "num_tokens": 161943729.0, + "step": 6252 + }, + { + "epoch": 0.6866900944432243, + "grad_norm": 2.118393898010254, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6955196857452393, + "num_tokens": 161970782.0, + "step": 6253 + }, + { + "epoch": 0.6867999121458379, + "grad_norm": 1.810607671737671, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6901966333389282, + "num_tokens": 162007033.0, + "step": 6254 + }, + { + "epoch": 0.6869097298484516, + "grad_norm": 2.471497058868408, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7133301496505737, + "num_tokens": 162027881.0, + "step": 6255 + }, + { + "epoch": 0.6870195475510652, + "grad_norm": 2.0234715938568115, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.7013236284255981, + "num_tokens": 162058289.0, + "step": 6256 + }, + { + "epoch": 0.6871293652536788, + "grad_norm": 2.394711971282959, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7279790639877319, + "num_tokens": 162079047.0, + "step": 6257 + }, + { + "epoch": 0.6872391829562926, + "grad_norm": 2.378236770629883, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7052116990089417, + "num_tokens": 162100921.0, + "step": 6258 + }, + { + "epoch": 0.6873490006589063, + "grad_norm": 2.2427401542663574, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7003009915351868, + "num_tokens": 162125491.0, + "step": 6259 + }, + { + "epoch": 0.6874588183615199, + "grad_norm": 2.2001845836639404, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6967028379440308, + "num_tokens": 162150785.0, + "step": 6260 + }, + { + "epoch": 0.6875686360641335, + "grad_norm": 2.3057632446289062, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.694808304309845, + "num_tokens": 162173188.0, + "step": 6261 + }, + { + "epoch": 0.6876784537667472, + "grad_norm": 1.934030294418335, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7226219177246094, + "num_tokens": 162203116.0, + "step": 6262 + }, + { + "epoch": 0.6877882714693608, + "grad_norm": 2.064709424972534, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.7005218267440796, + "num_tokens": 162230643.0, + "step": 6263 + }, + { + "epoch": 0.6878980891719745, + "grad_norm": 2.1568245887756348, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6940606236457825, + "num_tokens": 162259943.0, + "step": 6264 + }, + { + "epoch": 0.6880079068745882, + "grad_norm": 2.23337459564209, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7054135203361511, + "num_tokens": 162284553.0, + "step": 6265 + }, + { + "epoch": 0.6881177245772019, + "grad_norm": 2.3208868503570557, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6955481767654419, + "num_tokens": 162309858.0, + "step": 6266 + }, + { + "epoch": 0.6882275422798155, + "grad_norm": 2.1186070442199707, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7031828761100769, + "num_tokens": 162337622.0, + "step": 6267 + }, + { + "epoch": 0.6883373599824292, + "grad_norm": 2.0575201511383057, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6977229118347168, + "num_tokens": 162366547.0, + "step": 6268 + }, + { + "epoch": 0.6884471776850428, + "grad_norm": 2.0051655769348145, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.681966245174408, + "num_tokens": 162396141.0, + "step": 6269 + }, + { + "epoch": 0.6885569953876565, + "grad_norm": 2.2970235347747803, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7187361717224121, + "num_tokens": 162420169.0, + "step": 6270 + }, + { + "epoch": 0.6886668130902701, + "grad_norm": 2.2157723903656006, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6924355030059814, + "num_tokens": 162445872.0, + "step": 6271 + }, + { + "epoch": 0.6887766307928838, + "grad_norm": 2.0164458751678467, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7055419683456421, + "num_tokens": 162473104.0, + "step": 6272 + }, + { + "epoch": 0.6888864484954975, + "grad_norm": 2.1459169387817383, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7211180329322815, + "num_tokens": 162498431.0, + "step": 6273 + }, + { + "epoch": 0.6889962661981112, + "grad_norm": 1.9721486568450928, + "learning_rate": 1e-06, + "loss": 1.0648, + "mean_token_accuracy": 0.6813147664070129, + "num_tokens": 162528782.0, + "step": 6274 + }, + { + "epoch": 0.6891060839007248, + "grad_norm": 2.204190492630005, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7021999359130859, + "num_tokens": 162553110.0, + "step": 6275 + }, + { + "epoch": 0.6892159016033385, + "grad_norm": 2.0716140270233154, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6911725401878357, + "num_tokens": 162581838.0, + "step": 6276 + }, + { + "epoch": 0.6893257193059521, + "grad_norm": 2.2789361476898193, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6904057264328003, + "num_tokens": 162604965.0, + "step": 6277 + }, + { + "epoch": 0.6894355370085657, + "grad_norm": 2.00042462348938, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7124118804931641, + "num_tokens": 162634386.0, + "step": 6278 + }, + { + "epoch": 0.6895453547111794, + "grad_norm": 2.288966417312622, + "learning_rate": 1e-06, + "loss": 1.0699, + "mean_token_accuracy": 0.6829032897949219, + "num_tokens": 162660520.0, + "step": 6279 + }, + { + "epoch": 0.6896551724137931, + "grad_norm": 2.343928813934326, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7207416892051697, + "num_tokens": 162683482.0, + "step": 6280 + }, + { + "epoch": 0.6897649901164068, + "grad_norm": 2.0523386001586914, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.707654595375061, + "num_tokens": 162712341.0, + "step": 6281 + }, + { + "epoch": 0.6898748078190204, + "grad_norm": 2.232947826385498, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7195661067962646, + "num_tokens": 162735854.0, + "step": 6282 + }, + { + "epoch": 0.6899846255216341, + "grad_norm": 1.9645801782608032, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7045019865036011, + "num_tokens": 162767452.0, + "step": 6283 + }, + { + "epoch": 0.6900944432242477, + "grad_norm": 2.4522171020507812, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7219775915145874, + "num_tokens": 162788127.0, + "step": 6284 + }, + { + "epoch": 0.6902042609268614, + "grad_norm": 2.0499377250671387, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.707245409488678, + "num_tokens": 162817399.0, + "step": 6285 + }, + { + "epoch": 0.690314078629475, + "grad_norm": 2.275970458984375, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6856511235237122, + "num_tokens": 162842073.0, + "step": 6286 + }, + { + "epoch": 0.6904238963320888, + "grad_norm": 2.216358184814453, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7322749495506287, + "num_tokens": 162865066.0, + "step": 6287 + }, + { + "epoch": 0.6905337140347024, + "grad_norm": 2.2414698600769043, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.712051272392273, + "num_tokens": 162889656.0, + "step": 6288 + }, + { + "epoch": 0.6906435317373161, + "grad_norm": 2.2191061973571777, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7133574485778809, + "num_tokens": 162913759.0, + "step": 6289 + }, + { + "epoch": 0.6907533494399297, + "grad_norm": 2.161320447921753, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7120202779769897, + "num_tokens": 162939871.0, + "step": 6290 + }, + { + "epoch": 0.6908631671425434, + "grad_norm": 2.0814642906188965, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.7048051357269287, + "num_tokens": 162968388.0, + "step": 6291 + }, + { + "epoch": 0.690972984845157, + "grad_norm": 2.348550319671631, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7252934575080872, + "num_tokens": 162991016.0, + "step": 6292 + }, + { + "epoch": 0.6910828025477707, + "grad_norm": 2.2128779888153076, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7359560132026672, + "num_tokens": 163014896.0, + "step": 6293 + }, + { + "epoch": 0.6911926202503844, + "grad_norm": 2.4907023906707764, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7213621139526367, + "num_tokens": 163035705.0, + "step": 6294 + }, + { + "epoch": 0.6913024379529981, + "grad_norm": 2.1214513778686523, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6823886632919312, + "num_tokens": 163064708.0, + "step": 6295 + }, + { + "epoch": 0.6914122556556117, + "grad_norm": 2.1081879138946533, + "learning_rate": 1e-06, + "loss": 1.0513, + "mean_token_accuracy": 0.6794019937515259, + "num_tokens": 163092567.0, + "step": 6296 + }, + { + "epoch": 0.6915220733582254, + "grad_norm": 2.1546919345855713, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7169022560119629, + "num_tokens": 163117372.0, + "step": 6297 + }, + { + "epoch": 0.691631891060839, + "grad_norm": 2.14306902885437, + "learning_rate": 1e-06, + "loss": 1.0755, + "mean_token_accuracy": 0.6780157089233398, + "num_tokens": 163142418.0, + "step": 6298 + }, + { + "epoch": 0.6917417087634526, + "grad_norm": 2.029400110244751, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6940029859542847, + "num_tokens": 163172080.0, + "step": 6299 + }, + { + "epoch": 0.6918515264660663, + "grad_norm": 2.142286777496338, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.701347827911377, + "num_tokens": 163197984.0, + "step": 6300 + }, + { + "epoch": 0.6919613441686799, + "grad_norm": 2.086768865585327, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6955299973487854, + "num_tokens": 163226464.0, + "step": 6301 + }, + { + "epoch": 0.6920711618712937, + "grad_norm": 2.090705633163452, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7114639282226562, + "num_tokens": 163256158.0, + "step": 6302 + }, + { + "epoch": 0.6921809795739073, + "grad_norm": 2.4032371044158936, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7004034519195557, + "num_tokens": 163277834.0, + "step": 6303 + }, + { + "epoch": 0.692290797276521, + "grad_norm": 2.382615089416504, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7153391242027283, + "num_tokens": 163299028.0, + "step": 6304 + }, + { + "epoch": 0.6924006149791346, + "grad_norm": 1.9768224954605103, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.687667965888977, + "num_tokens": 163329095.0, + "step": 6305 + }, + { + "epoch": 0.6925104326817483, + "grad_norm": 2.119154691696167, + "learning_rate": 1e-06, + "loss": 1.0592, + "mean_token_accuracy": 0.6823133230209351, + "num_tokens": 163357278.0, + "step": 6306 + }, + { + "epoch": 0.6926202503843619, + "grad_norm": 2.209364175796509, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7100962400436401, + "num_tokens": 163381779.0, + "step": 6307 + }, + { + "epoch": 0.6927300680869756, + "grad_norm": 1.8713594675064087, + "learning_rate": 1e-06, + "loss": 1.0884, + "mean_token_accuracy": 0.6762996315956116, + "num_tokens": 163415146.0, + "step": 6308 + }, + { + "epoch": 0.6928398857895893, + "grad_norm": 2.216160297393799, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6943807601928711, + "num_tokens": 163440947.0, + "step": 6309 + }, + { + "epoch": 0.692949703492203, + "grad_norm": 2.4314825534820557, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7169603109359741, + "num_tokens": 163460802.0, + "step": 6310 + }, + { + "epoch": 0.6930595211948166, + "grad_norm": 1.823166012763977, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7010844349861145, + "num_tokens": 163496352.0, + "step": 6311 + }, + { + "epoch": 0.6931693388974303, + "grad_norm": 1.9824599027633667, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7096273899078369, + "num_tokens": 163526044.0, + "step": 6312 + }, + { + "epoch": 0.6932791566000439, + "grad_norm": 2.3796093463897705, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6912407875061035, + "num_tokens": 163550244.0, + "step": 6313 + }, + { + "epoch": 0.6933889743026576, + "grad_norm": 2.271610736846924, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6914956569671631, + "num_tokens": 163576341.0, + "step": 6314 + }, + { + "epoch": 0.6934987920052712, + "grad_norm": 2.086536169052124, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7053474187850952, + "num_tokens": 163604671.0, + "step": 6315 + }, + { + "epoch": 0.693608609707885, + "grad_norm": 2.286449670791626, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6980124711990356, + "num_tokens": 163629666.0, + "step": 6316 + }, + { + "epoch": 0.6937184274104986, + "grad_norm": 1.9754465818405151, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7112749814987183, + "num_tokens": 163658034.0, + "step": 6317 + }, + { + "epoch": 0.6938282451131123, + "grad_norm": 2.1897430419921875, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7123312950134277, + "num_tokens": 163683740.0, + "step": 6318 + }, + { + "epoch": 0.6939380628157259, + "grad_norm": 2.2182936668395996, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.728519082069397, + "num_tokens": 163706240.0, + "step": 6319 + }, + { + "epoch": 0.6940478805183395, + "grad_norm": 2.4937846660614014, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7207984924316406, + "num_tokens": 163726391.0, + "step": 6320 + }, + { + "epoch": 0.6941576982209532, + "grad_norm": 2.1001012325286865, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.714556097984314, + "num_tokens": 163754046.0, + "step": 6321 + }, + { + "epoch": 0.6942675159235668, + "grad_norm": 2.274561882019043, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7058199644088745, + "num_tokens": 163779277.0, + "step": 6322 + }, + { + "epoch": 0.6943773336261806, + "grad_norm": 2.016914129257202, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6884557604789734, + "num_tokens": 163811487.0, + "step": 6323 + }, + { + "epoch": 0.6944871513287942, + "grad_norm": 2.106414318084717, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6979807019233704, + "num_tokens": 163837897.0, + "step": 6324 + }, + { + "epoch": 0.6945969690314079, + "grad_norm": 2.0935826301574707, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6921394467353821, + "num_tokens": 163866188.0, + "step": 6325 + }, + { + "epoch": 0.6947067867340215, + "grad_norm": 2.0635342597961426, + "learning_rate": 1e-06, + "loss": 1.0898, + "mean_token_accuracy": 0.6822906136512756, + "num_tokens": 163897427.0, + "step": 6326 + }, + { + "epoch": 0.6948166044366352, + "grad_norm": 2.039830207824707, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.718993067741394, + "num_tokens": 163926436.0, + "step": 6327 + }, + { + "epoch": 0.6949264221392488, + "grad_norm": 2.0670101642608643, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6893305778503418, + "num_tokens": 163956225.0, + "step": 6328 + }, + { + "epoch": 0.6950362398418625, + "grad_norm": 2.2215793132781982, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.7090303897857666, + "num_tokens": 163982151.0, + "step": 6329 + }, + { + "epoch": 0.6951460575444761, + "grad_norm": 2.2504000663757324, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7004685997962952, + "num_tokens": 164007634.0, + "step": 6330 + }, + { + "epoch": 0.6952558752470899, + "grad_norm": 1.970019817352295, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7128413915634155, + "num_tokens": 164035131.0, + "step": 6331 + }, + { + "epoch": 0.6953656929497035, + "grad_norm": 1.995694637298584, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.700951874256134, + "num_tokens": 164063187.0, + "step": 6332 + }, + { + "epoch": 0.6954755106523172, + "grad_norm": 2.3925273418426514, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7214702367782593, + "num_tokens": 164085148.0, + "step": 6333 + }, + { + "epoch": 0.6955853283549308, + "grad_norm": 2.3035051822662354, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6909042000770569, + "num_tokens": 164108696.0, + "step": 6334 + }, + { + "epoch": 0.6956951460575445, + "grad_norm": 2.282694101333618, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7160733938217163, + "num_tokens": 164132496.0, + "step": 6335 + }, + { + "epoch": 0.6958049637601581, + "grad_norm": 2.2055447101593018, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7071845531463623, + "num_tokens": 164158020.0, + "step": 6336 + }, + { + "epoch": 0.6959147814627717, + "grad_norm": 2.4310953617095947, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7196775674819946, + "num_tokens": 164179998.0, + "step": 6337 + }, + { + "epoch": 0.6960245991653855, + "grad_norm": 2.763528823852539, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7282875776290894, + "num_tokens": 164197247.0, + "step": 6338 + }, + { + "epoch": 0.6961344168679992, + "grad_norm": 2.1434473991394043, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7066069841384888, + "num_tokens": 164222013.0, + "step": 6339 + }, + { + "epoch": 0.6962442345706128, + "grad_norm": 2.2660281658172607, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7205900549888611, + "num_tokens": 164244311.0, + "step": 6340 + }, + { + "epoch": 0.6963540522732264, + "grad_norm": 2.0213568210601807, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7043463587760925, + "num_tokens": 164272798.0, + "step": 6341 + }, + { + "epoch": 0.6964638699758401, + "grad_norm": 2.1598591804504395, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7080355882644653, + "num_tokens": 164298367.0, + "step": 6342 + }, + { + "epoch": 0.6965736876784537, + "grad_norm": 2.1070635318756104, + "learning_rate": 1e-06, + "loss": 1.0407, + "mean_token_accuracy": 0.6861642599105835, + "num_tokens": 164328980.0, + "step": 6343 + }, + { + "epoch": 0.6966835053810674, + "grad_norm": 2.239246129989624, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7025324106216431, + "num_tokens": 164354971.0, + "step": 6344 + }, + { + "epoch": 0.6967933230836811, + "grad_norm": 1.9980599880218506, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7099060416221619, + "num_tokens": 164382484.0, + "step": 6345 + }, + { + "epoch": 0.6969031407862948, + "grad_norm": 2.3934619426727295, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7133188247680664, + "num_tokens": 164406329.0, + "step": 6346 + }, + { + "epoch": 0.6970129584889084, + "grad_norm": 1.928698182106018, + "learning_rate": 1e-06, + "loss": 1.0979, + "mean_token_accuracy": 0.6784232258796692, + "num_tokens": 164438435.0, + "step": 6347 + }, + { + "epoch": 0.6971227761915221, + "grad_norm": 2.5094919204711914, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7012526988983154, + "num_tokens": 164458132.0, + "step": 6348 + }, + { + "epoch": 0.6972325938941357, + "grad_norm": 2.1851389408111572, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7181146144866943, + "num_tokens": 164483148.0, + "step": 6349 + }, + { + "epoch": 0.6973424115967494, + "grad_norm": 2.217247247695923, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7187070846557617, + "num_tokens": 164507414.0, + "step": 6350 + }, + { + "epoch": 0.697452229299363, + "grad_norm": 2.042585611343384, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7080059051513672, + "num_tokens": 164535855.0, + "step": 6351 + }, + { + "epoch": 0.6975620470019768, + "grad_norm": 2.300093412399292, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7137902975082397, + "num_tokens": 164558183.0, + "step": 6352 + }, + { + "epoch": 0.6976718647045904, + "grad_norm": 2.080472230911255, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7099786996841431, + "num_tokens": 164584290.0, + "step": 6353 + }, + { + "epoch": 0.6977816824072041, + "grad_norm": 2.661771774291992, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7150777578353882, + "num_tokens": 164601322.0, + "step": 6354 + }, + { + "epoch": 0.6978915001098177, + "grad_norm": 2.7415988445281982, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7315157651901245, + "num_tokens": 164618813.0, + "step": 6355 + }, + { + "epoch": 0.6980013178124314, + "grad_norm": 2.213374614715576, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7159190773963928, + "num_tokens": 164642036.0, + "step": 6356 + }, + { + "epoch": 0.698111135515045, + "grad_norm": 1.9102003574371338, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7116090059280396, + "num_tokens": 164674367.0, + "step": 6357 + }, + { + "epoch": 0.6982209532176586, + "grad_norm": 2.2705326080322266, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7133535146713257, + "num_tokens": 164696694.0, + "step": 6358 + }, + { + "epoch": 0.6983307709202724, + "grad_norm": 2.126826524734497, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7304605841636658, + "num_tokens": 164722792.0, + "step": 6359 + }, + { + "epoch": 0.698440588622886, + "grad_norm": 2.211866855621338, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7116717100143433, + "num_tokens": 164748014.0, + "step": 6360 + }, + { + "epoch": 0.6985504063254997, + "grad_norm": 2.2130138874053955, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6964345574378967, + "num_tokens": 164772019.0, + "step": 6361 + }, + { + "epoch": 0.6986602240281133, + "grad_norm": 2.0418355464935303, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6977788209915161, + "num_tokens": 164800535.0, + "step": 6362 + }, + { + "epoch": 0.698770041730727, + "grad_norm": 2.181396245956421, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6860119104385376, + "num_tokens": 164827307.0, + "step": 6363 + }, + { + "epoch": 0.6988798594333406, + "grad_norm": 2.416085720062256, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7190828323364258, + "num_tokens": 164850156.0, + "step": 6364 + }, + { + "epoch": 0.6989896771359543, + "grad_norm": 2.1830413341522217, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7092627882957458, + "num_tokens": 164876107.0, + "step": 6365 + }, + { + "epoch": 0.6990994948385679, + "grad_norm": 2.004260540008545, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7014214396476746, + "num_tokens": 164905536.0, + "step": 6366 + }, + { + "epoch": 0.6992093125411817, + "grad_norm": 2.239363193511963, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.719114363193512, + "num_tokens": 164930711.0, + "step": 6367 + }, + { + "epoch": 0.6993191302437953, + "grad_norm": 1.9143205881118774, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.6783277988433838, + "num_tokens": 164965215.0, + "step": 6368 + }, + { + "epoch": 0.699428947946409, + "grad_norm": 2.379488706588745, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7025055885314941, + "num_tokens": 164987382.0, + "step": 6369 + }, + { + "epoch": 0.6995387656490226, + "grad_norm": 2.1820507049560547, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6883686780929565, + "num_tokens": 165016576.0, + "step": 6370 + }, + { + "epoch": 0.6996485833516363, + "grad_norm": 2.164919137954712, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6966513395309448, + "num_tokens": 165043318.0, + "step": 6371 + }, + { + "epoch": 0.6997584010542499, + "grad_norm": 2.51202654838562, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7026427388191223, + "num_tokens": 165063912.0, + "step": 6372 + }, + { + "epoch": 0.6998682187568636, + "grad_norm": 2.097780704498291, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.7002785205841064, + "num_tokens": 165091183.0, + "step": 6373 + }, + { + "epoch": 0.6999780364594773, + "grad_norm": 2.0973587036132812, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7222665548324585, + "num_tokens": 165117875.0, + "step": 6374 + }, + { + "epoch": 0.700087854162091, + "grad_norm": 2.4464073181152344, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7208752036094666, + "num_tokens": 165139314.0, + "step": 6375 + }, + { + "epoch": 0.7001976718647046, + "grad_norm": 2.262502908706665, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7159053683280945, + "num_tokens": 165164950.0, + "step": 6376 + }, + { + "epoch": 0.7003074895673183, + "grad_norm": 2.2571780681610107, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.710760235786438, + "num_tokens": 165190399.0, + "step": 6377 + }, + { + "epoch": 0.7004173072699319, + "grad_norm": 2.495363235473633, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7069747447967529, + "num_tokens": 165213424.0, + "step": 6378 + }, + { + "epoch": 0.7005271249725455, + "grad_norm": 2.156092405319214, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7133045196533203, + "num_tokens": 165239322.0, + "step": 6379 + }, + { + "epoch": 0.7006369426751592, + "grad_norm": 1.8133705854415894, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.678263783454895, + "num_tokens": 165276415.0, + "step": 6380 + }, + { + "epoch": 0.700746760377773, + "grad_norm": 2.1503264904022217, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6874490976333618, + "num_tokens": 165304440.0, + "step": 6381 + }, + { + "epoch": 0.7008565780803866, + "grad_norm": 2.245851516723633, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7179780602455139, + "num_tokens": 165329692.0, + "step": 6382 + }, + { + "epoch": 0.7009663957830002, + "grad_norm": 1.9280301332473755, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7094290852546692, + "num_tokens": 165361978.0, + "step": 6383 + }, + { + "epoch": 0.7010762134856139, + "grad_norm": 2.0190811157226562, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7153756618499756, + "num_tokens": 165391661.0, + "step": 6384 + }, + { + "epoch": 0.7011860311882275, + "grad_norm": 2.445791006088257, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.70654296875, + "num_tokens": 165412857.0, + "step": 6385 + }, + { + "epoch": 0.7012958488908412, + "grad_norm": 2.4356212615966797, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7282223701477051, + "num_tokens": 165434553.0, + "step": 6386 + }, + { + "epoch": 0.7014056665934548, + "grad_norm": 2.217660427093506, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.723429799079895, + "num_tokens": 165458779.0, + "step": 6387 + }, + { + "epoch": 0.7015154842960686, + "grad_norm": 2.0499885082244873, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7279880046844482, + "num_tokens": 165484387.0, + "step": 6388 + }, + { + "epoch": 0.7016253019986822, + "grad_norm": 2.2305679321289062, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.711061418056488, + "num_tokens": 165508058.0, + "step": 6389 + }, + { + "epoch": 0.7017351197012959, + "grad_norm": 2.124828577041626, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6972461938858032, + "num_tokens": 165534481.0, + "step": 6390 + }, + { + "epoch": 0.7018449374039095, + "grad_norm": 2.0527405738830566, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7065105438232422, + "num_tokens": 165562167.0, + "step": 6391 + }, + { + "epoch": 0.7019547551065232, + "grad_norm": 2.637465000152588, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7292178869247437, + "num_tokens": 165580369.0, + "step": 6392 + }, + { + "epoch": 0.7020645728091368, + "grad_norm": 2.274296522140503, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7020031213760376, + "num_tokens": 165606836.0, + "step": 6393 + }, + { + "epoch": 0.7021743905117505, + "grad_norm": 2.411726713180542, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7029684782028198, + "num_tokens": 165629360.0, + "step": 6394 + }, + { + "epoch": 0.7022842082143641, + "grad_norm": 2.2790470123291016, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6944586038589478, + "num_tokens": 165653418.0, + "step": 6395 + }, + { + "epoch": 0.7023940259169779, + "grad_norm": 2.143188714981079, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7324520349502563, + "num_tokens": 165678775.0, + "step": 6396 + }, + { + "epoch": 0.7025038436195915, + "grad_norm": 2.155980110168457, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7114779353141785, + "num_tokens": 165704460.0, + "step": 6397 + }, + { + "epoch": 0.7026136613222052, + "grad_norm": 2.180102825164795, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7193776369094849, + "num_tokens": 165731850.0, + "step": 6398 + }, + { + "epoch": 0.7027234790248188, + "grad_norm": 2.224036931991577, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6994067430496216, + "num_tokens": 165757577.0, + "step": 6399 + }, + { + "epoch": 0.7028332967274324, + "grad_norm": 2.068089246749878, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6778500080108643, + "num_tokens": 165787726.0, + "step": 6400 + }, + { + "epoch": 0.7029431144300461, + "grad_norm": 2.407879114151001, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7064932584762573, + "num_tokens": 165810931.0, + "step": 6401 + }, + { + "epoch": 0.7030529321326597, + "grad_norm": 2.2907934188842773, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6929517388343811, + "num_tokens": 165837238.0, + "step": 6402 + }, + { + "epoch": 0.7031627498352735, + "grad_norm": 1.9800442457199097, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7074161767959595, + "num_tokens": 165867065.0, + "step": 6403 + }, + { + "epoch": 0.7032725675378871, + "grad_norm": 2.107633590698242, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6959970593452454, + "num_tokens": 165895370.0, + "step": 6404 + }, + { + "epoch": 0.7033823852405008, + "grad_norm": 1.9496147632598877, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6928598284721375, + "num_tokens": 165928520.0, + "step": 6405 + }, + { + "epoch": 0.7034922029431144, + "grad_norm": 2.301213026046753, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6919876337051392, + "num_tokens": 165952076.0, + "step": 6406 + }, + { + "epoch": 0.7036020206457281, + "grad_norm": 2.1140239238739014, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7058185338973999, + "num_tokens": 165978496.0, + "step": 6407 + }, + { + "epoch": 0.7037118383483417, + "grad_norm": 2.27394962310791, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7045000195503235, + "num_tokens": 166000883.0, + "step": 6408 + }, + { + "epoch": 0.7038216560509554, + "grad_norm": 2.0462512969970703, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6827424764633179, + "num_tokens": 166029905.0, + "step": 6409 + }, + { + "epoch": 0.7039314737535691, + "grad_norm": 2.274630308151245, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7037670016288757, + "num_tokens": 166052445.0, + "step": 6410 + }, + { + "epoch": 0.7040412914561828, + "grad_norm": 2.1684679985046387, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7028031349182129, + "num_tokens": 166080139.0, + "step": 6411 + }, + { + "epoch": 0.7041511091587964, + "grad_norm": 2.145355701446533, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7064119577407837, + "num_tokens": 166109391.0, + "step": 6412 + }, + { + "epoch": 0.7042609268614101, + "grad_norm": 2.0866847038269043, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6791452169418335, + "num_tokens": 166138729.0, + "step": 6413 + }, + { + "epoch": 0.7043707445640237, + "grad_norm": 2.321601629257202, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6866710782051086, + "num_tokens": 166164711.0, + "step": 6414 + }, + { + "epoch": 0.7044805622666374, + "grad_norm": 2.699862480163574, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7209039926528931, + "num_tokens": 166182875.0, + "step": 6415 + }, + { + "epoch": 0.704590379969251, + "grad_norm": 2.212097644805908, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7059398293495178, + "num_tokens": 166209426.0, + "step": 6416 + }, + { + "epoch": 0.7047001976718648, + "grad_norm": 2.2485384941101074, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.682783842086792, + "num_tokens": 166234077.0, + "step": 6417 + }, + { + "epoch": 0.7048100153744784, + "grad_norm": 2.057156801223755, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7035168409347534, + "num_tokens": 166263591.0, + "step": 6418 + }, + { + "epoch": 0.704919833077092, + "grad_norm": 2.206054925918579, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7097630500793457, + "num_tokens": 166287278.0, + "step": 6419 + }, + { + "epoch": 0.7050296507797057, + "grad_norm": 2.315227508544922, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7343827486038208, + "num_tokens": 166309647.0, + "step": 6420 + }, + { + "epoch": 0.7051394684823193, + "grad_norm": 2.131274700164795, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6867237091064453, + "num_tokens": 166336724.0, + "step": 6421 + }, + { + "epoch": 0.705249286184933, + "grad_norm": 2.3639681339263916, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7061401605606079, + "num_tokens": 166361211.0, + "step": 6422 + }, + { + "epoch": 0.7053591038875466, + "grad_norm": 2.095950126647949, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7283291816711426, + "num_tokens": 166385822.0, + "step": 6423 + }, + { + "epoch": 0.7054689215901603, + "grad_norm": 2.296771764755249, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7453771233558655, + "num_tokens": 166407055.0, + "step": 6424 + }, + { + "epoch": 0.705578739292774, + "grad_norm": 2.2751762866973877, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6984082460403442, + "num_tokens": 166430708.0, + "step": 6425 + }, + { + "epoch": 0.7056885569953877, + "grad_norm": 2.134730815887451, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7035821676254272, + "num_tokens": 166456779.0, + "step": 6426 + }, + { + "epoch": 0.7057983746980013, + "grad_norm": 2.1439530849456787, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7117190361022949, + "num_tokens": 166481233.0, + "step": 6427 + }, + { + "epoch": 0.705908192400615, + "grad_norm": 2.506154775619507, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7216054201126099, + "num_tokens": 166501202.0, + "step": 6428 + }, + { + "epoch": 0.7060180101032286, + "grad_norm": 2.0288262367248535, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7021584510803223, + "num_tokens": 166531152.0, + "step": 6429 + }, + { + "epoch": 0.7061278278058423, + "grad_norm": 2.3248345851898193, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7180428504943848, + "num_tokens": 166553589.0, + "step": 6430 + }, + { + "epoch": 0.7062376455084559, + "grad_norm": 2.1867125034332275, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7120374441146851, + "num_tokens": 166578512.0, + "step": 6431 + }, + { + "epoch": 0.7063474632110697, + "grad_norm": 2.4462718963623047, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7129477858543396, + "num_tokens": 166599150.0, + "step": 6432 + }, + { + "epoch": 0.7064572809136833, + "grad_norm": 2.189819097518921, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6884422302246094, + "num_tokens": 166626449.0, + "step": 6433 + }, + { + "epoch": 0.706567098616297, + "grad_norm": 2.4137134552001953, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7248350381851196, + "num_tokens": 166648587.0, + "step": 6434 + }, + { + "epoch": 0.7066769163189106, + "grad_norm": 2.335127353668213, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7220785617828369, + "num_tokens": 166670674.0, + "step": 6435 + }, + { + "epoch": 0.7067867340215243, + "grad_norm": 2.34454345703125, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7068513631820679, + "num_tokens": 166695813.0, + "step": 6436 + }, + { + "epoch": 0.7068965517241379, + "grad_norm": 2.4444758892059326, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7157018184661865, + "num_tokens": 166716119.0, + "step": 6437 + }, + { + "epoch": 0.7070063694267515, + "grad_norm": 1.9381890296936035, + "learning_rate": 1e-06, + "loss": 1.0869, + "mean_token_accuracy": 0.6740186810493469, + "num_tokens": 166750666.0, + "step": 6438 + }, + { + "epoch": 0.7071161871293653, + "grad_norm": 2.392895221710205, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7060679197311401, + "num_tokens": 166772775.0, + "step": 6439 + }, + { + "epoch": 0.707226004831979, + "grad_norm": 2.3990283012390137, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7238640785217285, + "num_tokens": 166792522.0, + "step": 6440 + }, + { + "epoch": 0.7073358225345926, + "grad_norm": 2.0692837238311768, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7021366953849792, + "num_tokens": 166821634.0, + "step": 6441 + }, + { + "epoch": 0.7074456402372062, + "grad_norm": 2.2415997982025146, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6940351128578186, + "num_tokens": 166847977.0, + "step": 6442 + }, + { + "epoch": 0.7075554579398199, + "grad_norm": 2.2923011779785156, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.701972246170044, + "num_tokens": 166871906.0, + "step": 6443 + }, + { + "epoch": 0.7076652756424335, + "grad_norm": 1.8579570055007935, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7046262621879578, + "num_tokens": 166903174.0, + "step": 6444 + }, + { + "epoch": 0.7077750933450472, + "grad_norm": 2.1380343437194824, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7262651324272156, + "num_tokens": 166927652.0, + "step": 6445 + }, + { + "epoch": 0.7078849110476609, + "grad_norm": 2.2016489505767822, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7093778252601624, + "num_tokens": 166951923.0, + "step": 6446 + }, + { + "epoch": 0.7079947287502746, + "grad_norm": 2.4642162322998047, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7100983262062073, + "num_tokens": 166973547.0, + "step": 6447 + }, + { + "epoch": 0.7081045464528882, + "grad_norm": 2.5141334533691406, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7283064723014832, + "num_tokens": 166992928.0, + "step": 6448 + }, + { + "epoch": 0.7082143641555019, + "grad_norm": 2.315009117126465, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.717703104019165, + "num_tokens": 167014443.0, + "step": 6449 + }, + { + "epoch": 0.7083241818581155, + "grad_norm": 2.080789804458618, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.702507734298706, + "num_tokens": 167040164.0, + "step": 6450 + }, + { + "epoch": 0.7084339995607292, + "grad_norm": 2.1017253398895264, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6981596946716309, + "num_tokens": 167065933.0, + "step": 6451 + }, + { + "epoch": 0.7085438172633428, + "grad_norm": 2.694464921951294, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7144004106521606, + "num_tokens": 167083429.0, + "step": 6452 + }, + { + "epoch": 0.7086536349659565, + "grad_norm": 2.1953065395355225, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7110889554023743, + "num_tokens": 167108697.0, + "step": 6453 + }, + { + "epoch": 0.7087634526685702, + "grad_norm": 1.9920523166656494, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7182552814483643, + "num_tokens": 167138333.0, + "step": 6454 + }, + { + "epoch": 0.7088732703711839, + "grad_norm": 2.0745601654052734, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7021996974945068, + "num_tokens": 167166755.0, + "step": 6455 + }, + { + "epoch": 0.7089830880737975, + "grad_norm": 2.363607168197632, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7162140011787415, + "num_tokens": 167189785.0, + "step": 6456 + }, + { + "epoch": 0.7090929057764112, + "grad_norm": 2.537316083908081, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7198063731193542, + "num_tokens": 167211804.0, + "step": 6457 + }, + { + "epoch": 0.7092027234790248, + "grad_norm": 2.4120676517486572, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.6986799836158752, + "num_tokens": 167233908.0, + "step": 6458 + }, + { + "epoch": 0.7093125411816384, + "grad_norm": 2.28155255317688, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7086176872253418, + "num_tokens": 167258245.0, + "step": 6459 + }, + { + "epoch": 0.7094223588842521, + "grad_norm": 2.110024929046631, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6999843120574951, + "num_tokens": 167284185.0, + "step": 6460 + }, + { + "epoch": 0.7095321765868658, + "grad_norm": 2.237422466278076, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.710595428943634, + "num_tokens": 167309018.0, + "step": 6461 + }, + { + "epoch": 0.7096419942894795, + "grad_norm": 1.9752197265625, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7127781510353088, + "num_tokens": 167337963.0, + "step": 6462 + }, + { + "epoch": 0.7097518119920931, + "grad_norm": 1.9407380819320679, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.684754490852356, + "num_tokens": 167368817.0, + "step": 6463 + }, + { + "epoch": 0.7098616296947068, + "grad_norm": 2.105268955230713, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7059998512268066, + "num_tokens": 167395048.0, + "step": 6464 + }, + { + "epoch": 0.7099714473973204, + "grad_norm": 1.9329246282577515, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7099299430847168, + "num_tokens": 167424637.0, + "step": 6465 + }, + { + "epoch": 0.7100812650999341, + "grad_norm": 1.8353450298309326, + "learning_rate": 1e-06, + "loss": 1.0366, + "mean_token_accuracy": 0.6837818622589111, + "num_tokens": 167459751.0, + "step": 6466 + }, + { + "epoch": 0.7101910828025477, + "grad_norm": 2.1155216693878174, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7198291420936584, + "num_tokens": 167484847.0, + "step": 6467 + }, + { + "epoch": 0.7103009005051615, + "grad_norm": 2.221007823944092, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7141849398612976, + "num_tokens": 167507780.0, + "step": 6468 + }, + { + "epoch": 0.7104107182077751, + "grad_norm": 2.286149501800537, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7205890417098999, + "num_tokens": 167530304.0, + "step": 6469 + }, + { + "epoch": 0.7105205359103888, + "grad_norm": 2.420039415359497, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6965756416320801, + "num_tokens": 167551363.0, + "step": 6470 + }, + { + "epoch": 0.7106303536130024, + "grad_norm": 2.026750087738037, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.683714747428894, + "num_tokens": 167582669.0, + "step": 6471 + }, + { + "epoch": 0.7107401713156161, + "grad_norm": 1.9132699966430664, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6892740726470947, + "num_tokens": 167613850.0, + "step": 6472 + }, + { + "epoch": 0.7108499890182297, + "grad_norm": 1.9062304496765137, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7064228653907776, + "num_tokens": 167646709.0, + "step": 6473 + }, + { + "epoch": 0.7109598067208434, + "grad_norm": 2.017334461212158, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.7124477624893188, + "num_tokens": 167675359.0, + "step": 6474 + }, + { + "epoch": 0.7110696244234571, + "grad_norm": 2.3097476959228516, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7028207778930664, + "num_tokens": 167698157.0, + "step": 6475 + }, + { + "epoch": 0.7111794421260708, + "grad_norm": 2.105623960494995, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6980910301208496, + "num_tokens": 167725839.0, + "step": 6476 + }, + { + "epoch": 0.7112892598286844, + "grad_norm": 2.245124101638794, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7251436710357666, + "num_tokens": 167749767.0, + "step": 6477 + }, + { + "epoch": 0.711399077531298, + "grad_norm": 2.015516757965088, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6961084604263306, + "num_tokens": 167777055.0, + "step": 6478 + }, + { + "epoch": 0.7115088952339117, + "grad_norm": 2.314314365386963, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7278797626495361, + "num_tokens": 167799343.0, + "step": 6479 + }, + { + "epoch": 0.7116187129365253, + "grad_norm": 2.2470996379852295, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7225579619407654, + "num_tokens": 167822339.0, + "step": 6480 + }, + { + "epoch": 0.711728530639139, + "grad_norm": 2.1501517295837402, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7268654108047485, + "num_tokens": 167847874.0, + "step": 6481 + }, + { + "epoch": 0.7118383483417526, + "grad_norm": 2.286634683609009, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.695968747138977, + "num_tokens": 167871087.0, + "step": 6482 + }, + { + "epoch": 0.7119481660443664, + "grad_norm": 2.1254677772521973, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6924793720245361, + "num_tokens": 167898170.0, + "step": 6483 + }, + { + "epoch": 0.71205798374698, + "grad_norm": 2.1424691677093506, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.707251787185669, + "num_tokens": 167923656.0, + "step": 6484 + }, + { + "epoch": 0.7121678014495937, + "grad_norm": 2.172307252883911, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6865884065628052, + "num_tokens": 167949266.0, + "step": 6485 + }, + { + "epoch": 0.7122776191522073, + "grad_norm": 2.2124218940734863, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7099460959434509, + "num_tokens": 167972994.0, + "step": 6486 + }, + { + "epoch": 0.712387436854821, + "grad_norm": 2.271878719329834, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7045673131942749, + "num_tokens": 167997392.0, + "step": 6487 + }, + { + "epoch": 0.7124972545574346, + "grad_norm": 2.3208305835723877, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6972507238388062, + "num_tokens": 168022751.0, + "step": 6488 + }, + { + "epoch": 0.7126070722600483, + "grad_norm": 2.108242988586426, + "learning_rate": 1e-06, + "loss": 1.0631, + "mean_token_accuracy": 0.6865568161010742, + "num_tokens": 168052042.0, + "step": 6489 + }, + { + "epoch": 0.712716889962662, + "grad_norm": 2.6046433448791504, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.708897054195404, + "num_tokens": 168071391.0, + "step": 6490 + }, + { + "epoch": 0.7128267076652757, + "grad_norm": 2.098924398422241, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6983685493469238, + "num_tokens": 168102465.0, + "step": 6491 + }, + { + "epoch": 0.7129365253678893, + "grad_norm": 2.661344528198242, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7007042169570923, + "num_tokens": 168122278.0, + "step": 6492 + }, + { + "epoch": 0.713046343070503, + "grad_norm": 2.0607900619506836, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7095426321029663, + "num_tokens": 168151102.0, + "step": 6493 + }, + { + "epoch": 0.7131561607731166, + "grad_norm": 2.2425334453582764, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7343859076499939, + "num_tokens": 168175128.0, + "step": 6494 + }, + { + "epoch": 0.7132659784757303, + "grad_norm": 1.9824655055999756, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.676498532295227, + "num_tokens": 168207639.0, + "step": 6495 + }, + { + "epoch": 0.7133757961783439, + "grad_norm": 2.2347733974456787, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7060043811798096, + "num_tokens": 168233243.0, + "step": 6496 + }, + { + "epoch": 0.7134856138809577, + "grad_norm": 2.228525400161743, + "learning_rate": 1e-06, + "loss": 1.0774, + "mean_token_accuracy": 0.678372859954834, + "num_tokens": 168260065.0, + "step": 6497 + }, + { + "epoch": 0.7135954315835713, + "grad_norm": 2.269918918609619, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7232384085655212, + "num_tokens": 168284638.0, + "step": 6498 + }, + { + "epoch": 0.713705249286185, + "grad_norm": 2.2682130336761475, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7063292264938354, + "num_tokens": 168310957.0, + "step": 6499 + }, + { + "epoch": 0.7138150669887986, + "grad_norm": 2.2789111137390137, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.6847754120826721, + "num_tokens": 168336000.0, + "step": 6500 + }, + { + "epoch": 0.7139248846914122, + "grad_norm": 2.4133830070495605, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7106152772903442, + "num_tokens": 168358893.0, + "step": 6501 + }, + { + "epoch": 0.7140347023940259, + "grad_norm": 2.212350845336914, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7267301082611084, + "num_tokens": 168382930.0, + "step": 6502 + }, + { + "epoch": 0.7141445200966395, + "grad_norm": 2.1261563301086426, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6926820278167725, + "num_tokens": 168410606.0, + "step": 6503 + }, + { + "epoch": 0.7142543377992533, + "grad_norm": 2.034785270690918, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7081507444381714, + "num_tokens": 168441184.0, + "step": 6504 + }, + { + "epoch": 0.7143641555018669, + "grad_norm": 2.433907985687256, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7164484858512878, + "num_tokens": 168464355.0, + "step": 6505 + }, + { + "epoch": 0.7144739732044806, + "grad_norm": 2.037471294403076, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7107206583023071, + "num_tokens": 168491266.0, + "step": 6506 + }, + { + "epoch": 0.7145837909070942, + "grad_norm": 2.3780477046966553, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6946096420288086, + "num_tokens": 168516585.0, + "step": 6507 + }, + { + "epoch": 0.7146936086097079, + "grad_norm": 2.7702183723449707, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6894540190696716, + "num_tokens": 168535196.0, + "step": 6508 + }, + { + "epoch": 0.7148034263123215, + "grad_norm": 2.1000149250030518, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.712082028388977, + "num_tokens": 168561111.0, + "step": 6509 + }, + { + "epoch": 0.7149132440149352, + "grad_norm": 2.7261319160461426, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7266750335693359, + "num_tokens": 168578428.0, + "step": 6510 + }, + { + "epoch": 0.7150230617175488, + "grad_norm": 2.022855520248413, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7092207670211792, + "num_tokens": 168607648.0, + "step": 6511 + }, + { + "epoch": 0.7151328794201626, + "grad_norm": 2.077824592590332, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6945762634277344, + "num_tokens": 168636005.0, + "step": 6512 + }, + { + "epoch": 0.7152426971227762, + "grad_norm": 2.096909761428833, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.6954032182693481, + "num_tokens": 168664419.0, + "step": 6513 + }, + { + "epoch": 0.7153525148253899, + "grad_norm": 2.2687571048736572, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7286374568939209, + "num_tokens": 168687639.0, + "step": 6514 + }, + { + "epoch": 0.7154623325280035, + "grad_norm": 2.253962278366089, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.706112265586853, + "num_tokens": 168712456.0, + "step": 6515 + }, + { + "epoch": 0.7155721502306172, + "grad_norm": 2.039820432662964, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7215580940246582, + "num_tokens": 168740946.0, + "step": 6516 + }, + { + "epoch": 0.7156819679332308, + "grad_norm": 2.075467824935913, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6891902685165405, + "num_tokens": 168769603.0, + "step": 6517 + }, + { + "epoch": 0.7157917856358444, + "grad_norm": 2.167343854904175, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7264629602432251, + "num_tokens": 168795783.0, + "step": 6518 + }, + { + "epoch": 0.7159016033384582, + "grad_norm": 2.719399929046631, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7312414050102234, + "num_tokens": 168813454.0, + "step": 6519 + }, + { + "epoch": 0.7160114210410718, + "grad_norm": 2.444489002227783, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7193769216537476, + "num_tokens": 168834759.0, + "step": 6520 + }, + { + "epoch": 0.7161212387436855, + "grad_norm": 2.1658072471618652, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6883794069290161, + "num_tokens": 168861563.0, + "step": 6521 + }, + { + "epoch": 0.7162310564462991, + "grad_norm": 2.092804431915283, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7068657875061035, + "num_tokens": 168888884.0, + "step": 6522 + }, + { + "epoch": 0.7163408741489128, + "grad_norm": 2.010681390762329, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7078658938407898, + "num_tokens": 168917774.0, + "step": 6523 + }, + { + "epoch": 0.7164506918515264, + "grad_norm": 2.5212929248809814, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7141933441162109, + "num_tokens": 168936506.0, + "step": 6524 + }, + { + "epoch": 0.7165605095541401, + "grad_norm": 2.1481828689575195, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6979013085365295, + "num_tokens": 168963308.0, + "step": 6525 + }, + { + "epoch": 0.7166703272567538, + "grad_norm": 2.4753050804138184, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7290069460868835, + "num_tokens": 168985502.0, + "step": 6526 + }, + { + "epoch": 0.7167801449593675, + "grad_norm": 2.358336925506592, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7387080192565918, + "num_tokens": 169006486.0, + "step": 6527 + }, + { + "epoch": 0.7168899626619811, + "grad_norm": 2.242872953414917, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7012773752212524, + "num_tokens": 169030895.0, + "step": 6528 + }, + { + "epoch": 0.7169997803645948, + "grad_norm": 2.3445723056793213, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7045886516571045, + "num_tokens": 169054078.0, + "step": 6529 + }, + { + "epoch": 0.7171095980672084, + "grad_norm": 2.5459721088409424, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7204010486602783, + "num_tokens": 169072668.0, + "step": 6530 + }, + { + "epoch": 0.7172194157698221, + "grad_norm": 2.165461778640747, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6945056915283203, + "num_tokens": 169099329.0, + "step": 6531 + }, + { + "epoch": 0.7173292334724357, + "grad_norm": 2.472228527069092, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7133443355560303, + "num_tokens": 169120903.0, + "step": 6532 + }, + { + "epoch": 0.7174390511750495, + "grad_norm": 2.104358434677124, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6892691850662231, + "num_tokens": 169149640.0, + "step": 6533 + }, + { + "epoch": 0.7175488688776631, + "grad_norm": 2.4110426902770996, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7127844095230103, + "num_tokens": 169170780.0, + "step": 6534 + }, + { + "epoch": 0.7176586865802768, + "grad_norm": 2.1183998584747314, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6936225891113281, + "num_tokens": 169200303.0, + "step": 6535 + }, + { + "epoch": 0.7177685042828904, + "grad_norm": 1.9723951816558838, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7251819372177124, + "num_tokens": 169229281.0, + "step": 6536 + }, + { + "epoch": 0.717878321985504, + "grad_norm": 2.4157838821411133, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7239630818367004, + "num_tokens": 169252296.0, + "step": 6537 + }, + { + "epoch": 0.7179881396881177, + "grad_norm": 2.248471260070801, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6904796361923218, + "num_tokens": 169279224.0, + "step": 6538 + }, + { + "epoch": 0.7180979573907313, + "grad_norm": 2.114722967147827, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6995601058006287, + "num_tokens": 169307045.0, + "step": 6539 + }, + { + "epoch": 0.7182077750933451, + "grad_norm": 2.1825215816497803, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7065259218215942, + "num_tokens": 169333309.0, + "step": 6540 + }, + { + "epoch": 0.7183175927959587, + "grad_norm": 2.1142821311950684, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6917808651924133, + "num_tokens": 169362714.0, + "step": 6541 + }, + { + "epoch": 0.7184274104985724, + "grad_norm": 2.2325987815856934, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6950930953025818, + "num_tokens": 169390450.0, + "step": 6542 + }, + { + "epoch": 0.718537228201186, + "grad_norm": 2.0824854373931885, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6962943077087402, + "num_tokens": 169419864.0, + "step": 6543 + }, + { + "epoch": 0.7186470459037997, + "grad_norm": 2.3170511722564697, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7366276979446411, + "num_tokens": 169441594.0, + "step": 6544 + }, + { + "epoch": 0.7187568636064133, + "grad_norm": 2.1680240631103516, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.705406904220581, + "num_tokens": 169467608.0, + "step": 6545 + }, + { + "epoch": 0.718866681309027, + "grad_norm": 2.1378860473632812, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.6985979676246643, + "num_tokens": 169494768.0, + "step": 6546 + }, + { + "epoch": 0.7189764990116406, + "grad_norm": 2.0649452209472656, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7019011378288269, + "num_tokens": 169521616.0, + "step": 6547 + }, + { + "epoch": 0.7190863167142544, + "grad_norm": 2.4474775791168213, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7223348021507263, + "num_tokens": 169542364.0, + "step": 6548 + }, + { + "epoch": 0.719196134416868, + "grad_norm": 2.0548486709594727, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.6789823770523071, + "num_tokens": 169574336.0, + "step": 6549 + }, + { + "epoch": 0.7193059521194817, + "grad_norm": 1.915356159210205, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7026644945144653, + "num_tokens": 169602383.0, + "step": 6550 + }, + { + "epoch": 0.7194157698220953, + "grad_norm": 2.2687082290649414, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7050724029541016, + "num_tokens": 169625680.0, + "step": 6551 + }, + { + "epoch": 0.719525587524709, + "grad_norm": 2.2229909896850586, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7143121957778931, + "num_tokens": 169649483.0, + "step": 6552 + }, + { + "epoch": 0.7196354052273226, + "grad_norm": 2.1314728260040283, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7173590660095215, + "num_tokens": 169676176.0, + "step": 6553 + }, + { + "epoch": 0.7197452229299363, + "grad_norm": 2.363060235977173, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7056493759155273, + "num_tokens": 169700230.0, + "step": 6554 + }, + { + "epoch": 0.71985504063255, + "grad_norm": 2.263812780380249, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6982563734054565, + "num_tokens": 169724249.0, + "step": 6555 + }, + { + "epoch": 0.7199648583351637, + "grad_norm": 2.190439224243164, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6996488571166992, + "num_tokens": 169750728.0, + "step": 6556 + }, + { + "epoch": 0.7200746760377773, + "grad_norm": 1.9563273191452026, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7016268968582153, + "num_tokens": 169782386.0, + "step": 6557 + }, + { + "epoch": 0.720184493740391, + "grad_norm": 2.0123825073242188, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.721307098865509, + "num_tokens": 169809532.0, + "step": 6558 + }, + { + "epoch": 0.7202943114430046, + "grad_norm": 2.075265884399414, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6991234421730042, + "num_tokens": 169838593.0, + "step": 6559 + }, + { + "epoch": 0.7204041291456182, + "grad_norm": 2.117614507675171, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7116847038269043, + "num_tokens": 169865464.0, + "step": 6560 + }, + { + "epoch": 0.7205139468482319, + "grad_norm": 2.0023193359375, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7241658568382263, + "num_tokens": 169894499.0, + "step": 6561 + }, + { + "epoch": 0.7206237645508456, + "grad_norm": 2.1083743572235107, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6956242918968201, + "num_tokens": 169922616.0, + "step": 6562 + }, + { + "epoch": 0.7207335822534593, + "grad_norm": 2.1009130477905273, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.687250018119812, + "num_tokens": 169950672.0, + "step": 6563 + }, + { + "epoch": 0.7208433999560729, + "grad_norm": 2.2406349182128906, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6987457275390625, + "num_tokens": 169974535.0, + "step": 6564 + }, + { + "epoch": 0.7209532176586866, + "grad_norm": 2.091644048690796, + "learning_rate": 1e-06, + "loss": 1.0794, + "mean_token_accuracy": 0.6815940141677856, + "num_tokens": 170002421.0, + "step": 6565 + }, + { + "epoch": 0.7210630353613002, + "grad_norm": 1.9450709819793701, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.718932032585144, + "num_tokens": 170032139.0, + "step": 6566 + }, + { + "epoch": 0.7211728530639139, + "grad_norm": 2.4527833461761475, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.705653190612793, + "num_tokens": 170051814.0, + "step": 6567 + }, + { + "epoch": 0.7212826707665275, + "grad_norm": 1.975307583808899, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.714223325252533, + "num_tokens": 170080831.0, + "step": 6568 + }, + { + "epoch": 0.7213924884691413, + "grad_norm": 2.029313564300537, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7027555704116821, + "num_tokens": 170108149.0, + "step": 6569 + }, + { + "epoch": 0.7215023061717549, + "grad_norm": 2.107203960418701, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6869747638702393, + "num_tokens": 170137356.0, + "step": 6570 + }, + { + "epoch": 0.7216121238743686, + "grad_norm": 2.284170389175415, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7099714279174805, + "num_tokens": 170161300.0, + "step": 6571 + }, + { + "epoch": 0.7217219415769822, + "grad_norm": 2.219137191772461, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.727253794670105, + "num_tokens": 170184667.0, + "step": 6572 + }, + { + "epoch": 0.7218317592795959, + "grad_norm": 1.9636949300765991, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6927852630615234, + "num_tokens": 170215863.0, + "step": 6573 + }, + { + "epoch": 0.7219415769822095, + "grad_norm": 2.2647011280059814, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7137281894683838, + "num_tokens": 170239781.0, + "step": 6574 + }, + { + "epoch": 0.7220513946848232, + "grad_norm": 2.0878050327301025, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6944255232810974, + "num_tokens": 170266272.0, + "step": 6575 + }, + { + "epoch": 0.7221612123874368, + "grad_norm": 2.283698558807373, + "learning_rate": 1e-06, + "loss": 1.0732, + "mean_token_accuracy": 0.675619900226593, + "num_tokens": 170293997.0, + "step": 6576 + }, + { + "epoch": 0.7222710300900506, + "grad_norm": 2.253014087677002, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7375991344451904, + "num_tokens": 170316894.0, + "step": 6577 + }, + { + "epoch": 0.7223808477926642, + "grad_norm": 2.1172449588775635, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7236030101776123, + "num_tokens": 170343035.0, + "step": 6578 + }, + { + "epoch": 0.7224906654952779, + "grad_norm": 2.241605043411255, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6991373896598816, + "num_tokens": 170368423.0, + "step": 6579 + }, + { + "epoch": 0.7226004831978915, + "grad_norm": 2.152207612991333, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7244887948036194, + "num_tokens": 170393599.0, + "step": 6580 + }, + { + "epoch": 0.7227103009005051, + "grad_norm": 2.1017532348632812, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7269641160964966, + "num_tokens": 170419722.0, + "step": 6581 + }, + { + "epoch": 0.7228201186031188, + "grad_norm": 2.266523838043213, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7117871642112732, + "num_tokens": 170443024.0, + "step": 6582 + }, + { + "epoch": 0.7229299363057324, + "grad_norm": 2.009005069732666, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7098356485366821, + "num_tokens": 170471462.0, + "step": 6583 + }, + { + "epoch": 0.7230397540083462, + "grad_norm": 2.0250632762908936, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.683917760848999, + "num_tokens": 170502437.0, + "step": 6584 + }, + { + "epoch": 0.7231495717109598, + "grad_norm": 2.295056104660034, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7192339897155762, + "num_tokens": 170525603.0, + "step": 6585 + }, + { + "epoch": 0.7232593894135735, + "grad_norm": 2.0359396934509277, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7039330005645752, + "num_tokens": 170555958.0, + "step": 6586 + }, + { + "epoch": 0.7233692071161871, + "grad_norm": 2.2393240928649902, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7029015421867371, + "num_tokens": 170579904.0, + "step": 6587 + }, + { + "epoch": 0.7234790248188008, + "grad_norm": 2.6172685623168945, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.718192458152771, + "num_tokens": 170599073.0, + "step": 6588 + }, + { + "epoch": 0.7235888425214144, + "grad_norm": 2.029716730117798, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7291582822799683, + "num_tokens": 170624812.0, + "step": 6589 + }, + { + "epoch": 0.7236986602240281, + "grad_norm": 2.2967324256896973, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6971309185028076, + "num_tokens": 170649253.0, + "step": 6590 + }, + { + "epoch": 0.7238084779266418, + "grad_norm": 2.281965970993042, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6950826644897461, + "num_tokens": 170672443.0, + "step": 6591 + }, + { + "epoch": 0.7239182956292555, + "grad_norm": 1.9893114566802979, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6927369832992554, + "num_tokens": 170701394.0, + "step": 6592 + }, + { + "epoch": 0.7240281133318691, + "grad_norm": 2.349247455596924, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7092386484146118, + "num_tokens": 170725227.0, + "step": 6593 + }, + { + "epoch": 0.7241379310344828, + "grad_norm": 2.4383552074432373, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7087442278862, + "num_tokens": 170747080.0, + "step": 6594 + }, + { + "epoch": 0.7242477487370964, + "grad_norm": 2.1486105918884277, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7239931225776672, + "num_tokens": 170769740.0, + "step": 6595 + }, + { + "epoch": 0.72435756643971, + "grad_norm": 2.237290859222412, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6920271515846252, + "num_tokens": 170795664.0, + "step": 6596 + }, + { + "epoch": 0.7244673841423237, + "grad_norm": 1.9542890787124634, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7169482707977295, + "num_tokens": 170824411.0, + "step": 6597 + }, + { + "epoch": 0.7245772018449375, + "grad_norm": 2.2025609016418457, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6890785098075867, + "num_tokens": 170850123.0, + "step": 6598 + }, + { + "epoch": 0.7246870195475511, + "grad_norm": 2.356480598449707, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7029724717140198, + "num_tokens": 170873265.0, + "step": 6599 + }, + { + "epoch": 0.7247968372501647, + "grad_norm": 2.1325244903564453, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.6989237666130066, + "num_tokens": 170899152.0, + "step": 6600 + }, + { + "epoch": 0.7249066549527784, + "grad_norm": 2.0685157775878906, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6982118487358093, + "num_tokens": 170929558.0, + "step": 6601 + }, + { + "epoch": 0.725016472655392, + "grad_norm": 2.175747871398926, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7190540432929993, + "num_tokens": 170955891.0, + "step": 6602 + }, + { + "epoch": 0.7251262903580057, + "grad_norm": 2.01259708404541, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7154697179794312, + "num_tokens": 170984853.0, + "step": 6603 + }, + { + "epoch": 0.7252361080606193, + "grad_norm": 1.892754316329956, + "learning_rate": 1e-06, + "loss": 1.0901, + "mean_token_accuracy": 0.6780647039413452, + "num_tokens": 171017608.0, + "step": 6604 + }, + { + "epoch": 0.725345925763233, + "grad_norm": 2.298304796218872, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.6994525194168091, + "num_tokens": 171043941.0, + "step": 6605 + }, + { + "epoch": 0.7254557434658467, + "grad_norm": 2.335739850997925, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6955779790878296, + "num_tokens": 171067231.0, + "step": 6606 + }, + { + "epoch": 0.7255655611684604, + "grad_norm": 2.027204990386963, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6995810270309448, + "num_tokens": 171094751.0, + "step": 6607 + }, + { + "epoch": 0.725675378871074, + "grad_norm": 2.213911533355713, + "learning_rate": 1e-06, + "loss": 1.0391, + "mean_token_accuracy": 0.6924858093261719, + "num_tokens": 171121793.0, + "step": 6608 + }, + { + "epoch": 0.7257851965736877, + "grad_norm": 2.2084226608276367, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6973967552185059, + "num_tokens": 171147151.0, + "step": 6609 + }, + { + "epoch": 0.7258950142763013, + "grad_norm": 2.1433651447296143, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7016392350196838, + "num_tokens": 171173165.0, + "step": 6610 + }, + { + "epoch": 0.726004831978915, + "grad_norm": 2.257035493850708, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7242976427078247, + "num_tokens": 171195083.0, + "step": 6611 + }, + { + "epoch": 0.7261146496815286, + "grad_norm": 1.9122368097305298, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6879273653030396, + "num_tokens": 171227814.0, + "step": 6612 + }, + { + "epoch": 0.7262244673841424, + "grad_norm": 2.170332908630371, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6918916702270508, + "num_tokens": 171255157.0, + "step": 6613 + }, + { + "epoch": 0.726334285086756, + "grad_norm": 2.1346311569213867, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.710138738155365, + "num_tokens": 171279921.0, + "step": 6614 + }, + { + "epoch": 0.7264441027893697, + "grad_norm": 2.269160509109497, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6908503770828247, + "num_tokens": 171306367.0, + "step": 6615 + }, + { + "epoch": 0.7265539204919833, + "grad_norm": 2.0875890254974365, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.690769374370575, + "num_tokens": 171335470.0, + "step": 6616 + }, + { + "epoch": 0.726663738194597, + "grad_norm": 2.2718071937561035, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6959962844848633, + "num_tokens": 171360291.0, + "step": 6617 + }, + { + "epoch": 0.7267735558972106, + "grad_norm": 2.2091493606567383, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.7097505331039429, + "num_tokens": 171386241.0, + "step": 6618 + }, + { + "epoch": 0.7268833735998242, + "grad_norm": 2.183598756790161, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6895036697387695, + "num_tokens": 171412587.0, + "step": 6619 + }, + { + "epoch": 0.726993191302438, + "grad_norm": 2.3910934925079346, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.705211341381073, + "num_tokens": 171434516.0, + "step": 6620 + }, + { + "epoch": 0.7271030090050516, + "grad_norm": 2.0241992473602295, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6870172619819641, + "num_tokens": 171464916.0, + "step": 6621 + }, + { + "epoch": 0.7272128267076653, + "grad_norm": 2.274887800216675, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6968658566474915, + "num_tokens": 171489467.0, + "step": 6622 + }, + { + "epoch": 0.7273226444102789, + "grad_norm": 2.615593910217285, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7391057014465332, + "num_tokens": 171506265.0, + "step": 6623 + }, + { + "epoch": 0.7274324621128926, + "grad_norm": 2.356276273727417, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7303932905197144, + "num_tokens": 171527160.0, + "step": 6624 + }, + { + "epoch": 0.7275422798155062, + "grad_norm": 2.1381638050079346, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.712786853313446, + "num_tokens": 171552198.0, + "step": 6625 + }, + { + "epoch": 0.7276520975181199, + "grad_norm": 2.4542927742004395, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7101707458496094, + "num_tokens": 171573001.0, + "step": 6626 + }, + { + "epoch": 0.7277619152207336, + "grad_norm": 1.9521913528442383, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6903769969940186, + "num_tokens": 171602431.0, + "step": 6627 + }, + { + "epoch": 0.7278717329233473, + "grad_norm": 2.0919559001922607, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7027682662010193, + "num_tokens": 171631809.0, + "step": 6628 + }, + { + "epoch": 0.7279815506259609, + "grad_norm": 2.298121452331543, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7037454843521118, + "num_tokens": 171655463.0, + "step": 6629 + }, + { + "epoch": 0.7280913683285746, + "grad_norm": 2.2491614818573, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7210122346878052, + "num_tokens": 171679233.0, + "step": 6630 + }, + { + "epoch": 0.7282011860311882, + "grad_norm": 2.248455286026001, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7139862179756165, + "num_tokens": 171704875.0, + "step": 6631 + }, + { + "epoch": 0.7283110037338019, + "grad_norm": 1.8919708728790283, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7155276536941528, + "num_tokens": 171736957.0, + "step": 6632 + }, + { + "epoch": 0.7284208214364155, + "grad_norm": 2.057124137878418, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6835113763809204, + "num_tokens": 171765231.0, + "step": 6633 + }, + { + "epoch": 0.7285306391390292, + "grad_norm": 2.3590900897979736, + "learning_rate": 1e-06, + "loss": 1.047, + "mean_token_accuracy": 0.689426064491272, + "num_tokens": 171790280.0, + "step": 6634 + }, + { + "epoch": 0.7286404568416429, + "grad_norm": 2.022498846054077, + "learning_rate": 1e-06, + "loss": 1.0713, + "mean_token_accuracy": 0.684787392616272, + "num_tokens": 171823396.0, + "step": 6635 + }, + { + "epoch": 0.7287502745442566, + "grad_norm": 1.9542784690856934, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7152692079544067, + "num_tokens": 171852932.0, + "step": 6636 + }, + { + "epoch": 0.7288600922468702, + "grad_norm": 2.030463218688965, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6772078275680542, + "num_tokens": 171882645.0, + "step": 6637 + }, + { + "epoch": 0.7289699099494839, + "grad_norm": 2.2595486640930176, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.6888211965560913, + "num_tokens": 171907999.0, + "step": 6638 + }, + { + "epoch": 0.7290797276520975, + "grad_norm": 2.46171236038208, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7193639278411865, + "num_tokens": 171929437.0, + "step": 6639 + }, + { + "epoch": 0.7291895453547111, + "grad_norm": 2.2585737705230713, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.720603346824646, + "num_tokens": 171952768.0, + "step": 6640 + }, + { + "epoch": 0.7292993630573248, + "grad_norm": 1.9954103231430054, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6985140442848206, + "num_tokens": 171982817.0, + "step": 6641 + }, + { + "epoch": 0.7294091807599385, + "grad_norm": 2.195566177368164, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7053636312484741, + "num_tokens": 172008264.0, + "step": 6642 + }, + { + "epoch": 0.7295189984625522, + "grad_norm": 2.0862338542938232, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7007976770401001, + "num_tokens": 172037187.0, + "step": 6643 + }, + { + "epoch": 0.7296288161651658, + "grad_norm": 2.3143515586853027, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.701356053352356, + "num_tokens": 172061265.0, + "step": 6644 + }, + { + "epoch": 0.7297386338677795, + "grad_norm": 2.233478307723999, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7123146653175354, + "num_tokens": 172088106.0, + "step": 6645 + }, + { + "epoch": 0.7298484515703931, + "grad_norm": 2.23612117767334, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.7002642154693604, + "num_tokens": 172111278.0, + "step": 6646 + }, + { + "epoch": 0.7299582692730068, + "grad_norm": 2.1161370277404785, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6950209736824036, + "num_tokens": 172139161.0, + "step": 6647 + }, + { + "epoch": 0.7300680869756204, + "grad_norm": 2.2081706523895264, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6863566040992737, + "num_tokens": 172164497.0, + "step": 6648 + }, + { + "epoch": 0.7301779046782342, + "grad_norm": 1.9213217496871948, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6901053786277771, + "num_tokens": 172195776.0, + "step": 6649 + }, + { + "epoch": 0.7302877223808478, + "grad_norm": 2.0918235778808594, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7153035998344421, + "num_tokens": 172222583.0, + "step": 6650 + }, + { + "epoch": 0.7303975400834615, + "grad_norm": 2.2283990383148193, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6941280364990234, + "num_tokens": 172247462.0, + "step": 6651 + }, + { + "epoch": 0.7305073577860751, + "grad_norm": 2.0827410221099854, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6923978924751282, + "num_tokens": 172275377.0, + "step": 6652 + }, + { + "epoch": 0.7306171754886888, + "grad_norm": 2.276963472366333, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7267279028892517, + "num_tokens": 172297610.0, + "step": 6653 + }, + { + "epoch": 0.7307269931913024, + "grad_norm": 2.1334688663482666, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.714179277420044, + "num_tokens": 172322826.0, + "step": 6654 + }, + { + "epoch": 0.7308368108939161, + "grad_norm": 2.290928840637207, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6979209184646606, + "num_tokens": 172347222.0, + "step": 6655 + }, + { + "epoch": 0.7309466285965298, + "grad_norm": 1.9935426712036133, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7274065017700195, + "num_tokens": 172373660.0, + "step": 6656 + }, + { + "epoch": 0.7310564462991435, + "grad_norm": 1.997296690940857, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6946890950202942, + "num_tokens": 172402897.0, + "step": 6657 + }, + { + "epoch": 0.7311662640017571, + "grad_norm": 2.288404941558838, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7137801647186279, + "num_tokens": 172424908.0, + "step": 6658 + }, + { + "epoch": 0.7312760817043708, + "grad_norm": 2.1414175033569336, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7025283575057983, + "num_tokens": 172452343.0, + "step": 6659 + }, + { + "epoch": 0.7313858994069844, + "grad_norm": 2.18664288520813, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7318522930145264, + "num_tokens": 172474687.0, + "step": 6660 + }, + { + "epoch": 0.731495717109598, + "grad_norm": 2.073462963104248, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7167308330535889, + "num_tokens": 172501083.0, + "step": 6661 + }, + { + "epoch": 0.7316055348122117, + "grad_norm": 2.0476737022399902, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6945909261703491, + "num_tokens": 172529837.0, + "step": 6662 + }, + { + "epoch": 0.7317153525148253, + "grad_norm": 1.880821704864502, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7017194628715515, + "num_tokens": 172565242.0, + "step": 6663 + }, + { + "epoch": 0.7318251702174391, + "grad_norm": 2.1424508094787598, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7233844995498657, + "num_tokens": 172590053.0, + "step": 6664 + }, + { + "epoch": 0.7319349879200527, + "grad_norm": 1.9465959072113037, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.717574954032898, + "num_tokens": 172617737.0, + "step": 6665 + }, + { + "epoch": 0.7320448056226664, + "grad_norm": 2.0975143909454346, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7150382399559021, + "num_tokens": 172643719.0, + "step": 6666 + }, + { + "epoch": 0.73215462332528, + "grad_norm": 2.0820882320404053, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7112289071083069, + "num_tokens": 172670802.0, + "step": 6667 + }, + { + "epoch": 0.7322644410278937, + "grad_norm": 1.9190744161605835, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7023041844367981, + "num_tokens": 172701329.0, + "step": 6668 + }, + { + "epoch": 0.7323742587305073, + "grad_norm": 2.353149890899658, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7015392184257507, + "num_tokens": 172723687.0, + "step": 6669 + }, + { + "epoch": 0.732484076433121, + "grad_norm": 1.9858652353286743, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7019702196121216, + "num_tokens": 172752354.0, + "step": 6670 + }, + { + "epoch": 0.7325938941357347, + "grad_norm": 2.2889535427093506, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.7006133198738098, + "num_tokens": 172776323.0, + "step": 6671 + }, + { + "epoch": 0.7327037118383484, + "grad_norm": 2.2304558753967285, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.715806782245636, + "num_tokens": 172800904.0, + "step": 6672 + }, + { + "epoch": 0.732813529540962, + "grad_norm": 2.2878172397613525, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7137762904167175, + "num_tokens": 172823370.0, + "step": 6673 + }, + { + "epoch": 0.7329233472435757, + "grad_norm": 2.3927390575408936, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7130171060562134, + "num_tokens": 172846258.0, + "step": 6674 + }, + { + "epoch": 0.7330331649461893, + "grad_norm": 2.4400298595428467, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7048784494400024, + "num_tokens": 172867940.0, + "step": 6675 + }, + { + "epoch": 0.733142982648803, + "grad_norm": 2.634507417678833, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7038372755050659, + "num_tokens": 172886700.0, + "step": 6676 + }, + { + "epoch": 0.7332528003514166, + "grad_norm": 2.1712090969085693, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7049305438995361, + "num_tokens": 172914941.0, + "step": 6677 + }, + { + "epoch": 0.7333626180540304, + "grad_norm": 2.1381888389587402, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7188048958778381, + "num_tokens": 172939971.0, + "step": 6678 + }, + { + "epoch": 0.733472435756644, + "grad_norm": 2.2050399780273438, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7099297642707825, + "num_tokens": 172968136.0, + "step": 6679 + }, + { + "epoch": 0.7335822534592576, + "grad_norm": 2.173815965652466, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7062753438949585, + "num_tokens": 172993683.0, + "step": 6680 + }, + { + "epoch": 0.7336920711618713, + "grad_norm": 2.2752513885498047, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7007932662963867, + "num_tokens": 173018448.0, + "step": 6681 + }, + { + "epoch": 0.7338018888644849, + "grad_norm": 2.174783706665039, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7026939392089844, + "num_tokens": 173044670.0, + "step": 6682 + }, + { + "epoch": 0.7339117065670986, + "grad_norm": 2.1684799194335938, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6951045989990234, + "num_tokens": 173072158.0, + "step": 6683 + }, + { + "epoch": 0.7340215242697122, + "grad_norm": 1.9502087831497192, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7115907073020935, + "num_tokens": 173101918.0, + "step": 6684 + }, + { + "epoch": 0.734131341972326, + "grad_norm": 2.2227602005004883, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7001222372055054, + "num_tokens": 173127160.0, + "step": 6685 + }, + { + "epoch": 0.7342411596749396, + "grad_norm": 2.0957703590393066, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6923692226409912, + "num_tokens": 173153811.0, + "step": 6686 + }, + { + "epoch": 0.7343509773775533, + "grad_norm": 2.2471530437469482, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6863453388214111, + "num_tokens": 173178814.0, + "step": 6687 + }, + { + "epoch": 0.7344607950801669, + "grad_norm": 2.381589651107788, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6931331753730774, + "num_tokens": 173200290.0, + "step": 6688 + }, + { + "epoch": 0.7345706127827806, + "grad_norm": 2.09917950630188, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6937131285667419, + "num_tokens": 173226366.0, + "step": 6689 + }, + { + "epoch": 0.7346804304853942, + "grad_norm": 2.266364336013794, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6983671188354492, + "num_tokens": 173253655.0, + "step": 6690 + }, + { + "epoch": 0.7347902481880079, + "grad_norm": 2.5201735496520996, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7183204293251038, + "num_tokens": 173275181.0, + "step": 6691 + }, + { + "epoch": 0.7349000658906216, + "grad_norm": 2.1351637840270996, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.711013674736023, + "num_tokens": 173302924.0, + "step": 6692 + }, + { + "epoch": 0.7350098835932353, + "grad_norm": 2.224623441696167, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6998182535171509, + "num_tokens": 173327534.0, + "step": 6693 + }, + { + "epoch": 0.7351197012958489, + "grad_norm": 2.2177159786224365, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7208199501037598, + "num_tokens": 173352614.0, + "step": 6694 + }, + { + "epoch": 0.7352295189984626, + "grad_norm": 2.38948655128479, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7067337036132812, + "num_tokens": 173374986.0, + "step": 6695 + }, + { + "epoch": 0.7353393367010762, + "grad_norm": 2.6194469928741455, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7050711512565613, + "num_tokens": 173394416.0, + "step": 6696 + }, + { + "epoch": 0.7354491544036899, + "grad_norm": 2.302459955215454, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7197726964950562, + "num_tokens": 173417035.0, + "step": 6697 + }, + { + "epoch": 0.7355589721063035, + "grad_norm": 1.9296612739562988, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7162176966667175, + "num_tokens": 173447210.0, + "step": 6698 + }, + { + "epoch": 0.7356687898089171, + "grad_norm": 2.157504081726074, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7040982842445374, + "num_tokens": 173473313.0, + "step": 6699 + }, + { + "epoch": 0.7357786075115309, + "grad_norm": 2.2995457649230957, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7401264905929565, + "num_tokens": 173495252.0, + "step": 6700 + }, + { + "epoch": 0.7358884252141445, + "grad_norm": 2.1747255325317383, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7258237600326538, + "num_tokens": 173521442.0, + "step": 6701 + }, + { + "epoch": 0.7359982429167582, + "grad_norm": 2.1140260696411133, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7231008410453796, + "num_tokens": 173547300.0, + "step": 6702 + }, + { + "epoch": 0.7361080606193718, + "grad_norm": 2.143791913986206, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7147305607795715, + "num_tokens": 173576514.0, + "step": 6703 + }, + { + "epoch": 0.7362178783219855, + "grad_norm": 2.333754539489746, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.7023166418075562, + "num_tokens": 173602159.0, + "step": 6704 + }, + { + "epoch": 0.7363276960245991, + "grad_norm": 2.168830394744873, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.710299015045166, + "num_tokens": 173627353.0, + "step": 6705 + }, + { + "epoch": 0.7364375137272128, + "grad_norm": 2.057453155517578, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7252127528190613, + "num_tokens": 173655914.0, + "step": 6706 + }, + { + "epoch": 0.7365473314298265, + "grad_norm": 2.1701607704162598, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7013096809387207, + "num_tokens": 173682089.0, + "step": 6707 + }, + { + "epoch": 0.7366571491324402, + "grad_norm": 2.066948890686035, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7224330902099609, + "num_tokens": 173710899.0, + "step": 6708 + }, + { + "epoch": 0.7367669668350538, + "grad_norm": 2.3770053386688232, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.6993509531021118, + "num_tokens": 173734020.0, + "step": 6709 + }, + { + "epoch": 0.7368767845376675, + "grad_norm": 1.9971723556518555, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7173944711685181, + "num_tokens": 173763730.0, + "step": 6710 + }, + { + "epoch": 0.7369866022402811, + "grad_norm": 2.314847946166992, + "learning_rate": 1e-06, + "loss": 1.0566, + "mean_token_accuracy": 0.6732786297798157, + "num_tokens": 173789322.0, + "step": 6711 + }, + { + "epoch": 0.7370964199428948, + "grad_norm": 1.9065707921981812, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6909366250038147, + "num_tokens": 173821838.0, + "step": 6712 + }, + { + "epoch": 0.7372062376455084, + "grad_norm": 2.114274501800537, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7192689776420593, + "num_tokens": 173848132.0, + "step": 6713 + }, + { + "epoch": 0.7373160553481222, + "grad_norm": 2.2044453620910645, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7308074831962585, + "num_tokens": 173872793.0, + "step": 6714 + }, + { + "epoch": 0.7374258730507358, + "grad_norm": 2.1795334815979004, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7186362743377686, + "num_tokens": 173897476.0, + "step": 6715 + }, + { + "epoch": 0.7375356907533495, + "grad_norm": 2.260895013809204, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7041877508163452, + "num_tokens": 173921544.0, + "step": 6716 + }, + { + "epoch": 0.7376455084559631, + "grad_norm": 2.4360525608062744, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7187983393669128, + "num_tokens": 173942604.0, + "step": 6717 + }, + { + "epoch": 0.7377553261585768, + "grad_norm": 2.2381227016448975, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7073426842689514, + "num_tokens": 173967385.0, + "step": 6718 + }, + { + "epoch": 0.7378651438611904, + "grad_norm": 2.353100061416626, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7165935039520264, + "num_tokens": 173989018.0, + "step": 6719 + }, + { + "epoch": 0.737974961563804, + "grad_norm": 2.1897342205047607, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6884959936141968, + "num_tokens": 174014756.0, + "step": 6720 + }, + { + "epoch": 0.7380847792664178, + "grad_norm": 2.60565447807312, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6953562498092651, + "num_tokens": 174036322.0, + "step": 6721 + }, + { + "epoch": 0.7381945969690314, + "grad_norm": 2.078964948654175, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7313518524169922, + "num_tokens": 174061966.0, + "step": 6722 + }, + { + "epoch": 0.7383044146716451, + "grad_norm": 2.509531021118164, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7005380392074585, + "num_tokens": 174082133.0, + "step": 6723 + }, + { + "epoch": 0.7384142323742587, + "grad_norm": 1.9126195907592773, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7030669450759888, + "num_tokens": 174114587.0, + "step": 6724 + }, + { + "epoch": 0.7385240500768724, + "grad_norm": 2.014941453933716, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6845773458480835, + "num_tokens": 174144910.0, + "step": 6725 + }, + { + "epoch": 0.738633867779486, + "grad_norm": 2.1770379543304443, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7061156034469604, + "num_tokens": 174170044.0, + "step": 6726 + }, + { + "epoch": 0.7387436854820997, + "grad_norm": 2.1628873348236084, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7297991514205933, + "num_tokens": 174194128.0, + "step": 6727 + }, + { + "epoch": 0.7388535031847133, + "grad_norm": 2.4349706172943115, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7311304807662964, + "num_tokens": 174213613.0, + "step": 6728 + }, + { + "epoch": 0.7389633208873271, + "grad_norm": 2.2072904109954834, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7219299077987671, + "num_tokens": 174239072.0, + "step": 6729 + }, + { + "epoch": 0.7390731385899407, + "grad_norm": 2.087523937225342, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6946083903312683, + "num_tokens": 174266710.0, + "step": 6730 + }, + { + "epoch": 0.7391829562925544, + "grad_norm": 2.3253962993621826, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7079919576644897, + "num_tokens": 174289578.0, + "step": 6731 + }, + { + "epoch": 0.739292773995168, + "grad_norm": 2.316406011581421, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7077579498291016, + "num_tokens": 174312357.0, + "step": 6732 + }, + { + "epoch": 0.7394025916977817, + "grad_norm": 2.515883445739746, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7126941084861755, + "num_tokens": 174333509.0, + "step": 6733 + }, + { + "epoch": 0.7395124094003953, + "grad_norm": 2.337545871734619, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7269522547721863, + "num_tokens": 174354178.0, + "step": 6734 + }, + { + "epoch": 0.739622227103009, + "grad_norm": 2.3394954204559326, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7045649886131287, + "num_tokens": 174377254.0, + "step": 6735 + }, + { + "epoch": 0.7397320448056227, + "grad_norm": 2.266268491744995, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7140514254570007, + "num_tokens": 174400581.0, + "step": 6736 + }, + { + "epoch": 0.7398418625082364, + "grad_norm": 1.918775200843811, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7009677886962891, + "num_tokens": 174431436.0, + "step": 6737 + }, + { + "epoch": 0.73995168021085, + "grad_norm": 2.3079965114593506, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.7069698572158813, + "num_tokens": 174456476.0, + "step": 6738 + }, + { + "epoch": 0.7400614979134637, + "grad_norm": 2.3490183353424072, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6897099018096924, + "num_tokens": 174479634.0, + "step": 6739 + }, + { + "epoch": 0.7401713156160773, + "grad_norm": 2.212517023086548, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.706476628780365, + "num_tokens": 174504223.0, + "step": 6740 + }, + { + "epoch": 0.7402811333186909, + "grad_norm": 2.0408096313476562, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7268068790435791, + "num_tokens": 174532251.0, + "step": 6741 + }, + { + "epoch": 0.7403909510213046, + "grad_norm": 2.5450901985168457, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.721865177154541, + "num_tokens": 174551363.0, + "step": 6742 + }, + { + "epoch": 0.7405007687239183, + "grad_norm": 1.8744332790374756, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7124196290969849, + "num_tokens": 174585552.0, + "step": 6743 + }, + { + "epoch": 0.740610586426532, + "grad_norm": 2.23429274559021, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6946254968643188, + "num_tokens": 174612006.0, + "step": 6744 + }, + { + "epoch": 0.7407204041291456, + "grad_norm": 2.315270185470581, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6992857456207275, + "num_tokens": 174635438.0, + "step": 6745 + }, + { + "epoch": 0.7408302218317593, + "grad_norm": 1.9980882406234741, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7010277509689331, + "num_tokens": 174663764.0, + "step": 6746 + }, + { + "epoch": 0.7409400395343729, + "grad_norm": 2.3107352256774902, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.711273193359375, + "num_tokens": 174686855.0, + "step": 6747 + }, + { + "epoch": 0.7410498572369866, + "grad_norm": 2.251725912094116, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6942903995513916, + "num_tokens": 174712408.0, + "step": 6748 + }, + { + "epoch": 0.7411596749396002, + "grad_norm": 1.9730650186538696, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7006750106811523, + "num_tokens": 174746301.0, + "step": 6749 + }, + { + "epoch": 0.741269492642214, + "grad_norm": 2.1434640884399414, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7077633142471313, + "num_tokens": 174772707.0, + "step": 6750 + }, + { + "epoch": 0.7413793103448276, + "grad_norm": 1.9998273849487305, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.701447606086731, + "num_tokens": 174800887.0, + "step": 6751 + }, + { + "epoch": 0.7414891280474413, + "grad_norm": 2.162184238433838, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.722146213054657, + "num_tokens": 174825484.0, + "step": 6752 + }, + { + "epoch": 0.7415989457500549, + "grad_norm": 2.0272116661071777, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6996662616729736, + "num_tokens": 174856788.0, + "step": 6753 + }, + { + "epoch": 0.7417087634526686, + "grad_norm": 2.076878309249878, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7196009159088135, + "num_tokens": 174883870.0, + "step": 6754 + }, + { + "epoch": 0.7418185811552822, + "grad_norm": 2.117802381515503, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7142010927200317, + "num_tokens": 174911277.0, + "step": 6755 + }, + { + "epoch": 0.7419283988578959, + "grad_norm": 2.1811447143554688, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.718951404094696, + "num_tokens": 174937395.0, + "step": 6756 + }, + { + "epoch": 0.7420382165605095, + "grad_norm": 2.220085859298706, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6868433952331543, + "num_tokens": 174962830.0, + "step": 6757 + }, + { + "epoch": 0.7421480342631233, + "grad_norm": 2.1199793815612793, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7018470168113708, + "num_tokens": 174989345.0, + "step": 6758 + }, + { + "epoch": 0.7422578519657369, + "grad_norm": 2.2785181999206543, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7198790311813354, + "num_tokens": 175012986.0, + "step": 6759 + }, + { + "epoch": 0.7423676696683505, + "grad_norm": 2.5802152156829834, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.71181321144104, + "num_tokens": 175032096.0, + "step": 6760 + }, + { + "epoch": 0.7424774873709642, + "grad_norm": 2.5283961296081543, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7331610918045044, + "num_tokens": 175051955.0, + "step": 6761 + }, + { + "epoch": 0.7425873050735778, + "grad_norm": 2.199967622756958, + "learning_rate": 1e-06, + "loss": 1.0712, + "mean_token_accuracy": 0.6803280115127563, + "num_tokens": 175078938.0, + "step": 6762 + }, + { + "epoch": 0.7426971227761915, + "grad_norm": 2.0981945991516113, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7223629951477051, + "num_tokens": 175104952.0, + "step": 6763 + }, + { + "epoch": 0.7428069404788051, + "grad_norm": 2.243560791015625, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7006756663322449, + "num_tokens": 175129499.0, + "step": 6764 + }, + { + "epoch": 0.7429167581814189, + "grad_norm": 2.3127245903015137, + "learning_rate": 1e-06, + "loss": 1.0696, + "mean_token_accuracy": 0.6790080070495605, + "num_tokens": 175155643.0, + "step": 6765 + }, + { + "epoch": 0.7430265758840325, + "grad_norm": 2.3121767044067383, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6991763710975647, + "num_tokens": 175179215.0, + "step": 6766 + }, + { + "epoch": 0.7431363935866462, + "grad_norm": 2.060131072998047, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6973452568054199, + "num_tokens": 175208654.0, + "step": 6767 + }, + { + "epoch": 0.7432462112892598, + "grad_norm": 2.141962766647339, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.7031893730163574, + "num_tokens": 175235570.0, + "step": 6768 + }, + { + "epoch": 0.7433560289918735, + "grad_norm": 2.1853222846984863, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6976622939109802, + "num_tokens": 175260521.0, + "step": 6769 + }, + { + "epoch": 0.7434658466944871, + "grad_norm": 2.1141152381896973, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7208790183067322, + "num_tokens": 175287083.0, + "step": 6770 + }, + { + "epoch": 0.7435756643971008, + "grad_norm": 2.383975028991699, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7177305221557617, + "num_tokens": 175309073.0, + "step": 6771 + }, + { + "epoch": 0.7436854820997145, + "grad_norm": 2.5722851753234863, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7104700207710266, + "num_tokens": 175327506.0, + "step": 6772 + }, + { + "epoch": 0.7437952998023282, + "grad_norm": 2.4996559619903564, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7109643220901489, + "num_tokens": 175350031.0, + "step": 6773 + }, + { + "epoch": 0.7439051175049418, + "grad_norm": 2.2991573810577393, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.71200031042099, + "num_tokens": 175371750.0, + "step": 6774 + }, + { + "epoch": 0.7440149352075555, + "grad_norm": 2.058643341064453, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6799081563949585, + "num_tokens": 175398988.0, + "step": 6775 + }, + { + "epoch": 0.7441247529101691, + "grad_norm": 2.025805950164795, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7102190256118774, + "num_tokens": 175427102.0, + "step": 6776 + }, + { + "epoch": 0.7442345706127828, + "grad_norm": 2.1442694664001465, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7069706916809082, + "num_tokens": 175454098.0, + "step": 6777 + }, + { + "epoch": 0.7443443883153964, + "grad_norm": 2.059512138366699, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6911953091621399, + "num_tokens": 175481485.0, + "step": 6778 + }, + { + "epoch": 0.7444542060180102, + "grad_norm": 2.258939027786255, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7078968286514282, + "num_tokens": 175505092.0, + "step": 6779 + }, + { + "epoch": 0.7445640237206238, + "grad_norm": 2.2412211894989014, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6981290578842163, + "num_tokens": 175529187.0, + "step": 6780 + }, + { + "epoch": 0.7446738414232374, + "grad_norm": 2.3492414951324463, + "learning_rate": 1e-06, + "loss": 1.0491, + "mean_token_accuracy": 0.6904446482658386, + "num_tokens": 175554714.0, + "step": 6781 + }, + { + "epoch": 0.7447836591258511, + "grad_norm": 2.1020820140838623, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.6991456747055054, + "num_tokens": 175582796.0, + "step": 6782 + }, + { + "epoch": 0.7448934768284647, + "grad_norm": 2.2981395721435547, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7028151154518127, + "num_tokens": 175606081.0, + "step": 6783 + }, + { + "epoch": 0.7450032945310784, + "grad_norm": 2.1228420734405518, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7229883074760437, + "num_tokens": 175632000.0, + "step": 6784 + }, + { + "epoch": 0.745113112233692, + "grad_norm": 2.0732758045196533, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7169715166091919, + "num_tokens": 175658364.0, + "step": 6785 + }, + { + "epoch": 0.7452229299363057, + "grad_norm": 2.391315221786499, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.712266743183136, + "num_tokens": 175683050.0, + "step": 6786 + }, + { + "epoch": 0.7453327476389194, + "grad_norm": 2.261716842651367, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7186763286590576, + "num_tokens": 175706060.0, + "step": 6787 + }, + { + "epoch": 0.7454425653415331, + "grad_norm": 2.023470878601074, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7163313627243042, + "num_tokens": 175734301.0, + "step": 6788 + }, + { + "epoch": 0.7455523830441467, + "grad_norm": 2.2307004928588867, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7240160703659058, + "num_tokens": 175757178.0, + "step": 6789 + }, + { + "epoch": 0.7456622007467604, + "grad_norm": 2.494279146194458, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6907519102096558, + "num_tokens": 175778863.0, + "step": 6790 + }, + { + "epoch": 0.745772018449374, + "grad_norm": 2.089555263519287, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7117389440536499, + "num_tokens": 175806023.0, + "step": 6791 + }, + { + "epoch": 0.7458818361519877, + "grad_norm": 2.380113363265991, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7070478200912476, + "num_tokens": 175828049.0, + "step": 6792 + }, + { + "epoch": 0.7459916538546013, + "grad_norm": 2.533752202987671, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7101840972900391, + "num_tokens": 175848135.0, + "step": 6793 + }, + { + "epoch": 0.7461014715572151, + "grad_norm": 2.082369327545166, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7035366296768188, + "num_tokens": 175876498.0, + "step": 6794 + }, + { + "epoch": 0.7462112892598287, + "grad_norm": 2.3180079460144043, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7006424069404602, + "num_tokens": 175902733.0, + "step": 6795 + }, + { + "epoch": 0.7463211069624424, + "grad_norm": 2.393620491027832, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.710825502872467, + "num_tokens": 175925566.0, + "step": 6796 + }, + { + "epoch": 0.746430924665056, + "grad_norm": 2.0282809734344482, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6882525682449341, + "num_tokens": 175955080.0, + "step": 6797 + }, + { + "epoch": 0.7465407423676697, + "grad_norm": 2.323310375213623, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7206614017486572, + "num_tokens": 175977085.0, + "step": 6798 + }, + { + "epoch": 0.7466505600702833, + "grad_norm": 2.2283501625061035, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7110790014266968, + "num_tokens": 176003377.0, + "step": 6799 + }, + { + "epoch": 0.7467603777728969, + "grad_norm": 1.978777289390564, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7092045545578003, + "num_tokens": 176030427.0, + "step": 6800 + }, + { + "epoch": 0.7468701954755107, + "grad_norm": 2.0244157314300537, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.687730073928833, + "num_tokens": 176058425.0, + "step": 6801 + }, + { + "epoch": 0.7469800131781243, + "grad_norm": 2.184178352355957, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7347785234451294, + "num_tokens": 176081969.0, + "step": 6802 + }, + { + "epoch": 0.747089830880738, + "grad_norm": 2.188220262527466, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.719779372215271, + "num_tokens": 176106404.0, + "step": 6803 + }, + { + "epoch": 0.7471996485833516, + "grad_norm": 2.0429184436798096, + "learning_rate": 1e-06, + "loss": 1.0648, + "mean_token_accuracy": 0.6726986169815063, + "num_tokens": 176137354.0, + "step": 6804 + }, + { + "epoch": 0.7473094662859653, + "grad_norm": 2.230621337890625, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7037962079048157, + "num_tokens": 176162288.0, + "step": 6805 + }, + { + "epoch": 0.7474192839885789, + "grad_norm": 2.1735222339630127, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7023691534996033, + "num_tokens": 176189523.0, + "step": 6806 + }, + { + "epoch": 0.7475291016911926, + "grad_norm": 2.308102607727051, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7074000835418701, + "num_tokens": 176213616.0, + "step": 6807 + }, + { + "epoch": 0.7476389193938063, + "grad_norm": 2.187640905380249, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7236518859863281, + "num_tokens": 176236767.0, + "step": 6808 + }, + { + "epoch": 0.74774873709642, + "grad_norm": 2.0773563385009766, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7259109020233154, + "num_tokens": 176263344.0, + "step": 6809 + }, + { + "epoch": 0.7478585547990336, + "grad_norm": 2.193624258041382, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7064318656921387, + "num_tokens": 176288047.0, + "step": 6810 + }, + { + "epoch": 0.7479683725016473, + "grad_norm": 2.01289963722229, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7018662095069885, + "num_tokens": 176317781.0, + "step": 6811 + }, + { + "epoch": 0.7480781902042609, + "grad_norm": 2.1680750846862793, + "learning_rate": 1e-06, + "loss": 1.0524, + "mean_token_accuracy": 0.6831550598144531, + "num_tokens": 176347624.0, + "step": 6812 + }, + { + "epoch": 0.7481880079068746, + "grad_norm": 2.562291145324707, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7169862389564514, + "num_tokens": 176367772.0, + "step": 6813 + }, + { + "epoch": 0.7482978256094882, + "grad_norm": 1.9467962980270386, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7116239666938782, + "num_tokens": 176400767.0, + "step": 6814 + }, + { + "epoch": 0.7484076433121019, + "grad_norm": 2.2032315731048584, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6870672106742859, + "num_tokens": 176425357.0, + "step": 6815 + }, + { + "epoch": 0.7485174610147156, + "grad_norm": 2.515704870223999, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6949551105499268, + "num_tokens": 176445214.0, + "step": 6816 + }, + { + "epoch": 0.7486272787173293, + "grad_norm": 2.1620540618896484, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7059919238090515, + "num_tokens": 176472967.0, + "step": 6817 + }, + { + "epoch": 0.7487370964199429, + "grad_norm": 2.0153512954711914, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7202188968658447, + "num_tokens": 176502370.0, + "step": 6818 + }, + { + "epoch": 0.7488469141225566, + "grad_norm": 2.2609031200408936, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7062342166900635, + "num_tokens": 176525646.0, + "step": 6819 + }, + { + "epoch": 0.7489567318251702, + "grad_norm": 2.0979743003845215, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6957108378410339, + "num_tokens": 176552922.0, + "step": 6820 + }, + { + "epoch": 0.7490665495277838, + "grad_norm": 2.103520393371582, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7103167772293091, + "num_tokens": 176580090.0, + "step": 6821 + }, + { + "epoch": 0.7491763672303975, + "grad_norm": 2.0332868099212646, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.70853590965271, + "num_tokens": 176607392.0, + "step": 6822 + }, + { + "epoch": 0.7492861849330112, + "grad_norm": 2.386526584625244, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7211055755615234, + "num_tokens": 176631020.0, + "step": 6823 + }, + { + "epoch": 0.7493960026356249, + "grad_norm": 2.080766201019287, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7262969017028809, + "num_tokens": 176659765.0, + "step": 6824 + }, + { + "epoch": 0.7495058203382385, + "grad_norm": 2.047058582305908, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7264832258224487, + "num_tokens": 176688642.0, + "step": 6825 + }, + { + "epoch": 0.7496156380408522, + "grad_norm": 2.3011789321899414, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7328188419342041, + "num_tokens": 176712306.0, + "step": 6826 + }, + { + "epoch": 0.7497254557434658, + "grad_norm": 2.3959004878997803, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7309894561767578, + "num_tokens": 176731830.0, + "step": 6827 + }, + { + "epoch": 0.7498352734460795, + "grad_norm": 2.0229625701904297, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7017989754676819, + "num_tokens": 176762912.0, + "step": 6828 + }, + { + "epoch": 0.7499450911486931, + "grad_norm": 2.1976442337036133, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7022032737731934, + "num_tokens": 176787808.0, + "step": 6829 + }, + { + "epoch": 0.7500549088513069, + "grad_norm": 2.409853458404541, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7145756483078003, + "num_tokens": 176809620.0, + "step": 6830 + }, + { + "epoch": 0.7501647265539205, + "grad_norm": 1.9707882404327393, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7019895315170288, + "num_tokens": 176838777.0, + "step": 6831 + }, + { + "epoch": 0.7502745442565342, + "grad_norm": 2.4437785148620605, + "learning_rate": 1e-06, + "loss": 1.0661, + "mean_token_accuracy": 0.6845378279685974, + "num_tokens": 176863801.0, + "step": 6832 + }, + { + "epoch": 0.7503843619591478, + "grad_norm": 2.0901997089385986, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7098723649978638, + "num_tokens": 176892098.0, + "step": 6833 + }, + { + "epoch": 0.7504941796617615, + "grad_norm": 2.0369629859924316, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7203383445739746, + "num_tokens": 176919806.0, + "step": 6834 + }, + { + "epoch": 0.7506039973643751, + "grad_norm": 2.201399803161621, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6872979402542114, + "num_tokens": 176944799.0, + "step": 6835 + }, + { + "epoch": 0.7507138150669888, + "grad_norm": 2.391489028930664, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6972111463546753, + "num_tokens": 176968499.0, + "step": 6836 + }, + { + "epoch": 0.7508236327696025, + "grad_norm": 2.235196590423584, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7131655216217041, + "num_tokens": 176992760.0, + "step": 6837 + }, + { + "epoch": 0.7509334504722162, + "grad_norm": 2.0652809143066406, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6954530477523804, + "num_tokens": 177019679.0, + "step": 6838 + }, + { + "epoch": 0.7510432681748298, + "grad_norm": 2.1931838989257812, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6956010460853577, + "num_tokens": 177048441.0, + "step": 6839 + }, + { + "epoch": 0.7511530858774434, + "grad_norm": 2.068186044692993, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7062636613845825, + "num_tokens": 177076460.0, + "step": 6840 + }, + { + "epoch": 0.7512629035800571, + "grad_norm": 2.346824884414673, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7043035626411438, + "num_tokens": 177099730.0, + "step": 6841 + }, + { + "epoch": 0.7513727212826707, + "grad_norm": 2.3571529388427734, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7064884305000305, + "num_tokens": 177122154.0, + "step": 6842 + }, + { + "epoch": 0.7514825389852844, + "grad_norm": 2.415681838989258, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7152094841003418, + "num_tokens": 177142945.0, + "step": 6843 + }, + { + "epoch": 0.7515923566878981, + "grad_norm": 2.096846580505371, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.7039166688919067, + "num_tokens": 177170514.0, + "step": 6844 + }, + { + "epoch": 0.7517021743905118, + "grad_norm": 2.1898670196533203, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.6993545293807983, + "num_tokens": 177196408.0, + "step": 6845 + }, + { + "epoch": 0.7518119920931254, + "grad_norm": 2.3384461402893066, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7243530750274658, + "num_tokens": 177220300.0, + "step": 6846 + }, + { + "epoch": 0.7519218097957391, + "grad_norm": 2.0008609294891357, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.6959326267242432, + "num_tokens": 177251674.0, + "step": 6847 + }, + { + "epoch": 0.7520316274983527, + "grad_norm": 1.937016248703003, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6967398524284363, + "num_tokens": 177284386.0, + "step": 6848 + }, + { + "epoch": 0.7521414452009664, + "grad_norm": 1.9476040601730347, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6988766193389893, + "num_tokens": 177314575.0, + "step": 6849 + }, + { + "epoch": 0.75225126290358, + "grad_norm": 2.171607494354248, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6931967735290527, + "num_tokens": 177340530.0, + "step": 6850 + }, + { + "epoch": 0.7523610806061937, + "grad_norm": 2.2940673828125, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7022395730018616, + "num_tokens": 177363114.0, + "step": 6851 + }, + { + "epoch": 0.7524708983088074, + "grad_norm": 2.048478603363037, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7044221758842468, + "num_tokens": 177392813.0, + "step": 6852 + }, + { + "epoch": 0.7525807160114211, + "grad_norm": 2.3659772872924805, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7086564302444458, + "num_tokens": 177416349.0, + "step": 6853 + }, + { + "epoch": 0.7526905337140347, + "grad_norm": 2.23643159866333, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6881258487701416, + "num_tokens": 177441313.0, + "step": 6854 + }, + { + "epoch": 0.7528003514166484, + "grad_norm": 2.371361494064331, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7461849451065063, + "num_tokens": 177462066.0, + "step": 6855 + }, + { + "epoch": 0.752910169119262, + "grad_norm": 2.234755277633667, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7079605460166931, + "num_tokens": 177488623.0, + "step": 6856 + }, + { + "epoch": 0.7530199868218757, + "grad_norm": 2.080322265625, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7099134922027588, + "num_tokens": 177517433.0, + "step": 6857 + }, + { + "epoch": 0.7531298045244893, + "grad_norm": 2.1115663051605225, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7017841339111328, + "num_tokens": 177546971.0, + "step": 6858 + }, + { + "epoch": 0.7532396222271031, + "grad_norm": 2.3652377128601074, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6910919547080994, + "num_tokens": 177568348.0, + "step": 6859 + }, + { + "epoch": 0.7533494399297167, + "grad_norm": 2.1972904205322266, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7114084959030151, + "num_tokens": 177595015.0, + "step": 6860 + }, + { + "epoch": 0.7534592576323303, + "grad_norm": 2.1502978801727295, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7111467123031616, + "num_tokens": 177621428.0, + "step": 6861 + }, + { + "epoch": 0.753569075334944, + "grad_norm": 2.117480993270874, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7111797332763672, + "num_tokens": 177648253.0, + "step": 6862 + }, + { + "epoch": 0.7536788930375576, + "grad_norm": 2.6573996543884277, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7355725765228271, + "num_tokens": 177666539.0, + "step": 6863 + }, + { + "epoch": 0.7537887107401713, + "grad_norm": 2.320645332336426, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7197800874710083, + "num_tokens": 177691319.0, + "step": 6864 + }, + { + "epoch": 0.7538985284427849, + "grad_norm": 2.1062936782836914, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7008923292160034, + "num_tokens": 177721356.0, + "step": 6865 + }, + { + "epoch": 0.7540083461453987, + "grad_norm": 1.9802780151367188, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.6831676363945007, + "num_tokens": 177753114.0, + "step": 6866 + }, + { + "epoch": 0.7541181638480123, + "grad_norm": 2.159144401550293, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7251114845275879, + "num_tokens": 177777685.0, + "step": 6867 + }, + { + "epoch": 0.754227981550626, + "grad_norm": 2.06877064704895, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7084060907363892, + "num_tokens": 177805844.0, + "step": 6868 + }, + { + "epoch": 0.7543377992532396, + "grad_norm": 2.107285499572754, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6958919763565063, + "num_tokens": 177835153.0, + "step": 6869 + }, + { + "epoch": 0.7544476169558533, + "grad_norm": 2.1009247303009033, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7125002145767212, + "num_tokens": 177861316.0, + "step": 6870 + }, + { + "epoch": 0.7545574346584669, + "grad_norm": 2.195173740386963, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7120952606201172, + "num_tokens": 177888119.0, + "step": 6871 + }, + { + "epoch": 0.7546672523610806, + "grad_norm": 2.0195631980895996, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.6831807494163513, + "num_tokens": 177920001.0, + "step": 6872 + }, + { + "epoch": 0.7547770700636943, + "grad_norm": 2.642518997192383, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7120683193206787, + "num_tokens": 177938096.0, + "step": 6873 + }, + { + "epoch": 0.754886887766308, + "grad_norm": 2.228267192840576, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7012906074523926, + "num_tokens": 177961331.0, + "step": 6874 + }, + { + "epoch": 0.7549967054689216, + "grad_norm": 2.1766836643218994, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7284179925918579, + "num_tokens": 177985902.0, + "step": 6875 + }, + { + "epoch": 0.7551065231715353, + "grad_norm": 2.3071789741516113, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6877448558807373, + "num_tokens": 178011077.0, + "step": 6876 + }, + { + "epoch": 0.7552163408741489, + "grad_norm": 2.370649576187134, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7198606729507446, + "num_tokens": 178033075.0, + "step": 6877 + }, + { + "epoch": 0.7553261585767626, + "grad_norm": 2.0227370262145996, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7061331272125244, + "num_tokens": 178060752.0, + "step": 6878 + }, + { + "epoch": 0.7554359762793762, + "grad_norm": 2.002075672149658, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7265586256980896, + "num_tokens": 178089622.0, + "step": 6879 + }, + { + "epoch": 0.7555457939819898, + "grad_norm": 1.991063117980957, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6793310046195984, + "num_tokens": 178120948.0, + "step": 6880 + }, + { + "epoch": 0.7556556116846036, + "grad_norm": 2.6493778228759766, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6926562786102295, + "num_tokens": 178143058.0, + "step": 6881 + }, + { + "epoch": 0.7557654293872172, + "grad_norm": 2.2563722133636475, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.704604983329773, + "num_tokens": 178166367.0, + "step": 6882 + }, + { + "epoch": 0.7558752470898309, + "grad_norm": 2.5439369678497314, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.714847207069397, + "num_tokens": 178185632.0, + "step": 6883 + }, + { + "epoch": 0.7559850647924445, + "grad_norm": 2.193943500518799, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.694645881652832, + "num_tokens": 178210509.0, + "step": 6884 + }, + { + "epoch": 0.7560948824950582, + "grad_norm": 1.873467206954956, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6945331692695618, + "num_tokens": 178246098.0, + "step": 6885 + }, + { + "epoch": 0.7562047001976718, + "grad_norm": 2.2542827129364014, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7035584449768066, + "num_tokens": 178269653.0, + "step": 6886 + }, + { + "epoch": 0.7563145179002855, + "grad_norm": 2.1686222553253174, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7071070075035095, + "num_tokens": 178297507.0, + "step": 6887 + }, + { + "epoch": 0.7564243356028992, + "grad_norm": 2.223499059677124, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7133673429489136, + "num_tokens": 178322444.0, + "step": 6888 + }, + { + "epoch": 0.7565341533055129, + "grad_norm": 2.401754379272461, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6902947425842285, + "num_tokens": 178345073.0, + "step": 6889 + }, + { + "epoch": 0.7566439710081265, + "grad_norm": 2.011885404586792, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6935392022132874, + "num_tokens": 178373921.0, + "step": 6890 + }, + { + "epoch": 0.7567537887107402, + "grad_norm": 2.066765785217285, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7074354887008667, + "num_tokens": 178401628.0, + "step": 6891 + }, + { + "epoch": 0.7568636064133538, + "grad_norm": 2.2653627395629883, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7325625419616699, + "num_tokens": 178424561.0, + "step": 6892 + }, + { + "epoch": 0.7569734241159675, + "grad_norm": 2.1306307315826416, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7314016819000244, + "num_tokens": 178447980.0, + "step": 6893 + }, + { + "epoch": 0.7570832418185811, + "grad_norm": 2.1451804637908936, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7183088064193726, + "num_tokens": 178471718.0, + "step": 6894 + }, + { + "epoch": 0.7571930595211949, + "grad_norm": 2.226447343826294, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.704700231552124, + "num_tokens": 178496847.0, + "step": 6895 + }, + { + "epoch": 0.7573028772238085, + "grad_norm": 2.020541191101074, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6889486312866211, + "num_tokens": 178527000.0, + "step": 6896 + }, + { + "epoch": 0.7574126949264222, + "grad_norm": 2.3718442916870117, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.7069573998451233, + "num_tokens": 178550133.0, + "step": 6897 + }, + { + "epoch": 0.7575225126290358, + "grad_norm": 2.284730911254883, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7012643218040466, + "num_tokens": 178577050.0, + "step": 6898 + }, + { + "epoch": 0.7576323303316495, + "grad_norm": 2.130194664001465, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7246057987213135, + "num_tokens": 178602725.0, + "step": 6899 + }, + { + "epoch": 0.7577421480342631, + "grad_norm": 1.9671088457107544, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7143319845199585, + "num_tokens": 178633785.0, + "step": 6900 + }, + { + "epoch": 0.7578519657368767, + "grad_norm": 2.5240747928619385, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.707835853099823, + "num_tokens": 178654144.0, + "step": 6901 + }, + { + "epoch": 0.7579617834394905, + "grad_norm": 2.1558730602264404, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.702394962310791, + "num_tokens": 178680025.0, + "step": 6902 + }, + { + "epoch": 0.7580716011421041, + "grad_norm": 2.3966548442840576, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7046237587928772, + "num_tokens": 178702119.0, + "step": 6903 + }, + { + "epoch": 0.7581814188447178, + "grad_norm": 1.9987658262252808, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6971418857574463, + "num_tokens": 178731672.0, + "step": 6904 + }, + { + "epoch": 0.7582912365473314, + "grad_norm": 2.0772459506988525, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7264525890350342, + "num_tokens": 178758582.0, + "step": 6905 + }, + { + "epoch": 0.7584010542499451, + "grad_norm": 1.941292643547058, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6894787549972534, + "num_tokens": 178787672.0, + "step": 6906 + }, + { + "epoch": 0.7585108719525587, + "grad_norm": 2.1249473094940186, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6882673501968384, + "num_tokens": 178814095.0, + "step": 6907 + }, + { + "epoch": 0.7586206896551724, + "grad_norm": 2.0224268436431885, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6943897008895874, + "num_tokens": 178843461.0, + "step": 6908 + }, + { + "epoch": 0.758730507357786, + "grad_norm": 2.0860586166381836, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.7021907567977905, + "num_tokens": 178870250.0, + "step": 6909 + }, + { + "epoch": 0.7588403250603998, + "grad_norm": 2.2644035816192627, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7192988395690918, + "num_tokens": 178893417.0, + "step": 6910 + }, + { + "epoch": 0.7589501427630134, + "grad_norm": 2.3098502159118652, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7028218507766724, + "num_tokens": 178915244.0, + "step": 6911 + }, + { + "epoch": 0.7590599604656271, + "grad_norm": 2.303779363632202, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7013096809387207, + "num_tokens": 178938120.0, + "step": 6912 + }, + { + "epoch": 0.7591697781682407, + "grad_norm": 2.1081438064575195, + "learning_rate": 1e-06, + "loss": 1.0545, + "mean_token_accuracy": 0.684944212436676, + "num_tokens": 178965558.0, + "step": 6913 + }, + { + "epoch": 0.7592795958708544, + "grad_norm": 2.046252727508545, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.716909646987915, + "num_tokens": 178994111.0, + "step": 6914 + }, + { + "epoch": 0.759389413573468, + "grad_norm": 2.3343112468719482, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7066762447357178, + "num_tokens": 179017532.0, + "step": 6915 + }, + { + "epoch": 0.7594992312760817, + "grad_norm": 2.6397926807403564, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7042175531387329, + "num_tokens": 179036660.0, + "step": 6916 + }, + { + "epoch": 0.7596090489786954, + "grad_norm": 2.0331249237060547, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7061530351638794, + "num_tokens": 179064753.0, + "step": 6917 + }, + { + "epoch": 0.7597188666813091, + "grad_norm": 2.264496326446533, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7186253070831299, + "num_tokens": 179087979.0, + "step": 6918 + }, + { + "epoch": 0.7598286843839227, + "grad_norm": 2.298288106918335, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.7002713680267334, + "num_tokens": 179113525.0, + "step": 6919 + }, + { + "epoch": 0.7599385020865363, + "grad_norm": 2.086228847503662, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7162100672721863, + "num_tokens": 179140003.0, + "step": 6920 + }, + { + "epoch": 0.76004831978915, + "grad_norm": 2.180649995803833, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.723764181137085, + "num_tokens": 179164191.0, + "step": 6921 + }, + { + "epoch": 0.7601581374917636, + "grad_norm": 2.0185654163360596, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.6845616698265076, + "num_tokens": 179194456.0, + "step": 6922 + }, + { + "epoch": 0.7602679551943773, + "grad_norm": 2.127595901489258, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7311872243881226, + "num_tokens": 179220806.0, + "step": 6923 + }, + { + "epoch": 0.760377772896991, + "grad_norm": 2.3950397968292236, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7047213315963745, + "num_tokens": 179243326.0, + "step": 6924 + }, + { + "epoch": 0.7604875905996047, + "grad_norm": 2.1278305053710938, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7096420526504517, + "num_tokens": 179270466.0, + "step": 6925 + }, + { + "epoch": 0.7605974083022183, + "grad_norm": 2.0065090656280518, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7193706035614014, + "num_tokens": 179300590.0, + "step": 6926 + }, + { + "epoch": 0.760707226004832, + "grad_norm": 2.256969690322876, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6895154714584351, + "num_tokens": 179324837.0, + "step": 6927 + }, + { + "epoch": 0.7608170437074456, + "grad_norm": 2.0002353191375732, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6987910270690918, + "num_tokens": 179356085.0, + "step": 6928 + }, + { + "epoch": 0.7609268614100593, + "grad_norm": 2.252305746078491, + "learning_rate": 1e-06, + "loss": 1.0634, + "mean_token_accuracy": 0.6808130741119385, + "num_tokens": 179380943.0, + "step": 6929 + }, + { + "epoch": 0.7610366791126729, + "grad_norm": 2.171189546585083, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6992745399475098, + "num_tokens": 179406636.0, + "step": 6930 + }, + { + "epoch": 0.7611464968152867, + "grad_norm": 2.165864944458008, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.698357343673706, + "num_tokens": 179432128.0, + "step": 6931 + }, + { + "epoch": 0.7612563145179003, + "grad_norm": 2.236395835876465, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.7030497789382935, + "num_tokens": 179457591.0, + "step": 6932 + }, + { + "epoch": 0.761366132220514, + "grad_norm": 2.251457691192627, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7121471166610718, + "num_tokens": 179483646.0, + "step": 6933 + }, + { + "epoch": 0.7614759499231276, + "grad_norm": 2.075746774673462, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7131258845329285, + "num_tokens": 179511204.0, + "step": 6934 + }, + { + "epoch": 0.7615857676257413, + "grad_norm": 2.149822473526001, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7063279151916504, + "num_tokens": 179536067.0, + "step": 6935 + }, + { + "epoch": 0.7616955853283549, + "grad_norm": 2.0107016563415527, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7380029559135437, + "num_tokens": 179563707.0, + "step": 6936 + }, + { + "epoch": 0.7618054030309686, + "grad_norm": 2.197725534439087, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7237720489501953, + "num_tokens": 179587191.0, + "step": 6937 + }, + { + "epoch": 0.7619152207335822, + "grad_norm": 2.172447443008423, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7344985604286194, + "num_tokens": 179610040.0, + "step": 6938 + }, + { + "epoch": 0.762025038436196, + "grad_norm": 2.26584792137146, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7027749419212341, + "num_tokens": 179635594.0, + "step": 6939 + }, + { + "epoch": 0.7621348561388096, + "grad_norm": 2.340444564819336, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7142430543899536, + "num_tokens": 179657892.0, + "step": 6940 + }, + { + "epoch": 0.7622446738414232, + "grad_norm": 2.0896782875061035, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.72926926612854, + "num_tokens": 179683394.0, + "step": 6941 + }, + { + "epoch": 0.7623544915440369, + "grad_norm": 2.0262277126312256, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6862519979476929, + "num_tokens": 179711403.0, + "step": 6942 + }, + { + "epoch": 0.7624643092466505, + "grad_norm": 2.5817975997924805, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7203785181045532, + "num_tokens": 179730406.0, + "step": 6943 + }, + { + "epoch": 0.7625741269492642, + "grad_norm": 2.0562968254089355, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.692777156829834, + "num_tokens": 179761665.0, + "step": 6944 + }, + { + "epoch": 0.7626839446518778, + "grad_norm": 2.2281031608581543, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7093021869659424, + "num_tokens": 179784254.0, + "step": 6945 + }, + { + "epoch": 0.7627937623544916, + "grad_norm": 2.468489408493042, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.703953742980957, + "num_tokens": 179807077.0, + "step": 6946 + }, + { + "epoch": 0.7629035800571052, + "grad_norm": 1.8655056953430176, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7177121639251709, + "num_tokens": 179838235.0, + "step": 6947 + }, + { + "epoch": 0.7630133977597189, + "grad_norm": 2.07724928855896, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7056371569633484, + "num_tokens": 179866103.0, + "step": 6948 + }, + { + "epoch": 0.7631232154623325, + "grad_norm": 2.3378913402557373, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7208790183067322, + "num_tokens": 179888382.0, + "step": 6949 + }, + { + "epoch": 0.7632330331649462, + "grad_norm": 2.3993539810180664, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7115212678909302, + "num_tokens": 179908263.0, + "step": 6950 + }, + { + "epoch": 0.7633428508675598, + "grad_norm": 2.140415906906128, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.716916024684906, + "num_tokens": 179935433.0, + "step": 6951 + }, + { + "epoch": 0.7634526685701735, + "grad_norm": 2.1511194705963135, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7013408541679382, + "num_tokens": 179961446.0, + "step": 6952 + }, + { + "epoch": 0.7635624862727872, + "grad_norm": 2.1004817485809326, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7197220325469971, + "num_tokens": 179987906.0, + "step": 6953 + }, + { + "epoch": 0.7636723039754009, + "grad_norm": 2.2673802375793457, + "learning_rate": 1e-06, + "loss": 1.057, + "mean_token_accuracy": 0.676104724407196, + "num_tokens": 180011551.0, + "step": 6954 + }, + { + "epoch": 0.7637821216780145, + "grad_norm": 2.121002435684204, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7145854234695435, + "num_tokens": 180037809.0, + "step": 6955 + }, + { + "epoch": 0.7638919393806282, + "grad_norm": 2.1732118129730225, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7134678363800049, + "num_tokens": 180063827.0, + "step": 6956 + }, + { + "epoch": 0.7640017570832418, + "grad_norm": 2.159963369369507, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6938676834106445, + "num_tokens": 180089028.0, + "step": 6957 + }, + { + "epoch": 0.7641115747858555, + "grad_norm": 2.31524920463562, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7120481729507446, + "num_tokens": 180113074.0, + "step": 6958 + }, + { + "epoch": 0.7642213924884691, + "grad_norm": 2.175043821334839, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6910535097122192, + "num_tokens": 180140288.0, + "step": 6959 + }, + { + "epoch": 0.7643312101910829, + "grad_norm": 2.1156318187713623, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6932252049446106, + "num_tokens": 180165848.0, + "step": 6960 + }, + { + "epoch": 0.7644410278936965, + "grad_norm": 2.059495687484741, + "learning_rate": 1e-06, + "loss": 1.0618, + "mean_token_accuracy": 0.6832889914512634, + "num_tokens": 180193944.0, + "step": 6961 + }, + { + "epoch": 0.7645508455963101, + "grad_norm": 2.348541259765625, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7031981945037842, + "num_tokens": 180217722.0, + "step": 6962 + }, + { + "epoch": 0.7646606632989238, + "grad_norm": 2.1402065753936768, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7257981300354004, + "num_tokens": 180242650.0, + "step": 6963 + }, + { + "epoch": 0.7647704810015374, + "grad_norm": 1.9013105630874634, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7353839874267578, + "num_tokens": 180272453.0, + "step": 6964 + }, + { + "epoch": 0.7648802987041511, + "grad_norm": 2.0349607467651367, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7178212404251099, + "num_tokens": 180300422.0, + "step": 6965 + }, + { + "epoch": 0.7649901164067647, + "grad_norm": 2.3904917240142822, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7310840487480164, + "num_tokens": 180322956.0, + "step": 6966 + }, + { + "epoch": 0.7650999341093784, + "grad_norm": 2.1778347492218018, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7031468152999878, + "num_tokens": 180348589.0, + "step": 6967 + }, + { + "epoch": 0.7652097518119921, + "grad_norm": 2.4732511043548584, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7062680721282959, + "num_tokens": 180369574.0, + "step": 6968 + }, + { + "epoch": 0.7653195695146058, + "grad_norm": 2.306560754776001, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.7046853303909302, + "num_tokens": 180391533.0, + "step": 6969 + }, + { + "epoch": 0.7654293872172194, + "grad_norm": 2.069720506668091, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6876827478408813, + "num_tokens": 180419677.0, + "step": 6970 + }, + { + "epoch": 0.7655392049198331, + "grad_norm": 2.078822612762451, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6878283619880676, + "num_tokens": 180449792.0, + "step": 6971 + }, + { + "epoch": 0.7656490226224467, + "grad_norm": 2.496079921722412, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7071864604949951, + "num_tokens": 180470960.0, + "step": 6972 + }, + { + "epoch": 0.7657588403250604, + "grad_norm": 2.0044748783111572, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.6818511486053467, + "num_tokens": 180501704.0, + "step": 6973 + }, + { + "epoch": 0.765868658027674, + "grad_norm": 2.453850746154785, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7053208351135254, + "num_tokens": 180522239.0, + "step": 6974 + }, + { + "epoch": 0.7659784757302878, + "grad_norm": 2.3127198219299316, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7069504261016846, + "num_tokens": 180545659.0, + "step": 6975 + }, + { + "epoch": 0.7660882934329014, + "grad_norm": 2.0915822982788086, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7253695726394653, + "num_tokens": 180570244.0, + "step": 6976 + }, + { + "epoch": 0.7661981111355151, + "grad_norm": 2.0096588134765625, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7217283248901367, + "num_tokens": 180597361.0, + "step": 6977 + }, + { + "epoch": 0.7663079288381287, + "grad_norm": 2.23366641998291, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7014613747596741, + "num_tokens": 180622400.0, + "step": 6978 + }, + { + "epoch": 0.7664177465407424, + "grad_norm": 2.061175584793091, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7097834944725037, + "num_tokens": 180649632.0, + "step": 6979 + }, + { + "epoch": 0.766527564243356, + "grad_norm": 2.032212257385254, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7165275812149048, + "num_tokens": 180678695.0, + "step": 6980 + }, + { + "epoch": 0.7666373819459696, + "grad_norm": 2.4502968788146973, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.722419023513794, + "num_tokens": 180700718.0, + "step": 6981 + }, + { + "epoch": 0.7667471996485834, + "grad_norm": 2.252633571624756, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7318877577781677, + "num_tokens": 180723553.0, + "step": 6982 + }, + { + "epoch": 0.766857017351197, + "grad_norm": 2.252467393875122, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.71409010887146, + "num_tokens": 180748214.0, + "step": 6983 + }, + { + "epoch": 0.7669668350538107, + "grad_norm": 2.1966640949249268, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.705935537815094, + "num_tokens": 180773817.0, + "step": 6984 + }, + { + "epoch": 0.7670766527564243, + "grad_norm": 2.3487253189086914, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7103629112243652, + "num_tokens": 180794784.0, + "step": 6985 + }, + { + "epoch": 0.767186470459038, + "grad_norm": 2.4331705570220947, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7023872137069702, + "num_tokens": 180815988.0, + "step": 6986 + }, + { + "epoch": 0.7672962881616516, + "grad_norm": 2.0737690925598145, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6995222568511963, + "num_tokens": 180842777.0, + "step": 6987 + }, + { + "epoch": 0.7674061058642653, + "grad_norm": 2.197018623352051, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7155150175094604, + "num_tokens": 180868011.0, + "step": 6988 + }, + { + "epoch": 0.767515923566879, + "grad_norm": 2.004004716873169, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6915531158447266, + "num_tokens": 180899163.0, + "step": 6989 + }, + { + "epoch": 0.7676257412694927, + "grad_norm": 2.186063766479492, + "learning_rate": 1e-06, + "loss": 1.0839, + "mean_token_accuracy": 0.673427402973175, + "num_tokens": 180926392.0, + "step": 6990 + }, + { + "epoch": 0.7677355589721063, + "grad_norm": 2.1693615913391113, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6966683864593506, + "num_tokens": 180951585.0, + "step": 6991 + }, + { + "epoch": 0.76784537667472, + "grad_norm": 2.2526116371154785, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7064157128334045, + "num_tokens": 180975363.0, + "step": 6992 + }, + { + "epoch": 0.7679551943773336, + "grad_norm": 2.1461753845214844, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6912009716033936, + "num_tokens": 181003116.0, + "step": 6993 + }, + { + "epoch": 0.7680650120799473, + "grad_norm": 2.172766923904419, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6925289630889893, + "num_tokens": 181029285.0, + "step": 6994 + }, + { + "epoch": 0.7681748297825609, + "grad_norm": 2.171018600463867, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7120254039764404, + "num_tokens": 181053653.0, + "step": 6995 + }, + { + "epoch": 0.7682846474851747, + "grad_norm": 2.157363176345825, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7066143155097961, + "num_tokens": 181079899.0, + "step": 6996 + }, + { + "epoch": 0.7683944651877883, + "grad_norm": 2.2223422527313232, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.693076491355896, + "num_tokens": 181105803.0, + "step": 6997 + }, + { + "epoch": 0.768504282890402, + "grad_norm": 2.152832508087158, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7075732946395874, + "num_tokens": 181132456.0, + "step": 6998 + }, + { + "epoch": 0.7686141005930156, + "grad_norm": 2.3889150619506836, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6953979134559631, + "num_tokens": 181155100.0, + "step": 6999 + }, + { + "epoch": 0.7687239182956292, + "grad_norm": 2.3922982215881348, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7062845826148987, + "num_tokens": 181176842.0, + "step": 7000 + }, + { + "epoch": 0.7688337359982429, + "grad_norm": 2.2534165382385254, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7014356255531311, + "num_tokens": 181200229.0, + "step": 7001 + }, + { + "epoch": 0.7689435537008565, + "grad_norm": 2.0732781887054443, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.7036975622177124, + "num_tokens": 181229968.0, + "step": 7002 + }, + { + "epoch": 0.7690533714034702, + "grad_norm": 2.1520750522613525, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6973362565040588, + "num_tokens": 181256302.0, + "step": 7003 + }, + { + "epoch": 0.7691631891060839, + "grad_norm": 1.910279393196106, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7331778407096863, + "num_tokens": 181286191.0, + "step": 7004 + }, + { + "epoch": 0.7692730068086976, + "grad_norm": 2.4264566898345947, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7040399312973022, + "num_tokens": 181307029.0, + "step": 7005 + }, + { + "epoch": 0.7693828245113112, + "grad_norm": 2.160149574279785, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7067712545394897, + "num_tokens": 181332716.0, + "step": 7006 + }, + { + "epoch": 0.7694926422139249, + "grad_norm": 1.9925971031188965, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7083573341369629, + "num_tokens": 181361598.0, + "step": 7007 + }, + { + "epoch": 0.7696024599165385, + "grad_norm": 2.2460289001464844, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6921284198760986, + "num_tokens": 181385434.0, + "step": 7008 + }, + { + "epoch": 0.7697122776191522, + "grad_norm": 2.2206833362579346, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6979304552078247, + "num_tokens": 181411830.0, + "step": 7009 + }, + { + "epoch": 0.7698220953217658, + "grad_norm": 1.952436923980713, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7021928429603577, + "num_tokens": 181442415.0, + "step": 7010 + }, + { + "epoch": 0.7699319130243796, + "grad_norm": 2.3131606578826904, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7033218145370483, + "num_tokens": 181466967.0, + "step": 7011 + }, + { + "epoch": 0.7700417307269932, + "grad_norm": 2.5483057498931885, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7216712236404419, + "num_tokens": 181487298.0, + "step": 7012 + }, + { + "epoch": 0.7701515484296069, + "grad_norm": 2.4433412551879883, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7393481731414795, + "num_tokens": 181507506.0, + "step": 7013 + }, + { + "epoch": 0.7702613661322205, + "grad_norm": 2.0246469974517822, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7051567435264587, + "num_tokens": 181536470.0, + "step": 7014 + }, + { + "epoch": 0.7703711838348342, + "grad_norm": 2.2025415897369385, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7069110870361328, + "num_tokens": 181562596.0, + "step": 7015 + }, + { + "epoch": 0.7704810015374478, + "grad_norm": 2.1519277095794678, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6896135210990906, + "num_tokens": 181591441.0, + "step": 7016 + }, + { + "epoch": 0.7705908192400615, + "grad_norm": 2.1563560962677, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6985354423522949, + "num_tokens": 181617441.0, + "step": 7017 + }, + { + "epoch": 0.7707006369426752, + "grad_norm": 2.160983085632324, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7125770449638367, + "num_tokens": 181644105.0, + "step": 7018 + }, + { + "epoch": 0.7708104546452889, + "grad_norm": 2.1354176998138428, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7142572999000549, + "num_tokens": 181669368.0, + "step": 7019 + }, + { + "epoch": 0.7709202723479025, + "grad_norm": 2.4102635383605957, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7057369351387024, + "num_tokens": 181690185.0, + "step": 7020 + }, + { + "epoch": 0.7710300900505161, + "grad_norm": 2.0325658321380615, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7179780602455139, + "num_tokens": 181717010.0, + "step": 7021 + }, + { + "epoch": 0.7711399077531298, + "grad_norm": 2.158806562423706, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6923527717590332, + "num_tokens": 181742561.0, + "step": 7022 + }, + { + "epoch": 0.7712497254557434, + "grad_norm": 2.3088133335113525, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6908430457115173, + "num_tokens": 181766667.0, + "step": 7023 + }, + { + "epoch": 0.7713595431583571, + "grad_norm": 2.2156288623809814, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7307730913162231, + "num_tokens": 181792346.0, + "step": 7024 + }, + { + "epoch": 0.7714693608609708, + "grad_norm": 2.2343099117279053, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6837354898452759, + "num_tokens": 181817478.0, + "step": 7025 + }, + { + "epoch": 0.7715791785635845, + "grad_norm": 2.2650113105773926, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6972266435623169, + "num_tokens": 181841823.0, + "step": 7026 + }, + { + "epoch": 0.7716889962661981, + "grad_norm": 2.340830087661743, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7000187039375305, + "num_tokens": 181863086.0, + "step": 7027 + }, + { + "epoch": 0.7717988139688118, + "grad_norm": 2.2168521881103516, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.684850811958313, + "num_tokens": 181888119.0, + "step": 7028 + }, + { + "epoch": 0.7719086316714254, + "grad_norm": 2.039665937423706, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6922744512557983, + "num_tokens": 181914649.0, + "step": 7029 + }, + { + "epoch": 0.7720184493740391, + "grad_norm": 2.2389707565307617, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6983870267868042, + "num_tokens": 181940137.0, + "step": 7030 + }, + { + "epoch": 0.7721282670766527, + "grad_norm": 2.2367336750030518, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7283123731613159, + "num_tokens": 181963552.0, + "step": 7031 + }, + { + "epoch": 0.7722380847792664, + "grad_norm": 2.4351933002471924, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7396799325942993, + "num_tokens": 181983514.0, + "step": 7032 + }, + { + "epoch": 0.7723479024818801, + "grad_norm": 2.0523266792297363, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.70815509557724, + "num_tokens": 182012525.0, + "step": 7033 + }, + { + "epoch": 0.7724577201844938, + "grad_norm": 2.273862361907959, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6908708810806274, + "num_tokens": 182036340.0, + "step": 7034 + }, + { + "epoch": 0.7725675378871074, + "grad_norm": 2.6259186267852783, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7150964736938477, + "num_tokens": 182055692.0, + "step": 7035 + }, + { + "epoch": 0.7726773555897211, + "grad_norm": 2.0222489833831787, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7098186612129211, + "num_tokens": 182084359.0, + "step": 7036 + }, + { + "epoch": 0.7727871732923347, + "grad_norm": 2.3667421340942383, + "learning_rate": 1e-06, + "loss": 1.0317, + "mean_token_accuracy": 0.6896967887878418, + "num_tokens": 182108290.0, + "step": 7037 + }, + { + "epoch": 0.7728969909949484, + "grad_norm": 2.0288503170013428, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6888775825500488, + "num_tokens": 182137242.0, + "step": 7038 + }, + { + "epoch": 0.773006808697562, + "grad_norm": 2.0433812141418457, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6925595998764038, + "num_tokens": 182166149.0, + "step": 7039 + }, + { + "epoch": 0.7731166264001758, + "grad_norm": 2.1773390769958496, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7152605056762695, + "num_tokens": 182189087.0, + "step": 7040 + }, + { + "epoch": 0.7732264441027894, + "grad_norm": 2.2997353076934814, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6940014362335205, + "num_tokens": 182213522.0, + "step": 7041 + }, + { + "epoch": 0.773336261805403, + "grad_norm": 2.342327833175659, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6877075433731079, + "num_tokens": 182236627.0, + "step": 7042 + }, + { + "epoch": 0.7734460795080167, + "grad_norm": 2.4334776401519775, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7316597104072571, + "num_tokens": 182259748.0, + "step": 7043 + }, + { + "epoch": 0.7735558972106303, + "grad_norm": 2.1196343898773193, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7053515911102295, + "num_tokens": 182288135.0, + "step": 7044 + }, + { + "epoch": 0.773665714913244, + "grad_norm": 2.080568552017212, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6957648396492004, + "num_tokens": 182313404.0, + "step": 7045 + }, + { + "epoch": 0.7737755326158576, + "grad_norm": 2.0026326179504395, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.713379442691803, + "num_tokens": 182342822.0, + "step": 7046 + }, + { + "epoch": 0.7738853503184714, + "grad_norm": 2.576629400253296, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7291146516799927, + "num_tokens": 182360758.0, + "step": 7047 + }, + { + "epoch": 0.773995168021085, + "grad_norm": 1.8981857299804688, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6939295530319214, + "num_tokens": 182393280.0, + "step": 7048 + }, + { + "epoch": 0.7741049857236987, + "grad_norm": 2.1988253593444824, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7044508457183838, + "num_tokens": 182417912.0, + "step": 7049 + }, + { + "epoch": 0.7742148034263123, + "grad_norm": 2.3623383045196533, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.679115891456604, + "num_tokens": 182442234.0, + "step": 7050 + }, + { + "epoch": 0.774324621128926, + "grad_norm": 2.2003979682922363, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7176967859268188, + "num_tokens": 182466737.0, + "step": 7051 + }, + { + "epoch": 0.7744344388315396, + "grad_norm": 2.065133571624756, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.6884132623672485, + "num_tokens": 182494786.0, + "step": 7052 + }, + { + "epoch": 0.7745442565341533, + "grad_norm": 2.3221731185913086, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7168830633163452, + "num_tokens": 182517728.0, + "step": 7053 + }, + { + "epoch": 0.774654074236767, + "grad_norm": 2.3722028732299805, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7253584861755371, + "num_tokens": 182539508.0, + "step": 7054 + }, + { + "epoch": 0.7747638919393807, + "grad_norm": 2.1247313022613525, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6922187805175781, + "num_tokens": 182565504.0, + "step": 7055 + }, + { + "epoch": 0.7748737096419943, + "grad_norm": 2.0713374614715576, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6951960921287537, + "num_tokens": 182593815.0, + "step": 7056 + }, + { + "epoch": 0.774983527344608, + "grad_norm": 2.0480306148529053, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7022595405578613, + "num_tokens": 182621249.0, + "step": 7057 + }, + { + "epoch": 0.7750933450472216, + "grad_norm": 2.209336042404175, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7230658531188965, + "num_tokens": 182645287.0, + "step": 7058 + }, + { + "epoch": 0.7752031627498353, + "grad_norm": 2.1517562866210938, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6976292133331299, + "num_tokens": 182670630.0, + "step": 7059 + }, + { + "epoch": 0.7753129804524489, + "grad_norm": 2.202336311340332, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7274383306503296, + "num_tokens": 182693031.0, + "step": 7060 + }, + { + "epoch": 0.7754227981550625, + "grad_norm": 2.1004631519317627, + "learning_rate": 1e-06, + "loss": 1.0672, + "mean_token_accuracy": 0.6788142323493958, + "num_tokens": 182720336.0, + "step": 7061 + }, + { + "epoch": 0.7755326158576763, + "grad_norm": 2.204347610473633, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7068523168563843, + "num_tokens": 182744926.0, + "step": 7062 + }, + { + "epoch": 0.77564243356029, + "grad_norm": 2.2029318809509277, + "learning_rate": 1e-06, + "loss": 1.0547, + "mean_token_accuracy": 0.684604823589325, + "num_tokens": 182771478.0, + "step": 7063 + }, + { + "epoch": 0.7757522512629036, + "grad_norm": 2.3927555084228516, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7031164765357971, + "num_tokens": 182795271.0, + "step": 7064 + }, + { + "epoch": 0.7758620689655172, + "grad_norm": 2.1872308254241943, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.6949464678764343, + "num_tokens": 182820752.0, + "step": 7065 + }, + { + "epoch": 0.7759718866681309, + "grad_norm": 2.032560348510742, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.691078782081604, + "num_tokens": 182850623.0, + "step": 7066 + }, + { + "epoch": 0.7760817043707445, + "grad_norm": 2.153125762939453, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6847873330116272, + "num_tokens": 182876812.0, + "step": 7067 + }, + { + "epoch": 0.7761915220733582, + "grad_norm": 2.1593122482299805, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7241230607032776, + "num_tokens": 182901447.0, + "step": 7068 + }, + { + "epoch": 0.7763013397759719, + "grad_norm": 2.580777406692505, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.6996640563011169, + "num_tokens": 182922621.0, + "step": 7069 + }, + { + "epoch": 0.7764111574785856, + "grad_norm": 2.263432264328003, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7209081649780273, + "num_tokens": 182945907.0, + "step": 7070 + }, + { + "epoch": 0.7765209751811992, + "grad_norm": 2.4089627265930176, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7072098255157471, + "num_tokens": 182967961.0, + "step": 7071 + }, + { + "epoch": 0.7766307928838129, + "grad_norm": 2.0676820278167725, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7009274959564209, + "num_tokens": 182997028.0, + "step": 7072 + }, + { + "epoch": 0.7767406105864265, + "grad_norm": 2.2037384510040283, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7175154685974121, + "num_tokens": 183020275.0, + "step": 7073 + }, + { + "epoch": 0.7768504282890402, + "grad_norm": 2.3020777702331543, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7069202065467834, + "num_tokens": 183042996.0, + "step": 7074 + }, + { + "epoch": 0.7769602459916538, + "grad_norm": 2.3034253120422363, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7047165632247925, + "num_tokens": 183069469.0, + "step": 7075 + }, + { + "epoch": 0.7770700636942676, + "grad_norm": 2.2261412143707275, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.7024906277656555, + "num_tokens": 183094259.0, + "step": 7076 + }, + { + "epoch": 0.7771798813968812, + "grad_norm": 2.3545944690704346, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.733046293258667, + "num_tokens": 183114980.0, + "step": 7077 + }, + { + "epoch": 0.7772896990994949, + "grad_norm": 2.2750566005706787, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.707786500453949, + "num_tokens": 183140267.0, + "step": 7078 + }, + { + "epoch": 0.7773995168021085, + "grad_norm": 2.229113817214966, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7196390628814697, + "num_tokens": 183165152.0, + "step": 7079 + }, + { + "epoch": 0.7775093345047221, + "grad_norm": 2.0171921253204346, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7133912444114685, + "num_tokens": 183192657.0, + "step": 7080 + }, + { + "epoch": 0.7776191522073358, + "grad_norm": 1.9697538614273071, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7365235686302185, + "num_tokens": 183221786.0, + "step": 7081 + }, + { + "epoch": 0.7777289699099494, + "grad_norm": 2.0442326068878174, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7282671332359314, + "num_tokens": 183248432.0, + "step": 7082 + }, + { + "epoch": 0.7778387876125632, + "grad_norm": 2.0811307430267334, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7136984467506409, + "num_tokens": 183276389.0, + "step": 7083 + }, + { + "epoch": 0.7779486053151768, + "grad_norm": 2.1442201137542725, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7119879722595215, + "num_tokens": 183302609.0, + "step": 7084 + }, + { + "epoch": 0.7780584230177905, + "grad_norm": 1.913198709487915, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6973487734794617, + "num_tokens": 183336102.0, + "step": 7085 + }, + { + "epoch": 0.7781682407204041, + "grad_norm": 1.9238144159317017, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.6979182958602905, + "num_tokens": 183365553.0, + "step": 7086 + }, + { + "epoch": 0.7782780584230178, + "grad_norm": 2.1272213459014893, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6974408030509949, + "num_tokens": 183391577.0, + "step": 7087 + }, + { + "epoch": 0.7783878761256314, + "grad_norm": 2.14436936378479, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6969308853149414, + "num_tokens": 183416801.0, + "step": 7088 + }, + { + "epoch": 0.7784976938282451, + "grad_norm": 2.5139424800872803, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7355871200561523, + "num_tokens": 183435573.0, + "step": 7089 + }, + { + "epoch": 0.7786075115308587, + "grad_norm": 1.9976410865783691, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6909494400024414, + "num_tokens": 183467300.0, + "step": 7090 + }, + { + "epoch": 0.7787173292334725, + "grad_norm": 2.3301830291748047, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7077818512916565, + "num_tokens": 183490195.0, + "step": 7091 + }, + { + "epoch": 0.7788271469360861, + "grad_norm": 2.3824377059936523, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7030270099639893, + "num_tokens": 183514431.0, + "step": 7092 + }, + { + "epoch": 0.7789369646386998, + "grad_norm": 2.2234368324279785, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7429946660995483, + "num_tokens": 183539682.0, + "step": 7093 + }, + { + "epoch": 0.7790467823413134, + "grad_norm": 2.3235273361206055, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7173038125038147, + "num_tokens": 183562472.0, + "step": 7094 + }, + { + "epoch": 0.7791566000439271, + "grad_norm": 2.1111745834350586, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7076345682144165, + "num_tokens": 183589502.0, + "step": 7095 + }, + { + "epoch": 0.7792664177465407, + "grad_norm": 2.4791393280029297, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7002306580543518, + "num_tokens": 183615986.0, + "step": 7096 + }, + { + "epoch": 0.7793762354491544, + "grad_norm": 2.0927319526672363, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6941201686859131, + "num_tokens": 183642905.0, + "step": 7097 + }, + { + "epoch": 0.7794860531517681, + "grad_norm": 2.590775728225708, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7174063920974731, + "num_tokens": 183664664.0, + "step": 7098 + }, + { + "epoch": 0.7795958708543818, + "grad_norm": 2.2272374629974365, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7304288148880005, + "num_tokens": 183688895.0, + "step": 7099 + }, + { + "epoch": 0.7797056885569954, + "grad_norm": 2.1472978591918945, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6990457773208618, + "num_tokens": 183716026.0, + "step": 7100 + }, + { + "epoch": 0.779815506259609, + "grad_norm": 1.9729640483856201, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7072583436965942, + "num_tokens": 183744988.0, + "step": 7101 + }, + { + "epoch": 0.7799253239622227, + "grad_norm": 2.1693379878997803, + "learning_rate": 1e-06, + "loss": 1.0602, + "mean_token_accuracy": 0.6772041916847229, + "num_tokens": 183771237.0, + "step": 7102 + }, + { + "epoch": 0.7800351416648363, + "grad_norm": 2.107262372970581, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.701768159866333, + "num_tokens": 183798604.0, + "step": 7103 + }, + { + "epoch": 0.78014495936745, + "grad_norm": 2.224320888519287, + "learning_rate": 1e-06, + "loss": 1.0491, + "mean_token_accuracy": 0.6814330220222473, + "num_tokens": 183824031.0, + "step": 7104 + }, + { + "epoch": 0.7802547770700637, + "grad_norm": 2.3324787616729736, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.711635947227478, + "num_tokens": 183847258.0, + "step": 7105 + }, + { + "epoch": 0.7803645947726774, + "grad_norm": 2.1718149185180664, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7178064584732056, + "num_tokens": 183872247.0, + "step": 7106 + }, + { + "epoch": 0.780474412475291, + "grad_norm": 2.1223256587982178, + "learning_rate": 1e-06, + "loss": 1.0602, + "mean_token_accuracy": 0.6901425719261169, + "num_tokens": 183899771.0, + "step": 7107 + }, + { + "epoch": 0.7805842301779047, + "grad_norm": 2.0402746200561523, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7047853469848633, + "num_tokens": 183927184.0, + "step": 7108 + }, + { + "epoch": 0.7806940478805183, + "grad_norm": 1.9788051843643188, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.705286979675293, + "num_tokens": 183957129.0, + "step": 7109 + }, + { + "epoch": 0.780803865583132, + "grad_norm": 2.077171564102173, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7053439617156982, + "num_tokens": 183985840.0, + "step": 7110 + }, + { + "epoch": 0.7809136832857456, + "grad_norm": 2.2003190517425537, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.692959725856781, + "num_tokens": 184012366.0, + "step": 7111 + }, + { + "epoch": 0.7810235009883594, + "grad_norm": 2.032207727432251, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6895464658737183, + "num_tokens": 184040951.0, + "step": 7112 + }, + { + "epoch": 0.781133318690973, + "grad_norm": 2.154297113418579, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7145910859107971, + "num_tokens": 184067588.0, + "step": 7113 + }, + { + "epoch": 0.7812431363935867, + "grad_norm": 2.316741466522217, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7339885234832764, + "num_tokens": 184089857.0, + "step": 7114 + }, + { + "epoch": 0.7813529540962003, + "grad_norm": 2.3532614707946777, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7240438461303711, + "num_tokens": 184113498.0, + "step": 7115 + }, + { + "epoch": 0.781462771798814, + "grad_norm": 2.2970478534698486, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6930299997329712, + "num_tokens": 184137404.0, + "step": 7116 + }, + { + "epoch": 0.7815725895014276, + "grad_norm": 2.425877571105957, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7006281614303589, + "num_tokens": 184159063.0, + "step": 7117 + }, + { + "epoch": 0.7816824072040413, + "grad_norm": 2.2386486530303955, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7145634889602661, + "num_tokens": 184181374.0, + "step": 7118 + }, + { + "epoch": 0.7817922249066549, + "grad_norm": 2.2438621520996094, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7185863256454468, + "num_tokens": 184204543.0, + "step": 7119 + }, + { + "epoch": 0.7819020426092687, + "grad_norm": 2.3427958488464355, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7131168246269226, + "num_tokens": 184227436.0, + "step": 7120 + }, + { + "epoch": 0.7820118603118823, + "grad_norm": 2.187190532684326, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7177369594573975, + "num_tokens": 184252789.0, + "step": 7121 + }, + { + "epoch": 0.782121678014496, + "grad_norm": 2.076791763305664, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7140810489654541, + "num_tokens": 184279574.0, + "step": 7122 + }, + { + "epoch": 0.7822314957171096, + "grad_norm": 1.9825479984283447, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6995121240615845, + "num_tokens": 184311071.0, + "step": 7123 + }, + { + "epoch": 0.7823413134197232, + "grad_norm": 2.09401798248291, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6856889128684998, + "num_tokens": 184339486.0, + "step": 7124 + }, + { + "epoch": 0.7824511311223369, + "grad_norm": 2.2120633125305176, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7275388240814209, + "num_tokens": 184362691.0, + "step": 7125 + }, + { + "epoch": 0.7825609488249505, + "grad_norm": 2.0021464824676514, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7047829627990723, + "num_tokens": 184391019.0, + "step": 7126 + }, + { + "epoch": 0.7826707665275643, + "grad_norm": 2.3012351989746094, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7194669246673584, + "num_tokens": 184412939.0, + "step": 7127 + }, + { + "epoch": 0.7827805842301779, + "grad_norm": 2.013226270675659, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7154145240783691, + "num_tokens": 184441180.0, + "step": 7128 + }, + { + "epoch": 0.7828904019327916, + "grad_norm": 1.9730850458145142, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7301707863807678, + "num_tokens": 184471774.0, + "step": 7129 + }, + { + "epoch": 0.7830002196354052, + "grad_norm": 2.327498435974121, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7097049951553345, + "num_tokens": 184494838.0, + "step": 7130 + }, + { + "epoch": 0.7831100373380189, + "grad_norm": 2.081369161605835, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7232722043991089, + "num_tokens": 184520574.0, + "step": 7131 + }, + { + "epoch": 0.7832198550406325, + "grad_norm": 2.3898894786834717, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7132607698440552, + "num_tokens": 184542020.0, + "step": 7132 + }, + { + "epoch": 0.7833296727432462, + "grad_norm": 2.2167880535125732, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6889707446098328, + "num_tokens": 184567106.0, + "step": 7133 + }, + { + "epoch": 0.7834394904458599, + "grad_norm": 2.092456817626953, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7143850326538086, + "num_tokens": 184592293.0, + "step": 7134 + }, + { + "epoch": 0.7835493081484736, + "grad_norm": 2.342514991760254, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6819868683815002, + "num_tokens": 184614675.0, + "step": 7135 + }, + { + "epoch": 0.7836591258510872, + "grad_norm": 2.0423245429992676, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6923916339874268, + "num_tokens": 184643311.0, + "step": 7136 + }, + { + "epoch": 0.7837689435537009, + "grad_norm": 2.0761945247650146, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6890354156494141, + "num_tokens": 184671028.0, + "step": 7137 + }, + { + "epoch": 0.7838787612563145, + "grad_norm": 2.1434483528137207, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7045561075210571, + "num_tokens": 184697887.0, + "step": 7138 + }, + { + "epoch": 0.7839885789589282, + "grad_norm": 2.010657548904419, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7125847935676575, + "num_tokens": 184728433.0, + "step": 7139 + }, + { + "epoch": 0.7840983966615418, + "grad_norm": 2.314281940460205, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7091989517211914, + "num_tokens": 184754313.0, + "step": 7140 + }, + { + "epoch": 0.7842082143641556, + "grad_norm": 1.9500254392623901, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6975995302200317, + "num_tokens": 184785843.0, + "step": 7141 + }, + { + "epoch": 0.7843180320667692, + "grad_norm": 2.1011674404144287, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7179363965988159, + "num_tokens": 184811996.0, + "step": 7142 + }, + { + "epoch": 0.7844278497693828, + "grad_norm": 2.424041271209717, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.691650390625, + "num_tokens": 184834675.0, + "step": 7143 + }, + { + "epoch": 0.7845376674719965, + "grad_norm": 2.021902084350586, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7093900442123413, + "num_tokens": 184863266.0, + "step": 7144 + }, + { + "epoch": 0.7846474851746101, + "grad_norm": 2.2965621948242188, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7234610319137573, + "num_tokens": 184886024.0, + "step": 7145 + }, + { + "epoch": 0.7847573028772238, + "grad_norm": 2.347468137741089, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6989395022392273, + "num_tokens": 184909965.0, + "step": 7146 + }, + { + "epoch": 0.7848671205798374, + "grad_norm": 2.2943291664123535, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7195142507553101, + "num_tokens": 184931186.0, + "step": 7147 + }, + { + "epoch": 0.7849769382824512, + "grad_norm": 2.162017345428467, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6899088621139526, + "num_tokens": 184959886.0, + "step": 7148 + }, + { + "epoch": 0.7850867559850648, + "grad_norm": 2.183950901031494, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7337456941604614, + "num_tokens": 184983109.0, + "step": 7149 + }, + { + "epoch": 0.7851965736876785, + "grad_norm": 2.165414333343506, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6883066892623901, + "num_tokens": 185012198.0, + "step": 7150 + }, + { + "epoch": 0.7853063913902921, + "grad_norm": 2.2639567852020264, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7389812469482422, + "num_tokens": 185033034.0, + "step": 7151 + }, + { + "epoch": 0.7854162090929058, + "grad_norm": 2.114811420440674, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6987031102180481, + "num_tokens": 185059299.0, + "step": 7152 + }, + { + "epoch": 0.7855260267955194, + "grad_norm": 2.074862003326416, + "learning_rate": 1e-06, + "loss": 1.0809, + "mean_token_accuracy": 0.6668286323547363, + "num_tokens": 185087271.0, + "step": 7153 + }, + { + "epoch": 0.7856358444981331, + "grad_norm": 2.081505060195923, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.70185387134552, + "num_tokens": 185113305.0, + "step": 7154 + }, + { + "epoch": 0.7857456622007467, + "grad_norm": 2.4336419105529785, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7160452604293823, + "num_tokens": 185134830.0, + "step": 7155 + }, + { + "epoch": 0.7858554799033605, + "grad_norm": 2.397857189178467, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7254659533500671, + "num_tokens": 185154646.0, + "step": 7156 + }, + { + "epoch": 0.7859652976059741, + "grad_norm": 2.002087116241455, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6975943446159363, + "num_tokens": 185183577.0, + "step": 7157 + }, + { + "epoch": 0.7860751153085878, + "grad_norm": 2.4695382118225098, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7220190763473511, + "num_tokens": 185203013.0, + "step": 7158 + }, + { + "epoch": 0.7861849330112014, + "grad_norm": 2.031601667404175, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6881484985351562, + "num_tokens": 185229869.0, + "step": 7159 + }, + { + "epoch": 0.786294750713815, + "grad_norm": 2.183839797973633, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7002360820770264, + "num_tokens": 185255496.0, + "step": 7160 + }, + { + "epoch": 0.7864045684164287, + "grad_norm": 2.0143473148345947, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6955481171607971, + "num_tokens": 185284771.0, + "step": 7161 + }, + { + "epoch": 0.7865143861190423, + "grad_norm": 2.4071364402770996, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7248615026473999, + "num_tokens": 185306814.0, + "step": 7162 + }, + { + "epoch": 0.7866242038216561, + "grad_norm": 2.2868220806121826, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6901575326919556, + "num_tokens": 185331860.0, + "step": 7163 + }, + { + "epoch": 0.7867340215242697, + "grad_norm": 2.039790153503418, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7017483711242676, + "num_tokens": 185359863.0, + "step": 7164 + }, + { + "epoch": 0.7868438392268834, + "grad_norm": 2.085400342941284, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.6995381116867065, + "num_tokens": 185387084.0, + "step": 7165 + }, + { + "epoch": 0.786953656929497, + "grad_norm": 2.141845941543579, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7274715900421143, + "num_tokens": 185413133.0, + "step": 7166 + }, + { + "epoch": 0.7870634746321107, + "grad_norm": 2.1387462615966797, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6952724456787109, + "num_tokens": 185437925.0, + "step": 7167 + }, + { + "epoch": 0.7871732923347243, + "grad_norm": 2.147874593734741, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6809852123260498, + "num_tokens": 185466086.0, + "step": 7168 + }, + { + "epoch": 0.787283110037338, + "grad_norm": 2.075617551803589, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7138092517852783, + "num_tokens": 185493799.0, + "step": 7169 + }, + { + "epoch": 0.7873929277399517, + "grad_norm": 2.105557441711426, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7023889422416687, + "num_tokens": 185522535.0, + "step": 7170 + }, + { + "epoch": 0.7875027454425654, + "grad_norm": 2.0462546348571777, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6834801435470581, + "num_tokens": 185556122.0, + "step": 7171 + }, + { + "epoch": 0.787612563145179, + "grad_norm": 2.22434663772583, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7191087603569031, + "num_tokens": 185582159.0, + "step": 7172 + }, + { + "epoch": 0.7877223808477927, + "grad_norm": 2.2525012493133545, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.681725263595581, + "num_tokens": 185607770.0, + "step": 7173 + }, + { + "epoch": 0.7878321985504063, + "grad_norm": 2.2397124767303467, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7172244787216187, + "num_tokens": 185633153.0, + "step": 7174 + }, + { + "epoch": 0.78794201625302, + "grad_norm": 2.029949426651001, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7133676409721375, + "num_tokens": 185662012.0, + "step": 7175 + }, + { + "epoch": 0.7880518339556336, + "grad_norm": 1.955962896347046, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6872902512550354, + "num_tokens": 185691755.0, + "step": 7176 + }, + { + "epoch": 0.7881616516582474, + "grad_norm": 2.214123249053955, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7034826874732971, + "num_tokens": 185715651.0, + "step": 7177 + }, + { + "epoch": 0.788271469360861, + "grad_norm": 2.18693208694458, + "learning_rate": 1e-06, + "loss": 1.0598, + "mean_token_accuracy": 0.6786118745803833, + "num_tokens": 185741311.0, + "step": 7178 + }, + { + "epoch": 0.7883812870634747, + "grad_norm": 2.080052137374878, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7243244051933289, + "num_tokens": 185768557.0, + "step": 7179 + }, + { + "epoch": 0.7884911047660883, + "grad_norm": 2.1767680644989014, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6917171478271484, + "num_tokens": 185795015.0, + "step": 7180 + }, + { + "epoch": 0.788600922468702, + "grad_norm": 2.214428663253784, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6927329301834106, + "num_tokens": 185819116.0, + "step": 7181 + }, + { + "epoch": 0.7887107401713156, + "grad_norm": 2.3452775478363037, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7100029587745667, + "num_tokens": 185841426.0, + "step": 7182 + }, + { + "epoch": 0.7888205578739292, + "grad_norm": 2.1774213314056396, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7044894695281982, + "num_tokens": 185866540.0, + "step": 7183 + }, + { + "epoch": 0.7889303755765429, + "grad_norm": 2.1792521476745605, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7043198347091675, + "num_tokens": 185892827.0, + "step": 7184 + }, + { + "epoch": 0.7890401932791566, + "grad_norm": 2.44938063621521, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.729281485080719, + "num_tokens": 185913702.0, + "step": 7185 + }, + { + "epoch": 0.7891500109817703, + "grad_norm": 2.1705784797668457, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7154836654663086, + "num_tokens": 185939475.0, + "step": 7186 + }, + { + "epoch": 0.7892598286843839, + "grad_norm": 1.9174309968948364, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6893821358680725, + "num_tokens": 185972425.0, + "step": 7187 + }, + { + "epoch": 0.7893696463869976, + "grad_norm": 1.8720312118530273, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6969132423400879, + "num_tokens": 186007050.0, + "step": 7188 + }, + { + "epoch": 0.7894794640896112, + "grad_norm": 2.2533726692199707, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7161052227020264, + "num_tokens": 186030479.0, + "step": 7189 + }, + { + "epoch": 0.7895892817922249, + "grad_norm": 1.9869959354400635, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.702927827835083, + "num_tokens": 186061037.0, + "step": 7190 + }, + { + "epoch": 0.7896990994948385, + "grad_norm": 2.161425828933716, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7053393125534058, + "num_tokens": 186086817.0, + "step": 7191 + }, + { + "epoch": 0.7898089171974523, + "grad_norm": 2.0619406700134277, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7135536670684814, + "num_tokens": 186114481.0, + "step": 7192 + }, + { + "epoch": 0.7899187349000659, + "grad_norm": 2.149547815322876, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6897412538528442, + "num_tokens": 186142585.0, + "step": 7193 + }, + { + "epoch": 0.7900285526026796, + "grad_norm": 2.1245741844177246, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.7026290893554688, + "num_tokens": 186170466.0, + "step": 7194 + }, + { + "epoch": 0.7901383703052932, + "grad_norm": 2.1631247997283936, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7084903717041016, + "num_tokens": 186197787.0, + "step": 7195 + }, + { + "epoch": 0.7902481880079069, + "grad_norm": 2.320181369781494, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7191709280014038, + "num_tokens": 186220283.0, + "step": 7196 + }, + { + "epoch": 0.7903580057105205, + "grad_norm": 1.9686387777328491, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7042988538742065, + "num_tokens": 186250044.0, + "step": 7197 + }, + { + "epoch": 0.7904678234131342, + "grad_norm": 2.0116658210754395, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6906975507736206, + "num_tokens": 186279918.0, + "step": 7198 + }, + { + "epoch": 0.7905776411157479, + "grad_norm": 2.230525016784668, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6972948312759399, + "num_tokens": 186306253.0, + "step": 7199 + }, + { + "epoch": 0.7906874588183616, + "grad_norm": 2.053166151046753, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7116948962211609, + "num_tokens": 186334988.0, + "step": 7200 + }, + { + "epoch": 0.7907972765209752, + "grad_norm": 2.185126304626465, + "learning_rate": 1e-06, + "loss": 1.0596, + "mean_token_accuracy": 0.6765882968902588, + "num_tokens": 186362393.0, + "step": 7201 + }, + { + "epoch": 0.7909070942235888, + "grad_norm": 1.9292981624603271, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7261422872543335, + "num_tokens": 186391206.0, + "step": 7202 + }, + { + "epoch": 0.7910169119262025, + "grad_norm": 2.010892152786255, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7059828042984009, + "num_tokens": 186420976.0, + "step": 7203 + }, + { + "epoch": 0.7911267296288161, + "grad_norm": 2.1760129928588867, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7120954990386963, + "num_tokens": 186447364.0, + "step": 7204 + }, + { + "epoch": 0.7912365473314298, + "grad_norm": 2.5023646354675293, + "learning_rate": 1e-06, + "loss": 1.0629, + "mean_token_accuracy": 0.6963053941726685, + "num_tokens": 186470996.0, + "step": 7205 + }, + { + "epoch": 0.7913463650340435, + "grad_norm": 2.15071439743042, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7275654077529907, + "num_tokens": 186496877.0, + "step": 7206 + }, + { + "epoch": 0.7914561827366572, + "grad_norm": 2.282114267349243, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7204872369766235, + "num_tokens": 186519265.0, + "step": 7207 + }, + { + "epoch": 0.7915660004392708, + "grad_norm": 2.0825862884521484, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.70069420337677, + "num_tokens": 186548132.0, + "step": 7208 + }, + { + "epoch": 0.7916758181418845, + "grad_norm": 2.057098388671875, + "learning_rate": 1e-06, + "loss": 1.1137, + "mean_token_accuracy": 0.6651146411895752, + "num_tokens": 186577678.0, + "step": 7209 + }, + { + "epoch": 0.7917856358444981, + "grad_norm": 1.9022979736328125, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.711122989654541, + "num_tokens": 186608047.0, + "step": 7210 + }, + { + "epoch": 0.7918954535471118, + "grad_norm": 1.8648011684417725, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6997413635253906, + "num_tokens": 186642725.0, + "step": 7211 + }, + { + "epoch": 0.7920052712497254, + "grad_norm": 1.9031000137329102, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7076584100723267, + "num_tokens": 186673867.0, + "step": 7212 + }, + { + "epoch": 0.7921150889523391, + "grad_norm": 1.9235771894454956, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6922037601470947, + "num_tokens": 186704477.0, + "step": 7213 + }, + { + "epoch": 0.7922249066549528, + "grad_norm": 2.1838014125823975, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.6992559432983398, + "num_tokens": 186730133.0, + "step": 7214 + }, + { + "epoch": 0.7923347243575665, + "grad_norm": 2.611001968383789, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.710383951663971, + "num_tokens": 186750994.0, + "step": 7215 + }, + { + "epoch": 0.7924445420601801, + "grad_norm": 2.5413737297058105, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7142837047576904, + "num_tokens": 186770534.0, + "step": 7216 + }, + { + "epoch": 0.7925543597627938, + "grad_norm": 2.0760273933410645, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6953674554824829, + "num_tokens": 186798520.0, + "step": 7217 + }, + { + "epoch": 0.7926641774654074, + "grad_norm": 1.999807596206665, + "learning_rate": 1e-06, + "loss": 1.0809, + "mean_token_accuracy": 0.6729060411453247, + "num_tokens": 186828948.0, + "step": 7218 + }, + { + "epoch": 0.792773995168021, + "grad_norm": 2.127230167388916, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.700237512588501, + "num_tokens": 186857082.0, + "step": 7219 + }, + { + "epoch": 0.7928838128706347, + "grad_norm": 2.4736812114715576, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6916115283966064, + "num_tokens": 186880348.0, + "step": 7220 + }, + { + "epoch": 0.7929936305732485, + "grad_norm": 2.6046881675720215, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7088573575019836, + "num_tokens": 186899871.0, + "step": 7221 + }, + { + "epoch": 0.7931034482758621, + "grad_norm": 2.2309823036193848, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6910343766212463, + "num_tokens": 186930246.0, + "step": 7222 + }, + { + "epoch": 0.7932132659784757, + "grad_norm": 2.1407930850982666, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.715632438659668, + "num_tokens": 186957223.0, + "step": 7223 + }, + { + "epoch": 0.7933230836810894, + "grad_norm": 2.30129337310791, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7085893154144287, + "num_tokens": 186980450.0, + "step": 7224 + }, + { + "epoch": 0.793432901383703, + "grad_norm": 2.2385644912719727, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7138438820838928, + "num_tokens": 187004492.0, + "step": 7225 + }, + { + "epoch": 0.7935427190863167, + "grad_norm": 2.166780710220337, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.71783846616745, + "num_tokens": 187030339.0, + "step": 7226 + }, + { + "epoch": 0.7936525367889303, + "grad_norm": 2.24010968208313, + "learning_rate": 1e-06, + "loss": 1.0368, + "mean_token_accuracy": 0.6908540725708008, + "num_tokens": 187057485.0, + "step": 7227 + }, + { + "epoch": 0.7937623544915441, + "grad_norm": 2.509364366531372, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.708589494228363, + "num_tokens": 187077206.0, + "step": 7228 + }, + { + "epoch": 0.7938721721941577, + "grad_norm": 2.1020209789276123, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7115681171417236, + "num_tokens": 187104941.0, + "step": 7229 + }, + { + "epoch": 0.7939819898967714, + "grad_norm": 2.4000961780548096, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7238866090774536, + "num_tokens": 187127950.0, + "step": 7230 + }, + { + "epoch": 0.794091807599385, + "grad_norm": 2.1731505393981934, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6936050653457642, + "num_tokens": 187153998.0, + "step": 7231 + }, + { + "epoch": 0.7942016253019987, + "grad_norm": 2.3123679161071777, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7251409292221069, + "num_tokens": 187174998.0, + "step": 7232 + }, + { + "epoch": 0.7943114430046123, + "grad_norm": 2.115903615951538, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.708876371383667, + "num_tokens": 187201294.0, + "step": 7233 + }, + { + "epoch": 0.794421260707226, + "grad_norm": 2.0285110473632812, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6999384760856628, + "num_tokens": 187229083.0, + "step": 7234 + }, + { + "epoch": 0.7945310784098397, + "grad_norm": 2.321171522140503, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.71454918384552, + "num_tokens": 187250489.0, + "step": 7235 + }, + { + "epoch": 0.7946408961124534, + "grad_norm": 2.098217248916626, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7212376594543457, + "num_tokens": 187277302.0, + "step": 7236 + }, + { + "epoch": 0.794750713815067, + "grad_norm": 2.2754762172698975, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7131286859512329, + "num_tokens": 187300062.0, + "step": 7237 + }, + { + "epoch": 0.7948605315176807, + "grad_norm": 2.5282139778137207, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7198019027709961, + "num_tokens": 187319118.0, + "step": 7238 + }, + { + "epoch": 0.7949703492202943, + "grad_norm": 2.153877019882202, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7094309329986572, + "num_tokens": 187345374.0, + "step": 7239 + }, + { + "epoch": 0.795080166922908, + "grad_norm": 2.117980718612671, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7284069061279297, + "num_tokens": 187370379.0, + "step": 7240 + }, + { + "epoch": 0.7951899846255216, + "grad_norm": 2.1467807292938232, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7212916612625122, + "num_tokens": 187395194.0, + "step": 7241 + }, + { + "epoch": 0.7952998023281352, + "grad_norm": 2.3188538551330566, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7050236463546753, + "num_tokens": 187418656.0, + "step": 7242 + }, + { + "epoch": 0.795409620030749, + "grad_norm": 2.180668592453003, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7014338970184326, + "num_tokens": 187442710.0, + "step": 7243 + }, + { + "epoch": 0.7955194377333626, + "grad_norm": 2.057866334915161, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7129713296890259, + "num_tokens": 187470929.0, + "step": 7244 + }, + { + "epoch": 0.7956292554359763, + "grad_norm": 2.0108022689819336, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7197670936584473, + "num_tokens": 187499503.0, + "step": 7245 + }, + { + "epoch": 0.7957390731385899, + "grad_norm": 2.2255845069885254, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7245672941207886, + "num_tokens": 187523173.0, + "step": 7246 + }, + { + "epoch": 0.7958488908412036, + "grad_norm": 2.010174036026001, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7092505693435669, + "num_tokens": 187551392.0, + "step": 7247 + }, + { + "epoch": 0.7959587085438172, + "grad_norm": 1.875681757926941, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6887036561965942, + "num_tokens": 187583865.0, + "step": 7248 + }, + { + "epoch": 0.7960685262464309, + "grad_norm": 2.3536524772644043, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7137632966041565, + "num_tokens": 187605803.0, + "step": 7249 + }, + { + "epoch": 0.7961783439490446, + "grad_norm": 2.378002882003784, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.704033613204956, + "num_tokens": 187628523.0, + "step": 7250 + }, + { + "epoch": 0.7962881616516583, + "grad_norm": 2.5474653244018555, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.703690767288208, + "num_tokens": 187648844.0, + "step": 7251 + }, + { + "epoch": 0.7963979793542719, + "grad_norm": 2.589958906173706, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.6999871730804443, + "num_tokens": 187668170.0, + "step": 7252 + }, + { + "epoch": 0.7965077970568856, + "grad_norm": 2.1396069526672363, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7087773084640503, + "num_tokens": 187695231.0, + "step": 7253 + }, + { + "epoch": 0.7966176147594992, + "grad_norm": 2.3908627033233643, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7497475147247314, + "num_tokens": 187716433.0, + "step": 7254 + }, + { + "epoch": 0.7967274324621129, + "grad_norm": 2.373872995376587, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7428322434425354, + "num_tokens": 187736921.0, + "step": 7255 + }, + { + "epoch": 0.7968372501647265, + "grad_norm": 2.130286455154419, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6984506845474243, + "num_tokens": 187761659.0, + "step": 7256 + }, + { + "epoch": 0.7969470678673403, + "grad_norm": 1.9336525201797485, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.6787999272346497, + "num_tokens": 187794860.0, + "step": 7257 + }, + { + "epoch": 0.7970568855699539, + "grad_norm": 2.0924718379974365, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7052043676376343, + "num_tokens": 187822637.0, + "step": 7258 + }, + { + "epoch": 0.7971667032725676, + "grad_norm": 1.9708166122436523, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.7037422060966492, + "num_tokens": 187853797.0, + "step": 7259 + }, + { + "epoch": 0.7972765209751812, + "grad_norm": 2.4411089420318604, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7171163558959961, + "num_tokens": 187874537.0, + "step": 7260 + }, + { + "epoch": 0.7973863386777948, + "grad_norm": 1.9242194890975952, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6727033853530884, + "num_tokens": 187907296.0, + "step": 7261 + }, + { + "epoch": 0.7974961563804085, + "grad_norm": 2.1008408069610596, + "learning_rate": 1e-06, + "loss": 1.1128, + "mean_token_accuracy": 0.6654011011123657, + "num_tokens": 187935828.0, + "step": 7262 + }, + { + "epoch": 0.7976059740830221, + "grad_norm": 2.2513837814331055, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7279629707336426, + "num_tokens": 187958585.0, + "step": 7263 + }, + { + "epoch": 0.7977157917856359, + "grad_norm": 2.3506195545196533, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7187942266464233, + "num_tokens": 187979118.0, + "step": 7264 + }, + { + "epoch": 0.7978256094882495, + "grad_norm": 2.233121395111084, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7100399732589722, + "num_tokens": 188005438.0, + "step": 7265 + }, + { + "epoch": 0.7979354271908632, + "grad_norm": 2.360846519470215, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7091883420944214, + "num_tokens": 188029090.0, + "step": 7266 + }, + { + "epoch": 0.7980452448934768, + "grad_norm": 2.1718602180480957, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.679743766784668, + "num_tokens": 188053893.0, + "step": 7267 + }, + { + "epoch": 0.7981550625960905, + "grad_norm": 2.1709601879119873, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6995806097984314, + "num_tokens": 188078874.0, + "step": 7268 + }, + { + "epoch": 0.7982648802987041, + "grad_norm": 2.222470760345459, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7027905583381653, + "num_tokens": 188105436.0, + "step": 7269 + }, + { + "epoch": 0.7983746980013178, + "grad_norm": 1.9781818389892578, + "learning_rate": 1e-06, + "loss": 1.0672, + "mean_token_accuracy": 0.688006579875946, + "num_tokens": 188139579.0, + "step": 7270 + }, + { + "epoch": 0.7984845157039314, + "grad_norm": 2.178161144256592, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7191929817199707, + "num_tokens": 188165248.0, + "step": 7271 + }, + { + "epoch": 0.7985943334065452, + "grad_norm": 2.078784465789795, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6954445838928223, + "num_tokens": 188192813.0, + "step": 7272 + }, + { + "epoch": 0.7987041511091588, + "grad_norm": 2.49615478515625, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7164422273635864, + "num_tokens": 188212555.0, + "step": 7273 + }, + { + "epoch": 0.7988139688117725, + "grad_norm": 2.0845673084259033, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7045154571533203, + "num_tokens": 188240737.0, + "step": 7274 + }, + { + "epoch": 0.7989237865143861, + "grad_norm": 1.9971171617507935, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7070623636245728, + "num_tokens": 188268722.0, + "step": 7275 + }, + { + "epoch": 0.7990336042169998, + "grad_norm": 2.2997093200683594, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6866844296455383, + "num_tokens": 188295061.0, + "step": 7276 + }, + { + "epoch": 0.7991434219196134, + "grad_norm": 2.1326136589050293, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7188708782196045, + "num_tokens": 188322008.0, + "step": 7277 + }, + { + "epoch": 0.799253239622227, + "grad_norm": 2.024484872817993, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6903747916221619, + "num_tokens": 188351369.0, + "step": 7278 + }, + { + "epoch": 0.7993630573248408, + "grad_norm": 2.293896198272705, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.722567081451416, + "num_tokens": 188374342.0, + "step": 7279 + }, + { + "epoch": 0.7994728750274545, + "grad_norm": 2.102720260620117, + "learning_rate": 1e-06, + "loss": 1.0682, + "mean_token_accuracy": 0.6823684573173523, + "num_tokens": 188403080.0, + "step": 7280 + }, + { + "epoch": 0.7995826927300681, + "grad_norm": 1.902053952217102, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6886470317840576, + "num_tokens": 188434808.0, + "step": 7281 + }, + { + "epoch": 0.7996925104326817, + "grad_norm": 2.0435943603515625, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7132983207702637, + "num_tokens": 188465375.0, + "step": 7282 + }, + { + "epoch": 0.7998023281352954, + "grad_norm": 1.935660481452942, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7103093266487122, + "num_tokens": 188495748.0, + "step": 7283 + }, + { + "epoch": 0.799912145837909, + "grad_norm": 2.4034125804901123, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7066713571548462, + "num_tokens": 188517345.0, + "step": 7284 + }, + { + "epoch": 0.8000219635405227, + "grad_norm": 2.1603081226348877, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7098448276519775, + "num_tokens": 188543347.0, + "step": 7285 + }, + { + "epoch": 0.8001317812431364, + "grad_norm": 1.8398702144622803, + "learning_rate": 1e-06, + "loss": 1.0854, + "mean_token_accuracy": 0.6788573265075684, + "num_tokens": 188577493.0, + "step": 7286 + }, + { + "epoch": 0.8002415989457501, + "grad_norm": 2.075857639312744, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7142651081085205, + "num_tokens": 188602905.0, + "step": 7287 + }, + { + "epoch": 0.8003514166483637, + "grad_norm": 2.132030963897705, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.708319365978241, + "num_tokens": 188629470.0, + "step": 7288 + }, + { + "epoch": 0.8004612343509774, + "grad_norm": 2.1195788383483887, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6951623558998108, + "num_tokens": 188657305.0, + "step": 7289 + }, + { + "epoch": 0.800571052053591, + "grad_norm": 2.108630895614624, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6823672652244568, + "num_tokens": 188685231.0, + "step": 7290 + }, + { + "epoch": 0.8006808697562047, + "grad_norm": 1.9301890134811401, + "learning_rate": 1e-06, + "loss": 1.0592, + "mean_token_accuracy": 0.6838688850402832, + "num_tokens": 188719881.0, + "step": 7291 + }, + { + "epoch": 0.8007906874588183, + "grad_norm": 2.4466359615325928, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7352226972579956, + "num_tokens": 188740271.0, + "step": 7292 + }, + { + "epoch": 0.8009005051614321, + "grad_norm": 1.8749245405197144, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7196812629699707, + "num_tokens": 188771459.0, + "step": 7293 + }, + { + "epoch": 0.8010103228640457, + "grad_norm": 2.2321529388427734, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7218121290206909, + "num_tokens": 188796163.0, + "step": 7294 + }, + { + "epoch": 0.8011201405666594, + "grad_norm": 2.46403169631958, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7107763290405273, + "num_tokens": 188818117.0, + "step": 7295 + }, + { + "epoch": 0.801229958269273, + "grad_norm": 2.367694854736328, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7118167877197266, + "num_tokens": 188839835.0, + "step": 7296 + }, + { + "epoch": 0.8013397759718867, + "grad_norm": 2.2268872261047363, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7059921026229858, + "num_tokens": 188864548.0, + "step": 7297 + }, + { + "epoch": 0.8014495936745003, + "grad_norm": 2.0022709369659424, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6987137794494629, + "num_tokens": 188894610.0, + "step": 7298 + }, + { + "epoch": 0.801559411377114, + "grad_norm": 2.4753832817077637, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7237663269042969, + "num_tokens": 188915161.0, + "step": 7299 + }, + { + "epoch": 0.8016692290797276, + "grad_norm": 2.16857647895813, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6911593675613403, + "num_tokens": 188944775.0, + "step": 7300 + }, + { + "epoch": 0.8017790467823414, + "grad_norm": 2.0939667224884033, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7113878726959229, + "num_tokens": 188973964.0, + "step": 7301 + }, + { + "epoch": 0.801888864484955, + "grad_norm": 2.23191499710083, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7129027843475342, + "num_tokens": 189000178.0, + "step": 7302 + }, + { + "epoch": 0.8019986821875686, + "grad_norm": 2.0142159461975098, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7030297517776489, + "num_tokens": 189030193.0, + "step": 7303 + }, + { + "epoch": 0.8021084998901823, + "grad_norm": 2.314436912536621, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6879521608352661, + "num_tokens": 189053274.0, + "step": 7304 + }, + { + "epoch": 0.8022183175927959, + "grad_norm": 2.085240364074707, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6900932788848877, + "num_tokens": 189082363.0, + "step": 7305 + }, + { + "epoch": 0.8023281352954096, + "grad_norm": 2.1092703342437744, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6990206241607666, + "num_tokens": 189109955.0, + "step": 7306 + }, + { + "epoch": 0.8024379529980232, + "grad_norm": 2.097445249557495, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7248893976211548, + "num_tokens": 189134395.0, + "step": 7307 + }, + { + "epoch": 0.802547770700637, + "grad_norm": 2.0815672874450684, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7050850987434387, + "num_tokens": 189162058.0, + "step": 7308 + }, + { + "epoch": 0.8026575884032506, + "grad_norm": 2.1368110179901123, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.7005321979522705, + "num_tokens": 189187227.0, + "step": 7309 + }, + { + "epoch": 0.8027674061058643, + "grad_norm": 2.192235231399536, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7185807228088379, + "num_tokens": 189212907.0, + "step": 7310 + }, + { + "epoch": 0.8028772238084779, + "grad_norm": 2.0548593997955322, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.711530327796936, + "num_tokens": 189241018.0, + "step": 7311 + }, + { + "epoch": 0.8029870415110916, + "grad_norm": 2.3193342685699463, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7051951885223389, + "num_tokens": 189263735.0, + "step": 7312 + }, + { + "epoch": 0.8030968592137052, + "grad_norm": 2.205258846282959, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7247756123542786, + "num_tokens": 189293717.0, + "step": 7313 + }, + { + "epoch": 0.8032066769163189, + "grad_norm": 2.314335584640503, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.7007149457931519, + "num_tokens": 189316183.0, + "step": 7314 + }, + { + "epoch": 0.8033164946189326, + "grad_norm": 2.112374782562256, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7056600451469421, + "num_tokens": 189343617.0, + "step": 7315 + }, + { + "epoch": 0.8034263123215463, + "grad_norm": 2.2062063217163086, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6999971866607666, + "num_tokens": 189371355.0, + "step": 7316 + }, + { + "epoch": 0.8035361300241599, + "grad_norm": 2.5912230014801025, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7068046927452087, + "num_tokens": 189391749.0, + "step": 7317 + }, + { + "epoch": 0.8036459477267736, + "grad_norm": 2.0389809608459473, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6922814846038818, + "num_tokens": 189421161.0, + "step": 7318 + }, + { + "epoch": 0.8037557654293872, + "grad_norm": 2.5008633136749268, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7210431098937988, + "num_tokens": 189440703.0, + "step": 7319 + }, + { + "epoch": 0.8038655831320008, + "grad_norm": 2.496730089187622, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.710914134979248, + "num_tokens": 189460939.0, + "step": 7320 + }, + { + "epoch": 0.8039754008346145, + "grad_norm": 2.0049784183502197, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6958656907081604, + "num_tokens": 189492736.0, + "step": 7321 + }, + { + "epoch": 0.8040852185372283, + "grad_norm": 2.247527837753296, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7025469541549683, + "num_tokens": 189517571.0, + "step": 7322 + }, + { + "epoch": 0.8041950362398419, + "grad_norm": 2.0960235595703125, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7148244380950928, + "num_tokens": 189544763.0, + "step": 7323 + }, + { + "epoch": 0.8043048539424555, + "grad_norm": 2.103663444519043, + "learning_rate": 1e-06, + "loss": 1.0755, + "mean_token_accuracy": 0.6851674914360046, + "num_tokens": 189571734.0, + "step": 7324 + }, + { + "epoch": 0.8044146716450692, + "grad_norm": 2.078756332397461, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7170624732971191, + "num_tokens": 189599544.0, + "step": 7325 + }, + { + "epoch": 0.8045244893476828, + "grad_norm": 2.467456579208374, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6961257457733154, + "num_tokens": 189620801.0, + "step": 7326 + }, + { + "epoch": 0.8046343070502965, + "grad_norm": 2.4153170585632324, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.69602370262146, + "num_tokens": 189642798.0, + "step": 7327 + }, + { + "epoch": 0.8047441247529101, + "grad_norm": 2.348839282989502, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7153944373130798, + "num_tokens": 189667045.0, + "step": 7328 + }, + { + "epoch": 0.8048539424555239, + "grad_norm": 2.198279619216919, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7180144786834717, + "num_tokens": 189693170.0, + "step": 7329 + }, + { + "epoch": 0.8049637601581375, + "grad_norm": 2.4360194206237793, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6966468691825867, + "num_tokens": 189717385.0, + "step": 7330 + }, + { + "epoch": 0.8050735778607512, + "grad_norm": 2.172520160675049, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7064789533615112, + "num_tokens": 189743773.0, + "step": 7331 + }, + { + "epoch": 0.8051833955633648, + "grad_norm": 2.0324387550354004, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6844033002853394, + "num_tokens": 189775152.0, + "step": 7332 + }, + { + "epoch": 0.8052932132659785, + "grad_norm": 2.103389263153076, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7384817600250244, + "num_tokens": 189801396.0, + "step": 7333 + }, + { + "epoch": 0.8054030309685921, + "grad_norm": 2.4816222190856934, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6922147274017334, + "num_tokens": 189824889.0, + "step": 7334 + }, + { + "epoch": 0.8055128486712058, + "grad_norm": 2.453030586242676, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7141808271408081, + "num_tokens": 189845178.0, + "step": 7335 + }, + { + "epoch": 0.8056226663738194, + "grad_norm": 2.149677276611328, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7102327346801758, + "num_tokens": 189869934.0, + "step": 7336 + }, + { + "epoch": 0.8057324840764332, + "grad_norm": 2.123964786529541, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6910138130187988, + "num_tokens": 189897957.0, + "step": 7337 + }, + { + "epoch": 0.8058423017790468, + "grad_norm": 2.1077451705932617, + "learning_rate": 1e-06, + "loss": 1.0501, + "mean_token_accuracy": 0.6825471520423889, + "num_tokens": 189926967.0, + "step": 7338 + }, + { + "epoch": 0.8059521194816605, + "grad_norm": 2.3597004413604736, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7239009737968445, + "num_tokens": 189948959.0, + "step": 7339 + }, + { + "epoch": 0.8060619371842741, + "grad_norm": 1.9581676721572876, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7023460865020752, + "num_tokens": 189979548.0, + "step": 7340 + }, + { + "epoch": 0.8061717548868877, + "grad_norm": 2.3185696601867676, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6937414407730103, + "num_tokens": 190003615.0, + "step": 7341 + }, + { + "epoch": 0.8062815725895014, + "grad_norm": 2.111210346221924, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7305629253387451, + "num_tokens": 190027255.0, + "step": 7342 + }, + { + "epoch": 0.806391390292115, + "grad_norm": 2.0543124675750732, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7068545818328857, + "num_tokens": 190054949.0, + "step": 7343 + }, + { + "epoch": 0.8065012079947288, + "grad_norm": 2.1460931301116943, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7313804626464844, + "num_tokens": 190077434.0, + "step": 7344 + }, + { + "epoch": 0.8066110256973424, + "grad_norm": 2.1450586318969727, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.710308313369751, + "num_tokens": 190105048.0, + "step": 7345 + }, + { + "epoch": 0.8067208433999561, + "grad_norm": 1.9809236526489258, + "learning_rate": 1e-06, + "loss": 1.1262, + "mean_token_accuracy": 0.6669880151748657, + "num_tokens": 190135929.0, + "step": 7346 + }, + { + "epoch": 0.8068306611025697, + "grad_norm": 2.0027854442596436, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6925007104873657, + "num_tokens": 190164509.0, + "step": 7347 + }, + { + "epoch": 0.8069404788051834, + "grad_norm": 2.4092535972595215, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7230356931686401, + "num_tokens": 190186916.0, + "step": 7348 + }, + { + "epoch": 0.807050296507797, + "grad_norm": 2.1352076530456543, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6975297927856445, + "num_tokens": 190215352.0, + "step": 7349 + }, + { + "epoch": 0.8071601142104107, + "grad_norm": 2.1414501667022705, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.716039776802063, + "num_tokens": 190240627.0, + "step": 7350 + }, + { + "epoch": 0.8072699319130244, + "grad_norm": 2.1659185886383057, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.701438844203949, + "num_tokens": 190266976.0, + "step": 7351 + }, + { + "epoch": 0.8073797496156381, + "grad_norm": 2.32550311088562, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7281094789505005, + "num_tokens": 190290414.0, + "step": 7352 + }, + { + "epoch": 0.8074895673182517, + "grad_norm": 2.4624035358428955, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6919836401939392, + "num_tokens": 190314442.0, + "step": 7353 + }, + { + "epoch": 0.8075993850208654, + "grad_norm": 2.0895814895629883, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7289107441902161, + "num_tokens": 190341818.0, + "step": 7354 + }, + { + "epoch": 0.807709202723479, + "grad_norm": 2.067949056625366, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7281294465065002, + "num_tokens": 190368152.0, + "step": 7355 + }, + { + "epoch": 0.8078190204260927, + "grad_norm": 2.41552996635437, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7086250185966492, + "num_tokens": 190387620.0, + "step": 7356 + }, + { + "epoch": 0.8079288381287063, + "grad_norm": 2.2339272499084473, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7095023393630981, + "num_tokens": 190411877.0, + "step": 7357 + }, + { + "epoch": 0.8080386558313201, + "grad_norm": 2.213942289352417, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7157886624336243, + "num_tokens": 190436207.0, + "step": 7358 + }, + { + "epoch": 0.8081484735339337, + "grad_norm": 1.8734400272369385, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7058056592941284, + "num_tokens": 190469819.0, + "step": 7359 + }, + { + "epoch": 0.8082582912365474, + "grad_norm": 2.0412003993988037, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7137489318847656, + "num_tokens": 190497336.0, + "step": 7360 + }, + { + "epoch": 0.808368108939161, + "grad_norm": 2.0553348064422607, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6860859394073486, + "num_tokens": 190526504.0, + "step": 7361 + }, + { + "epoch": 0.8084779266417746, + "grad_norm": 1.9588677883148193, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6980657577514648, + "num_tokens": 190559033.0, + "step": 7362 + }, + { + "epoch": 0.8085877443443883, + "grad_norm": 2.646009922027588, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7297479510307312, + "num_tokens": 190576453.0, + "step": 7363 + }, + { + "epoch": 0.8086975620470019, + "grad_norm": 2.3156578540802, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6998707056045532, + "num_tokens": 190601617.0, + "step": 7364 + }, + { + "epoch": 0.8088073797496156, + "grad_norm": 2.4611947536468506, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.6984333992004395, + "num_tokens": 190622769.0, + "step": 7365 + }, + { + "epoch": 0.8089171974522293, + "grad_norm": 1.9396464824676514, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7044540047645569, + "num_tokens": 190654941.0, + "step": 7366 + }, + { + "epoch": 0.809027015154843, + "grad_norm": 2.268902063369751, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6936632990837097, + "num_tokens": 190679722.0, + "step": 7367 + }, + { + "epoch": 0.8091368328574566, + "grad_norm": 1.9965835809707642, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7015645503997803, + "num_tokens": 190708192.0, + "step": 7368 + }, + { + "epoch": 0.8092466505600703, + "grad_norm": 2.2795917987823486, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6979206204414368, + "num_tokens": 190731374.0, + "step": 7369 + }, + { + "epoch": 0.8093564682626839, + "grad_norm": 2.1646535396575928, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6986729502677917, + "num_tokens": 190759008.0, + "step": 7370 + }, + { + "epoch": 0.8094662859652976, + "grad_norm": 2.1305108070373535, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7084490656852722, + "num_tokens": 190784306.0, + "step": 7371 + }, + { + "epoch": 0.8095761036679112, + "grad_norm": 2.158039093017578, + "learning_rate": 1e-06, + "loss": 1.0681, + "mean_token_accuracy": 0.6834347248077393, + "num_tokens": 190812971.0, + "step": 7372 + }, + { + "epoch": 0.809685921370525, + "grad_norm": 2.2895026206970215, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.718682050704956, + "num_tokens": 190835592.0, + "step": 7373 + }, + { + "epoch": 0.8097957390731386, + "grad_norm": 2.15020489692688, + "learning_rate": 1e-06, + "loss": 1.0631, + "mean_token_accuracy": 0.6826794147491455, + "num_tokens": 190864363.0, + "step": 7374 + }, + { + "epoch": 0.8099055567757523, + "grad_norm": 2.7163078784942627, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7077188491821289, + "num_tokens": 190882458.0, + "step": 7375 + }, + { + "epoch": 0.8100153744783659, + "grad_norm": 2.1679019927978516, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6967507600784302, + "num_tokens": 190909073.0, + "step": 7376 + }, + { + "epoch": 0.8101251921809796, + "grad_norm": 2.1880033016204834, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.719194769859314, + "num_tokens": 190933418.0, + "step": 7377 + }, + { + "epoch": 0.8102350098835932, + "grad_norm": 2.3184866905212402, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6931508779525757, + "num_tokens": 190957448.0, + "step": 7378 + }, + { + "epoch": 0.8103448275862069, + "grad_norm": 2.0654356479644775, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.720923662185669, + "num_tokens": 190984754.0, + "step": 7379 + }, + { + "epoch": 0.8104546452888206, + "grad_norm": 2.391916513442993, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6993448734283447, + "num_tokens": 191007254.0, + "step": 7380 + }, + { + "epoch": 0.8105644629914343, + "grad_norm": 2.2419159412384033, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.709143340587616, + "num_tokens": 191030203.0, + "step": 7381 + }, + { + "epoch": 0.8106742806940479, + "grad_norm": 2.2519445419311523, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7029191255569458, + "num_tokens": 191053705.0, + "step": 7382 + }, + { + "epoch": 0.8107840983966615, + "grad_norm": 2.084726572036743, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7114965915679932, + "num_tokens": 191081547.0, + "step": 7383 + }, + { + "epoch": 0.8108939160992752, + "grad_norm": 2.7511417865753174, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.725670576095581, + "num_tokens": 191100151.0, + "step": 7384 + }, + { + "epoch": 0.8110037338018888, + "grad_norm": 2.005466938018799, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7180508375167847, + "num_tokens": 191128014.0, + "step": 7385 + }, + { + "epoch": 0.8111135515045025, + "grad_norm": 2.0997865200042725, + "learning_rate": 1e-06, + "loss": 1.0592, + "mean_token_accuracy": 0.6848277449607849, + "num_tokens": 191154703.0, + "step": 7386 + }, + { + "epoch": 0.8112233692071162, + "grad_norm": 2.1993248462677, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7078555226325989, + "num_tokens": 191180285.0, + "step": 7387 + }, + { + "epoch": 0.8113331869097299, + "grad_norm": 2.3567020893096924, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7021958827972412, + "num_tokens": 191201760.0, + "step": 7388 + }, + { + "epoch": 0.8114430046123435, + "grad_norm": 2.021402359008789, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6981955766677856, + "num_tokens": 191231226.0, + "step": 7389 + }, + { + "epoch": 0.8115528223149572, + "grad_norm": 2.3605942726135254, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6838180422782898, + "num_tokens": 191254797.0, + "step": 7390 + }, + { + "epoch": 0.8116626400175708, + "grad_norm": 2.387065887451172, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7248692512512207, + "num_tokens": 191278029.0, + "step": 7391 + }, + { + "epoch": 0.8117724577201845, + "grad_norm": 1.9436396360397339, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7034211754798889, + "num_tokens": 191306205.0, + "step": 7392 + }, + { + "epoch": 0.8118822754227981, + "grad_norm": 2.220102310180664, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6910527348518372, + "num_tokens": 191331158.0, + "step": 7393 + }, + { + "epoch": 0.8119920931254118, + "grad_norm": 2.0592401027679443, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6945226788520813, + "num_tokens": 191362487.0, + "step": 7394 + }, + { + "epoch": 0.8121019108280255, + "grad_norm": 2.152857780456543, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.702078104019165, + "num_tokens": 191389257.0, + "step": 7395 + }, + { + "epoch": 0.8122117285306392, + "grad_norm": 2.262017250061035, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7002817392349243, + "num_tokens": 191414774.0, + "step": 7396 + }, + { + "epoch": 0.8123215462332528, + "grad_norm": 2.3173532485961914, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7139983177185059, + "num_tokens": 191438055.0, + "step": 7397 + }, + { + "epoch": 0.8124313639358665, + "grad_norm": 2.2717275619506836, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7020561695098877, + "num_tokens": 191461414.0, + "step": 7398 + }, + { + "epoch": 0.8125411816384801, + "grad_norm": 2.3013832569122314, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.720725417137146, + "num_tokens": 191485426.0, + "step": 7399 + }, + { + "epoch": 0.8126509993410937, + "grad_norm": 2.12750244140625, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7049503326416016, + "num_tokens": 191513727.0, + "step": 7400 + }, + { + "epoch": 0.8127608170437074, + "grad_norm": 2.2240118980407715, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6872239112854004, + "num_tokens": 191541044.0, + "step": 7401 + }, + { + "epoch": 0.8128706347463212, + "grad_norm": 1.7519422769546509, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.733349084854126, + "num_tokens": 191575243.0, + "step": 7402 + }, + { + "epoch": 0.8129804524489348, + "grad_norm": 2.1728391647338867, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7147805690765381, + "num_tokens": 191599817.0, + "step": 7403 + }, + { + "epoch": 0.8130902701515484, + "grad_norm": 1.9482574462890625, + "learning_rate": 1e-06, + "loss": 1.0751, + "mean_token_accuracy": 0.6744199395179749, + "num_tokens": 191633627.0, + "step": 7404 + }, + { + "epoch": 0.8132000878541621, + "grad_norm": 2.1326351165771484, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6888912916183472, + "num_tokens": 191661966.0, + "step": 7405 + }, + { + "epoch": 0.8133099055567757, + "grad_norm": 2.097107410430908, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7015496492385864, + "num_tokens": 191689951.0, + "step": 7406 + }, + { + "epoch": 0.8134197232593894, + "grad_norm": 2.1786904335021973, + "learning_rate": 1e-06, + "loss": 1.0547, + "mean_token_accuracy": 0.6837586164474487, + "num_tokens": 191716482.0, + "step": 7407 + }, + { + "epoch": 0.813529540962003, + "grad_norm": 2.272063732147217, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7033811807632446, + "num_tokens": 191741044.0, + "step": 7408 + }, + { + "epoch": 0.8136393586646168, + "grad_norm": 2.063291549682617, + "learning_rate": 1e-06, + "loss": 1.109, + "mean_token_accuracy": 0.6731430292129517, + "num_tokens": 191772667.0, + "step": 7409 + }, + { + "epoch": 0.8137491763672304, + "grad_norm": 2.0019302368164062, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7121943235397339, + "num_tokens": 191801028.0, + "step": 7410 + }, + { + "epoch": 0.8138589940698441, + "grad_norm": 2.1933770179748535, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7066409587860107, + "num_tokens": 191825693.0, + "step": 7411 + }, + { + "epoch": 0.8139688117724577, + "grad_norm": 2.1698973178863525, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.717527449131012, + "num_tokens": 191850044.0, + "step": 7412 + }, + { + "epoch": 0.8140786294750714, + "grad_norm": 2.3208398818969727, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7081639170646667, + "num_tokens": 191873183.0, + "step": 7413 + }, + { + "epoch": 0.814188447177685, + "grad_norm": 2.0593903064727783, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7086114883422852, + "num_tokens": 191899944.0, + "step": 7414 + }, + { + "epoch": 0.8142982648802987, + "grad_norm": 2.100008010864258, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6783491373062134, + "num_tokens": 191927847.0, + "step": 7415 + }, + { + "epoch": 0.8144080825829124, + "grad_norm": 2.1697678565979004, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7044932842254639, + "num_tokens": 191953831.0, + "step": 7416 + }, + { + "epoch": 0.8145179002855261, + "grad_norm": 2.1649014949798584, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6991183757781982, + "num_tokens": 191979703.0, + "step": 7417 + }, + { + "epoch": 0.8146277179881397, + "grad_norm": 2.065793991088867, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7240082025527954, + "num_tokens": 192007014.0, + "step": 7418 + }, + { + "epoch": 0.8147375356907534, + "grad_norm": 2.1181910037994385, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6863025426864624, + "num_tokens": 192034396.0, + "step": 7419 + }, + { + "epoch": 0.814847353393367, + "grad_norm": 1.9886194467544556, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6913114786148071, + "num_tokens": 192065478.0, + "step": 7420 + }, + { + "epoch": 0.8149571710959806, + "grad_norm": 2.1880483627319336, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.6930251121520996, + "num_tokens": 192091541.0, + "step": 7421 + }, + { + "epoch": 0.8150669887985943, + "grad_norm": 2.382181406021118, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7190878987312317, + "num_tokens": 192113111.0, + "step": 7422 + }, + { + "epoch": 0.8151768065012079, + "grad_norm": 2.178985118865967, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6953598260879517, + "num_tokens": 192139874.0, + "step": 7423 + }, + { + "epoch": 0.8152866242038217, + "grad_norm": 1.9599227905273438, + "learning_rate": 1e-06, + "loss": 1.0566, + "mean_token_accuracy": 0.6860108375549316, + "num_tokens": 192173021.0, + "step": 7424 + }, + { + "epoch": 0.8153964419064353, + "grad_norm": 2.093330144882202, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7256476283073425, + "num_tokens": 192200042.0, + "step": 7425 + }, + { + "epoch": 0.815506259609049, + "grad_norm": 1.9148366451263428, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.7048858404159546, + "num_tokens": 192233652.0, + "step": 7426 + }, + { + "epoch": 0.8156160773116626, + "grad_norm": 2.209557294845581, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6945548057556152, + "num_tokens": 192259177.0, + "step": 7427 + }, + { + "epoch": 0.8157258950142763, + "grad_norm": 2.0235018730163574, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6904478669166565, + "num_tokens": 192289215.0, + "step": 7428 + }, + { + "epoch": 0.8158357127168899, + "grad_norm": 2.1864120960235596, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7284176349639893, + "num_tokens": 192311361.0, + "step": 7429 + }, + { + "epoch": 0.8159455304195036, + "grad_norm": 2.1790919303894043, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7191693782806396, + "num_tokens": 192336211.0, + "step": 7430 + }, + { + "epoch": 0.8160553481221173, + "grad_norm": 2.2406725883483887, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7161747217178345, + "num_tokens": 192359588.0, + "step": 7431 + }, + { + "epoch": 0.816165165824731, + "grad_norm": 2.172593832015991, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6940127015113831, + "num_tokens": 192387109.0, + "step": 7432 + }, + { + "epoch": 0.8162749835273446, + "grad_norm": 2.1717965602874756, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.7046480178833008, + "num_tokens": 192414021.0, + "step": 7433 + }, + { + "epoch": 0.8163848012299583, + "grad_norm": 2.008965015411377, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.694998562335968, + "num_tokens": 192444406.0, + "step": 7434 + }, + { + "epoch": 0.8164946189325719, + "grad_norm": 2.046761989593506, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.689051628112793, + "num_tokens": 192472462.0, + "step": 7435 + }, + { + "epoch": 0.8166044366351856, + "grad_norm": 2.0703322887420654, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.6991750001907349, + "num_tokens": 192501003.0, + "step": 7436 + }, + { + "epoch": 0.8167142543377992, + "grad_norm": 2.324773073196411, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7214888334274292, + "num_tokens": 192521744.0, + "step": 7437 + }, + { + "epoch": 0.816824072040413, + "grad_norm": 1.922958493232727, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7213175296783447, + "num_tokens": 192551757.0, + "step": 7438 + }, + { + "epoch": 0.8169338897430266, + "grad_norm": 1.917873740196228, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6928887367248535, + "num_tokens": 192585787.0, + "step": 7439 + }, + { + "epoch": 0.8170437074456403, + "grad_norm": 2.509507417678833, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7191996574401855, + "num_tokens": 192605935.0, + "step": 7440 + }, + { + "epoch": 0.8171535251482539, + "grad_norm": 2.258255958557129, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7131853103637695, + "num_tokens": 192629921.0, + "step": 7441 + }, + { + "epoch": 0.8172633428508675, + "grad_norm": 1.960950493812561, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7181938886642456, + "num_tokens": 192660071.0, + "step": 7442 + }, + { + "epoch": 0.8173731605534812, + "grad_norm": 2.3806533813476562, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.6792689561843872, + "num_tokens": 192686540.0, + "step": 7443 + }, + { + "epoch": 0.8174829782560948, + "grad_norm": 1.9998958110809326, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6886505484580994, + "num_tokens": 192718722.0, + "step": 7444 + }, + { + "epoch": 0.8175927959587086, + "grad_norm": 2.3754138946533203, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.70004802942276, + "num_tokens": 192743632.0, + "step": 7445 + }, + { + "epoch": 0.8177026136613222, + "grad_norm": 2.137861728668213, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.731353759765625, + "num_tokens": 192768444.0, + "step": 7446 + }, + { + "epoch": 0.8178124313639359, + "grad_norm": 2.180853843688965, + "learning_rate": 1e-06, + "loss": 1.064, + "mean_token_accuracy": 0.6863934397697449, + "num_tokens": 192795814.0, + "step": 7447 + }, + { + "epoch": 0.8179222490665495, + "grad_norm": 2.0659453868865967, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7091925144195557, + "num_tokens": 192821027.0, + "step": 7448 + }, + { + "epoch": 0.8180320667691632, + "grad_norm": 2.218287467956543, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6923494935035706, + "num_tokens": 192846075.0, + "step": 7449 + }, + { + "epoch": 0.8181418844717768, + "grad_norm": 1.819705843925476, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.718811571598053, + "num_tokens": 192875195.0, + "step": 7450 + }, + { + "epoch": 0.8182517021743905, + "grad_norm": 2.197192907333374, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7034986019134521, + "num_tokens": 192899409.0, + "step": 7451 + }, + { + "epoch": 0.8183615198770041, + "grad_norm": 2.099163293838501, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7092952132225037, + "num_tokens": 192924224.0, + "step": 7452 + }, + { + "epoch": 0.8184713375796179, + "grad_norm": 2.480865955352783, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7070130109786987, + "num_tokens": 192944988.0, + "step": 7453 + }, + { + "epoch": 0.8185811552822315, + "grad_norm": 2.296860694885254, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7315950393676758, + "num_tokens": 192967089.0, + "step": 7454 + }, + { + "epoch": 0.8186909729848452, + "grad_norm": 2.1412200927734375, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.704494297504425, + "num_tokens": 192993389.0, + "step": 7455 + }, + { + "epoch": 0.8188007906874588, + "grad_norm": 2.321823835372925, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7110005617141724, + "num_tokens": 193016217.0, + "step": 7456 + }, + { + "epoch": 0.8189106083900725, + "grad_norm": 2.124075174331665, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6970601081848145, + "num_tokens": 193043656.0, + "step": 7457 + }, + { + "epoch": 0.8190204260926861, + "grad_norm": 2.3145458698272705, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7112839221954346, + "num_tokens": 193066359.0, + "step": 7458 + }, + { + "epoch": 0.8191302437952998, + "grad_norm": 2.0602071285247803, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6905080676078796, + "num_tokens": 193096540.0, + "step": 7459 + }, + { + "epoch": 0.8192400614979135, + "grad_norm": 2.3053743839263916, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7026386260986328, + "num_tokens": 193119171.0, + "step": 7460 + }, + { + "epoch": 0.8193498792005272, + "grad_norm": 1.9751185178756714, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7011844515800476, + "num_tokens": 193149032.0, + "step": 7461 + }, + { + "epoch": 0.8194596969031408, + "grad_norm": 2.1026883125305176, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7098491787910461, + "num_tokens": 193174665.0, + "step": 7462 + }, + { + "epoch": 0.8195695146057544, + "grad_norm": 2.310786724090576, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7098600268363953, + "num_tokens": 193198821.0, + "step": 7463 + }, + { + "epoch": 0.8196793323083681, + "grad_norm": 2.328979730606079, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.752899706363678, + "num_tokens": 193219550.0, + "step": 7464 + }, + { + "epoch": 0.8197891500109817, + "grad_norm": 2.129154682159424, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7196763157844543, + "num_tokens": 193245055.0, + "step": 7465 + }, + { + "epoch": 0.8198989677135954, + "grad_norm": 1.9284929037094116, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7169356346130371, + "num_tokens": 193274167.0, + "step": 7466 + }, + { + "epoch": 0.8200087854162091, + "grad_norm": 2.2439157962799072, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7232789397239685, + "num_tokens": 193297555.0, + "step": 7467 + }, + { + "epoch": 0.8201186031188228, + "grad_norm": 2.1507649421691895, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7098748087882996, + "num_tokens": 193324070.0, + "step": 7468 + }, + { + "epoch": 0.8202284208214364, + "grad_norm": 2.4124770164489746, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.6993833780288696, + "num_tokens": 193347653.0, + "step": 7469 + }, + { + "epoch": 0.8203382385240501, + "grad_norm": 2.181288957595825, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7100570797920227, + "num_tokens": 193372170.0, + "step": 7470 + }, + { + "epoch": 0.8204480562266637, + "grad_norm": 2.1284680366516113, + "learning_rate": 1e-06, + "loss": 1.0928, + "mean_token_accuracy": 0.6707533001899719, + "num_tokens": 193404706.0, + "step": 7471 + }, + { + "epoch": 0.8205578739292774, + "grad_norm": 2.120743989944458, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7172589302062988, + "num_tokens": 193431115.0, + "step": 7472 + }, + { + "epoch": 0.820667691631891, + "grad_norm": 2.188765287399292, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6936613917350769, + "num_tokens": 193458507.0, + "step": 7473 + }, + { + "epoch": 0.8207775093345048, + "grad_norm": 2.0622098445892334, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.7013883590698242, + "num_tokens": 193488381.0, + "step": 7474 + }, + { + "epoch": 0.8208873270371184, + "grad_norm": 2.067492723464966, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6963555812835693, + "num_tokens": 193519142.0, + "step": 7475 + }, + { + "epoch": 0.8209971447397321, + "grad_norm": 2.261983871459961, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7166795134544373, + "num_tokens": 193543492.0, + "step": 7476 + }, + { + "epoch": 0.8211069624423457, + "grad_norm": 2.2411367893218994, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.699418306350708, + "num_tokens": 193567255.0, + "step": 7477 + }, + { + "epoch": 0.8212167801449594, + "grad_norm": 2.446274757385254, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.710669219493866, + "num_tokens": 193588674.0, + "step": 7478 + }, + { + "epoch": 0.821326597847573, + "grad_norm": 2.0722415447235107, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6960418224334717, + "num_tokens": 193617385.0, + "step": 7479 + }, + { + "epoch": 0.8214364155501866, + "grad_norm": 2.1379098892211914, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.732352077960968, + "num_tokens": 193641889.0, + "step": 7480 + }, + { + "epoch": 0.8215462332528004, + "grad_norm": 2.4403035640716553, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6908948421478271, + "num_tokens": 193663586.0, + "step": 7481 + }, + { + "epoch": 0.821656050955414, + "grad_norm": 2.3769071102142334, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7096226811408997, + "num_tokens": 193685218.0, + "step": 7482 + }, + { + "epoch": 0.8217658686580277, + "grad_norm": 2.223552942276001, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7187699675559998, + "num_tokens": 193708146.0, + "step": 7483 + }, + { + "epoch": 0.8218756863606413, + "grad_norm": 2.4636030197143555, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7130419611930847, + "num_tokens": 193727761.0, + "step": 7484 + }, + { + "epoch": 0.821985504063255, + "grad_norm": 2.2644803524017334, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7004368305206299, + "num_tokens": 193751345.0, + "step": 7485 + }, + { + "epoch": 0.8220953217658686, + "grad_norm": 2.040851593017578, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7309087514877319, + "num_tokens": 193781026.0, + "step": 7486 + }, + { + "epoch": 0.8222051394684823, + "grad_norm": 2.169691562652588, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7148027420043945, + "num_tokens": 193807025.0, + "step": 7487 + }, + { + "epoch": 0.8223149571710959, + "grad_norm": 2.1119911670684814, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7336016297340393, + "num_tokens": 193831146.0, + "step": 7488 + }, + { + "epoch": 0.8224247748737097, + "grad_norm": 2.201185941696167, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7309931516647339, + "num_tokens": 193856325.0, + "step": 7489 + }, + { + "epoch": 0.8225345925763233, + "grad_norm": 2.1366963386535645, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.6989016532897949, + "num_tokens": 193881662.0, + "step": 7490 + }, + { + "epoch": 0.822644410278937, + "grad_norm": 2.097238063812256, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7023271918296814, + "num_tokens": 193909299.0, + "step": 7491 + }, + { + "epoch": 0.8227542279815506, + "grad_norm": 1.8665698766708374, + "learning_rate": 1e-06, + "loss": 1.1051, + "mean_token_accuracy": 0.6687865853309631, + "num_tokens": 193945491.0, + "step": 7492 + }, + { + "epoch": 0.8228640456841643, + "grad_norm": 1.9853960275650024, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6996041536331177, + "num_tokens": 193972892.0, + "step": 7493 + }, + { + "epoch": 0.8229738633867779, + "grad_norm": 2.4147748947143555, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.6989434957504272, + "num_tokens": 193994630.0, + "step": 7494 + }, + { + "epoch": 0.8230836810893916, + "grad_norm": 2.3423635959625244, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7256250381469727, + "num_tokens": 194016765.0, + "step": 7495 + }, + { + "epoch": 0.8231934987920053, + "grad_norm": 2.1210124492645264, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7144217491149902, + "num_tokens": 194041630.0, + "step": 7496 + }, + { + "epoch": 0.823303316494619, + "grad_norm": 2.6608948707580566, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7068925499916077, + "num_tokens": 194059924.0, + "step": 7497 + }, + { + "epoch": 0.8234131341972326, + "grad_norm": 2.1110081672668457, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7052909135818481, + "num_tokens": 194087610.0, + "step": 7498 + }, + { + "epoch": 0.8235229518998463, + "grad_norm": 2.1325886249542236, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7218737602233887, + "num_tokens": 194112974.0, + "step": 7499 + }, + { + "epoch": 0.8236327696024599, + "grad_norm": 1.8853694200515747, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7254271507263184, + "num_tokens": 194145084.0, + "step": 7500 + }, + { + "epoch": 0.8237425873050735, + "grad_norm": 2.039166212081909, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7094339728355408, + "num_tokens": 194170949.0, + "step": 7501 + }, + { + "epoch": 0.8238524050076872, + "grad_norm": 2.348360300064087, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.710997462272644, + "num_tokens": 194194584.0, + "step": 7502 + }, + { + "epoch": 0.823962222710301, + "grad_norm": 2.2040224075317383, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7024021148681641, + "num_tokens": 194219238.0, + "step": 7503 + }, + { + "epoch": 0.8240720404129146, + "grad_norm": 2.204709768295288, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.71767258644104, + "num_tokens": 194245490.0, + "step": 7504 + }, + { + "epoch": 0.8241818581155282, + "grad_norm": 1.9945894479751587, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6983612179756165, + "num_tokens": 194277489.0, + "step": 7505 + }, + { + "epoch": 0.8242916758181419, + "grad_norm": 1.999735713005066, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7278327941894531, + "num_tokens": 194304550.0, + "step": 7506 + }, + { + "epoch": 0.8244014935207555, + "grad_norm": 2.263308048248291, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7241398692131042, + "num_tokens": 194328241.0, + "step": 7507 + }, + { + "epoch": 0.8245113112233692, + "grad_norm": 2.257840633392334, + "learning_rate": 1e-06, + "loss": 1.0853, + "mean_token_accuracy": 0.6895194053649902, + "num_tokens": 194353809.0, + "step": 7508 + }, + { + "epoch": 0.8246211289259828, + "grad_norm": 2.4573965072631836, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7416139245033264, + "num_tokens": 194374441.0, + "step": 7509 + }, + { + "epoch": 0.8247309466285966, + "grad_norm": 2.108490228652954, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7180171608924866, + "num_tokens": 194401876.0, + "step": 7510 + }, + { + "epoch": 0.8248407643312102, + "grad_norm": 2.348365545272827, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7286481261253357, + "num_tokens": 194423064.0, + "step": 7511 + }, + { + "epoch": 0.8249505820338239, + "grad_norm": 2.1551644802093506, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7211835384368896, + "num_tokens": 194449636.0, + "step": 7512 + }, + { + "epoch": 0.8250603997364375, + "grad_norm": 2.1227002143859863, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7174011468887329, + "num_tokens": 194476248.0, + "step": 7513 + }, + { + "epoch": 0.8251702174390512, + "grad_norm": 2.068105697631836, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7010906338691711, + "num_tokens": 194504609.0, + "step": 7514 + }, + { + "epoch": 0.8252800351416648, + "grad_norm": 2.2495038509368896, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7107052803039551, + "num_tokens": 194526147.0, + "step": 7515 + }, + { + "epoch": 0.8253898528442785, + "grad_norm": 2.147050142288208, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.7027051448822021, + "num_tokens": 194552878.0, + "step": 7516 + }, + { + "epoch": 0.8254996705468921, + "grad_norm": 1.9986504316329956, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.7002747058868408, + "num_tokens": 194581847.0, + "step": 7517 + }, + { + "epoch": 0.8256094882495059, + "grad_norm": 1.9853459596633911, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7072553634643555, + "num_tokens": 194610977.0, + "step": 7518 + }, + { + "epoch": 0.8257193059521195, + "grad_norm": 2.2139012813568115, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.755836546421051, + "num_tokens": 194633139.0, + "step": 7519 + }, + { + "epoch": 0.8258291236547332, + "grad_norm": 2.202986478805542, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6961950063705444, + "num_tokens": 194658417.0, + "step": 7520 + }, + { + "epoch": 0.8259389413573468, + "grad_norm": 2.290950059890747, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7073700428009033, + "num_tokens": 194682723.0, + "step": 7521 + }, + { + "epoch": 0.8260487590599604, + "grad_norm": 2.2004053592681885, + "learning_rate": 1e-06, + "loss": 1.0821, + "mean_token_accuracy": 0.6744760274887085, + "num_tokens": 194709719.0, + "step": 7522 + }, + { + "epoch": 0.8261585767625741, + "grad_norm": 2.2489898204803467, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.723491907119751, + "num_tokens": 194735545.0, + "step": 7523 + }, + { + "epoch": 0.8262683944651877, + "grad_norm": 2.071909189224243, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7097560167312622, + "num_tokens": 194764463.0, + "step": 7524 + }, + { + "epoch": 0.8263782121678015, + "grad_norm": 2.2295830249786377, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7062077522277832, + "num_tokens": 194788477.0, + "step": 7525 + }, + { + "epoch": 0.8264880298704151, + "grad_norm": 2.0471136569976807, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.71427321434021, + "num_tokens": 194818075.0, + "step": 7526 + }, + { + "epoch": 0.8265978475730288, + "grad_norm": 2.117392063140869, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7114670872688293, + "num_tokens": 194844525.0, + "step": 7527 + }, + { + "epoch": 0.8267076652756424, + "grad_norm": 1.9247819185256958, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6928336024284363, + "num_tokens": 194875526.0, + "step": 7528 + }, + { + "epoch": 0.8268174829782561, + "grad_norm": 2.0299019813537598, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7072875499725342, + "num_tokens": 194902455.0, + "step": 7529 + }, + { + "epoch": 0.8269273006808697, + "grad_norm": 2.298330307006836, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7285367250442505, + "num_tokens": 194924955.0, + "step": 7530 + }, + { + "epoch": 0.8270371183834834, + "grad_norm": 1.904958724975586, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.71608567237854, + "num_tokens": 194956923.0, + "step": 7531 + }, + { + "epoch": 0.8271469360860971, + "grad_norm": 2.107295513153076, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7127828598022461, + "num_tokens": 194982918.0, + "step": 7532 + }, + { + "epoch": 0.8272567537887108, + "grad_norm": 2.2099173069000244, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6964403390884399, + "num_tokens": 195006390.0, + "step": 7533 + }, + { + "epoch": 0.8273665714913244, + "grad_norm": 2.203847646713257, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.721828043460846, + "num_tokens": 195030450.0, + "step": 7534 + }, + { + "epoch": 0.8274763891939381, + "grad_norm": 2.117738723754883, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6858358383178711, + "num_tokens": 195057157.0, + "step": 7535 + }, + { + "epoch": 0.8275862068965517, + "grad_norm": 2.125607967376709, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6885482668876648, + "num_tokens": 195086748.0, + "step": 7536 + }, + { + "epoch": 0.8276960245991654, + "grad_norm": 1.9623489379882812, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7052282094955444, + "num_tokens": 195116699.0, + "step": 7537 + }, + { + "epoch": 0.827805842301779, + "grad_norm": 2.043968439102173, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7096553444862366, + "num_tokens": 195144108.0, + "step": 7538 + }, + { + "epoch": 0.8279156600043928, + "grad_norm": 2.205227851867676, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7136139869689941, + "num_tokens": 195167984.0, + "step": 7539 + }, + { + "epoch": 0.8280254777070064, + "grad_norm": 2.613186836242676, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7427786588668823, + "num_tokens": 195186092.0, + "step": 7540 + }, + { + "epoch": 0.82813529540962, + "grad_norm": 2.0811984539031982, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7055683135986328, + "num_tokens": 195213490.0, + "step": 7541 + }, + { + "epoch": 0.8282451131122337, + "grad_norm": 2.4890294075012207, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7063024044036865, + "num_tokens": 195235215.0, + "step": 7542 + }, + { + "epoch": 0.8283549308148473, + "grad_norm": 1.8934041261672974, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6888831853866577, + "num_tokens": 195268719.0, + "step": 7543 + }, + { + "epoch": 0.828464748517461, + "grad_norm": 2.0147557258605957, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6855263710021973, + "num_tokens": 195302496.0, + "step": 7544 + }, + { + "epoch": 0.8285745662200746, + "grad_norm": 2.215477705001831, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7056180238723755, + "num_tokens": 195328694.0, + "step": 7545 + }, + { + "epoch": 0.8286843839226883, + "grad_norm": 2.211869478225708, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7182926535606384, + "num_tokens": 195352880.0, + "step": 7546 + }, + { + "epoch": 0.828794201625302, + "grad_norm": 1.9037648439407349, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7348895072937012, + "num_tokens": 195381758.0, + "step": 7547 + }, + { + "epoch": 0.8289040193279157, + "grad_norm": 2.2325668334960938, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6940088272094727, + "num_tokens": 195406281.0, + "step": 7548 + }, + { + "epoch": 0.8290138370305293, + "grad_norm": 2.350738048553467, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6942823529243469, + "num_tokens": 195431651.0, + "step": 7549 + }, + { + "epoch": 0.829123654733143, + "grad_norm": 2.0403265953063965, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7121124863624573, + "num_tokens": 195460148.0, + "step": 7550 + }, + { + "epoch": 0.8292334724357566, + "grad_norm": 2.221177339553833, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7454944849014282, + "num_tokens": 195484865.0, + "step": 7551 + }, + { + "epoch": 0.8293432901383703, + "grad_norm": 2.206807851791382, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7125567197799683, + "num_tokens": 195510871.0, + "step": 7552 + }, + { + "epoch": 0.8294531078409839, + "grad_norm": 2.0510480403900146, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6974741220474243, + "num_tokens": 195540329.0, + "step": 7553 + }, + { + "epoch": 0.8295629255435977, + "grad_norm": 2.227745532989502, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7307108640670776, + "num_tokens": 195563569.0, + "step": 7554 + }, + { + "epoch": 0.8296727432462113, + "grad_norm": 2.515698194503784, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7115327715873718, + "num_tokens": 195584615.0, + "step": 7555 + }, + { + "epoch": 0.829782560948825, + "grad_norm": 2.0576460361480713, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6869940757751465, + "num_tokens": 195612512.0, + "step": 7556 + }, + { + "epoch": 0.8298923786514386, + "grad_norm": 2.4657154083251953, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7180243730545044, + "num_tokens": 195632874.0, + "step": 7557 + }, + { + "epoch": 0.8300021963540523, + "grad_norm": 2.4173221588134766, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6963536143302917, + "num_tokens": 195655467.0, + "step": 7558 + }, + { + "epoch": 0.8301120140566659, + "grad_norm": 1.9397144317626953, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7089053392410278, + "num_tokens": 195687886.0, + "step": 7559 + }, + { + "epoch": 0.8302218317592795, + "grad_norm": 2.2800135612487793, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7054775953292847, + "num_tokens": 195711794.0, + "step": 7560 + }, + { + "epoch": 0.8303316494618933, + "grad_norm": 2.0344390869140625, + "learning_rate": 1e-06, + "loss": 1.0672, + "mean_token_accuracy": 0.6790509223937988, + "num_tokens": 195742931.0, + "step": 7561 + }, + { + "epoch": 0.830441467164507, + "grad_norm": 2.556152820587158, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7165466547012329, + "num_tokens": 195762619.0, + "step": 7562 + }, + { + "epoch": 0.8305512848671206, + "grad_norm": 2.2730183601379395, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7303769588470459, + "num_tokens": 195784610.0, + "step": 7563 + }, + { + "epoch": 0.8306611025697342, + "grad_norm": 2.137357711791992, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7057675123214722, + "num_tokens": 195810897.0, + "step": 7564 + }, + { + "epoch": 0.8307709202723479, + "grad_norm": 1.8329155445098877, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6866580247879028, + "num_tokens": 195846421.0, + "step": 7565 + }, + { + "epoch": 0.8308807379749615, + "grad_norm": 2.3180079460144043, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7015379667282104, + "num_tokens": 195868398.0, + "step": 7566 + }, + { + "epoch": 0.8309905556775752, + "grad_norm": 2.3989198207855225, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.6994651556015015, + "num_tokens": 195889234.0, + "step": 7567 + }, + { + "epoch": 0.8311003733801889, + "grad_norm": 2.528521776199341, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.705621600151062, + "num_tokens": 195908887.0, + "step": 7568 + }, + { + "epoch": 0.8312101910828026, + "grad_norm": 2.413151979446411, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6909666657447815, + "num_tokens": 195932480.0, + "step": 7569 + }, + { + "epoch": 0.8313200087854162, + "grad_norm": 2.1508753299713135, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7038147449493408, + "num_tokens": 195959682.0, + "step": 7570 + }, + { + "epoch": 0.8314298264880299, + "grad_norm": 1.9561519622802734, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6881933212280273, + "num_tokens": 195992913.0, + "step": 7571 + }, + { + "epoch": 0.8315396441906435, + "grad_norm": 2.3899967670440674, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7106549739837646, + "num_tokens": 196013906.0, + "step": 7572 + }, + { + "epoch": 0.8316494618932572, + "grad_norm": 2.351654529571533, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6867039203643799, + "num_tokens": 196039076.0, + "step": 7573 + }, + { + "epoch": 0.8317592795958708, + "grad_norm": 1.955642580986023, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7035394906997681, + "num_tokens": 196070521.0, + "step": 7574 + }, + { + "epoch": 0.8318690972984845, + "grad_norm": 2.1128926277160645, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6784495115280151, + "num_tokens": 196097730.0, + "step": 7575 + }, + { + "epoch": 0.8319789150010982, + "grad_norm": 2.3523263931274414, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7393546104431152, + "num_tokens": 196117974.0, + "step": 7576 + }, + { + "epoch": 0.8320887327037119, + "grad_norm": 2.123711585998535, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7059170603752136, + "num_tokens": 196146404.0, + "step": 7577 + }, + { + "epoch": 0.8321985504063255, + "grad_norm": 2.1728219985961914, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7157413959503174, + "num_tokens": 196171219.0, + "step": 7578 + }, + { + "epoch": 0.8323083681089392, + "grad_norm": 2.3786966800689697, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6836066246032715, + "num_tokens": 196194458.0, + "step": 7579 + }, + { + "epoch": 0.8324181858115528, + "grad_norm": 2.156256675720215, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7261532545089722, + "num_tokens": 196219471.0, + "step": 7580 + }, + { + "epoch": 0.8325280035141664, + "grad_norm": 2.205883502960205, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6948368549346924, + "num_tokens": 196243947.0, + "step": 7581 + }, + { + "epoch": 0.8326378212167801, + "grad_norm": 2.1145687103271484, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7028055191040039, + "num_tokens": 196270203.0, + "step": 7582 + }, + { + "epoch": 0.8327476389193939, + "grad_norm": 2.1586198806762695, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7116609215736389, + "num_tokens": 196297066.0, + "step": 7583 + }, + { + "epoch": 0.8328574566220075, + "grad_norm": 2.3711013793945312, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7153217792510986, + "num_tokens": 196318379.0, + "step": 7584 + }, + { + "epoch": 0.8329672743246211, + "grad_norm": 2.453951835632324, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7194710373878479, + "num_tokens": 196338137.0, + "step": 7585 + }, + { + "epoch": 0.8330770920272348, + "grad_norm": 2.040170669555664, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6971920728683472, + "num_tokens": 196367424.0, + "step": 7586 + }, + { + "epoch": 0.8331869097298484, + "grad_norm": 2.1440036296844482, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7181785106658936, + "num_tokens": 196391314.0, + "step": 7587 + }, + { + "epoch": 0.8332967274324621, + "grad_norm": 2.1608777046203613, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7007224559783936, + "num_tokens": 196416374.0, + "step": 7588 + }, + { + "epoch": 0.8334065451350757, + "grad_norm": 2.131544828414917, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7297612428665161, + "num_tokens": 196441639.0, + "step": 7589 + }, + { + "epoch": 0.8335163628376895, + "grad_norm": 1.903573989868164, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7166780829429626, + "num_tokens": 196471450.0, + "step": 7590 + }, + { + "epoch": 0.8336261805403031, + "grad_norm": 1.9745900630950928, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.690269947052002, + "num_tokens": 196500994.0, + "step": 7591 + }, + { + "epoch": 0.8337359982429168, + "grad_norm": 2.0789637565612793, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7123460173606873, + "num_tokens": 196526947.0, + "step": 7592 + }, + { + "epoch": 0.8338458159455304, + "grad_norm": 2.3062570095062256, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7076829075813293, + "num_tokens": 196550249.0, + "step": 7593 + }, + { + "epoch": 0.8339556336481441, + "grad_norm": 2.2506608963012695, + "learning_rate": 1e-06, + "loss": 1.0684, + "mean_token_accuracy": 0.6863083839416504, + "num_tokens": 196577281.0, + "step": 7594 + }, + { + "epoch": 0.8340654513507577, + "grad_norm": 2.2033467292785645, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7077872157096863, + "num_tokens": 196602095.0, + "step": 7595 + }, + { + "epoch": 0.8341752690533714, + "grad_norm": 2.0009050369262695, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6954444646835327, + "num_tokens": 196629629.0, + "step": 7596 + }, + { + "epoch": 0.8342850867559851, + "grad_norm": 2.456157922744751, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7085498571395874, + "num_tokens": 196652009.0, + "step": 7597 + }, + { + "epoch": 0.8343949044585988, + "grad_norm": 2.157421350479126, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7255959510803223, + "num_tokens": 196677268.0, + "step": 7598 + }, + { + "epoch": 0.8345047221612124, + "grad_norm": 2.1837563514709473, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.69706130027771, + "num_tokens": 196703110.0, + "step": 7599 + }, + { + "epoch": 0.834614539863826, + "grad_norm": 1.8563331365585327, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7030068635940552, + "num_tokens": 196735845.0, + "step": 7600 + }, + { + "epoch": 0.8347243575664397, + "grad_norm": 2.0668375492095947, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7028594613075256, + "num_tokens": 196764382.0, + "step": 7601 + }, + { + "epoch": 0.8348341752690533, + "grad_norm": 2.3080992698669434, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7231242656707764, + "num_tokens": 196785955.0, + "step": 7602 + }, + { + "epoch": 0.834943992971667, + "grad_norm": 2.045029401779175, + "learning_rate": 1e-06, + "loss": 1.0848, + "mean_token_accuracy": 0.6763736009597778, + "num_tokens": 196815121.0, + "step": 7603 + }, + { + "epoch": 0.8350538106742806, + "grad_norm": 1.9056224822998047, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7132308483123779, + "num_tokens": 196844968.0, + "step": 7604 + }, + { + "epoch": 0.8351636283768944, + "grad_norm": 2.1454343795776367, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7040137052536011, + "num_tokens": 196872347.0, + "step": 7605 + }, + { + "epoch": 0.835273446079508, + "grad_norm": 2.290372133255005, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6975159645080566, + "num_tokens": 196896899.0, + "step": 7606 + }, + { + "epoch": 0.8353832637821217, + "grad_norm": 2.069606065750122, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7047576904296875, + "num_tokens": 196925898.0, + "step": 7607 + }, + { + "epoch": 0.8354930814847353, + "grad_norm": 2.2348885536193848, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7009451389312744, + "num_tokens": 196949512.0, + "step": 7608 + }, + { + "epoch": 0.835602899187349, + "grad_norm": 2.516672134399414, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7234595417976379, + "num_tokens": 196969131.0, + "step": 7609 + }, + { + "epoch": 0.8357127168899626, + "grad_norm": 2.1130447387695312, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6947942972183228, + "num_tokens": 196996637.0, + "step": 7610 + }, + { + "epoch": 0.8358225345925763, + "grad_norm": 2.4254400730133057, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7097615003585815, + "num_tokens": 197017944.0, + "step": 7611 + }, + { + "epoch": 0.83593235229519, + "grad_norm": 2.382863759994507, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.717889666557312, + "num_tokens": 197039417.0, + "step": 7612 + }, + { + "epoch": 0.8360421699978037, + "grad_norm": 2.1443586349487305, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7105097770690918, + "num_tokens": 197063264.0, + "step": 7613 + }, + { + "epoch": 0.8361519877004173, + "grad_norm": 2.3501665592193604, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7311036586761475, + "num_tokens": 197085752.0, + "step": 7614 + }, + { + "epoch": 0.836261805403031, + "grad_norm": 2.0941858291625977, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6922620534896851, + "num_tokens": 197114138.0, + "step": 7615 + }, + { + "epoch": 0.8363716231056446, + "grad_norm": 2.384263753890991, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6827576160430908, + "num_tokens": 197136693.0, + "step": 7616 + }, + { + "epoch": 0.8364814408082583, + "grad_norm": 2.124363899230957, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7239765524864197, + "num_tokens": 197161869.0, + "step": 7617 + }, + { + "epoch": 0.8365912585108719, + "grad_norm": 2.042060613632202, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7150431871414185, + "num_tokens": 197192424.0, + "step": 7618 + }, + { + "epoch": 0.8367010762134857, + "grad_norm": 1.9193192720413208, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7109682559967041, + "num_tokens": 197221800.0, + "step": 7619 + }, + { + "epoch": 0.8368108939160993, + "grad_norm": 2.08968186378479, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7088167667388916, + "num_tokens": 197249437.0, + "step": 7620 + }, + { + "epoch": 0.836920711618713, + "grad_norm": 2.339463233947754, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7165237069129944, + "num_tokens": 197271744.0, + "step": 7621 + }, + { + "epoch": 0.8370305293213266, + "grad_norm": 2.2449283599853516, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7122330069541931, + "num_tokens": 197295207.0, + "step": 7622 + }, + { + "epoch": 0.8371403470239402, + "grad_norm": 2.2892346382141113, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7088735103607178, + "num_tokens": 197319632.0, + "step": 7623 + }, + { + "epoch": 0.8372501647265539, + "grad_norm": 2.097484827041626, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6849455237388611, + "num_tokens": 197350363.0, + "step": 7624 + }, + { + "epoch": 0.8373599824291675, + "grad_norm": 2.1862947940826416, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7244304418563843, + "num_tokens": 197375405.0, + "step": 7625 + }, + { + "epoch": 0.8374698001317813, + "grad_norm": 2.1479477882385254, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.6948971152305603, + "num_tokens": 197401800.0, + "step": 7626 + }, + { + "epoch": 0.8375796178343949, + "grad_norm": 2.2467217445373535, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6929030418395996, + "num_tokens": 197425361.0, + "step": 7627 + }, + { + "epoch": 0.8376894355370086, + "grad_norm": 2.1000618934631348, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.711097002029419, + "num_tokens": 197451855.0, + "step": 7628 + }, + { + "epoch": 0.8377992532396222, + "grad_norm": 2.340715169906616, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6917078495025635, + "num_tokens": 197475433.0, + "step": 7629 + }, + { + "epoch": 0.8379090709422359, + "grad_norm": 2.5861611366271973, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7175332903862, + "num_tokens": 197495708.0, + "step": 7630 + }, + { + "epoch": 0.8380188886448495, + "grad_norm": 1.896994709968567, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.697121262550354, + "num_tokens": 197529731.0, + "step": 7631 + }, + { + "epoch": 0.8381287063474632, + "grad_norm": 2.249394655227661, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6891764402389526, + "num_tokens": 197555343.0, + "step": 7632 + }, + { + "epoch": 0.8382385240500769, + "grad_norm": 2.0040829181671143, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7140089273452759, + "num_tokens": 197585097.0, + "step": 7633 + }, + { + "epoch": 0.8383483417526906, + "grad_norm": 2.0669615268707275, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7414405345916748, + "num_tokens": 197610696.0, + "step": 7634 + }, + { + "epoch": 0.8384581594553042, + "grad_norm": 2.0264768600463867, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7023625373840332, + "num_tokens": 197638076.0, + "step": 7635 + }, + { + "epoch": 0.8385679771579179, + "grad_norm": 2.5090739727020264, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7090271711349487, + "num_tokens": 197659044.0, + "step": 7636 + }, + { + "epoch": 0.8386777948605315, + "grad_norm": 2.1252057552337646, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7073062658309937, + "num_tokens": 197684359.0, + "step": 7637 + }, + { + "epoch": 0.8387876125631452, + "grad_norm": 2.3863418102264404, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.711704671382904, + "num_tokens": 197706907.0, + "step": 7638 + }, + { + "epoch": 0.8388974302657588, + "grad_norm": 2.1091551780700684, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7086554169654846, + "num_tokens": 197734448.0, + "step": 7639 + }, + { + "epoch": 0.8390072479683724, + "grad_norm": 2.0882363319396973, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.707213282585144, + "num_tokens": 197760172.0, + "step": 7640 + }, + { + "epoch": 0.8391170656709862, + "grad_norm": 2.215266227722168, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7051388621330261, + "num_tokens": 197786884.0, + "step": 7641 + }, + { + "epoch": 0.8392268833735999, + "grad_norm": 2.3044979572296143, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7121518850326538, + "num_tokens": 197809095.0, + "step": 7642 + }, + { + "epoch": 0.8393367010762135, + "grad_norm": 2.1238033771514893, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7122111320495605, + "num_tokens": 197835389.0, + "step": 7643 + }, + { + "epoch": 0.8394465187788271, + "grad_norm": 2.439171552658081, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7326357960700989, + "num_tokens": 197857238.0, + "step": 7644 + }, + { + "epoch": 0.8395563364814408, + "grad_norm": 1.848892331123352, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6954513192176819, + "num_tokens": 197891332.0, + "step": 7645 + }, + { + "epoch": 0.8396661541840544, + "grad_norm": 2.1804003715515137, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6920524835586548, + "num_tokens": 197917516.0, + "step": 7646 + }, + { + "epoch": 0.8397759718866681, + "grad_norm": 2.346445083618164, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7229450941085815, + "num_tokens": 197937808.0, + "step": 7647 + }, + { + "epoch": 0.8398857895892818, + "grad_norm": 1.9904711246490479, + "learning_rate": 1e-06, + "loss": 1.0809, + "mean_token_accuracy": 0.6728299856185913, + "num_tokens": 197970058.0, + "step": 7648 + }, + { + "epoch": 0.8399956072918955, + "grad_norm": 2.0560553073883057, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.693016767501831, + "num_tokens": 198000588.0, + "step": 7649 + }, + { + "epoch": 0.8401054249945091, + "grad_norm": 1.9566086530685425, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6939784288406372, + "num_tokens": 198030599.0, + "step": 7650 + }, + { + "epoch": 0.8402152426971228, + "grad_norm": 2.1653897762298584, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7097156643867493, + "num_tokens": 198055412.0, + "step": 7651 + }, + { + "epoch": 0.8403250603997364, + "grad_norm": 2.1816160678863525, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7234323024749756, + "num_tokens": 198079265.0, + "step": 7652 + }, + { + "epoch": 0.8404348781023501, + "grad_norm": 2.566004753112793, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7132025957107544, + "num_tokens": 198099381.0, + "step": 7653 + }, + { + "epoch": 0.8405446958049637, + "grad_norm": 2.2613818645477295, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7017562389373779, + "num_tokens": 198122592.0, + "step": 7654 + }, + { + "epoch": 0.8406545135075775, + "grad_norm": 2.382546901702881, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7104785442352295, + "num_tokens": 198146233.0, + "step": 7655 + }, + { + "epoch": 0.8407643312101911, + "grad_norm": 2.3117589950561523, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.710149347782135, + "num_tokens": 198170128.0, + "step": 7656 + }, + { + "epoch": 0.8408741489128048, + "grad_norm": 2.2001655101776123, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7269597053527832, + "num_tokens": 198195310.0, + "step": 7657 + }, + { + "epoch": 0.8409839666154184, + "grad_norm": 2.1306495666503906, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.6995788812637329, + "num_tokens": 198221303.0, + "step": 7658 + }, + { + "epoch": 0.8410937843180321, + "grad_norm": 2.1476857662200928, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6911466121673584, + "num_tokens": 198247562.0, + "step": 7659 + }, + { + "epoch": 0.8412036020206457, + "grad_norm": 2.0554568767547607, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6899930834770203, + "num_tokens": 198276906.0, + "step": 7660 + }, + { + "epoch": 0.8413134197232593, + "grad_norm": 2.623629331588745, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7140917778015137, + "num_tokens": 198295104.0, + "step": 7661 + }, + { + "epoch": 0.8414232374258731, + "grad_norm": 2.0575404167175293, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7022677659988403, + "num_tokens": 198323331.0, + "step": 7662 + }, + { + "epoch": 0.8415330551284868, + "grad_norm": 2.1736156940460205, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7059774994850159, + "num_tokens": 198350668.0, + "step": 7663 + }, + { + "epoch": 0.8416428728311004, + "grad_norm": 1.9106284379959106, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7152101993560791, + "num_tokens": 198381292.0, + "step": 7664 + }, + { + "epoch": 0.841752690533714, + "grad_norm": 2.2476963996887207, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7117494344711304, + "num_tokens": 198406274.0, + "step": 7665 + }, + { + "epoch": 0.8418625082363277, + "grad_norm": 2.2548797130584717, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6990898251533508, + "num_tokens": 198430130.0, + "step": 7666 + }, + { + "epoch": 0.8419723259389413, + "grad_norm": 1.950021743774414, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7265514731407166, + "num_tokens": 198459777.0, + "step": 7667 + }, + { + "epoch": 0.842082143641555, + "grad_norm": 2.184495449066162, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7012447118759155, + "num_tokens": 198485192.0, + "step": 7668 + }, + { + "epoch": 0.8421919613441686, + "grad_norm": 2.497084617614746, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7296745777130127, + "num_tokens": 198503231.0, + "step": 7669 + }, + { + "epoch": 0.8423017790467824, + "grad_norm": 2.259589672088623, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7207695841789246, + "num_tokens": 198528863.0, + "step": 7670 + }, + { + "epoch": 0.842411596749396, + "grad_norm": 2.19228458404541, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7165935039520264, + "num_tokens": 198553857.0, + "step": 7671 + }, + { + "epoch": 0.8425214144520097, + "grad_norm": 2.045576810836792, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.7005224227905273, + "num_tokens": 198583558.0, + "step": 7672 + }, + { + "epoch": 0.8426312321546233, + "grad_norm": 2.176802158355713, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7277654409408569, + "num_tokens": 198609291.0, + "step": 7673 + }, + { + "epoch": 0.842741049857237, + "grad_norm": 2.079319953918457, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7120922803878784, + "num_tokens": 198636900.0, + "step": 7674 + }, + { + "epoch": 0.8428508675598506, + "grad_norm": 1.9246392250061035, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6922265291213989, + "num_tokens": 198668433.0, + "step": 7675 + }, + { + "epoch": 0.8429606852624643, + "grad_norm": 2.2038893699645996, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7154784202575684, + "num_tokens": 198694562.0, + "step": 7676 + }, + { + "epoch": 0.843070502965078, + "grad_norm": 2.3803281784057617, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7067172527313232, + "num_tokens": 198717369.0, + "step": 7677 + }, + { + "epoch": 0.8431803206676917, + "grad_norm": 2.063774347305298, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7118128538131714, + "num_tokens": 198744856.0, + "step": 7678 + }, + { + "epoch": 0.8432901383703053, + "grad_norm": 2.159560441970825, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.691077709197998, + "num_tokens": 198771538.0, + "step": 7679 + }, + { + "epoch": 0.843399956072919, + "grad_norm": 2.249044418334961, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6933596134185791, + "num_tokens": 198796885.0, + "step": 7680 + }, + { + "epoch": 0.8435097737755326, + "grad_norm": 2.2894785404205322, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7168916463851929, + "num_tokens": 198818940.0, + "step": 7681 + }, + { + "epoch": 0.8436195914781462, + "grad_norm": 2.1749093532562256, + "learning_rate": 1e-06, + "loss": 1.047, + "mean_token_accuracy": 0.6776913404464722, + "num_tokens": 198845612.0, + "step": 7682 + }, + { + "epoch": 0.8437294091807599, + "grad_norm": 2.0794713497161865, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7096002101898193, + "num_tokens": 198874429.0, + "step": 7683 + }, + { + "epoch": 0.8438392268833736, + "grad_norm": 2.3347270488739014, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7126610279083252, + "num_tokens": 198899257.0, + "step": 7684 + }, + { + "epoch": 0.8439490445859873, + "grad_norm": 2.251094102859497, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7132594585418701, + "num_tokens": 198923303.0, + "step": 7685 + }, + { + "epoch": 0.8440588622886009, + "grad_norm": 2.175013780593872, + "learning_rate": 1e-06, + "loss": 1.1112, + "mean_token_accuracy": 0.6616224050521851, + "num_tokens": 198953795.0, + "step": 7686 + }, + { + "epoch": 0.8441686799912146, + "grad_norm": 2.1338086128234863, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7030047178268433, + "num_tokens": 198981725.0, + "step": 7687 + }, + { + "epoch": 0.8442784976938282, + "grad_norm": 2.222036361694336, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7123705148696899, + "num_tokens": 199005153.0, + "step": 7688 + }, + { + "epoch": 0.8443883153964419, + "grad_norm": 2.3697149753570557, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7283526659011841, + "num_tokens": 199027042.0, + "step": 7689 + }, + { + "epoch": 0.8444981330990555, + "grad_norm": 2.408141851425171, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7126454710960388, + "num_tokens": 199050146.0, + "step": 7690 + }, + { + "epoch": 0.8446079508016693, + "grad_norm": 2.44558048248291, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6980798840522766, + "num_tokens": 199073404.0, + "step": 7691 + }, + { + "epoch": 0.8447177685042829, + "grad_norm": 2.1437695026397705, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.700391411781311, + "num_tokens": 199101008.0, + "step": 7692 + }, + { + "epoch": 0.8448275862068966, + "grad_norm": 2.2491042613983154, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6943404674530029, + "num_tokens": 199126165.0, + "step": 7693 + }, + { + "epoch": 0.8449374039095102, + "grad_norm": 2.063490867614746, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7072274088859558, + "num_tokens": 199153533.0, + "step": 7694 + }, + { + "epoch": 0.8450472216121239, + "grad_norm": 2.033290147781372, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7391141653060913, + "num_tokens": 199179424.0, + "step": 7695 + }, + { + "epoch": 0.8451570393147375, + "grad_norm": 2.4409377574920654, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7014370560646057, + "num_tokens": 199202764.0, + "step": 7696 + }, + { + "epoch": 0.8452668570173512, + "grad_norm": 2.263315439224243, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7179455757141113, + "num_tokens": 199227940.0, + "step": 7697 + }, + { + "epoch": 0.8453766747199648, + "grad_norm": 2.1845932006835938, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.711653470993042, + "num_tokens": 199252255.0, + "step": 7698 + }, + { + "epoch": 0.8454864924225786, + "grad_norm": 2.041118621826172, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7143620848655701, + "num_tokens": 199278817.0, + "step": 7699 + }, + { + "epoch": 0.8455963101251922, + "grad_norm": 2.175135612487793, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7105715870857239, + "num_tokens": 199304937.0, + "step": 7700 + }, + { + "epoch": 0.8457061278278059, + "grad_norm": 1.8190797567367554, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7314069271087646, + "num_tokens": 199337317.0, + "step": 7701 + }, + { + "epoch": 0.8458159455304195, + "grad_norm": 2.186674118041992, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7079153656959534, + "num_tokens": 199363824.0, + "step": 7702 + }, + { + "epoch": 0.8459257632330331, + "grad_norm": 2.2160537242889404, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7073224186897278, + "num_tokens": 199388320.0, + "step": 7703 + }, + { + "epoch": 0.8460355809356468, + "grad_norm": 2.1758313179016113, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.6992165446281433, + "num_tokens": 199415523.0, + "step": 7704 + }, + { + "epoch": 0.8461453986382604, + "grad_norm": 2.267319679260254, + "learning_rate": 1e-06, + "loss": 1.0802, + "mean_token_accuracy": 0.6903390288352966, + "num_tokens": 199441645.0, + "step": 7705 + }, + { + "epoch": 0.8462552163408742, + "grad_norm": 2.5331575870513916, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7189741134643555, + "num_tokens": 199462863.0, + "step": 7706 + }, + { + "epoch": 0.8463650340434878, + "grad_norm": 2.394944667816162, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7306092977523804, + "num_tokens": 199484387.0, + "step": 7707 + }, + { + "epoch": 0.8464748517461015, + "grad_norm": 2.1275441646575928, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7055933475494385, + "num_tokens": 199513679.0, + "step": 7708 + }, + { + "epoch": 0.8465846694487151, + "grad_norm": 2.1527304649353027, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6879959106445312, + "num_tokens": 199542808.0, + "step": 7709 + }, + { + "epoch": 0.8466944871513288, + "grad_norm": 2.3360085487365723, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.693238377571106, + "num_tokens": 199566968.0, + "step": 7710 + }, + { + "epoch": 0.8468043048539424, + "grad_norm": 2.1608145236968994, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7185324430465698, + "num_tokens": 199593692.0, + "step": 7711 + }, + { + "epoch": 0.8469141225565561, + "grad_norm": 2.233456611633301, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7046751976013184, + "num_tokens": 199622402.0, + "step": 7712 + }, + { + "epoch": 0.8470239402591698, + "grad_norm": 2.6846840381622314, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7194774150848389, + "num_tokens": 199641522.0, + "step": 7713 + }, + { + "epoch": 0.8471337579617835, + "grad_norm": 2.1342217922210693, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7108792066574097, + "num_tokens": 199667927.0, + "step": 7714 + }, + { + "epoch": 0.8472435756643971, + "grad_norm": 2.3640387058258057, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.689639687538147, + "num_tokens": 199693552.0, + "step": 7715 + }, + { + "epoch": 0.8473533933670108, + "grad_norm": 2.275334358215332, + "learning_rate": 1e-06, + "loss": 1.0535, + "mean_token_accuracy": 0.6855926513671875, + "num_tokens": 199716886.0, + "step": 7716 + }, + { + "epoch": 0.8474632110696244, + "grad_norm": 2.4273195266723633, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.712237536907196, + "num_tokens": 199737928.0, + "step": 7717 + }, + { + "epoch": 0.8475730287722381, + "grad_norm": 2.007161855697632, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.711965799331665, + "num_tokens": 199766914.0, + "step": 7718 + }, + { + "epoch": 0.8476828464748517, + "grad_norm": 1.9435882568359375, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7090425491333008, + "num_tokens": 199797223.0, + "step": 7719 + }, + { + "epoch": 0.8477926641774655, + "grad_norm": 2.2649612426757812, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.737492561340332, + "num_tokens": 199819060.0, + "step": 7720 + }, + { + "epoch": 0.8479024818800791, + "grad_norm": 2.2163095474243164, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7126399278640747, + "num_tokens": 199845241.0, + "step": 7721 + }, + { + "epoch": 0.8480122995826928, + "grad_norm": 2.173534393310547, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7255052328109741, + "num_tokens": 199868686.0, + "step": 7722 + }, + { + "epoch": 0.8481221172853064, + "grad_norm": 2.5179502964019775, + "learning_rate": 1e-06, + "loss": 1.0664, + "mean_token_accuracy": 0.6825001239776611, + "num_tokens": 199892312.0, + "step": 7723 + }, + { + "epoch": 0.84823193498792, + "grad_norm": 2.2644243240356445, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7099002599716187, + "num_tokens": 199917331.0, + "step": 7724 + }, + { + "epoch": 0.8483417526905337, + "grad_norm": 2.214574098587036, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7083148956298828, + "num_tokens": 199942069.0, + "step": 7725 + }, + { + "epoch": 0.8484515703931473, + "grad_norm": 2.178997755050659, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6932294368743896, + "num_tokens": 199967921.0, + "step": 7726 + }, + { + "epoch": 0.848561388095761, + "grad_norm": 2.1302499771118164, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7193416357040405, + "num_tokens": 199993258.0, + "step": 7727 + }, + { + "epoch": 0.8486712057983747, + "grad_norm": 1.9597344398498535, + "learning_rate": 1e-06, + "loss": 1.0602, + "mean_token_accuracy": 0.6758697032928467, + "num_tokens": 200025395.0, + "step": 7728 + }, + { + "epoch": 0.8487810235009884, + "grad_norm": 2.575895071029663, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6938669681549072, + "num_tokens": 200047088.0, + "step": 7729 + }, + { + "epoch": 0.848890841203602, + "grad_norm": 2.0921266078948975, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7179899215698242, + "num_tokens": 200073044.0, + "step": 7730 + }, + { + "epoch": 0.8490006589062157, + "grad_norm": 1.9844584465026855, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7047573924064636, + "num_tokens": 200103693.0, + "step": 7731 + }, + { + "epoch": 0.8491104766088293, + "grad_norm": 2.090233325958252, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7187089920043945, + "num_tokens": 200129605.0, + "step": 7732 + }, + { + "epoch": 0.849220294311443, + "grad_norm": 2.058297872543335, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.7018899917602539, + "num_tokens": 200159419.0, + "step": 7733 + }, + { + "epoch": 0.8493301120140566, + "grad_norm": 2.266052484512329, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6866150498390198, + "num_tokens": 200183858.0, + "step": 7734 + }, + { + "epoch": 0.8494399297166704, + "grad_norm": 2.0405890941619873, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6860762238502502, + "num_tokens": 200215080.0, + "step": 7735 + }, + { + "epoch": 0.849549747419284, + "grad_norm": 2.4096710681915283, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6922121047973633, + "num_tokens": 200236944.0, + "step": 7736 + }, + { + "epoch": 0.8496595651218977, + "grad_norm": 2.12155818939209, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7040351629257202, + "num_tokens": 200265185.0, + "step": 7737 + }, + { + "epoch": 0.8497693828245113, + "grad_norm": 2.0049076080322266, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7059138417243958, + "num_tokens": 200293716.0, + "step": 7738 + }, + { + "epoch": 0.849879200527125, + "grad_norm": 2.1834864616394043, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7030360102653503, + "num_tokens": 200318403.0, + "step": 7739 + }, + { + "epoch": 0.8499890182297386, + "grad_norm": 2.0472095012664795, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7343511581420898, + "num_tokens": 200343584.0, + "step": 7740 + }, + { + "epoch": 0.8500988359323522, + "grad_norm": 1.8904802799224854, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7075070142745972, + "num_tokens": 200375251.0, + "step": 7741 + }, + { + "epoch": 0.850208653634966, + "grad_norm": 2.053926706314087, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6938807964324951, + "num_tokens": 200406054.0, + "step": 7742 + }, + { + "epoch": 0.8503184713375797, + "grad_norm": 2.420488119125366, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.7081727981567383, + "num_tokens": 200427875.0, + "step": 7743 + }, + { + "epoch": 0.8504282890401933, + "grad_norm": 2.0239102840423584, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7266629338264465, + "num_tokens": 200454987.0, + "step": 7744 + }, + { + "epoch": 0.8505381067428069, + "grad_norm": 2.2030413150787354, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.731055498123169, + "num_tokens": 200478848.0, + "step": 7745 + }, + { + "epoch": 0.8506479244454206, + "grad_norm": 2.195291519165039, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7174317836761475, + "num_tokens": 200502938.0, + "step": 7746 + }, + { + "epoch": 0.8507577421480342, + "grad_norm": 2.197269916534424, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7316566705703735, + "num_tokens": 200527962.0, + "step": 7747 + }, + { + "epoch": 0.8508675598506479, + "grad_norm": 2.2114319801330566, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7061089277267456, + "num_tokens": 200554384.0, + "step": 7748 + }, + { + "epoch": 0.8509773775532616, + "grad_norm": 2.166443347930908, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.6969047784805298, + "num_tokens": 200580256.0, + "step": 7749 + }, + { + "epoch": 0.8510871952558753, + "grad_norm": 2.419767379760742, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7033851146697998, + "num_tokens": 200602913.0, + "step": 7750 + }, + { + "epoch": 0.8511970129584889, + "grad_norm": 2.0021417140960693, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6971917748451233, + "num_tokens": 200637997.0, + "step": 7751 + }, + { + "epoch": 0.8513068306611026, + "grad_norm": 2.2542202472686768, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.719348132610321, + "num_tokens": 200663680.0, + "step": 7752 + }, + { + "epoch": 0.8514166483637162, + "grad_norm": 2.128286361694336, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7230292558670044, + "num_tokens": 200687698.0, + "step": 7753 + }, + { + "epoch": 0.8515264660663299, + "grad_norm": 2.0502474308013916, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6936888098716736, + "num_tokens": 200718240.0, + "step": 7754 + }, + { + "epoch": 0.8516362837689435, + "grad_norm": 2.224259853363037, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6856319904327393, + "num_tokens": 200743003.0, + "step": 7755 + }, + { + "epoch": 0.8517461014715572, + "grad_norm": 2.1112802028656006, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6847085952758789, + "num_tokens": 200772187.0, + "step": 7756 + }, + { + "epoch": 0.8518559191741709, + "grad_norm": 2.2186434268951416, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7248547077178955, + "num_tokens": 200800494.0, + "step": 7757 + }, + { + "epoch": 0.8519657368767846, + "grad_norm": 2.4277894496917725, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7219668626785278, + "num_tokens": 200820127.0, + "step": 7758 + }, + { + "epoch": 0.8520755545793982, + "grad_norm": 2.632737636566162, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7365485429763794, + "num_tokens": 200838496.0, + "step": 7759 + }, + { + "epoch": 0.8521853722820119, + "grad_norm": 2.23815655708313, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7057569026947021, + "num_tokens": 200863720.0, + "step": 7760 + }, + { + "epoch": 0.8522951899846255, + "grad_norm": 2.2292685508728027, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.709856390953064, + "num_tokens": 200890496.0, + "step": 7761 + }, + { + "epoch": 0.8524050076872391, + "grad_norm": 2.0420548915863037, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7120273113250732, + "num_tokens": 200920864.0, + "step": 7762 + }, + { + "epoch": 0.8525148253898528, + "grad_norm": 2.2809879779815674, + "learning_rate": 1e-06, + "loss": 1.0461, + "mean_token_accuracy": 0.6826698184013367, + "num_tokens": 200945141.0, + "step": 7763 + }, + { + "epoch": 0.8526246430924665, + "grad_norm": 2.1145853996276855, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7087130546569824, + "num_tokens": 200973178.0, + "step": 7764 + }, + { + "epoch": 0.8527344607950802, + "grad_norm": 2.1771323680877686, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7198408842086792, + "num_tokens": 200996047.0, + "step": 7765 + }, + { + "epoch": 0.8528442784976938, + "grad_norm": 2.224602699279785, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7024008631706238, + "num_tokens": 201021146.0, + "step": 7766 + }, + { + "epoch": 0.8529540962003075, + "grad_norm": 2.2129483222961426, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7219623923301697, + "num_tokens": 201046614.0, + "step": 7767 + }, + { + "epoch": 0.8530639139029211, + "grad_norm": 2.2483596801757812, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7363796234130859, + "num_tokens": 201068484.0, + "step": 7768 + }, + { + "epoch": 0.8531737316055348, + "grad_norm": 2.3268446922302246, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7145387530326843, + "num_tokens": 201089170.0, + "step": 7769 + }, + { + "epoch": 0.8532835493081484, + "grad_norm": 2.3219172954559326, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6941717863082886, + "num_tokens": 201112542.0, + "step": 7770 + }, + { + "epoch": 0.8533933670107622, + "grad_norm": 2.1938846111297607, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7356240749359131, + "num_tokens": 201136180.0, + "step": 7771 + }, + { + "epoch": 0.8535031847133758, + "grad_norm": 2.383925437927246, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.706087589263916, + "num_tokens": 201159136.0, + "step": 7772 + }, + { + "epoch": 0.8536130024159895, + "grad_norm": 2.0273497104644775, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6957537531852722, + "num_tokens": 201188466.0, + "step": 7773 + }, + { + "epoch": 0.8537228201186031, + "grad_norm": 2.1196351051330566, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6904503703117371, + "num_tokens": 201214870.0, + "step": 7774 + }, + { + "epoch": 0.8538326378212168, + "grad_norm": 2.309671640396118, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7162234783172607, + "num_tokens": 201238516.0, + "step": 7775 + }, + { + "epoch": 0.8539424555238304, + "grad_norm": 2.172668933868408, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7098884582519531, + "num_tokens": 201264492.0, + "step": 7776 + }, + { + "epoch": 0.8540522732264441, + "grad_norm": 1.7889569997787476, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6880710124969482, + "num_tokens": 201300458.0, + "step": 7777 + }, + { + "epoch": 0.8541620909290578, + "grad_norm": 1.7218894958496094, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7138996124267578, + "num_tokens": 201337032.0, + "step": 7778 + }, + { + "epoch": 0.8542719086316715, + "grad_norm": 2.3405208587646484, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7104700207710266, + "num_tokens": 201360124.0, + "step": 7779 + }, + { + "epoch": 0.8543817263342851, + "grad_norm": 2.5749599933624268, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.707095205783844, + "num_tokens": 201379881.0, + "step": 7780 + }, + { + "epoch": 0.8544915440368988, + "grad_norm": 2.1234610080718994, + "learning_rate": 1e-06, + "loss": 1.0801, + "mean_token_accuracy": 0.6697231531143188, + "num_tokens": 201407020.0, + "step": 7781 + }, + { + "epoch": 0.8546013617395124, + "grad_norm": 2.4179773330688477, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7171578407287598, + "num_tokens": 201430308.0, + "step": 7782 + }, + { + "epoch": 0.854711179442126, + "grad_norm": 2.2060022354125977, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.692131757736206, + "num_tokens": 201456423.0, + "step": 7783 + }, + { + "epoch": 0.8548209971447397, + "grad_norm": 2.156064033508301, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7187501192092896, + "num_tokens": 201479965.0, + "step": 7784 + }, + { + "epoch": 0.8549308148473534, + "grad_norm": 2.1395764350891113, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6864637136459351, + "num_tokens": 201506398.0, + "step": 7785 + }, + { + "epoch": 0.8550406325499671, + "grad_norm": 2.0791232585906982, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7065950632095337, + "num_tokens": 201533902.0, + "step": 7786 + }, + { + "epoch": 0.8551504502525807, + "grad_norm": 2.3391456604003906, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.721729576587677, + "num_tokens": 201555508.0, + "step": 7787 + }, + { + "epoch": 0.8552602679551944, + "grad_norm": 2.2273850440979004, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6776821613311768, + "num_tokens": 201583580.0, + "step": 7788 + }, + { + "epoch": 0.855370085657808, + "grad_norm": 2.106766939163208, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6999919414520264, + "num_tokens": 201614497.0, + "step": 7789 + }, + { + "epoch": 0.8554799033604217, + "grad_norm": 2.2176952362060547, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.715313196182251, + "num_tokens": 201637724.0, + "step": 7790 + }, + { + "epoch": 0.8555897210630353, + "grad_norm": 2.0926880836486816, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7219217419624329, + "num_tokens": 201663407.0, + "step": 7791 + }, + { + "epoch": 0.855699538765649, + "grad_norm": 2.5206151008605957, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7025014162063599, + "num_tokens": 201683632.0, + "step": 7792 + }, + { + "epoch": 0.8558093564682627, + "grad_norm": 2.3943045139312744, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7193623185157776, + "num_tokens": 201704739.0, + "step": 7793 + }, + { + "epoch": 0.8559191741708764, + "grad_norm": 2.306436777114868, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.709775447845459, + "num_tokens": 201727432.0, + "step": 7794 + }, + { + "epoch": 0.85602899187349, + "grad_norm": 2.3575663566589355, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7056062817573547, + "num_tokens": 201751819.0, + "step": 7795 + }, + { + "epoch": 0.8561388095761037, + "grad_norm": 1.8203015327453613, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7129917144775391, + "num_tokens": 201783478.0, + "step": 7796 + }, + { + "epoch": 0.8562486272787173, + "grad_norm": 2.1206512451171875, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7111630439758301, + "num_tokens": 201808626.0, + "step": 7797 + }, + { + "epoch": 0.856358444981331, + "grad_norm": 2.313678026199341, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7162818312644958, + "num_tokens": 201830947.0, + "step": 7798 + }, + { + "epoch": 0.8564682626839446, + "grad_norm": 2.277979612350464, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6968979835510254, + "num_tokens": 201855299.0, + "step": 7799 + }, + { + "epoch": 0.8565780803865584, + "grad_norm": 2.2058236598968506, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7146531343460083, + "num_tokens": 201881901.0, + "step": 7800 + }, + { + "epoch": 0.856687898089172, + "grad_norm": 1.9766640663146973, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7018532752990723, + "num_tokens": 201914176.0, + "step": 7801 + }, + { + "epoch": 0.8567977157917857, + "grad_norm": 2.0148799419403076, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7024348974227905, + "num_tokens": 201941746.0, + "step": 7802 + }, + { + "epoch": 0.8569075334943993, + "grad_norm": 2.0931105613708496, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7233270406723022, + "num_tokens": 201968023.0, + "step": 7803 + }, + { + "epoch": 0.8570173511970129, + "grad_norm": 2.1150026321411133, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7120140194892883, + "num_tokens": 201992326.0, + "step": 7804 + }, + { + "epoch": 0.8571271688996266, + "grad_norm": 1.7686283588409424, + "learning_rate": 1e-06, + "loss": 1.0795, + "mean_token_accuracy": 0.6708263754844666, + "num_tokens": 202030782.0, + "step": 7805 + }, + { + "epoch": 0.8572369866022402, + "grad_norm": 2.0644190311431885, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7061530351638794, + "num_tokens": 202058757.0, + "step": 7806 + }, + { + "epoch": 0.857346804304854, + "grad_norm": 2.275315046310425, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7040435671806335, + "num_tokens": 202083103.0, + "step": 7807 + }, + { + "epoch": 0.8574566220074676, + "grad_norm": 2.0635762214660645, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6954578161239624, + "num_tokens": 202112823.0, + "step": 7808 + }, + { + "epoch": 0.8575664397100813, + "grad_norm": 1.9036800861358643, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7260122895240784, + "num_tokens": 202143862.0, + "step": 7809 + }, + { + "epoch": 0.8576762574126949, + "grad_norm": 1.8983193635940552, + "learning_rate": 1e-06, + "loss": 1.0305, + "mean_token_accuracy": 0.6853234171867371, + "num_tokens": 202176427.0, + "step": 7810 + }, + { + "epoch": 0.8577860751153086, + "grad_norm": 2.0273711681365967, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7231414914131165, + "num_tokens": 202205462.0, + "step": 7811 + }, + { + "epoch": 0.8578958928179222, + "grad_norm": 2.3129141330718994, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7111408710479736, + "num_tokens": 202228124.0, + "step": 7812 + }, + { + "epoch": 0.8580057105205359, + "grad_norm": 2.2267813682556152, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7125087976455688, + "num_tokens": 202253879.0, + "step": 7813 + }, + { + "epoch": 0.8581155282231496, + "grad_norm": 1.9646892547607422, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7012271285057068, + "num_tokens": 202283863.0, + "step": 7814 + }, + { + "epoch": 0.8582253459257633, + "grad_norm": 2.2012436389923096, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7108967304229736, + "num_tokens": 202308816.0, + "step": 7815 + }, + { + "epoch": 0.8583351636283769, + "grad_norm": 2.5963921546936035, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.723625659942627, + "num_tokens": 202328376.0, + "step": 7816 + }, + { + "epoch": 0.8584449813309906, + "grad_norm": 2.0813279151916504, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7117922306060791, + "num_tokens": 202355544.0, + "step": 7817 + }, + { + "epoch": 0.8585547990336042, + "grad_norm": 2.0413811206817627, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7287145853042603, + "num_tokens": 202381441.0, + "step": 7818 + }, + { + "epoch": 0.8586646167362179, + "grad_norm": 2.184345006942749, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7306252717971802, + "num_tokens": 202408014.0, + "step": 7819 + }, + { + "epoch": 0.8587744344388315, + "grad_norm": 2.053760290145874, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7468545436859131, + "num_tokens": 202435919.0, + "step": 7820 + }, + { + "epoch": 0.8588842521414451, + "grad_norm": 2.2315707206726074, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7204698324203491, + "num_tokens": 202458627.0, + "step": 7821 + }, + { + "epoch": 0.8589940698440589, + "grad_norm": 2.1667134761810303, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7102960348129272, + "num_tokens": 202484299.0, + "step": 7822 + }, + { + "epoch": 0.8591038875466726, + "grad_norm": 2.1432559490203857, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7117329835891724, + "num_tokens": 202510414.0, + "step": 7823 + }, + { + "epoch": 0.8592137052492862, + "grad_norm": 2.2792375087738037, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.714724063873291, + "num_tokens": 202533513.0, + "step": 7824 + }, + { + "epoch": 0.8593235229518998, + "grad_norm": 2.2295379638671875, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7008765935897827, + "num_tokens": 202559780.0, + "step": 7825 + }, + { + "epoch": 0.8594333406545135, + "grad_norm": 1.9535138607025146, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7038793563842773, + "num_tokens": 202589175.0, + "step": 7826 + }, + { + "epoch": 0.8595431583571271, + "grad_norm": 2.4288253784179688, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.717241644859314, + "num_tokens": 202611568.0, + "step": 7827 + }, + { + "epoch": 0.8596529760597408, + "grad_norm": 2.0036072731018066, + "learning_rate": 1e-06, + "loss": 1.0431, + "mean_token_accuracy": 0.6790505051612854, + "num_tokens": 202640644.0, + "step": 7828 + }, + { + "epoch": 0.8597627937623545, + "grad_norm": 1.8515747785568237, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.702069103717804, + "num_tokens": 202674639.0, + "step": 7829 + }, + { + "epoch": 0.8598726114649682, + "grad_norm": 2.181584119796753, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6933908462524414, + "num_tokens": 202700287.0, + "step": 7830 + }, + { + "epoch": 0.8599824291675818, + "grad_norm": 1.8513010740280151, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6830115914344788, + "num_tokens": 202736629.0, + "step": 7831 + }, + { + "epoch": 0.8600922468701955, + "grad_norm": 2.3003652095794678, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7182670831680298, + "num_tokens": 202757926.0, + "step": 7832 + }, + { + "epoch": 0.8602020645728091, + "grad_norm": 2.1809444427490234, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6988679766654968, + "num_tokens": 202784249.0, + "step": 7833 + }, + { + "epoch": 0.8603118822754228, + "grad_norm": 2.2203474044799805, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6989619731903076, + "num_tokens": 202809701.0, + "step": 7834 + }, + { + "epoch": 0.8604216999780364, + "grad_norm": 2.054832935333252, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7012417316436768, + "num_tokens": 202837782.0, + "step": 7835 + }, + { + "epoch": 0.8605315176806502, + "grad_norm": 2.136202335357666, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7254709005355835, + "num_tokens": 202864615.0, + "step": 7836 + }, + { + "epoch": 0.8606413353832638, + "grad_norm": 2.058842182159424, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7089135050773621, + "num_tokens": 202891672.0, + "step": 7837 + }, + { + "epoch": 0.8607511530858775, + "grad_norm": 2.1005196571350098, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.6986252069473267, + "num_tokens": 202922336.0, + "step": 7838 + }, + { + "epoch": 0.8608609707884911, + "grad_norm": 2.1023898124694824, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.6979106664657593, + "num_tokens": 202949143.0, + "step": 7839 + }, + { + "epoch": 0.8609707884911048, + "grad_norm": 2.410781145095825, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7421834468841553, + "num_tokens": 202967791.0, + "step": 7840 + }, + { + "epoch": 0.8610806061937184, + "grad_norm": 2.235048770904541, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6936004757881165, + "num_tokens": 202990516.0, + "step": 7841 + }, + { + "epoch": 0.861190423896332, + "grad_norm": 2.2026031017303467, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7013250589370728, + "num_tokens": 203014639.0, + "step": 7842 + }, + { + "epoch": 0.8613002415989458, + "grad_norm": 2.062281370162964, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7122927904129028, + "num_tokens": 203042275.0, + "step": 7843 + }, + { + "epoch": 0.8614100593015594, + "grad_norm": 2.2767539024353027, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6981076002120972, + "num_tokens": 203066848.0, + "step": 7844 + }, + { + "epoch": 0.8615198770041731, + "grad_norm": 2.000795364379883, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6860365867614746, + "num_tokens": 203097751.0, + "step": 7845 + }, + { + "epoch": 0.8616296947067867, + "grad_norm": 2.160590887069702, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6924363970756531, + "num_tokens": 203124755.0, + "step": 7846 + }, + { + "epoch": 0.8617395124094004, + "grad_norm": 2.392474889755249, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.7054734230041504, + "num_tokens": 203146497.0, + "step": 7847 + }, + { + "epoch": 0.861849330112014, + "grad_norm": 2.3943052291870117, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7151633501052856, + "num_tokens": 203167468.0, + "step": 7848 + }, + { + "epoch": 0.8619591478146277, + "grad_norm": 1.9339755773544312, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7190574407577515, + "num_tokens": 203198814.0, + "step": 7849 + }, + { + "epoch": 0.8620689655172413, + "grad_norm": 2.120102643966675, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7055661678314209, + "num_tokens": 203224152.0, + "step": 7850 + }, + { + "epoch": 0.8621787832198551, + "grad_norm": 2.1453258991241455, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7340127229690552, + "num_tokens": 203248679.0, + "step": 7851 + }, + { + "epoch": 0.8622886009224687, + "grad_norm": 1.9665664434432983, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6852909326553345, + "num_tokens": 203279320.0, + "step": 7852 + }, + { + "epoch": 0.8623984186250824, + "grad_norm": 2.1887731552124023, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7180635929107666, + "num_tokens": 203303420.0, + "step": 7853 + }, + { + "epoch": 0.862508236327696, + "grad_norm": 2.2876904010772705, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7171306014060974, + "num_tokens": 203327137.0, + "step": 7854 + }, + { + "epoch": 0.8626180540303097, + "grad_norm": 2.167022943496704, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7174514532089233, + "num_tokens": 203353839.0, + "step": 7855 + }, + { + "epoch": 0.8627278717329233, + "grad_norm": 2.002777099609375, + "learning_rate": 1e-06, + "loss": 1.0293, + "mean_token_accuracy": 0.6903653144836426, + "num_tokens": 203383795.0, + "step": 7856 + }, + { + "epoch": 0.862837689435537, + "grad_norm": 2.3135576248168945, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7151355147361755, + "num_tokens": 203407162.0, + "step": 7857 + }, + { + "epoch": 0.8629475071381507, + "grad_norm": 2.1889872550964355, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7012836337089539, + "num_tokens": 203434099.0, + "step": 7858 + }, + { + "epoch": 0.8630573248407644, + "grad_norm": 2.347026824951172, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7184926271438599, + "num_tokens": 203456548.0, + "step": 7859 + }, + { + "epoch": 0.863167142543378, + "grad_norm": 2.1157779693603516, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7181147336959839, + "num_tokens": 203484169.0, + "step": 7860 + }, + { + "epoch": 0.8632769602459917, + "grad_norm": 2.2434113025665283, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6957078576087952, + "num_tokens": 203509125.0, + "step": 7861 + }, + { + "epoch": 0.8633867779486053, + "grad_norm": 2.093855619430542, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.716629147529602, + "num_tokens": 203534755.0, + "step": 7862 + }, + { + "epoch": 0.863496595651219, + "grad_norm": 2.241860866546631, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6969039440155029, + "num_tokens": 203558568.0, + "step": 7863 + }, + { + "epoch": 0.8636064133538326, + "grad_norm": 2.237921714782715, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.709374189376831, + "num_tokens": 203582837.0, + "step": 7864 + }, + { + "epoch": 0.8637162310564463, + "grad_norm": 2.076033592224121, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7177646160125732, + "num_tokens": 203611467.0, + "step": 7865 + }, + { + "epoch": 0.86382604875906, + "grad_norm": 2.255657434463501, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7129331231117249, + "num_tokens": 203636023.0, + "step": 7866 + }, + { + "epoch": 0.8639358664616736, + "grad_norm": 2.2429592609405518, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7031410336494446, + "num_tokens": 203660568.0, + "step": 7867 + }, + { + "epoch": 0.8640456841642873, + "grad_norm": 2.1608338356018066, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6924833059310913, + "num_tokens": 203688505.0, + "step": 7868 + }, + { + "epoch": 0.8641555018669009, + "grad_norm": 2.053901195526123, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.682781457901001, + "num_tokens": 203718239.0, + "step": 7869 + }, + { + "epoch": 0.8642653195695146, + "grad_norm": 2.145859718322754, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7170580625534058, + "num_tokens": 203743844.0, + "step": 7870 + }, + { + "epoch": 0.8643751372721282, + "grad_norm": 2.2918827533721924, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7101679444313049, + "num_tokens": 203767153.0, + "step": 7871 + }, + { + "epoch": 0.864484954974742, + "grad_norm": 2.0106656551361084, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7075402736663818, + "num_tokens": 203794917.0, + "step": 7872 + }, + { + "epoch": 0.8645947726773556, + "grad_norm": 2.344668388366699, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7061368227005005, + "num_tokens": 203817801.0, + "step": 7873 + }, + { + "epoch": 0.8647045903799693, + "grad_norm": 2.23008394241333, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7007544040679932, + "num_tokens": 203841656.0, + "step": 7874 + }, + { + "epoch": 0.8648144080825829, + "grad_norm": 2.3824679851531982, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7219841480255127, + "num_tokens": 203864128.0, + "step": 7875 + }, + { + "epoch": 0.8649242257851966, + "grad_norm": 2.248830795288086, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7260633707046509, + "num_tokens": 203886915.0, + "step": 7876 + }, + { + "epoch": 0.8650340434878102, + "grad_norm": 2.116842269897461, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7175363898277283, + "num_tokens": 203912550.0, + "step": 7877 + }, + { + "epoch": 0.8651438611904239, + "grad_norm": 2.3376359939575195, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7117817997932434, + "num_tokens": 203935089.0, + "step": 7878 + }, + { + "epoch": 0.8652536788930375, + "grad_norm": 2.0035929679870605, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7376134395599365, + "num_tokens": 203963748.0, + "step": 7879 + }, + { + "epoch": 0.8653634965956513, + "grad_norm": 2.474480152130127, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7244401574134827, + "num_tokens": 203984878.0, + "step": 7880 + }, + { + "epoch": 0.8654733142982649, + "grad_norm": 2.0270650386810303, + "learning_rate": 1e-06, + "loss": 1.0758, + "mean_token_accuracy": 0.6828088164329529, + "num_tokens": 204014695.0, + "step": 7881 + }, + { + "epoch": 0.8655831320008786, + "grad_norm": 2.3985326290130615, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7221398949623108, + "num_tokens": 204035281.0, + "step": 7882 + }, + { + "epoch": 0.8656929497034922, + "grad_norm": 2.211545944213867, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7181216478347778, + "num_tokens": 204059679.0, + "step": 7883 + }, + { + "epoch": 0.8658027674061058, + "grad_norm": 2.367994546890259, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.716829776763916, + "num_tokens": 204081454.0, + "step": 7884 + }, + { + "epoch": 0.8659125851087195, + "grad_norm": 2.311866044998169, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.724661111831665, + "num_tokens": 204103325.0, + "step": 7885 + }, + { + "epoch": 0.8660224028113331, + "grad_norm": 2.5764687061309814, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7077320218086243, + "num_tokens": 204124511.0, + "step": 7886 + }, + { + "epoch": 0.8661322205139469, + "grad_norm": 2.100181818008423, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7086777091026306, + "num_tokens": 204151632.0, + "step": 7887 + }, + { + "epoch": 0.8662420382165605, + "grad_norm": 2.118669271469116, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.688255786895752, + "num_tokens": 204177986.0, + "step": 7888 + }, + { + "epoch": 0.8663518559191742, + "grad_norm": 2.1670351028442383, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7012518644332886, + "num_tokens": 204205111.0, + "step": 7889 + }, + { + "epoch": 0.8664616736217878, + "grad_norm": 2.266550064086914, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7051642537117004, + "num_tokens": 204227601.0, + "step": 7890 + }, + { + "epoch": 0.8665714913244015, + "grad_norm": 2.2699246406555176, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.723901629447937, + "num_tokens": 204250245.0, + "step": 7891 + }, + { + "epoch": 0.8666813090270151, + "grad_norm": 2.304302453994751, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7097599506378174, + "num_tokens": 204274328.0, + "step": 7892 + }, + { + "epoch": 0.8667911267296288, + "grad_norm": 2.356419086456299, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7213283777236938, + "num_tokens": 204297123.0, + "step": 7893 + }, + { + "epoch": 0.8669009444322425, + "grad_norm": 1.9350321292877197, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7097107768058777, + "num_tokens": 204327408.0, + "step": 7894 + }, + { + "epoch": 0.8670107621348562, + "grad_norm": 2.434415102005005, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7191717624664307, + "num_tokens": 204349246.0, + "step": 7895 + }, + { + "epoch": 0.8671205798374698, + "grad_norm": 2.5728461742401123, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7283698916435242, + "num_tokens": 204367333.0, + "step": 7896 + }, + { + "epoch": 0.8672303975400835, + "grad_norm": 1.859381079673767, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7274753451347351, + "num_tokens": 204400348.0, + "step": 7897 + }, + { + "epoch": 0.8673402152426971, + "grad_norm": 2.240652084350586, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.688888669013977, + "num_tokens": 204425388.0, + "step": 7898 + }, + { + "epoch": 0.8674500329453108, + "grad_norm": 2.4166903495788574, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6945832967758179, + "num_tokens": 204449858.0, + "step": 7899 + }, + { + "epoch": 0.8675598506479244, + "grad_norm": 2.1474246978759766, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6879684925079346, + "num_tokens": 204476824.0, + "step": 7900 + }, + { + "epoch": 0.8676696683505382, + "grad_norm": 1.9706743955612183, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6859414577484131, + "num_tokens": 204505730.0, + "step": 7901 + }, + { + "epoch": 0.8677794860531518, + "grad_norm": 2.5978047847747803, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7070938348770142, + "num_tokens": 204526003.0, + "step": 7902 + }, + { + "epoch": 0.8678893037557655, + "grad_norm": 2.4237005710601807, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6979737877845764, + "num_tokens": 204548218.0, + "step": 7903 + }, + { + "epoch": 0.8679991214583791, + "grad_norm": 2.106182098388672, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6961584687232971, + "num_tokens": 204576206.0, + "step": 7904 + }, + { + "epoch": 0.8681089391609927, + "grad_norm": 1.9525973796844482, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.6987783312797546, + "num_tokens": 204608378.0, + "step": 7905 + }, + { + "epoch": 0.8682187568636064, + "grad_norm": 2.5494494438171387, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6899840831756592, + "num_tokens": 204628158.0, + "step": 7906 + }, + { + "epoch": 0.86832857456622, + "grad_norm": 2.35189151763916, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7167865037918091, + "num_tokens": 204650319.0, + "step": 7907 + }, + { + "epoch": 0.8684383922688337, + "grad_norm": 2.609135389328003, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7201622724533081, + "num_tokens": 204668863.0, + "step": 7908 + }, + { + "epoch": 0.8685482099714474, + "grad_norm": 2.2424356937408447, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7207496762275696, + "num_tokens": 204692670.0, + "step": 7909 + }, + { + "epoch": 0.8686580276740611, + "grad_norm": 2.3870232105255127, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6910537481307983, + "num_tokens": 204715673.0, + "step": 7910 + }, + { + "epoch": 0.8687678453766747, + "grad_norm": 2.358150005340576, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7339597940444946, + "num_tokens": 204738523.0, + "step": 7911 + }, + { + "epoch": 0.8688776630792884, + "grad_norm": 2.054356336593628, + "learning_rate": 1e-06, + "loss": 1.065, + "mean_token_accuracy": 0.673919141292572, + "num_tokens": 204767872.0, + "step": 7912 + }, + { + "epoch": 0.868987480781902, + "grad_norm": 2.1721293926239014, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.697390079498291, + "num_tokens": 204795170.0, + "step": 7913 + }, + { + "epoch": 0.8690972984845157, + "grad_norm": 2.0334489345550537, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6842021942138672, + "num_tokens": 204825372.0, + "step": 7914 + }, + { + "epoch": 0.8692071161871293, + "grad_norm": 2.0505034923553467, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.6999950408935547, + "num_tokens": 204853010.0, + "step": 7915 + }, + { + "epoch": 0.8693169338897431, + "grad_norm": 2.0196032524108887, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7051861882209778, + "num_tokens": 204883722.0, + "step": 7916 + }, + { + "epoch": 0.8694267515923567, + "grad_norm": 2.200117349624634, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.710413932800293, + "num_tokens": 204908704.0, + "step": 7917 + }, + { + "epoch": 0.8695365692949704, + "grad_norm": 2.259716272354126, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7146902680397034, + "num_tokens": 204931118.0, + "step": 7918 + }, + { + "epoch": 0.869646386997584, + "grad_norm": 2.0601823329925537, + "learning_rate": 1e-06, + "loss": 1.0407, + "mean_token_accuracy": 0.6787432432174683, + "num_tokens": 204959245.0, + "step": 7919 + }, + { + "epoch": 0.8697562047001977, + "grad_norm": 2.496689558029175, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7053720951080322, + "num_tokens": 204980960.0, + "step": 7920 + }, + { + "epoch": 0.8698660224028113, + "grad_norm": 1.9518190622329712, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7146172523498535, + "num_tokens": 205012385.0, + "step": 7921 + }, + { + "epoch": 0.869975840105425, + "grad_norm": 1.9967385530471802, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7074193954467773, + "num_tokens": 205042316.0, + "step": 7922 + }, + { + "epoch": 0.8700856578080387, + "grad_norm": 2.112656354904175, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7058439254760742, + "num_tokens": 205069593.0, + "step": 7923 + }, + { + "epoch": 0.8701954755106523, + "grad_norm": 2.1270318031311035, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7118927240371704, + "num_tokens": 205095869.0, + "step": 7924 + }, + { + "epoch": 0.870305293213266, + "grad_norm": 2.2777099609375, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6907458305358887, + "num_tokens": 205120650.0, + "step": 7925 + }, + { + "epoch": 0.8704151109158796, + "grad_norm": 2.3231656551361084, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6993211507797241, + "num_tokens": 205145191.0, + "step": 7926 + }, + { + "epoch": 0.8705249286184933, + "grad_norm": 2.2358205318450928, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7169616222381592, + "num_tokens": 205171048.0, + "step": 7927 + }, + { + "epoch": 0.8706347463211069, + "grad_norm": 2.104116678237915, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7180922031402588, + "num_tokens": 205197124.0, + "step": 7928 + }, + { + "epoch": 0.8707445640237206, + "grad_norm": 1.9038527011871338, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7076520919799805, + "num_tokens": 205228334.0, + "step": 7929 + }, + { + "epoch": 0.8708543817263343, + "grad_norm": 2.185621976852417, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7236402630805969, + "num_tokens": 205252781.0, + "step": 7930 + }, + { + "epoch": 0.870964199428948, + "grad_norm": 2.151855230331421, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.7019643783569336, + "num_tokens": 205279568.0, + "step": 7931 + }, + { + "epoch": 0.8710740171315616, + "grad_norm": 2.4649689197540283, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7194075584411621, + "num_tokens": 205298739.0, + "step": 7932 + }, + { + "epoch": 0.8711838348341753, + "grad_norm": 2.21242094039917, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7270362973213196, + "num_tokens": 205323281.0, + "step": 7933 + }, + { + "epoch": 0.8712936525367889, + "grad_norm": 2.395068883895874, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7067668437957764, + "num_tokens": 205346345.0, + "step": 7934 + }, + { + "epoch": 0.8714034702394026, + "grad_norm": 2.377262830734253, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7157002687454224, + "num_tokens": 205367389.0, + "step": 7935 + }, + { + "epoch": 0.8715132879420162, + "grad_norm": 2.256180763244629, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7016158103942871, + "num_tokens": 205390695.0, + "step": 7936 + }, + { + "epoch": 0.87162310564463, + "grad_norm": 2.2223541736602783, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.694672703742981, + "num_tokens": 205415769.0, + "step": 7937 + }, + { + "epoch": 0.8717329233472436, + "grad_norm": 2.2355902194976807, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7195871472358704, + "num_tokens": 205438970.0, + "step": 7938 + }, + { + "epoch": 0.8718427410498573, + "grad_norm": 2.1704390048980713, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7077559232711792, + "num_tokens": 205463899.0, + "step": 7939 + }, + { + "epoch": 0.8719525587524709, + "grad_norm": 2.041046619415283, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.6994144916534424, + "num_tokens": 205493986.0, + "step": 7940 + }, + { + "epoch": 0.8720623764550846, + "grad_norm": 2.4573557376861572, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7069413661956787, + "num_tokens": 205516520.0, + "step": 7941 + }, + { + "epoch": 0.8721721941576982, + "grad_norm": 2.1638917922973633, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.692314624786377, + "num_tokens": 205544956.0, + "step": 7942 + }, + { + "epoch": 0.8722820118603118, + "grad_norm": 2.4747979640960693, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7044767141342163, + "num_tokens": 205565308.0, + "step": 7943 + }, + { + "epoch": 0.8723918295629255, + "grad_norm": 2.052642822265625, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6992537975311279, + "num_tokens": 205592370.0, + "step": 7944 + }, + { + "epoch": 0.8725016472655392, + "grad_norm": 1.969635009765625, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6983909606933594, + "num_tokens": 205622412.0, + "step": 7945 + }, + { + "epoch": 0.8726114649681529, + "grad_norm": 2.317115068435669, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7060335278511047, + "num_tokens": 205646197.0, + "step": 7946 + }, + { + "epoch": 0.8727212826707665, + "grad_norm": 2.2566094398498535, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7121446132659912, + "num_tokens": 205669733.0, + "step": 7947 + }, + { + "epoch": 0.8728311003733802, + "grad_norm": 1.9850834608078003, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7078068256378174, + "num_tokens": 205697984.0, + "step": 7948 + }, + { + "epoch": 0.8729409180759938, + "grad_norm": 2.3234457969665527, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6881780624389648, + "num_tokens": 205722254.0, + "step": 7949 + }, + { + "epoch": 0.8730507357786075, + "grad_norm": 2.1691174507141113, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7065292000770569, + "num_tokens": 205748538.0, + "step": 7950 + }, + { + "epoch": 0.8731605534812211, + "grad_norm": 2.224266529083252, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7356635332107544, + "num_tokens": 205772601.0, + "step": 7951 + }, + { + "epoch": 0.8732703711838349, + "grad_norm": 2.2109148502349854, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.6953524947166443, + "num_tokens": 205796423.0, + "step": 7952 + }, + { + "epoch": 0.8733801888864485, + "grad_norm": 2.0074827671051025, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7268120050430298, + "num_tokens": 205826257.0, + "step": 7953 + }, + { + "epoch": 0.8734900065890622, + "grad_norm": 2.1742467880249023, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6975671648979187, + "num_tokens": 205850383.0, + "step": 7954 + }, + { + "epoch": 0.8735998242916758, + "grad_norm": 2.583670139312744, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7062666416168213, + "num_tokens": 205869212.0, + "step": 7955 + }, + { + "epoch": 0.8737096419942895, + "grad_norm": 2.380754232406616, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7130875587463379, + "num_tokens": 205891962.0, + "step": 7956 + }, + { + "epoch": 0.8738194596969031, + "grad_norm": 2.3269712924957275, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7164673805236816, + "num_tokens": 205916887.0, + "step": 7957 + }, + { + "epoch": 0.8739292773995168, + "grad_norm": 2.0415148735046387, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7107656598091125, + "num_tokens": 205943337.0, + "step": 7958 + }, + { + "epoch": 0.8740390951021305, + "grad_norm": 2.116985321044922, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7022446393966675, + "num_tokens": 205969059.0, + "step": 7959 + }, + { + "epoch": 0.8741489128047442, + "grad_norm": 1.9794033765792847, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.722355306148529, + "num_tokens": 205998041.0, + "step": 7960 + }, + { + "epoch": 0.8742587305073578, + "grad_norm": 2.243565559387207, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7181576490402222, + "num_tokens": 206022857.0, + "step": 7961 + }, + { + "epoch": 0.8743685482099715, + "grad_norm": 2.040597915649414, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6855412125587463, + "num_tokens": 206051159.0, + "step": 7962 + }, + { + "epoch": 0.8744783659125851, + "grad_norm": 2.0106987953186035, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7205082178115845, + "num_tokens": 206079039.0, + "step": 7963 + }, + { + "epoch": 0.8745881836151987, + "grad_norm": 2.1992881298065186, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7053186893463135, + "num_tokens": 206103595.0, + "step": 7964 + }, + { + "epoch": 0.8746980013178124, + "grad_norm": 2.2294862270355225, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7008763551712036, + "num_tokens": 206128831.0, + "step": 7965 + }, + { + "epoch": 0.8748078190204261, + "grad_norm": 2.2151639461517334, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7032879590988159, + "num_tokens": 206153310.0, + "step": 7966 + }, + { + "epoch": 0.8749176367230398, + "grad_norm": 2.0267369747161865, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7264524698257446, + "num_tokens": 206179791.0, + "step": 7967 + }, + { + "epoch": 0.8750274544256534, + "grad_norm": 2.5636298656463623, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7177520990371704, + "num_tokens": 206198392.0, + "step": 7968 + }, + { + "epoch": 0.8751372721282671, + "grad_norm": 2.094204902648926, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.7083046436309814, + "num_tokens": 206226572.0, + "step": 7969 + }, + { + "epoch": 0.8752470898308807, + "grad_norm": 2.328484058380127, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7349532842636108, + "num_tokens": 206247127.0, + "step": 7970 + }, + { + "epoch": 0.8753569075334944, + "grad_norm": 2.2090532779693604, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7143468856811523, + "num_tokens": 206271724.0, + "step": 7971 + }, + { + "epoch": 0.875466725236108, + "grad_norm": 1.9225167036056519, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7001221179962158, + "num_tokens": 206302578.0, + "step": 7972 + }, + { + "epoch": 0.8755765429387217, + "grad_norm": 2.2828257083892822, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7142328023910522, + "num_tokens": 206328020.0, + "step": 7973 + }, + { + "epoch": 0.8756863606413354, + "grad_norm": 2.0901622772216797, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7205886244773865, + "num_tokens": 206353497.0, + "step": 7974 + }, + { + "epoch": 0.8757961783439491, + "grad_norm": 2.1695704460144043, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7090069055557251, + "num_tokens": 206377663.0, + "step": 7975 + }, + { + "epoch": 0.8759059960465627, + "grad_norm": 2.096424102783203, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7053998708724976, + "num_tokens": 206404111.0, + "step": 7976 + }, + { + "epoch": 0.8760158137491764, + "grad_norm": 2.355360269546509, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7080554962158203, + "num_tokens": 206425714.0, + "step": 7977 + }, + { + "epoch": 0.87612563145179, + "grad_norm": 2.0743536949157715, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7000285387039185, + "num_tokens": 206454801.0, + "step": 7978 + }, + { + "epoch": 0.8762354491544037, + "grad_norm": 2.241611957550049, + "learning_rate": 1e-06, + "loss": 1.0593, + "mean_token_accuracy": 0.6873026490211487, + "num_tokens": 206480724.0, + "step": 7979 + }, + { + "epoch": 0.8763452668570173, + "grad_norm": 2.0685877799987793, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7155965566635132, + "num_tokens": 206508803.0, + "step": 7980 + }, + { + "epoch": 0.8764550845596311, + "grad_norm": 2.2290663719177246, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7052494883537292, + "num_tokens": 206532791.0, + "step": 7981 + }, + { + "epoch": 0.8765649022622447, + "grad_norm": 2.6955459117889404, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.710934579372406, + "num_tokens": 206550452.0, + "step": 7982 + }, + { + "epoch": 0.8766747199648584, + "grad_norm": 2.128823757171631, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7045758366584778, + "num_tokens": 206575311.0, + "step": 7983 + }, + { + "epoch": 0.876784537667472, + "grad_norm": 1.9252324104309082, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6936126351356506, + "num_tokens": 206606387.0, + "step": 7984 + }, + { + "epoch": 0.8768943553700856, + "grad_norm": 2.099156379699707, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7242550849914551, + "num_tokens": 206631078.0, + "step": 7985 + }, + { + "epoch": 0.8770041730726993, + "grad_norm": 2.2555015087127686, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7038931846618652, + "num_tokens": 206656326.0, + "step": 7986 + }, + { + "epoch": 0.8771139907753129, + "grad_norm": 1.9759666919708252, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6990103721618652, + "num_tokens": 206687052.0, + "step": 7987 + }, + { + "epoch": 0.8772238084779267, + "grad_norm": 2.4651236534118652, + "learning_rate": 1e-06, + "loss": 1.06, + "mean_token_accuracy": 0.6767359972000122, + "num_tokens": 206708479.0, + "step": 7988 + }, + { + "epoch": 0.8773336261805403, + "grad_norm": 2.095716953277588, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6935222148895264, + "num_tokens": 206734578.0, + "step": 7989 + }, + { + "epoch": 0.877443443883154, + "grad_norm": 2.027251958847046, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7216238379478455, + "num_tokens": 206761012.0, + "step": 7990 + }, + { + "epoch": 0.8775532615857676, + "grad_norm": 2.224940299987793, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7278589010238647, + "num_tokens": 206782739.0, + "step": 7991 + }, + { + "epoch": 0.8776630792883813, + "grad_norm": 1.914066195487976, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6797930002212524, + "num_tokens": 206816363.0, + "step": 7992 + }, + { + "epoch": 0.8777728969909949, + "grad_norm": 2.211848020553589, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.722258985042572, + "num_tokens": 206840364.0, + "step": 7993 + }, + { + "epoch": 0.8778827146936086, + "grad_norm": 2.290764808654785, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6896471381187439, + "num_tokens": 206868088.0, + "step": 7994 + }, + { + "epoch": 0.8779925323962223, + "grad_norm": 2.4955127239227295, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7069513201713562, + "num_tokens": 206889482.0, + "step": 7995 + }, + { + "epoch": 0.878102350098836, + "grad_norm": 2.3064725399017334, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7073792219161987, + "num_tokens": 206912349.0, + "step": 7996 + }, + { + "epoch": 0.8782121678014496, + "grad_norm": 2.077542304992676, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7204477787017822, + "num_tokens": 206937852.0, + "step": 7997 + }, + { + "epoch": 0.8783219855040633, + "grad_norm": 2.219435930252075, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7312004566192627, + "num_tokens": 206961008.0, + "step": 7998 + }, + { + "epoch": 0.8784318032066769, + "grad_norm": 2.2961909770965576, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7011398077011108, + "num_tokens": 206983322.0, + "step": 7999 + }, + { + "epoch": 0.8785416209092906, + "grad_norm": 2.27229380607605, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7044875621795654, + "num_tokens": 207007351.0, + "step": 8000 + }, + { + "epoch": 0.8786514386119042, + "grad_norm": 1.9230117797851562, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7107217311859131, + "num_tokens": 207038395.0, + "step": 8001 + }, + { + "epoch": 0.8787612563145178, + "grad_norm": 2.3006367683410645, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7109148502349854, + "num_tokens": 207062632.0, + "step": 8002 + }, + { + "epoch": 0.8788710740171316, + "grad_norm": 1.9359912872314453, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.6881349682807922, + "num_tokens": 207092243.0, + "step": 8003 + }, + { + "epoch": 0.8789808917197452, + "grad_norm": 1.9427579641342163, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6998142004013062, + "num_tokens": 207122819.0, + "step": 8004 + }, + { + "epoch": 0.8790907094223589, + "grad_norm": 2.2475829124450684, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7148555517196655, + "num_tokens": 207147033.0, + "step": 8005 + }, + { + "epoch": 0.8792005271249725, + "grad_norm": 2.1791274547576904, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6902819871902466, + "num_tokens": 207172566.0, + "step": 8006 + }, + { + "epoch": 0.8793103448275862, + "grad_norm": 1.9928537607192993, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7036837935447693, + "num_tokens": 207201224.0, + "step": 8007 + }, + { + "epoch": 0.8794201625301998, + "grad_norm": 2.057886838912964, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7223511338233948, + "num_tokens": 207228077.0, + "step": 8008 + }, + { + "epoch": 0.8795299802328135, + "grad_norm": 2.0562288761138916, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6815998554229736, + "num_tokens": 207256588.0, + "step": 8009 + }, + { + "epoch": 0.8796397979354272, + "grad_norm": 2.2422566413879395, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.719193696975708, + "num_tokens": 207278042.0, + "step": 8010 + }, + { + "epoch": 0.8797496156380409, + "grad_norm": 2.1527788639068604, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6890535354614258, + "num_tokens": 207305518.0, + "step": 8011 + }, + { + "epoch": 0.8798594333406545, + "grad_norm": 1.9468668699264526, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7101272344589233, + "num_tokens": 207335128.0, + "step": 8012 + }, + { + "epoch": 0.8799692510432682, + "grad_norm": 2.457637071609497, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7178581953048706, + "num_tokens": 207357774.0, + "step": 8013 + }, + { + "epoch": 0.8800790687458818, + "grad_norm": 2.192768096923828, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7126692533493042, + "num_tokens": 207383683.0, + "step": 8014 + }, + { + "epoch": 0.8801888864484955, + "grad_norm": 1.9594717025756836, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7128596901893616, + "num_tokens": 207413913.0, + "step": 8015 + }, + { + "epoch": 0.8802987041511091, + "grad_norm": 1.8485444784164429, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.708937406539917, + "num_tokens": 207449426.0, + "step": 8016 + }, + { + "epoch": 0.8804085218537229, + "grad_norm": 2.1333837509155273, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7078745365142822, + "num_tokens": 207476016.0, + "step": 8017 + }, + { + "epoch": 0.8805183395563365, + "grad_norm": 2.357722520828247, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7155399918556213, + "num_tokens": 207497955.0, + "step": 8018 + }, + { + "epoch": 0.8806281572589502, + "grad_norm": 2.461422920227051, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7416911125183105, + "num_tokens": 207516932.0, + "step": 8019 + }, + { + "epoch": 0.8807379749615638, + "grad_norm": 2.102752208709717, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.689487099647522, + "num_tokens": 207545331.0, + "step": 8020 + }, + { + "epoch": 0.8808477926641775, + "grad_norm": 2.341221570968628, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7135196328163147, + "num_tokens": 207568324.0, + "step": 8021 + }, + { + "epoch": 0.8809576103667911, + "grad_norm": 2.25716233253479, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7229468822479248, + "num_tokens": 207590390.0, + "step": 8022 + }, + { + "epoch": 0.8810674280694047, + "grad_norm": 2.1099424362182617, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7274062633514404, + "num_tokens": 207614861.0, + "step": 8023 + }, + { + "epoch": 0.8811772457720185, + "grad_norm": 2.049015760421753, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7036393880844116, + "num_tokens": 207643294.0, + "step": 8024 + }, + { + "epoch": 0.8812870634746321, + "grad_norm": 2.500054359436035, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7034424543380737, + "num_tokens": 207663473.0, + "step": 8025 + }, + { + "epoch": 0.8813968811772458, + "grad_norm": 2.126502513885498, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.712925374507904, + "num_tokens": 207693923.0, + "step": 8026 + }, + { + "epoch": 0.8815066988798594, + "grad_norm": 2.09273362159729, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7309246063232422, + "num_tokens": 207722358.0, + "step": 8027 + }, + { + "epoch": 0.8816165165824731, + "grad_norm": 2.183868885040283, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7096148133277893, + "num_tokens": 207748579.0, + "step": 8028 + }, + { + "epoch": 0.8817263342850867, + "grad_norm": 2.333070755004883, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.723349392414093, + "num_tokens": 207771070.0, + "step": 8029 + }, + { + "epoch": 0.8818361519877004, + "grad_norm": 2.5738511085510254, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7206343412399292, + "num_tokens": 207789767.0, + "step": 8030 + }, + { + "epoch": 0.881945969690314, + "grad_norm": 2.3178470134735107, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7218410968780518, + "num_tokens": 207810612.0, + "step": 8031 + }, + { + "epoch": 0.8820557873929278, + "grad_norm": 2.202056646347046, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6965447664260864, + "num_tokens": 207836096.0, + "step": 8032 + }, + { + "epoch": 0.8821656050955414, + "grad_norm": 1.84405517578125, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.70355623960495, + "num_tokens": 207868083.0, + "step": 8033 + }, + { + "epoch": 0.8822754227981551, + "grad_norm": 2.212554693222046, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7215671539306641, + "num_tokens": 207892881.0, + "step": 8034 + }, + { + "epoch": 0.8823852405007687, + "grad_norm": 1.9296218156814575, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.695573091506958, + "num_tokens": 207924786.0, + "step": 8035 + }, + { + "epoch": 0.8824950582033824, + "grad_norm": 2.5476672649383545, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7226752042770386, + "num_tokens": 207946705.0, + "step": 8036 + }, + { + "epoch": 0.882604875905996, + "grad_norm": 2.2782130241394043, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7332320809364319, + "num_tokens": 207968344.0, + "step": 8037 + }, + { + "epoch": 0.8827146936086097, + "grad_norm": 2.287964344024658, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7118489146232605, + "num_tokens": 207991588.0, + "step": 8038 + }, + { + "epoch": 0.8828245113112234, + "grad_norm": 2.0650293827056885, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6962276697158813, + "num_tokens": 208020156.0, + "step": 8039 + }, + { + "epoch": 0.8829343290138371, + "grad_norm": 2.5638315677642822, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.712033748626709, + "num_tokens": 208039776.0, + "step": 8040 + }, + { + "epoch": 0.8830441467164507, + "grad_norm": 1.7925188541412354, + "learning_rate": 1e-06, + "loss": 1.0671, + "mean_token_accuracy": 0.6838158369064331, + "num_tokens": 208078399.0, + "step": 8041 + }, + { + "epoch": 0.8831539644190644, + "grad_norm": 2.4105310440063477, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7147550582885742, + "num_tokens": 208098993.0, + "step": 8042 + }, + { + "epoch": 0.883263782121678, + "grad_norm": 1.9387788772583008, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.708929181098938, + "num_tokens": 208132064.0, + "step": 8043 + }, + { + "epoch": 0.8833735998242916, + "grad_norm": 2.0927090644836426, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7312199473381042, + "num_tokens": 208158784.0, + "step": 8044 + }, + { + "epoch": 0.8834834175269053, + "grad_norm": 2.1654891967773438, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7039421796798706, + "num_tokens": 208184449.0, + "step": 8045 + }, + { + "epoch": 0.883593235229519, + "grad_norm": 2.4299652576446533, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7226141095161438, + "num_tokens": 208204624.0, + "step": 8046 + }, + { + "epoch": 0.8837030529321327, + "grad_norm": 2.4302656650543213, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7006898522377014, + "num_tokens": 208227598.0, + "step": 8047 + }, + { + "epoch": 0.8838128706347463, + "grad_norm": 2.35756254196167, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7276909351348877, + "num_tokens": 208251223.0, + "step": 8048 + }, + { + "epoch": 0.88392268833736, + "grad_norm": 2.5331103801727295, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7030031085014343, + "num_tokens": 208271542.0, + "step": 8049 + }, + { + "epoch": 0.8840325060399736, + "grad_norm": 2.102571964263916, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.702094554901123, + "num_tokens": 208300464.0, + "step": 8050 + }, + { + "epoch": 0.8841423237425873, + "grad_norm": 2.4700512886047363, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7175444960594177, + "num_tokens": 208320189.0, + "step": 8051 + }, + { + "epoch": 0.8842521414452009, + "grad_norm": 1.9176160097122192, + "learning_rate": 1e-06, + "loss": 1.0643, + "mean_token_accuracy": 0.691569447517395, + "num_tokens": 208355152.0, + "step": 8052 + }, + { + "epoch": 0.8843619591478147, + "grad_norm": 2.0229663848876953, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6888449192047119, + "num_tokens": 208383984.0, + "step": 8053 + }, + { + "epoch": 0.8844717768504283, + "grad_norm": 2.074399709701538, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7164567708969116, + "num_tokens": 208410915.0, + "step": 8054 + }, + { + "epoch": 0.884581594553042, + "grad_norm": 2.141284465789795, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7459815144538879, + "num_tokens": 208436148.0, + "step": 8055 + }, + { + "epoch": 0.8846914122556556, + "grad_norm": 2.13895320892334, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7110744714736938, + "num_tokens": 208465224.0, + "step": 8056 + }, + { + "epoch": 0.8848012299582693, + "grad_norm": 1.9053723812103271, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7345536947250366, + "num_tokens": 208494111.0, + "step": 8057 + }, + { + "epoch": 0.8849110476608829, + "grad_norm": 2.340374231338501, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7269583344459534, + "num_tokens": 208514390.0, + "step": 8058 + }, + { + "epoch": 0.8850208653634966, + "grad_norm": 1.9876230955123901, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6993352770805359, + "num_tokens": 208543421.0, + "step": 8059 + }, + { + "epoch": 0.8851306830661102, + "grad_norm": 2.116687536239624, + "learning_rate": 1e-06, + "loss": 1.0513, + "mean_token_accuracy": 0.6876469850540161, + "num_tokens": 208570744.0, + "step": 8060 + }, + { + "epoch": 0.885240500768724, + "grad_norm": 2.1575825214385986, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7186885476112366, + "num_tokens": 208596170.0, + "step": 8061 + }, + { + "epoch": 0.8853503184713376, + "grad_norm": 2.080714702606201, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6978182792663574, + "num_tokens": 208625206.0, + "step": 8062 + }, + { + "epoch": 0.8854601361739513, + "grad_norm": 2.4054763317108154, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.726388156414032, + "num_tokens": 208645910.0, + "step": 8063 + }, + { + "epoch": 0.8855699538765649, + "grad_norm": 2.0454089641571045, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7073225378990173, + "num_tokens": 208672805.0, + "step": 8064 + }, + { + "epoch": 0.8856797715791785, + "grad_norm": 1.981127381324768, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7062845826148987, + "num_tokens": 208703429.0, + "step": 8065 + }, + { + "epoch": 0.8857895892817922, + "grad_norm": 2.139350175857544, + "learning_rate": 1e-06, + "loss": 1.0609, + "mean_token_accuracy": 0.6834040880203247, + "num_tokens": 208732360.0, + "step": 8066 + }, + { + "epoch": 0.8858994069844058, + "grad_norm": 2.163205146789551, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.7019343376159668, + "num_tokens": 208759178.0, + "step": 8067 + }, + { + "epoch": 0.8860092246870196, + "grad_norm": 2.525941848754883, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7163751125335693, + "num_tokens": 208778917.0, + "step": 8068 + }, + { + "epoch": 0.8861190423896332, + "grad_norm": 2.0223255157470703, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7114641070365906, + "num_tokens": 208808384.0, + "step": 8069 + }, + { + "epoch": 0.8862288600922469, + "grad_norm": 2.44680118560791, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.714413046836853, + "num_tokens": 208829418.0, + "step": 8070 + }, + { + "epoch": 0.8863386777948605, + "grad_norm": 2.07711124420166, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.7043255567550659, + "num_tokens": 208857521.0, + "step": 8071 + }, + { + "epoch": 0.8864484954974742, + "grad_norm": 2.2234106063842773, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6969624161720276, + "num_tokens": 208881906.0, + "step": 8072 + }, + { + "epoch": 0.8865583132000878, + "grad_norm": 1.837926983833313, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7345104813575745, + "num_tokens": 208914194.0, + "step": 8073 + }, + { + "epoch": 0.8866681309027015, + "grad_norm": 2.247861385345459, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.6851593255996704, + "num_tokens": 208939704.0, + "step": 8074 + }, + { + "epoch": 0.8867779486053152, + "grad_norm": 2.1580803394317627, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7315274477005005, + "num_tokens": 208963714.0, + "step": 8075 + }, + { + "epoch": 0.8868877663079289, + "grad_norm": 2.1146979331970215, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.690793514251709, + "num_tokens": 208992718.0, + "step": 8076 + }, + { + "epoch": 0.8869975840105425, + "grad_norm": 2.014312267303467, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7180676460266113, + "num_tokens": 209021300.0, + "step": 8077 + }, + { + "epoch": 0.8871074017131562, + "grad_norm": 2.1574196815490723, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.6991660594940186, + "num_tokens": 209048287.0, + "step": 8078 + }, + { + "epoch": 0.8872172194157698, + "grad_norm": 2.459961175918579, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7060164213180542, + "num_tokens": 209069545.0, + "step": 8079 + }, + { + "epoch": 0.8873270371183835, + "grad_norm": 2.114121437072754, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7029777765274048, + "num_tokens": 209095028.0, + "step": 8080 + }, + { + "epoch": 0.8874368548209971, + "grad_norm": 2.2154996395111084, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7240757346153259, + "num_tokens": 209118193.0, + "step": 8081 + }, + { + "epoch": 0.8875466725236109, + "grad_norm": 2.0027475357055664, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7079195976257324, + "num_tokens": 209146980.0, + "step": 8082 + }, + { + "epoch": 0.8876564902262245, + "grad_norm": 2.0641345977783203, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7011422514915466, + "num_tokens": 209174812.0, + "step": 8083 + }, + { + "epoch": 0.8877663079288381, + "grad_norm": 2.2021002769470215, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7207695245742798, + "num_tokens": 209200809.0, + "step": 8084 + }, + { + "epoch": 0.8878761256314518, + "grad_norm": 2.1170244216918945, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7314106225967407, + "num_tokens": 209225031.0, + "step": 8085 + }, + { + "epoch": 0.8879859433340654, + "grad_norm": 2.166835308074951, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6804795265197754, + "num_tokens": 209253581.0, + "step": 8086 + }, + { + "epoch": 0.8880957610366791, + "grad_norm": 2.1027791500091553, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6847978830337524, + "num_tokens": 209283539.0, + "step": 8087 + }, + { + "epoch": 0.8882055787392927, + "grad_norm": 1.964232325553894, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.689587414264679, + "num_tokens": 209313896.0, + "step": 8088 + }, + { + "epoch": 0.8883153964419065, + "grad_norm": 2.0506415367126465, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7228372097015381, + "num_tokens": 209338742.0, + "step": 8089 + }, + { + "epoch": 0.8884252141445201, + "grad_norm": 2.445403575897217, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7134466171264648, + "num_tokens": 209358816.0, + "step": 8090 + }, + { + "epoch": 0.8885350318471338, + "grad_norm": 2.1907780170440674, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7124037742614746, + "num_tokens": 209383834.0, + "step": 8091 + }, + { + "epoch": 0.8886448495497474, + "grad_norm": 2.120854616165161, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6964498162269592, + "num_tokens": 209410554.0, + "step": 8092 + }, + { + "epoch": 0.8887546672523611, + "grad_norm": 2.4800617694854736, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7167210578918457, + "num_tokens": 209430014.0, + "step": 8093 + }, + { + "epoch": 0.8888644849549747, + "grad_norm": 2.357206106185913, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7089890241622925, + "num_tokens": 209450901.0, + "step": 8094 + }, + { + "epoch": 0.8889743026575884, + "grad_norm": 2.0430097579956055, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6995928287506104, + "num_tokens": 209480423.0, + "step": 8095 + }, + { + "epoch": 0.889084120360202, + "grad_norm": 2.0024871826171875, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7188581228256226, + "num_tokens": 209509147.0, + "step": 8096 + }, + { + "epoch": 0.8891939380628158, + "grad_norm": 1.9174401760101318, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7144430875778198, + "num_tokens": 209539620.0, + "step": 8097 + }, + { + "epoch": 0.8893037557654294, + "grad_norm": 2.3362793922424316, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7321146726608276, + "num_tokens": 209560177.0, + "step": 8098 + }, + { + "epoch": 0.8894135734680431, + "grad_norm": 2.1565403938293457, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7154273390769958, + "num_tokens": 209585869.0, + "step": 8099 + }, + { + "epoch": 0.8895233911706567, + "grad_norm": 2.4051806926727295, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6982929706573486, + "num_tokens": 209607506.0, + "step": 8100 + }, + { + "epoch": 0.8896332088732704, + "grad_norm": 2.2039451599121094, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7038149833679199, + "num_tokens": 209634975.0, + "step": 8101 + }, + { + "epoch": 0.889743026575884, + "grad_norm": 2.1430466175079346, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7098614573478699, + "num_tokens": 209661090.0, + "step": 8102 + }, + { + "epoch": 0.8898528442784976, + "grad_norm": 2.1344826221466064, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7038822174072266, + "num_tokens": 209687591.0, + "step": 8103 + }, + { + "epoch": 0.8899626619811114, + "grad_norm": 2.369381904602051, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.718338131904602, + "num_tokens": 209709367.0, + "step": 8104 + }, + { + "epoch": 0.890072479683725, + "grad_norm": 1.9445743560791016, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7089446187019348, + "num_tokens": 209740280.0, + "step": 8105 + }, + { + "epoch": 0.8901822973863387, + "grad_norm": 2.069214105606079, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.726531982421875, + "num_tokens": 209766365.0, + "step": 8106 + }, + { + "epoch": 0.8902921150889523, + "grad_norm": 2.1811933517456055, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7091568112373352, + "num_tokens": 209792007.0, + "step": 8107 + }, + { + "epoch": 0.890401932791566, + "grad_norm": 1.9334042072296143, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7088532447814941, + "num_tokens": 209820382.0, + "step": 8108 + }, + { + "epoch": 0.8905117504941796, + "grad_norm": 2.345867872238159, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7016226649284363, + "num_tokens": 209844582.0, + "step": 8109 + }, + { + "epoch": 0.8906215681967933, + "grad_norm": 1.9922025203704834, + "learning_rate": 1e-06, + "loss": 1.0895, + "mean_token_accuracy": 0.6720679998397827, + "num_tokens": 209876586.0, + "step": 8110 + }, + { + "epoch": 0.890731385899407, + "grad_norm": 2.256394386291504, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7209552526473999, + "num_tokens": 209901208.0, + "step": 8111 + }, + { + "epoch": 0.8908412036020207, + "grad_norm": 2.288607358932495, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7414270043373108, + "num_tokens": 209922889.0, + "step": 8112 + }, + { + "epoch": 0.8909510213046343, + "grad_norm": 2.1198811531066895, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7083967328071594, + "num_tokens": 209948901.0, + "step": 8113 + }, + { + "epoch": 0.891060839007248, + "grad_norm": 2.2646656036376953, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7078437805175781, + "num_tokens": 209972783.0, + "step": 8114 + }, + { + "epoch": 0.8911706567098616, + "grad_norm": 1.9700829982757568, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6800085306167603, + "num_tokens": 210003463.0, + "step": 8115 + }, + { + "epoch": 0.8912804744124753, + "grad_norm": 2.2792413234710693, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7123235464096069, + "num_tokens": 210026671.0, + "step": 8116 + }, + { + "epoch": 0.8913902921150889, + "grad_norm": 2.157766103744507, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.727996826171875, + "num_tokens": 210049798.0, + "step": 8117 + }, + { + "epoch": 0.8915001098177027, + "grad_norm": 2.1153290271759033, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6989272832870483, + "num_tokens": 210076020.0, + "step": 8118 + }, + { + "epoch": 0.8916099275203163, + "grad_norm": 1.9801143407821655, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6920528411865234, + "num_tokens": 210106761.0, + "step": 8119 + }, + { + "epoch": 0.89171974522293, + "grad_norm": 2.178213357925415, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7239615321159363, + "num_tokens": 210132344.0, + "step": 8120 + }, + { + "epoch": 0.8918295629255436, + "grad_norm": 2.189650058746338, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7062913179397583, + "num_tokens": 210159175.0, + "step": 8121 + }, + { + "epoch": 0.8919393806281573, + "grad_norm": 2.0680222511291504, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6916015148162842, + "num_tokens": 210189422.0, + "step": 8122 + }, + { + "epoch": 0.8920491983307709, + "grad_norm": 1.9876397848129272, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6884564757347107, + "num_tokens": 210220355.0, + "step": 8123 + }, + { + "epoch": 0.8921590160333845, + "grad_norm": 2.2328977584838867, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7155564427375793, + "num_tokens": 210243366.0, + "step": 8124 + }, + { + "epoch": 0.8922688337359982, + "grad_norm": 1.9997835159301758, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7020941972732544, + "num_tokens": 210273857.0, + "step": 8125 + }, + { + "epoch": 0.892378651438612, + "grad_norm": 2.3283042907714844, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.694501519203186, + "num_tokens": 210298244.0, + "step": 8126 + }, + { + "epoch": 0.8924884691412256, + "grad_norm": 2.071877956390381, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6855015754699707, + "num_tokens": 210329923.0, + "step": 8127 + }, + { + "epoch": 0.8925982868438392, + "grad_norm": 2.049734592437744, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7020986080169678, + "num_tokens": 210358538.0, + "step": 8128 + }, + { + "epoch": 0.8927081045464529, + "grad_norm": 2.006843090057373, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7090083360671997, + "num_tokens": 210386966.0, + "step": 8129 + }, + { + "epoch": 0.8928179222490665, + "grad_norm": 2.125490665435791, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6953287124633789, + "num_tokens": 210414246.0, + "step": 8130 + }, + { + "epoch": 0.8929277399516802, + "grad_norm": 2.3152482509613037, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6953765749931335, + "num_tokens": 210437784.0, + "step": 8131 + }, + { + "epoch": 0.8930375576542938, + "grad_norm": 2.14652943611145, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7014658451080322, + "num_tokens": 210463775.0, + "step": 8132 + }, + { + "epoch": 0.8931473753569076, + "grad_norm": 2.059718370437622, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6956595778465271, + "num_tokens": 210492247.0, + "step": 8133 + }, + { + "epoch": 0.8932571930595212, + "grad_norm": 2.272501230239868, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7003465294837952, + "num_tokens": 210516236.0, + "step": 8134 + }, + { + "epoch": 0.8933670107621349, + "grad_norm": 2.2966291904449463, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7244288921356201, + "num_tokens": 210539167.0, + "step": 8135 + }, + { + "epoch": 0.8934768284647485, + "grad_norm": 2.278517246246338, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7172675132751465, + "num_tokens": 210562684.0, + "step": 8136 + }, + { + "epoch": 0.8935866461673622, + "grad_norm": 1.8969130516052246, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7219330668449402, + "num_tokens": 210596516.0, + "step": 8137 + }, + { + "epoch": 0.8936964638699758, + "grad_norm": 2.100226640701294, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7234683036804199, + "num_tokens": 210623295.0, + "step": 8138 + }, + { + "epoch": 0.8938062815725895, + "grad_norm": 2.3992068767547607, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7252872586250305, + "num_tokens": 210643221.0, + "step": 8139 + }, + { + "epoch": 0.8939160992752032, + "grad_norm": 2.1418354511260986, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6985665559768677, + "num_tokens": 210670898.0, + "step": 8140 + }, + { + "epoch": 0.8940259169778169, + "grad_norm": 2.390815496444702, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7072720527648926, + "num_tokens": 210694162.0, + "step": 8141 + }, + { + "epoch": 0.8941357346804305, + "grad_norm": 2.158353090286255, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6969704627990723, + "num_tokens": 210719930.0, + "step": 8142 + }, + { + "epoch": 0.8942455523830442, + "grad_norm": 2.293684959411621, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6895020008087158, + "num_tokens": 210747367.0, + "step": 8143 + }, + { + "epoch": 0.8943553700856578, + "grad_norm": 2.187321901321411, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7093358039855957, + "num_tokens": 210773069.0, + "step": 8144 + }, + { + "epoch": 0.8944651877882714, + "grad_norm": 2.0840556621551514, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7130606174468994, + "num_tokens": 210801344.0, + "step": 8145 + }, + { + "epoch": 0.8945750054908851, + "grad_norm": 1.8496021032333374, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7174540758132935, + "num_tokens": 210835548.0, + "step": 8146 + }, + { + "epoch": 0.8946848231934988, + "grad_norm": 2.286823034286499, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7268452048301697, + "num_tokens": 210857281.0, + "step": 8147 + }, + { + "epoch": 0.8947946408961125, + "grad_norm": 2.1299421787261963, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7086937427520752, + "num_tokens": 210884292.0, + "step": 8148 + }, + { + "epoch": 0.8949044585987261, + "grad_norm": 2.213083028793335, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7086108326911926, + "num_tokens": 210908448.0, + "step": 8149 + }, + { + "epoch": 0.8950142763013398, + "grad_norm": 2.327451705932617, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6930859088897705, + "num_tokens": 210932876.0, + "step": 8150 + }, + { + "epoch": 0.8951240940039534, + "grad_norm": 1.9097295999526978, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7153490781784058, + "num_tokens": 210963703.0, + "step": 8151 + }, + { + "epoch": 0.8952339117065671, + "grad_norm": 2.1733286380767822, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7109230160713196, + "num_tokens": 210990458.0, + "step": 8152 + }, + { + "epoch": 0.8953437294091807, + "grad_norm": 1.9301140308380127, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6956554651260376, + "num_tokens": 211021700.0, + "step": 8153 + }, + { + "epoch": 0.8954535471117944, + "grad_norm": 2.340150833129883, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7057753801345825, + "num_tokens": 211044523.0, + "step": 8154 + }, + { + "epoch": 0.8955633648144081, + "grad_norm": 2.187556028366089, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7126188278198242, + "num_tokens": 211067809.0, + "step": 8155 + }, + { + "epoch": 0.8956731825170218, + "grad_norm": 2.1954877376556396, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7047702074050903, + "num_tokens": 211093559.0, + "step": 8156 + }, + { + "epoch": 0.8957830002196354, + "grad_norm": 2.1836578845977783, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7208726406097412, + "num_tokens": 211120767.0, + "step": 8157 + }, + { + "epoch": 0.8958928179222491, + "grad_norm": 2.1842172145843506, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.717137336730957, + "num_tokens": 211144996.0, + "step": 8158 + }, + { + "epoch": 0.8960026356248627, + "grad_norm": 1.9775205850601196, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7178151607513428, + "num_tokens": 211175424.0, + "step": 8159 + }, + { + "epoch": 0.8961124533274764, + "grad_norm": 1.9707139730453491, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7050655484199524, + "num_tokens": 211204392.0, + "step": 8160 + }, + { + "epoch": 0.89622227103009, + "grad_norm": 2.2395265102386475, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7038783431053162, + "num_tokens": 211229550.0, + "step": 8161 + }, + { + "epoch": 0.8963320887327038, + "grad_norm": 2.1492044925689697, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7220295667648315, + "num_tokens": 211255545.0, + "step": 8162 + }, + { + "epoch": 0.8964419064353174, + "grad_norm": 2.0097808837890625, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7006880044937134, + "num_tokens": 211283602.0, + "step": 8163 + }, + { + "epoch": 0.896551724137931, + "grad_norm": 2.533937692642212, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7082410454750061, + "num_tokens": 211302935.0, + "step": 8164 + }, + { + "epoch": 0.8966615418405447, + "grad_norm": 2.6484310626983643, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.708325207233429, + "num_tokens": 211321835.0, + "step": 8165 + }, + { + "epoch": 0.8967713595431583, + "grad_norm": 2.0148537158966064, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7010492086410522, + "num_tokens": 211351891.0, + "step": 8166 + }, + { + "epoch": 0.896881177245772, + "grad_norm": 2.367406129837036, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7014462351799011, + "num_tokens": 211372892.0, + "step": 8167 + }, + { + "epoch": 0.8969909949483856, + "grad_norm": 2.235459327697754, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7191336154937744, + "num_tokens": 211394671.0, + "step": 8168 + }, + { + "epoch": 0.8971008126509994, + "grad_norm": 1.995661973953247, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7154476046562195, + "num_tokens": 211422879.0, + "step": 8169 + }, + { + "epoch": 0.897210630353613, + "grad_norm": 2.015078067779541, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7047725319862366, + "num_tokens": 211450565.0, + "step": 8170 + }, + { + "epoch": 0.8973204480562267, + "grad_norm": 2.371649742126465, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.704323410987854, + "num_tokens": 211471829.0, + "step": 8171 + }, + { + "epoch": 0.8974302657588403, + "grad_norm": 2.3096609115600586, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7257947325706482, + "num_tokens": 211495522.0, + "step": 8172 + }, + { + "epoch": 0.897540083461454, + "grad_norm": 2.1132633686065674, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7147631049156189, + "num_tokens": 211522860.0, + "step": 8173 + }, + { + "epoch": 0.8976499011640676, + "grad_norm": 2.177800178527832, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6974934339523315, + "num_tokens": 211548583.0, + "step": 8174 + }, + { + "epoch": 0.8977597188666813, + "grad_norm": 2.1801741123199463, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7084981203079224, + "num_tokens": 211572444.0, + "step": 8175 + }, + { + "epoch": 0.897869536569295, + "grad_norm": 2.3920340538024902, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.725590705871582, + "num_tokens": 211594694.0, + "step": 8176 + }, + { + "epoch": 0.8979793542719087, + "grad_norm": 2.517054796218872, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7061444520950317, + "num_tokens": 211614529.0, + "step": 8177 + }, + { + "epoch": 0.8980891719745223, + "grad_norm": 2.1885182857513428, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7155019044876099, + "num_tokens": 211640705.0, + "step": 8178 + }, + { + "epoch": 0.898198989677136, + "grad_norm": 2.2774438858032227, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7151232957839966, + "num_tokens": 211664393.0, + "step": 8179 + }, + { + "epoch": 0.8983088073797496, + "grad_norm": 2.115560531616211, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7164928317070007, + "num_tokens": 211689982.0, + "step": 8180 + }, + { + "epoch": 0.8984186250823633, + "grad_norm": 2.5684187412261963, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7131080031394958, + "num_tokens": 211708693.0, + "step": 8181 + }, + { + "epoch": 0.8985284427849769, + "grad_norm": 2.311966896057129, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6966173052787781, + "num_tokens": 211732231.0, + "step": 8182 + }, + { + "epoch": 0.8986382604875905, + "grad_norm": 2.0024642944335938, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7184278964996338, + "num_tokens": 211760191.0, + "step": 8183 + }, + { + "epoch": 0.8987480781902043, + "grad_norm": 2.144253730773926, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6979502439498901, + "num_tokens": 211786463.0, + "step": 8184 + }, + { + "epoch": 0.898857895892818, + "grad_norm": 2.185439348220825, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7135612368583679, + "num_tokens": 211812926.0, + "step": 8185 + }, + { + "epoch": 0.8989677135954316, + "grad_norm": 2.275298833847046, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7321590185165405, + "num_tokens": 211834368.0, + "step": 8186 + }, + { + "epoch": 0.8990775312980452, + "grad_norm": 2.5815775394439697, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7168057560920715, + "num_tokens": 211853863.0, + "step": 8187 + }, + { + "epoch": 0.8991873490006589, + "grad_norm": 2.302248477935791, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7001345157623291, + "num_tokens": 211877620.0, + "step": 8188 + }, + { + "epoch": 0.8992971667032725, + "grad_norm": 2.032858371734619, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6955116987228394, + "num_tokens": 211908189.0, + "step": 8189 + }, + { + "epoch": 0.8994069844058862, + "grad_norm": 2.1832261085510254, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.6991056799888611, + "num_tokens": 211935489.0, + "step": 8190 + }, + { + "epoch": 0.8995168021084999, + "grad_norm": 2.3979034423828125, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7140139937400818, + "num_tokens": 211958611.0, + "step": 8191 + }, + { + "epoch": 0.8996266198111136, + "grad_norm": 2.104170560836792, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7090454697608948, + "num_tokens": 211983462.0, + "step": 8192 + }, + { + "epoch": 0.8997364375137272, + "grad_norm": 2.0066449642181396, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7267102003097534, + "num_tokens": 212011557.0, + "step": 8193 + }, + { + "epoch": 0.8998462552163409, + "grad_norm": 1.9269092082977295, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7268632650375366, + "num_tokens": 212042302.0, + "step": 8194 + }, + { + "epoch": 0.8999560729189545, + "grad_norm": 2.3211560249328613, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7063695192337036, + "num_tokens": 212065357.0, + "step": 8195 + }, + { + "epoch": 0.9000658906215682, + "grad_norm": 2.1588335037231445, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.708387017250061, + "num_tokens": 212092573.0, + "step": 8196 + }, + { + "epoch": 0.9001757083241818, + "grad_norm": 2.069136619567871, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6812113523483276, + "num_tokens": 212120452.0, + "step": 8197 + }, + { + "epoch": 0.9002855260267956, + "grad_norm": 2.1178348064422607, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7194710969924927, + "num_tokens": 212146205.0, + "step": 8198 + }, + { + "epoch": 0.9003953437294092, + "grad_norm": 2.256596565246582, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7195600867271423, + "num_tokens": 212170022.0, + "step": 8199 + }, + { + "epoch": 0.9005051614320229, + "grad_norm": 2.089550733566284, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7052129507064819, + "num_tokens": 212197763.0, + "step": 8200 + }, + { + "epoch": 0.9006149791346365, + "grad_norm": 2.275041341781616, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6994742751121521, + "num_tokens": 212223065.0, + "step": 8201 + }, + { + "epoch": 0.9007247968372502, + "grad_norm": 2.306180953979492, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7080285549163818, + "num_tokens": 212245488.0, + "step": 8202 + }, + { + "epoch": 0.9008346145398638, + "grad_norm": 2.0852231979370117, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6970101594924927, + "num_tokens": 212274724.0, + "step": 8203 + }, + { + "epoch": 0.9009444322424774, + "grad_norm": 2.200881242752075, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7357311248779297, + "num_tokens": 212296850.0, + "step": 8204 + }, + { + "epoch": 0.9010542499450912, + "grad_norm": 2.0941898822784424, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6857203841209412, + "num_tokens": 212324861.0, + "step": 8205 + }, + { + "epoch": 0.9011640676477048, + "grad_norm": 2.247706651687622, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6966074705123901, + "num_tokens": 212350599.0, + "step": 8206 + }, + { + "epoch": 0.9012738853503185, + "grad_norm": 2.3080687522888184, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7148757576942444, + "num_tokens": 212374498.0, + "step": 8207 + }, + { + "epoch": 0.9013837030529321, + "grad_norm": 2.2746427059173584, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6926832795143127, + "num_tokens": 212400734.0, + "step": 8208 + }, + { + "epoch": 0.9014935207555458, + "grad_norm": 2.067690849304199, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7282934188842773, + "num_tokens": 212428540.0, + "step": 8209 + }, + { + "epoch": 0.9016033384581594, + "grad_norm": 2.3697357177734375, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7225662469863892, + "num_tokens": 212450911.0, + "step": 8210 + }, + { + "epoch": 0.9017131561607731, + "grad_norm": 2.314718246459961, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7342160940170288, + "num_tokens": 212473053.0, + "step": 8211 + }, + { + "epoch": 0.9018229738633867, + "grad_norm": 2.2120490074157715, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7000914812088013, + "num_tokens": 212498830.0, + "step": 8212 + }, + { + "epoch": 0.9019327915660005, + "grad_norm": 1.9085023403167725, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6983019709587097, + "num_tokens": 212530401.0, + "step": 8213 + }, + { + "epoch": 0.9020426092686141, + "grad_norm": 2.1000008583068848, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6902588605880737, + "num_tokens": 212558520.0, + "step": 8214 + }, + { + "epoch": 0.9021524269712278, + "grad_norm": 2.0954136848449707, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6941218972206116, + "num_tokens": 212586735.0, + "step": 8215 + }, + { + "epoch": 0.9022622446738414, + "grad_norm": 2.0873665809631348, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.698827862739563, + "num_tokens": 212613134.0, + "step": 8216 + }, + { + "epoch": 0.9023720623764551, + "grad_norm": 2.1705613136291504, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.7009292840957642, + "num_tokens": 212638391.0, + "step": 8217 + }, + { + "epoch": 0.9024818800790687, + "grad_norm": 2.2366113662719727, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7193182706832886, + "num_tokens": 212663725.0, + "step": 8218 + }, + { + "epoch": 0.9025916977816824, + "grad_norm": 2.1701266765594482, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.711069643497467, + "num_tokens": 212693559.0, + "step": 8219 + }, + { + "epoch": 0.9027015154842961, + "grad_norm": 2.248938798904419, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7108298540115356, + "num_tokens": 212716262.0, + "step": 8220 + }, + { + "epoch": 0.9028113331869098, + "grad_norm": 1.9731013774871826, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7165130376815796, + "num_tokens": 212747179.0, + "step": 8221 + }, + { + "epoch": 0.9029211508895234, + "grad_norm": 2.075618267059326, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7114320397377014, + "num_tokens": 212774828.0, + "step": 8222 + }, + { + "epoch": 0.903030968592137, + "grad_norm": 1.9609769582748413, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6934698224067688, + "num_tokens": 212808125.0, + "step": 8223 + }, + { + "epoch": 0.9031407862947507, + "grad_norm": 2.1734931468963623, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6872159838676453, + "num_tokens": 212834200.0, + "step": 8224 + }, + { + "epoch": 0.9032506039973643, + "grad_norm": 2.1407477855682373, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6843544244766235, + "num_tokens": 212861783.0, + "step": 8225 + }, + { + "epoch": 0.903360421699978, + "grad_norm": 1.9442672729492188, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6926445960998535, + "num_tokens": 212893688.0, + "step": 8226 + }, + { + "epoch": 0.9034702394025917, + "grad_norm": 1.9917314052581787, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.710436999797821, + "num_tokens": 212923304.0, + "step": 8227 + }, + { + "epoch": 0.9035800571052054, + "grad_norm": 2.0188965797424316, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6878833770751953, + "num_tokens": 212955307.0, + "step": 8228 + }, + { + "epoch": 0.903689874807819, + "grad_norm": 2.2665083408355713, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7110607028007507, + "num_tokens": 212976112.0, + "step": 8229 + }, + { + "epoch": 0.9037996925104327, + "grad_norm": 2.5804519653320312, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7166422605514526, + "num_tokens": 212995197.0, + "step": 8230 + }, + { + "epoch": 0.9039095102130463, + "grad_norm": 2.101996898651123, + "learning_rate": 1e-06, + "loss": 1.0918, + "mean_token_accuracy": 0.6768366098403931, + "num_tokens": 213024273.0, + "step": 8231 + }, + { + "epoch": 0.90401932791566, + "grad_norm": 1.9319816827774048, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7170195579528809, + "num_tokens": 213053233.0, + "step": 8232 + }, + { + "epoch": 0.9041291456182736, + "grad_norm": 2.4095394611358643, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7286695837974548, + "num_tokens": 213074531.0, + "step": 8233 + }, + { + "epoch": 0.9042389633208874, + "grad_norm": 2.049222946166992, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7049383521080017, + "num_tokens": 213103128.0, + "step": 8234 + }, + { + "epoch": 0.904348781023501, + "grad_norm": 2.3300983905792236, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.715693473815918, + "num_tokens": 213124284.0, + "step": 8235 + }, + { + "epoch": 0.9044585987261147, + "grad_norm": 1.955196499824524, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7400112748146057, + "num_tokens": 213153427.0, + "step": 8236 + }, + { + "epoch": 0.9045684164287283, + "grad_norm": 2.08378529548645, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6978585720062256, + "num_tokens": 213180649.0, + "step": 8237 + }, + { + "epoch": 0.904678234131342, + "grad_norm": 1.9318815469741821, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6951062679290771, + "num_tokens": 213212215.0, + "step": 8238 + }, + { + "epoch": 0.9047880518339556, + "grad_norm": 1.9889469146728516, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6909804344177246, + "num_tokens": 213242563.0, + "step": 8239 + }, + { + "epoch": 0.9048978695365693, + "grad_norm": 2.1271133422851562, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6884111762046814, + "num_tokens": 213271046.0, + "step": 8240 + }, + { + "epoch": 0.9050076872391829, + "grad_norm": 1.965545892715454, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7217307686805725, + "num_tokens": 213301255.0, + "step": 8241 + }, + { + "epoch": 0.9051175049417967, + "grad_norm": 2.312913417816162, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7277272343635559, + "num_tokens": 213324833.0, + "step": 8242 + }, + { + "epoch": 0.9052273226444103, + "grad_norm": 2.061473846435547, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6946706771850586, + "num_tokens": 213351024.0, + "step": 8243 + }, + { + "epoch": 0.905337140347024, + "grad_norm": 2.064228057861328, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6903457641601562, + "num_tokens": 213379797.0, + "step": 8244 + }, + { + "epoch": 0.9054469580496376, + "grad_norm": 2.034674644470215, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6888542175292969, + "num_tokens": 213409556.0, + "step": 8245 + }, + { + "epoch": 0.9055567757522512, + "grad_norm": 2.157379388809204, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7101362943649292, + "num_tokens": 213433519.0, + "step": 8246 + }, + { + "epoch": 0.9056665934548649, + "grad_norm": 2.0370445251464844, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6937525272369385, + "num_tokens": 213462154.0, + "step": 8247 + }, + { + "epoch": 0.9057764111574785, + "grad_norm": 2.2040979862213135, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7115024924278259, + "num_tokens": 213485521.0, + "step": 8248 + }, + { + "epoch": 0.9058862288600923, + "grad_norm": 2.0512683391571045, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7025271654129028, + "num_tokens": 213514699.0, + "step": 8249 + }, + { + "epoch": 0.9059960465627059, + "grad_norm": 2.220686912536621, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6963827610015869, + "num_tokens": 213539877.0, + "step": 8250 + }, + { + "epoch": 0.9061058642653196, + "grad_norm": 2.117786169052124, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6985419392585754, + "num_tokens": 213567267.0, + "step": 8251 + }, + { + "epoch": 0.9062156819679332, + "grad_norm": 2.145475387573242, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.719914436340332, + "num_tokens": 213593174.0, + "step": 8252 + }, + { + "epoch": 0.9063254996705469, + "grad_norm": 2.1619622707366943, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7042723894119263, + "num_tokens": 213619039.0, + "step": 8253 + }, + { + "epoch": 0.9064353173731605, + "grad_norm": 2.2079739570617676, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7179511785507202, + "num_tokens": 213644412.0, + "step": 8254 + }, + { + "epoch": 0.9065451350757742, + "grad_norm": 2.2553870677948, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7142535448074341, + "num_tokens": 213668556.0, + "step": 8255 + }, + { + "epoch": 0.9066549527783879, + "grad_norm": 1.9626424312591553, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7456986904144287, + "num_tokens": 213694420.0, + "step": 8256 + }, + { + "epoch": 0.9067647704810016, + "grad_norm": 2.229835033416748, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6908683180809021, + "num_tokens": 213721246.0, + "step": 8257 + }, + { + "epoch": 0.9068745881836152, + "grad_norm": 2.091019630432129, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7178760766983032, + "num_tokens": 213747046.0, + "step": 8258 + }, + { + "epoch": 0.9069844058862289, + "grad_norm": 2.2692415714263916, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7015550136566162, + "num_tokens": 213771105.0, + "step": 8259 + }, + { + "epoch": 0.9070942235888425, + "grad_norm": 2.162760019302368, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.72542405128479, + "num_tokens": 213795878.0, + "step": 8260 + }, + { + "epoch": 0.9072040412914562, + "grad_norm": 2.0448148250579834, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6942204236984253, + "num_tokens": 213824083.0, + "step": 8261 + }, + { + "epoch": 0.9073138589940698, + "grad_norm": 2.0659422874450684, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7185065150260925, + "num_tokens": 213851743.0, + "step": 8262 + }, + { + "epoch": 0.9074236766966836, + "grad_norm": 2.447955846786499, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7090834975242615, + "num_tokens": 213872726.0, + "step": 8263 + }, + { + "epoch": 0.9075334943992972, + "grad_norm": 1.9073255062103271, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.700482964515686, + "num_tokens": 213904422.0, + "step": 8264 + }, + { + "epoch": 0.9076433121019108, + "grad_norm": 2.221702814102173, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7098376750946045, + "num_tokens": 213929673.0, + "step": 8265 + }, + { + "epoch": 0.9077531298045245, + "grad_norm": 2.2507574558258057, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7157930731773376, + "num_tokens": 213952516.0, + "step": 8266 + }, + { + "epoch": 0.9078629475071381, + "grad_norm": 1.903501033782959, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.693192720413208, + "num_tokens": 213982810.0, + "step": 8267 + }, + { + "epoch": 0.9079727652097518, + "grad_norm": 2.0783920288085938, + "learning_rate": 1e-06, + "loss": 1.0496, + "mean_token_accuracy": 0.6862009763717651, + "num_tokens": 214010916.0, + "step": 8268 + }, + { + "epoch": 0.9080825829123654, + "grad_norm": 2.2818665504455566, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7049044370651245, + "num_tokens": 214035488.0, + "step": 8269 + }, + { + "epoch": 0.9081924006149792, + "grad_norm": 1.9795788526535034, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7239803075790405, + "num_tokens": 214064658.0, + "step": 8270 + }, + { + "epoch": 0.9083022183175928, + "grad_norm": 2.1597096920013428, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.703586220741272, + "num_tokens": 214090493.0, + "step": 8271 + }, + { + "epoch": 0.9084120360202065, + "grad_norm": 2.4731626510620117, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7230914831161499, + "num_tokens": 214109319.0, + "step": 8272 + }, + { + "epoch": 0.9085218537228201, + "grad_norm": 2.116245985031128, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7080072164535522, + "num_tokens": 214135566.0, + "step": 8273 + }, + { + "epoch": 0.9086316714254338, + "grad_norm": 2.2799062728881836, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7181337475776672, + "num_tokens": 214158638.0, + "step": 8274 + }, + { + "epoch": 0.9087414891280474, + "grad_norm": 2.0249102115631104, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6969708204269409, + "num_tokens": 214186129.0, + "step": 8275 + }, + { + "epoch": 0.9088513068306611, + "grad_norm": 2.016676187515259, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6982207894325256, + "num_tokens": 214216102.0, + "step": 8276 + }, + { + "epoch": 0.9089611245332747, + "grad_norm": 2.120375633239746, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7052711248397827, + "num_tokens": 214244267.0, + "step": 8277 + }, + { + "epoch": 0.9090709422358885, + "grad_norm": 1.9700357913970947, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6962413191795349, + "num_tokens": 214274438.0, + "step": 8278 + }, + { + "epoch": 0.9091807599385021, + "grad_norm": 1.909244179725647, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6950109004974365, + "num_tokens": 214306660.0, + "step": 8279 + }, + { + "epoch": 0.9092905776411158, + "grad_norm": 2.2295172214508057, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7449629306793213, + "num_tokens": 214331101.0, + "step": 8280 + }, + { + "epoch": 0.9094003953437294, + "grad_norm": 2.0916082859039307, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7215980291366577, + "num_tokens": 214358743.0, + "step": 8281 + }, + { + "epoch": 0.909510213046343, + "grad_norm": 2.138639211654663, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7006697654724121, + "num_tokens": 214386075.0, + "step": 8282 + }, + { + "epoch": 0.9096200307489567, + "grad_norm": 2.677347183227539, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7090119123458862, + "num_tokens": 214405293.0, + "step": 8283 + }, + { + "epoch": 0.9097298484515703, + "grad_norm": 2.0532400608062744, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.707631528377533, + "num_tokens": 214432685.0, + "step": 8284 + }, + { + "epoch": 0.9098396661541841, + "grad_norm": 2.1861796379089355, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7029446959495544, + "num_tokens": 214458723.0, + "step": 8285 + }, + { + "epoch": 0.9099494838567977, + "grad_norm": 2.0426175594329834, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7159380316734314, + "num_tokens": 214486020.0, + "step": 8286 + }, + { + "epoch": 0.9100593015594114, + "grad_norm": 2.3473496437072754, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.704332709312439, + "num_tokens": 214509254.0, + "step": 8287 + }, + { + "epoch": 0.910169119262025, + "grad_norm": 2.031663656234741, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6892900466918945, + "num_tokens": 214540063.0, + "step": 8288 + }, + { + "epoch": 0.9102789369646387, + "grad_norm": 2.22763991355896, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6980161070823669, + "num_tokens": 214565092.0, + "step": 8289 + }, + { + "epoch": 0.9103887546672523, + "grad_norm": 1.8595057725906372, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.705536961555481, + "num_tokens": 214598113.0, + "step": 8290 + }, + { + "epoch": 0.910498572369866, + "grad_norm": 2.2595808506011963, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7098528742790222, + "num_tokens": 214621759.0, + "step": 8291 + }, + { + "epoch": 0.9106083900724797, + "grad_norm": 2.197448253631592, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7222391366958618, + "num_tokens": 214646977.0, + "step": 8292 + }, + { + "epoch": 0.9107182077750934, + "grad_norm": 2.3285293579101562, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.736289918422699, + "num_tokens": 214667009.0, + "step": 8293 + }, + { + "epoch": 0.910828025477707, + "grad_norm": 2.294902801513672, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6952006220817566, + "num_tokens": 214690203.0, + "step": 8294 + }, + { + "epoch": 0.9109378431803207, + "grad_norm": 2.4208250045776367, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7289636135101318, + "num_tokens": 214711257.0, + "step": 8295 + }, + { + "epoch": 0.9110476608829343, + "grad_norm": 1.9622451066970825, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7059298753738403, + "num_tokens": 214740590.0, + "step": 8296 + }, + { + "epoch": 0.911157478585548, + "grad_norm": 1.9535918235778809, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7157022356987, + "num_tokens": 214772673.0, + "step": 8297 + }, + { + "epoch": 0.9112672962881616, + "grad_norm": 2.26316237449646, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6970186829566956, + "num_tokens": 214796315.0, + "step": 8298 + }, + { + "epoch": 0.9113771139907754, + "grad_norm": 2.2669715881347656, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6875721216201782, + "num_tokens": 214822372.0, + "step": 8299 + }, + { + "epoch": 0.911486931693389, + "grad_norm": 1.9188767671585083, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6937475204467773, + "num_tokens": 214855213.0, + "step": 8300 + }, + { + "epoch": 0.9115967493960027, + "grad_norm": 1.8232308626174927, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7059324979782104, + "num_tokens": 214890209.0, + "step": 8301 + }, + { + "epoch": 0.9117065670986163, + "grad_norm": 1.926050066947937, + "learning_rate": 1e-06, + "loss": 1.0563, + "mean_token_accuracy": 0.6868032217025757, + "num_tokens": 214924274.0, + "step": 8302 + }, + { + "epoch": 0.91181638480123, + "grad_norm": 2.342512607574463, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7009788751602173, + "num_tokens": 214946732.0, + "step": 8303 + }, + { + "epoch": 0.9119262025038436, + "grad_norm": 2.531890869140625, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7339901328086853, + "num_tokens": 214965979.0, + "step": 8304 + }, + { + "epoch": 0.9120360202064572, + "grad_norm": 2.3422369956970215, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7041347026824951, + "num_tokens": 214989147.0, + "step": 8305 + }, + { + "epoch": 0.9121458379090709, + "grad_norm": 2.321648120880127, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7150415778160095, + "num_tokens": 215012035.0, + "step": 8306 + }, + { + "epoch": 0.9122556556116846, + "grad_norm": 2.271878480911255, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7245626449584961, + "num_tokens": 215036677.0, + "step": 8307 + }, + { + "epoch": 0.9123654733142983, + "grad_norm": 2.3310470581054688, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.726674497127533, + "num_tokens": 215057133.0, + "step": 8308 + }, + { + "epoch": 0.9124752910169119, + "grad_norm": 2.1218924522399902, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6977574825286865, + "num_tokens": 215082831.0, + "step": 8309 + }, + { + "epoch": 0.9125851087195256, + "grad_norm": 2.1924078464508057, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7073125243186951, + "num_tokens": 215106983.0, + "step": 8310 + }, + { + "epoch": 0.9126949264221392, + "grad_norm": 2.186309814453125, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7216767072677612, + "num_tokens": 215131950.0, + "step": 8311 + }, + { + "epoch": 0.9128047441247529, + "grad_norm": 2.361316442489624, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7080650925636292, + "num_tokens": 215153394.0, + "step": 8312 + }, + { + "epoch": 0.9129145618273665, + "grad_norm": 2.0157036781311035, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6782879829406738, + "num_tokens": 215184587.0, + "step": 8313 + }, + { + "epoch": 0.9130243795299803, + "grad_norm": 2.1303889751434326, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7000752091407776, + "num_tokens": 215212894.0, + "step": 8314 + }, + { + "epoch": 0.9131341972325939, + "grad_norm": 2.1593360900878906, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6969195008277893, + "num_tokens": 215240789.0, + "step": 8315 + }, + { + "epoch": 0.9132440149352076, + "grad_norm": 2.2811007499694824, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6872072815895081, + "num_tokens": 215265364.0, + "step": 8316 + }, + { + "epoch": 0.9133538326378212, + "grad_norm": 2.510678768157959, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7080057859420776, + "num_tokens": 215285525.0, + "step": 8317 + }, + { + "epoch": 0.9134636503404349, + "grad_norm": 2.2752487659454346, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7208467125892639, + "num_tokens": 215308355.0, + "step": 8318 + }, + { + "epoch": 0.9135734680430485, + "grad_norm": 2.341409206390381, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6972332000732422, + "num_tokens": 215336239.0, + "step": 8319 + }, + { + "epoch": 0.9136832857456622, + "grad_norm": 2.367596387863159, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7085788249969482, + "num_tokens": 215360536.0, + "step": 8320 + }, + { + "epoch": 0.9137931034482759, + "grad_norm": 2.197850227355957, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7107583284378052, + "num_tokens": 215386149.0, + "step": 8321 + }, + { + "epoch": 0.9139029211508896, + "grad_norm": 2.370682954788208, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7221698760986328, + "num_tokens": 215407934.0, + "step": 8322 + }, + { + "epoch": 0.9140127388535032, + "grad_norm": 2.206218957901001, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7295438051223755, + "num_tokens": 215435377.0, + "step": 8323 + }, + { + "epoch": 0.9141225565561168, + "grad_norm": 2.3072969913482666, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7001900672912598, + "num_tokens": 215461261.0, + "step": 8324 + }, + { + "epoch": 0.9142323742587305, + "grad_norm": 2.304738759994507, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7248811721801758, + "num_tokens": 215483834.0, + "step": 8325 + }, + { + "epoch": 0.9143421919613441, + "grad_norm": 2.1549789905548096, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6983846426010132, + "num_tokens": 215508844.0, + "step": 8326 + }, + { + "epoch": 0.9144520096639578, + "grad_norm": 1.9722182750701904, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7316122055053711, + "num_tokens": 215536103.0, + "step": 8327 + }, + { + "epoch": 0.9145618273665715, + "grad_norm": 2.4862070083618164, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7301274538040161, + "num_tokens": 215555919.0, + "step": 8328 + }, + { + "epoch": 0.9146716450691852, + "grad_norm": 2.2670464515686035, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7074213027954102, + "num_tokens": 215580581.0, + "step": 8329 + }, + { + "epoch": 0.9147814627717988, + "grad_norm": 2.089646339416504, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7247555255889893, + "num_tokens": 215605680.0, + "step": 8330 + }, + { + "epoch": 0.9148912804744125, + "grad_norm": 2.020202398300171, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6931620836257935, + "num_tokens": 215634982.0, + "step": 8331 + }, + { + "epoch": 0.9150010981770261, + "grad_norm": 1.925525188446045, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6923599243164062, + "num_tokens": 215667249.0, + "step": 8332 + }, + { + "epoch": 0.9151109158796398, + "grad_norm": 2.2856831550598145, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7231075167655945, + "num_tokens": 215690494.0, + "step": 8333 + }, + { + "epoch": 0.9152207335822534, + "grad_norm": 2.100625991821289, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6982724666595459, + "num_tokens": 215716541.0, + "step": 8334 + }, + { + "epoch": 0.9153305512848671, + "grad_norm": 2.205457925796509, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7080321311950684, + "num_tokens": 215741018.0, + "step": 8335 + }, + { + "epoch": 0.9154403689874808, + "grad_norm": 2.2543435096740723, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7159496545791626, + "num_tokens": 215765566.0, + "step": 8336 + }, + { + "epoch": 0.9155501866900945, + "grad_norm": 2.3175973892211914, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7195426225662231, + "num_tokens": 215789434.0, + "step": 8337 + }, + { + "epoch": 0.9156600043927081, + "grad_norm": 2.0587592124938965, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7172555923461914, + "num_tokens": 215816457.0, + "step": 8338 + }, + { + "epoch": 0.9157698220953218, + "grad_norm": 2.3313329219818115, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7257852554321289, + "num_tokens": 215839046.0, + "step": 8339 + }, + { + "epoch": 0.9158796397979354, + "grad_norm": 2.1819937229156494, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7274409532546997, + "num_tokens": 215861389.0, + "step": 8340 + }, + { + "epoch": 0.915989457500549, + "grad_norm": 2.440918445587158, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7162061929702759, + "num_tokens": 215881662.0, + "step": 8341 + }, + { + "epoch": 0.9160992752031627, + "grad_norm": 2.430546760559082, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7243719100952148, + "num_tokens": 215901617.0, + "step": 8342 + }, + { + "epoch": 0.9162090929057765, + "grad_norm": 2.1719815731048584, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7151166200637817, + "num_tokens": 215927452.0, + "step": 8343 + }, + { + "epoch": 0.9163189106083901, + "grad_norm": 2.2067205905914307, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7150865197181702, + "num_tokens": 215953475.0, + "step": 8344 + }, + { + "epoch": 0.9164287283110037, + "grad_norm": 2.3302814960479736, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7026957273483276, + "num_tokens": 215978069.0, + "step": 8345 + }, + { + "epoch": 0.9165385460136174, + "grad_norm": 2.07021427154541, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7157609462738037, + "num_tokens": 216005963.0, + "step": 8346 + }, + { + "epoch": 0.916648363716231, + "grad_norm": 2.1431376934051514, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7162020206451416, + "num_tokens": 216032217.0, + "step": 8347 + }, + { + "epoch": 0.9167581814188447, + "grad_norm": 2.0668551921844482, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6821847558021545, + "num_tokens": 216061479.0, + "step": 8348 + }, + { + "epoch": 0.9168679991214583, + "grad_norm": 2.0218889713287354, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7064799666404724, + "num_tokens": 216090079.0, + "step": 8349 + }, + { + "epoch": 0.9169778168240721, + "grad_norm": 2.163472890853882, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6879382133483887, + "num_tokens": 216117217.0, + "step": 8350 + }, + { + "epoch": 0.9170876345266857, + "grad_norm": 2.071866750717163, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7256133556365967, + "num_tokens": 216144407.0, + "step": 8351 + }, + { + "epoch": 0.9171974522292994, + "grad_norm": 2.289463758468628, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7265570163726807, + "num_tokens": 216168373.0, + "step": 8352 + }, + { + "epoch": 0.917307269931913, + "grad_norm": 2.3186581134796143, + "learning_rate": 1e-06, + "loss": 1.0727, + "mean_token_accuracy": 0.6706082820892334, + "num_tokens": 216193251.0, + "step": 8353 + }, + { + "epoch": 0.9174170876345267, + "grad_norm": 2.454897880554199, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.726967453956604, + "num_tokens": 216213242.0, + "step": 8354 + }, + { + "epoch": 0.9175269053371403, + "grad_norm": 2.013712167739868, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7181057929992676, + "num_tokens": 216240426.0, + "step": 8355 + }, + { + "epoch": 0.917636723039754, + "grad_norm": 1.9867764711380005, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6982097029685974, + "num_tokens": 216270015.0, + "step": 8356 + }, + { + "epoch": 0.9177465407423677, + "grad_norm": 1.990235686302185, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7047539949417114, + "num_tokens": 216300090.0, + "step": 8357 + }, + { + "epoch": 0.9178563584449814, + "grad_norm": 1.9331556558609009, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7121814489364624, + "num_tokens": 216330103.0, + "step": 8358 + }, + { + "epoch": 0.917966176147595, + "grad_norm": 2.067135810852051, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7207484245300293, + "num_tokens": 216356619.0, + "step": 8359 + }, + { + "epoch": 0.9180759938502087, + "grad_norm": 2.3484888076782227, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7090783715248108, + "num_tokens": 216379171.0, + "step": 8360 + }, + { + "epoch": 0.9181858115528223, + "grad_norm": 1.9938260316848755, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6950109004974365, + "num_tokens": 216409387.0, + "step": 8361 + }, + { + "epoch": 0.918295629255436, + "grad_norm": 2.225994825363159, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7163450717926025, + "num_tokens": 216433635.0, + "step": 8362 + }, + { + "epoch": 0.9184054469580496, + "grad_norm": 1.9256397485733032, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6989311575889587, + "num_tokens": 216466485.0, + "step": 8363 + }, + { + "epoch": 0.9185152646606632, + "grad_norm": 2.2573587894439697, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7157028913497925, + "num_tokens": 216488907.0, + "step": 8364 + }, + { + "epoch": 0.918625082363277, + "grad_norm": 1.896246075630188, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6997179985046387, + "num_tokens": 216521259.0, + "step": 8365 + }, + { + "epoch": 0.9187349000658906, + "grad_norm": 2.6175122261047363, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7294645309448242, + "num_tokens": 216538864.0, + "step": 8366 + }, + { + "epoch": 0.9188447177685043, + "grad_norm": 2.024435043334961, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7080250978469849, + "num_tokens": 216568362.0, + "step": 8367 + }, + { + "epoch": 0.9189545354711179, + "grad_norm": 1.94144606590271, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7051669359207153, + "num_tokens": 216598722.0, + "step": 8368 + }, + { + "epoch": 0.9190643531737316, + "grad_norm": 2.198361873626709, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.698601484298706, + "num_tokens": 216623241.0, + "step": 8369 + }, + { + "epoch": 0.9191741708763452, + "grad_norm": 2.05004620552063, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7234338521957397, + "num_tokens": 216650065.0, + "step": 8370 + }, + { + "epoch": 0.9192839885789589, + "grad_norm": 2.1822898387908936, + "learning_rate": 1e-06, + "loss": 1.04, + "mean_token_accuracy": 0.6900880336761475, + "num_tokens": 216675984.0, + "step": 8371 + }, + { + "epoch": 0.9193938062815726, + "grad_norm": 1.9601500034332275, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7049440145492554, + "num_tokens": 216706335.0, + "step": 8372 + }, + { + "epoch": 0.9195036239841863, + "grad_norm": 1.8999511003494263, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6863082647323608, + "num_tokens": 216738678.0, + "step": 8373 + }, + { + "epoch": 0.9196134416867999, + "grad_norm": 1.9631383419036865, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7651147842407227, + "num_tokens": 216764265.0, + "step": 8374 + }, + { + "epoch": 0.9197232593894136, + "grad_norm": 2.2873525619506836, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7311750650405884, + "num_tokens": 216787019.0, + "step": 8375 + }, + { + "epoch": 0.9198330770920272, + "grad_norm": 2.1774864196777344, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7079672813415527, + "num_tokens": 216810780.0, + "step": 8376 + }, + { + "epoch": 0.9199428947946409, + "grad_norm": 2.3805196285247803, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6925478577613831, + "num_tokens": 216833727.0, + "step": 8377 + }, + { + "epoch": 0.9200527124972545, + "grad_norm": 2.1839346885681152, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6882680654525757, + "num_tokens": 216860498.0, + "step": 8378 + }, + { + "epoch": 0.9201625301998683, + "grad_norm": 2.5989818572998047, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7237825989723206, + "num_tokens": 216880677.0, + "step": 8379 + }, + { + "epoch": 0.9202723479024819, + "grad_norm": 2.260857343673706, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7144448757171631, + "num_tokens": 216903748.0, + "step": 8380 + }, + { + "epoch": 0.9203821656050956, + "grad_norm": 2.1399426460266113, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.690881609916687, + "num_tokens": 216935680.0, + "step": 8381 + }, + { + "epoch": 0.9204919833077092, + "grad_norm": 2.1557695865631104, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7087088823318481, + "num_tokens": 216960966.0, + "step": 8382 + }, + { + "epoch": 0.9206018010103229, + "grad_norm": 2.2241690158843994, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7159762978553772, + "num_tokens": 216984655.0, + "step": 8383 + }, + { + "epoch": 0.9207116187129365, + "grad_norm": 1.972260594367981, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7280230522155762, + "num_tokens": 217013859.0, + "step": 8384 + }, + { + "epoch": 0.9208214364155501, + "grad_norm": 2.317410945892334, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6923526525497437, + "num_tokens": 217039191.0, + "step": 8385 + }, + { + "epoch": 0.9209312541181639, + "grad_norm": 2.1776061058044434, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7050585746765137, + "num_tokens": 217065249.0, + "step": 8386 + }, + { + "epoch": 0.9210410718207775, + "grad_norm": 2.1680076122283936, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6998353004455566, + "num_tokens": 217092553.0, + "step": 8387 + }, + { + "epoch": 0.9211508895233912, + "grad_norm": 2.203413486480713, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7125951647758484, + "num_tokens": 217117661.0, + "step": 8388 + }, + { + "epoch": 0.9212607072260048, + "grad_norm": 2.1240453720092773, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6826602220535278, + "num_tokens": 217146021.0, + "step": 8389 + }, + { + "epoch": 0.9213705249286185, + "grad_norm": 2.19130539894104, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.697783350944519, + "num_tokens": 217171003.0, + "step": 8390 + }, + { + "epoch": 0.9214803426312321, + "grad_norm": 2.183622121810913, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7004488706588745, + "num_tokens": 217194716.0, + "step": 8391 + }, + { + "epoch": 0.9215901603338458, + "grad_norm": 2.177626132965088, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6936290264129639, + "num_tokens": 217220352.0, + "step": 8392 + }, + { + "epoch": 0.9216999780364594, + "grad_norm": 2.255406141281128, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7290289402008057, + "num_tokens": 217241491.0, + "step": 8393 + }, + { + "epoch": 0.9218097957390732, + "grad_norm": 2.3075242042541504, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.724061906337738, + "num_tokens": 217261562.0, + "step": 8394 + }, + { + "epoch": 0.9219196134416868, + "grad_norm": 1.8443713188171387, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7098609805107117, + "num_tokens": 217294792.0, + "step": 8395 + }, + { + "epoch": 0.9220294311443005, + "grad_norm": 2.0132529735565186, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7137452363967896, + "num_tokens": 217322645.0, + "step": 8396 + }, + { + "epoch": 0.9221392488469141, + "grad_norm": 1.8588695526123047, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7025719881057739, + "num_tokens": 217357087.0, + "step": 8397 + }, + { + "epoch": 0.9222490665495278, + "grad_norm": 2.2520525455474854, + "learning_rate": 1e-06, + "loss": 1.0961, + "mean_token_accuracy": 0.67401123046875, + "num_tokens": 217383181.0, + "step": 8398 + }, + { + "epoch": 0.9223588842521414, + "grad_norm": 2.663830041885376, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7140264511108398, + "num_tokens": 217401323.0, + "step": 8399 + }, + { + "epoch": 0.922468701954755, + "grad_norm": 1.9900907278060913, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7025920152664185, + "num_tokens": 217430911.0, + "step": 8400 + }, + { + "epoch": 0.9225785196573688, + "grad_norm": 1.9755669832229614, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6933863162994385, + "num_tokens": 217460849.0, + "step": 8401 + }, + { + "epoch": 0.9226883373599825, + "grad_norm": 2.108563184738159, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7166846394538879, + "num_tokens": 217487254.0, + "step": 8402 + }, + { + "epoch": 0.9227981550625961, + "grad_norm": 2.1125824451446533, + "learning_rate": 1e-06, + "loss": 1.0678, + "mean_token_accuracy": 0.6777656674385071, + "num_tokens": 217515029.0, + "step": 8403 + }, + { + "epoch": 0.9229079727652097, + "grad_norm": 2.2064425945281982, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7001495361328125, + "num_tokens": 217539472.0, + "step": 8404 + }, + { + "epoch": 0.9230177904678234, + "grad_norm": 2.110703229904175, + "learning_rate": 1e-06, + "loss": 1.0787, + "mean_token_accuracy": 0.6979705095291138, + "num_tokens": 217565277.0, + "step": 8405 + }, + { + "epoch": 0.923127608170437, + "grad_norm": 2.129169225692749, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7022290825843811, + "num_tokens": 217591244.0, + "step": 8406 + }, + { + "epoch": 0.9232374258730507, + "grad_norm": 2.011298418045044, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7070097923278809, + "num_tokens": 217619176.0, + "step": 8407 + }, + { + "epoch": 0.9233472435756644, + "grad_norm": 2.13926100730896, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7217274904251099, + "num_tokens": 217646909.0, + "step": 8408 + }, + { + "epoch": 0.9234570612782781, + "grad_norm": 2.4430384635925293, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7117798924446106, + "num_tokens": 217667989.0, + "step": 8409 + }, + { + "epoch": 0.9235668789808917, + "grad_norm": 2.0385236740112305, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7244992256164551, + "num_tokens": 217694095.0, + "step": 8410 + }, + { + "epoch": 0.9236766966835054, + "grad_norm": 2.031975507736206, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.685795247554779, + "num_tokens": 217723548.0, + "step": 8411 + }, + { + "epoch": 0.923786514386119, + "grad_norm": 2.2900595664978027, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7089581489562988, + "num_tokens": 217747359.0, + "step": 8412 + }, + { + "epoch": 0.9238963320887327, + "grad_norm": 2.2253735065460205, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7003588676452637, + "num_tokens": 217772721.0, + "step": 8413 + }, + { + "epoch": 0.9240061497913463, + "grad_norm": 2.2271456718444824, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6916214823722839, + "num_tokens": 217797517.0, + "step": 8414 + }, + { + "epoch": 0.9241159674939601, + "grad_norm": 2.0015830993652344, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6941628456115723, + "num_tokens": 217826190.0, + "step": 8415 + }, + { + "epoch": 0.9242257851965737, + "grad_norm": 2.2580833435058594, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6855948567390442, + "num_tokens": 217852227.0, + "step": 8416 + }, + { + "epoch": 0.9243356028991874, + "grad_norm": 2.0798771381378174, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7027044892311096, + "num_tokens": 217880720.0, + "step": 8417 + }, + { + "epoch": 0.924445420601801, + "grad_norm": 2.4151642322540283, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7086275815963745, + "num_tokens": 217901861.0, + "step": 8418 + }, + { + "epoch": 0.9245552383044147, + "grad_norm": 2.0236408710479736, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6944162845611572, + "num_tokens": 217931798.0, + "step": 8419 + }, + { + "epoch": 0.9246650560070283, + "grad_norm": 2.1056087017059326, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.691857099533081, + "num_tokens": 217959279.0, + "step": 8420 + }, + { + "epoch": 0.924774873709642, + "grad_norm": 2.174006223678589, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7067430019378662, + "num_tokens": 217984120.0, + "step": 8421 + }, + { + "epoch": 0.9248846914122557, + "grad_norm": 2.10740065574646, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.729475200176239, + "num_tokens": 218010085.0, + "step": 8422 + }, + { + "epoch": 0.9249945091148694, + "grad_norm": 1.926679253578186, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7047688961029053, + "num_tokens": 218042423.0, + "step": 8423 + }, + { + "epoch": 0.925104326817483, + "grad_norm": 2.26997971534729, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.6998723745346069, + "num_tokens": 218067588.0, + "step": 8424 + }, + { + "epoch": 0.9252141445200966, + "grad_norm": 1.90260910987854, + "learning_rate": 1e-06, + "loss": 1.1293, + "mean_token_accuracy": 0.6649148464202881, + "num_tokens": 218101759.0, + "step": 8425 + }, + { + "epoch": 0.9253239622227103, + "grad_norm": 2.109900951385498, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7025733590126038, + "num_tokens": 218129303.0, + "step": 8426 + }, + { + "epoch": 0.9254337799253239, + "grad_norm": 2.2655553817749023, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7032498121261597, + "num_tokens": 218154835.0, + "step": 8427 + }, + { + "epoch": 0.9255435976279376, + "grad_norm": 2.0496394634246826, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6847264766693115, + "num_tokens": 218183636.0, + "step": 8428 + }, + { + "epoch": 0.9256534153305512, + "grad_norm": 2.357090950012207, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7132214307785034, + "num_tokens": 218207019.0, + "step": 8429 + }, + { + "epoch": 0.925763233033165, + "grad_norm": 2.249821662902832, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7195809483528137, + "num_tokens": 218229583.0, + "step": 8430 + }, + { + "epoch": 0.9258730507357786, + "grad_norm": 2.1416642665863037, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.70287024974823, + "num_tokens": 218256194.0, + "step": 8431 + }, + { + "epoch": 0.9259828684383923, + "grad_norm": 2.3210384845733643, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6850250959396362, + "num_tokens": 218280046.0, + "step": 8432 + }, + { + "epoch": 0.9260926861410059, + "grad_norm": 2.4720752239227295, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7178422212600708, + "num_tokens": 218299251.0, + "step": 8433 + }, + { + "epoch": 0.9262025038436196, + "grad_norm": 2.2590136528015137, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7139123678207397, + "num_tokens": 218323170.0, + "step": 8434 + }, + { + "epoch": 0.9263123215462332, + "grad_norm": 2.2982547283172607, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.713947057723999, + "num_tokens": 218346012.0, + "step": 8435 + }, + { + "epoch": 0.9264221392488469, + "grad_norm": 2.3920586109161377, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7211520671844482, + "num_tokens": 218368155.0, + "step": 8436 + }, + { + "epoch": 0.9265319569514606, + "grad_norm": 2.0275778770446777, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7004764080047607, + "num_tokens": 218394595.0, + "step": 8437 + }, + { + "epoch": 0.9266417746540743, + "grad_norm": 2.2503135204315186, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7063589096069336, + "num_tokens": 218419929.0, + "step": 8438 + }, + { + "epoch": 0.9267515923566879, + "grad_norm": 2.332447052001953, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7350181937217712, + "num_tokens": 218441566.0, + "step": 8439 + }, + { + "epoch": 0.9268614100593016, + "grad_norm": 2.330263137817383, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7184149026870728, + "num_tokens": 218464451.0, + "step": 8440 + }, + { + "epoch": 0.9269712277619152, + "grad_norm": 2.0576789379119873, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7018070220947266, + "num_tokens": 218492548.0, + "step": 8441 + }, + { + "epoch": 0.9270810454645289, + "grad_norm": 2.218060255050659, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7101607322692871, + "num_tokens": 218519220.0, + "step": 8442 + }, + { + "epoch": 0.9271908631671425, + "grad_norm": 2.3577358722686768, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.6993756294250488, + "num_tokens": 218541685.0, + "step": 8443 + }, + { + "epoch": 0.9273006808697563, + "grad_norm": 2.2736434936523438, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7255451679229736, + "num_tokens": 218565855.0, + "step": 8444 + }, + { + "epoch": 0.9274104985723699, + "grad_norm": 2.0950379371643066, + "learning_rate": 1e-06, + "loss": 1.0406, + "mean_token_accuracy": 0.6938419342041016, + "num_tokens": 218592633.0, + "step": 8445 + }, + { + "epoch": 0.9275203162749835, + "grad_norm": 2.09102725982666, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7094465494155884, + "num_tokens": 218618417.0, + "step": 8446 + }, + { + "epoch": 0.9276301339775972, + "grad_norm": 2.223322629928589, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7206816077232361, + "num_tokens": 218641999.0, + "step": 8447 + }, + { + "epoch": 0.9277399516802108, + "grad_norm": 2.0900182723999023, + "learning_rate": 1e-06, + "loss": 1.0496, + "mean_token_accuracy": 0.6860427856445312, + "num_tokens": 218670529.0, + "step": 8448 + }, + { + "epoch": 0.9278497693828245, + "grad_norm": 2.3935701847076416, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6847387552261353, + "num_tokens": 218695484.0, + "step": 8449 + }, + { + "epoch": 0.9279595870854381, + "grad_norm": 1.9596437215805054, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.7067800760269165, + "num_tokens": 218727725.0, + "step": 8450 + }, + { + "epoch": 0.9280694047880519, + "grad_norm": 2.015031576156616, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7217361927032471, + "num_tokens": 218754505.0, + "step": 8451 + }, + { + "epoch": 0.9281792224906655, + "grad_norm": 2.1882269382476807, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7081009149551392, + "num_tokens": 218778201.0, + "step": 8452 + }, + { + "epoch": 0.9282890401932792, + "grad_norm": 2.266587734222412, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7197549939155579, + "num_tokens": 218801071.0, + "step": 8453 + }, + { + "epoch": 0.9283988578958928, + "grad_norm": 2.3569624423980713, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7031888961791992, + "num_tokens": 218821290.0, + "step": 8454 + }, + { + "epoch": 0.9285086755985065, + "grad_norm": 2.0423378944396973, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7218933701515198, + "num_tokens": 218847790.0, + "step": 8455 + }, + { + "epoch": 0.9286184933011201, + "grad_norm": 2.157808780670166, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6816637516021729, + "num_tokens": 218872798.0, + "step": 8456 + }, + { + "epoch": 0.9287283110037338, + "grad_norm": 2.210183620452881, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.6965879797935486, + "num_tokens": 218896411.0, + "step": 8457 + }, + { + "epoch": 0.9288381287063474, + "grad_norm": 2.250260829925537, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7025633454322815, + "num_tokens": 218919199.0, + "step": 8458 + }, + { + "epoch": 0.9289479464089612, + "grad_norm": 2.110266923904419, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7075783610343933, + "num_tokens": 218947282.0, + "step": 8459 + }, + { + "epoch": 0.9290577641115748, + "grad_norm": 2.239349603652954, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7057807445526123, + "num_tokens": 218970788.0, + "step": 8460 + }, + { + "epoch": 0.9291675818141885, + "grad_norm": 1.9889731407165527, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7174230813980103, + "num_tokens": 219001070.0, + "step": 8461 + }, + { + "epoch": 0.9292773995168021, + "grad_norm": 2.1881167888641357, + "learning_rate": 1e-06, + "loss": 1.0466, + "mean_token_accuracy": 0.6781399250030518, + "num_tokens": 219029358.0, + "step": 8462 + }, + { + "epoch": 0.9293872172194158, + "grad_norm": 2.1103549003601074, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7292649149894714, + "num_tokens": 219055155.0, + "step": 8463 + }, + { + "epoch": 0.9294970349220294, + "grad_norm": 1.851409912109375, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6904827356338501, + "num_tokens": 219088784.0, + "step": 8464 + }, + { + "epoch": 0.929606852624643, + "grad_norm": 1.8876725435256958, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7238904237747192, + "num_tokens": 219119790.0, + "step": 8465 + }, + { + "epoch": 0.9297166703272568, + "grad_norm": 2.203052043914795, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7132376432418823, + "num_tokens": 219142801.0, + "step": 8466 + }, + { + "epoch": 0.9298264880298704, + "grad_norm": 2.0912065505981445, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7323414087295532, + "num_tokens": 219168134.0, + "step": 8467 + }, + { + "epoch": 0.9299363057324841, + "grad_norm": 2.2703778743743896, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7329989075660706, + "num_tokens": 219191748.0, + "step": 8468 + }, + { + "epoch": 0.9300461234350977, + "grad_norm": 2.49137806892395, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6990509033203125, + "num_tokens": 219214834.0, + "step": 8469 + }, + { + "epoch": 0.9301559411377114, + "grad_norm": 2.1627347469329834, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.71503746509552, + "num_tokens": 219243265.0, + "step": 8470 + }, + { + "epoch": 0.930265758840325, + "grad_norm": 1.8304733037948608, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7171820402145386, + "num_tokens": 219275706.0, + "step": 8471 + }, + { + "epoch": 0.9303755765429387, + "grad_norm": 2.0292887687683105, + "learning_rate": 1e-06, + "loss": 1.0599, + "mean_token_accuracy": 0.6747612953186035, + "num_tokens": 219305391.0, + "step": 8472 + }, + { + "epoch": 0.9304853942455524, + "grad_norm": 2.1935641765594482, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6987901329994202, + "num_tokens": 219331399.0, + "step": 8473 + }, + { + "epoch": 0.9305952119481661, + "grad_norm": 2.037750244140625, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7298643589019775, + "num_tokens": 219358967.0, + "step": 8474 + }, + { + "epoch": 0.9307050296507797, + "grad_norm": 1.8327726125717163, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7026887536048889, + "num_tokens": 219394839.0, + "step": 8475 + }, + { + "epoch": 0.9308148473533934, + "grad_norm": 2.5337517261505127, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7190589904785156, + "num_tokens": 219413954.0, + "step": 8476 + }, + { + "epoch": 0.930924665056007, + "grad_norm": 2.1605618000030518, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7048538327217102, + "num_tokens": 219440917.0, + "step": 8477 + }, + { + "epoch": 0.9310344827586207, + "grad_norm": 2.1255929470062256, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6879045963287354, + "num_tokens": 219469897.0, + "step": 8478 + }, + { + "epoch": 0.9311443004612343, + "grad_norm": 2.279961109161377, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7123494148254395, + "num_tokens": 219491251.0, + "step": 8479 + }, + { + "epoch": 0.9312541181638481, + "grad_norm": 2.3480608463287354, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7093135714530945, + "num_tokens": 219512466.0, + "step": 8480 + }, + { + "epoch": 0.9313639358664617, + "grad_norm": 2.128690481185913, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7071618437767029, + "num_tokens": 219539230.0, + "step": 8481 + }, + { + "epoch": 0.9314737535690754, + "grad_norm": 2.0864996910095215, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7138113975524902, + "num_tokens": 219567470.0, + "step": 8482 + }, + { + "epoch": 0.931583571271689, + "grad_norm": 2.0413687229156494, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.714146077632904, + "num_tokens": 219595741.0, + "step": 8483 + }, + { + "epoch": 0.9316933889743026, + "grad_norm": 2.1829802989959717, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7251070141792297, + "num_tokens": 219620741.0, + "step": 8484 + }, + { + "epoch": 0.9318032066769163, + "grad_norm": 2.1432082653045654, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.703153669834137, + "num_tokens": 219648414.0, + "step": 8485 + }, + { + "epoch": 0.9319130243795299, + "grad_norm": 2.046640396118164, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6964932084083557, + "num_tokens": 219675933.0, + "step": 8486 + }, + { + "epoch": 0.9320228420821436, + "grad_norm": 2.1478517055511475, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7010445594787598, + "num_tokens": 219702289.0, + "step": 8487 + }, + { + "epoch": 0.9321326597847573, + "grad_norm": 2.3104329109191895, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7011373043060303, + "num_tokens": 219724969.0, + "step": 8488 + }, + { + "epoch": 0.932242477487371, + "grad_norm": 2.445558786392212, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7271783351898193, + "num_tokens": 219745247.0, + "step": 8489 + }, + { + "epoch": 0.9323522951899846, + "grad_norm": 2.374250650405884, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7210094332695007, + "num_tokens": 219767121.0, + "step": 8490 + }, + { + "epoch": 0.9324621128925983, + "grad_norm": 2.094594955444336, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.707758367061615, + "num_tokens": 219793110.0, + "step": 8491 + }, + { + "epoch": 0.9325719305952119, + "grad_norm": 2.1901285648345947, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7193900346755981, + "num_tokens": 219817431.0, + "step": 8492 + }, + { + "epoch": 0.9326817482978256, + "grad_norm": 2.326632022857666, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7143040895462036, + "num_tokens": 219840958.0, + "step": 8493 + }, + { + "epoch": 0.9327915660004392, + "grad_norm": 2.3659842014312744, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7036999464035034, + "num_tokens": 219864220.0, + "step": 8494 + }, + { + "epoch": 0.932901383703053, + "grad_norm": 2.2955825328826904, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7278696298599243, + "num_tokens": 219885350.0, + "step": 8495 + }, + { + "epoch": 0.9330112014056666, + "grad_norm": 2.141641139984131, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7060929536819458, + "num_tokens": 219910183.0, + "step": 8496 + }, + { + "epoch": 0.9331210191082803, + "grad_norm": 2.2901158332824707, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.713241457939148, + "num_tokens": 219935143.0, + "step": 8497 + }, + { + "epoch": 0.9332308368108939, + "grad_norm": 2.1062965393066406, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6897244453430176, + "num_tokens": 219962271.0, + "step": 8498 + }, + { + "epoch": 0.9333406545135076, + "grad_norm": 1.961898684501648, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6854844093322754, + "num_tokens": 219996977.0, + "step": 8499 + }, + { + "epoch": 0.9334504722161212, + "grad_norm": 2.347158193588257, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7074998617172241, + "num_tokens": 220019017.0, + "step": 8500 + }, + { + "epoch": 0.9335602899187349, + "grad_norm": 2.067643165588379, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7221549153327942, + "num_tokens": 220044241.0, + "step": 8501 + }, + { + "epoch": 0.9336701076213486, + "grad_norm": 2.4450390338897705, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6980568170547485, + "num_tokens": 220066675.0, + "step": 8502 + }, + { + "epoch": 0.9337799253239623, + "grad_norm": 2.412374258041382, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7225131988525391, + "num_tokens": 220087579.0, + "step": 8503 + }, + { + "epoch": 0.9338897430265759, + "grad_norm": 2.666933298110962, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7150993347167969, + "num_tokens": 220105762.0, + "step": 8504 + }, + { + "epoch": 0.9339995607291895, + "grad_norm": 2.3582725524902344, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7111775875091553, + "num_tokens": 220127377.0, + "step": 8505 + }, + { + "epoch": 0.9341093784318032, + "grad_norm": 1.987039566040039, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7184607982635498, + "num_tokens": 220156035.0, + "step": 8506 + }, + { + "epoch": 0.9342191961344168, + "grad_norm": 1.9779157638549805, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.710448682308197, + "num_tokens": 220186364.0, + "step": 8507 + }, + { + "epoch": 0.9343290138370305, + "grad_norm": 2.0131633281707764, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6911896467208862, + "num_tokens": 220216923.0, + "step": 8508 + }, + { + "epoch": 0.9344388315396442, + "grad_norm": 2.1220200061798096, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6848410964012146, + "num_tokens": 220243427.0, + "step": 8509 + }, + { + "epoch": 0.9345486492422579, + "grad_norm": 2.110065460205078, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6824390888214111, + "num_tokens": 220270601.0, + "step": 8510 + }, + { + "epoch": 0.9346584669448715, + "grad_norm": 2.1226139068603516, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.7029380202293396, + "num_tokens": 220299537.0, + "step": 8511 + }, + { + "epoch": 0.9347682846474852, + "grad_norm": 2.3204991817474365, + "learning_rate": 1e-06, + "loss": 1.0614, + "mean_token_accuracy": 0.6792999505996704, + "num_tokens": 220325252.0, + "step": 8512 + }, + { + "epoch": 0.9348781023500988, + "grad_norm": 1.9934426546096802, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7136276364326477, + "num_tokens": 220351982.0, + "step": 8513 + }, + { + "epoch": 0.9349879200527125, + "grad_norm": 1.9612373113632202, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7049959897994995, + "num_tokens": 220381152.0, + "step": 8514 + }, + { + "epoch": 0.9350977377553261, + "grad_norm": 2.5463056564331055, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.733001708984375, + "num_tokens": 220399949.0, + "step": 8515 + }, + { + "epoch": 0.9352075554579398, + "grad_norm": 1.8939871788024902, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7114613056182861, + "num_tokens": 220432231.0, + "step": 8516 + }, + { + "epoch": 0.9353173731605535, + "grad_norm": 2.256704568862915, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7177598476409912, + "num_tokens": 220455363.0, + "step": 8517 + }, + { + "epoch": 0.9354271908631672, + "grad_norm": 2.0178141593933105, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.696244478225708, + "num_tokens": 220485388.0, + "step": 8518 + }, + { + "epoch": 0.9355370085657808, + "grad_norm": 2.122640371322632, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7000089287757874, + "num_tokens": 220510686.0, + "step": 8519 + }, + { + "epoch": 0.9356468262683945, + "grad_norm": 2.1092255115509033, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7009768486022949, + "num_tokens": 220538470.0, + "step": 8520 + }, + { + "epoch": 0.9357566439710081, + "grad_norm": 2.400656223297119, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.73004150390625, + "num_tokens": 220559701.0, + "step": 8521 + }, + { + "epoch": 0.9358664616736218, + "grad_norm": 2.057976245880127, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7270547747612, + "num_tokens": 220587287.0, + "step": 8522 + }, + { + "epoch": 0.9359762793762354, + "grad_norm": 2.197878360748291, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7044886350631714, + "num_tokens": 220612885.0, + "step": 8523 + }, + { + "epoch": 0.9360860970788492, + "grad_norm": 2.015238046646118, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6868615746498108, + "num_tokens": 220642379.0, + "step": 8524 + }, + { + "epoch": 0.9361959147814628, + "grad_norm": 2.025164842605591, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6999900937080383, + "num_tokens": 220670345.0, + "step": 8525 + }, + { + "epoch": 0.9363057324840764, + "grad_norm": 2.241260051727295, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6874611377716064, + "num_tokens": 220695748.0, + "step": 8526 + }, + { + "epoch": 0.9364155501866901, + "grad_norm": 2.0616278648376465, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6977220177650452, + "num_tokens": 220725066.0, + "step": 8527 + }, + { + "epoch": 0.9365253678893037, + "grad_norm": 2.281858444213867, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6918996572494507, + "num_tokens": 220750281.0, + "step": 8528 + }, + { + "epoch": 0.9366351855919174, + "grad_norm": 2.154513120651245, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7078652381896973, + "num_tokens": 220778300.0, + "step": 8529 + }, + { + "epoch": 0.936745003294531, + "grad_norm": 2.176262378692627, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7084265947341919, + "num_tokens": 220803943.0, + "step": 8530 + }, + { + "epoch": 0.9368548209971448, + "grad_norm": 2.147260904312134, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7230820655822754, + "num_tokens": 220827576.0, + "step": 8531 + }, + { + "epoch": 0.9369646386997584, + "grad_norm": 2.5187160968780518, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6961398124694824, + "num_tokens": 220849305.0, + "step": 8532 + }, + { + "epoch": 0.9370744564023721, + "grad_norm": 2.220846652984619, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6908928751945496, + "num_tokens": 220875563.0, + "step": 8533 + }, + { + "epoch": 0.9371842741049857, + "grad_norm": 2.259178638458252, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7108398675918579, + "num_tokens": 220898919.0, + "step": 8534 + }, + { + "epoch": 0.9372940918075994, + "grad_norm": 2.1189663410186768, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.6948827505111694, + "num_tokens": 220925388.0, + "step": 8535 + }, + { + "epoch": 0.937403909510213, + "grad_norm": 2.4005796909332275, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7089428901672363, + "num_tokens": 220946587.0, + "step": 8536 + }, + { + "epoch": 0.9375137272128267, + "grad_norm": 2.2851319313049316, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6988115310668945, + "num_tokens": 220970097.0, + "step": 8537 + }, + { + "epoch": 0.9376235449154404, + "grad_norm": 2.000267505645752, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6961918473243713, + "num_tokens": 220999343.0, + "step": 8538 + }, + { + "epoch": 0.9377333626180541, + "grad_norm": 2.0410568714141846, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.705951452255249, + "num_tokens": 221027252.0, + "step": 8539 + }, + { + "epoch": 0.9378431803206677, + "grad_norm": 2.3272242546081543, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.7030763030052185, + "num_tokens": 221052625.0, + "step": 8540 + }, + { + "epoch": 0.9379529980232814, + "grad_norm": 2.268444299697876, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7076141238212585, + "num_tokens": 221076412.0, + "step": 8541 + }, + { + "epoch": 0.938062815725895, + "grad_norm": 2.156336545944214, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7243346571922302, + "num_tokens": 221104149.0, + "step": 8542 + }, + { + "epoch": 0.9381726334285087, + "grad_norm": 2.0713248252868652, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7006609439849854, + "num_tokens": 221132434.0, + "step": 8543 + }, + { + "epoch": 0.9382824511311223, + "grad_norm": 2.1753110885620117, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7063868045806885, + "num_tokens": 221157135.0, + "step": 8544 + }, + { + "epoch": 0.9383922688337359, + "grad_norm": 2.3576574325561523, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7232987284660339, + "num_tokens": 221177932.0, + "step": 8545 + }, + { + "epoch": 0.9385020865363497, + "grad_norm": 2.3465445041656494, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7130219340324402, + "num_tokens": 221200677.0, + "step": 8546 + }, + { + "epoch": 0.9386119042389633, + "grad_norm": 2.0555262565612793, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7128691673278809, + "num_tokens": 221226510.0, + "step": 8547 + }, + { + "epoch": 0.938721721941577, + "grad_norm": 2.073937177658081, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6823406219482422, + "num_tokens": 221254321.0, + "step": 8548 + }, + { + "epoch": 0.9388315396441906, + "grad_norm": 2.3628616333007812, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7053890824317932, + "num_tokens": 221277301.0, + "step": 8549 + }, + { + "epoch": 0.9389413573468043, + "grad_norm": 2.417125701904297, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.725173830986023, + "num_tokens": 221298457.0, + "step": 8550 + }, + { + "epoch": 0.9390511750494179, + "grad_norm": 2.156804084777832, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7118663787841797, + "num_tokens": 221323999.0, + "step": 8551 + }, + { + "epoch": 0.9391609927520316, + "grad_norm": 2.0158393383026123, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7209286093711853, + "num_tokens": 221350861.0, + "step": 8552 + }, + { + "epoch": 0.9392708104546453, + "grad_norm": 2.0708000659942627, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6990238428115845, + "num_tokens": 221379066.0, + "step": 8553 + }, + { + "epoch": 0.939380628157259, + "grad_norm": 2.158734083175659, + "learning_rate": 1e-06, + "loss": 1.0854, + "mean_token_accuracy": 0.6761987209320068, + "num_tokens": 221405645.0, + "step": 8554 + }, + { + "epoch": 0.9394904458598726, + "grad_norm": 1.9267746210098267, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7079650163650513, + "num_tokens": 221434456.0, + "step": 8555 + }, + { + "epoch": 0.9396002635624863, + "grad_norm": 2.316093683242798, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7258675694465637, + "num_tokens": 221455517.0, + "step": 8556 + }, + { + "epoch": 0.9397100812650999, + "grad_norm": 2.471383571624756, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7124860882759094, + "num_tokens": 221475373.0, + "step": 8557 + }, + { + "epoch": 0.9398198989677136, + "grad_norm": 2.1212034225463867, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7117899060249329, + "num_tokens": 221503028.0, + "step": 8558 + }, + { + "epoch": 0.9399297166703272, + "grad_norm": 2.3483779430389404, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7162606716156006, + "num_tokens": 221526144.0, + "step": 8559 + }, + { + "epoch": 0.940039534372941, + "grad_norm": 2.495269536972046, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6954610347747803, + "num_tokens": 221547183.0, + "step": 8560 + }, + { + "epoch": 0.9401493520755546, + "grad_norm": 2.2836623191833496, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7504847645759583, + "num_tokens": 221569164.0, + "step": 8561 + }, + { + "epoch": 0.9402591697781683, + "grad_norm": 2.067476272583008, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.706638753414154, + "num_tokens": 221597868.0, + "step": 8562 + }, + { + "epoch": 0.9403689874807819, + "grad_norm": 2.1904659271240234, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7049902677536011, + "num_tokens": 221624517.0, + "step": 8563 + }, + { + "epoch": 0.9404788051833955, + "grad_norm": 1.990261435508728, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7030433416366577, + "num_tokens": 221654198.0, + "step": 8564 + }, + { + "epoch": 0.9405886228860092, + "grad_norm": 2.4125003814697266, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7128669023513794, + "num_tokens": 221676354.0, + "step": 8565 + }, + { + "epoch": 0.9406984405886228, + "grad_norm": 1.9015995264053345, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6791846752166748, + "num_tokens": 221711475.0, + "step": 8566 + }, + { + "epoch": 0.9408082582912366, + "grad_norm": 2.168349266052246, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7200438976287842, + "num_tokens": 221738072.0, + "step": 8567 + }, + { + "epoch": 0.9409180759938502, + "grad_norm": 2.1351375579833984, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6942386031150818, + "num_tokens": 221765201.0, + "step": 8568 + }, + { + "epoch": 0.9410278936964639, + "grad_norm": 2.1063179969787598, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7055213451385498, + "num_tokens": 221789983.0, + "step": 8569 + }, + { + "epoch": 0.9411377113990775, + "grad_norm": 2.0152077674865723, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7133258581161499, + "num_tokens": 221818675.0, + "step": 8570 + }, + { + "epoch": 0.9412475291016912, + "grad_norm": 1.946625828742981, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6970318555831909, + "num_tokens": 221849966.0, + "step": 8571 + }, + { + "epoch": 0.9413573468043048, + "grad_norm": 2.0976457595825195, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7088333368301392, + "num_tokens": 221877967.0, + "step": 8572 + }, + { + "epoch": 0.9414671645069185, + "grad_norm": 2.2367122173309326, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7317154407501221, + "num_tokens": 221901832.0, + "step": 8573 + }, + { + "epoch": 0.9415769822095322, + "grad_norm": 2.308361291885376, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.709450364112854, + "num_tokens": 221926364.0, + "step": 8574 + }, + { + "epoch": 0.9416867999121459, + "grad_norm": 2.2354137897491455, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7227399349212646, + "num_tokens": 221949997.0, + "step": 8575 + }, + { + "epoch": 0.9417966176147595, + "grad_norm": 2.583292245864868, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7146443724632263, + "num_tokens": 221972166.0, + "step": 8576 + }, + { + "epoch": 0.9419064353173732, + "grad_norm": 2.0268263816833496, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.7018091678619385, + "num_tokens": 222000577.0, + "step": 8577 + }, + { + "epoch": 0.9420162530199868, + "grad_norm": 2.0994248390197754, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7146714925765991, + "num_tokens": 222028566.0, + "step": 8578 + }, + { + "epoch": 0.9421260707226005, + "grad_norm": 2.260140895843506, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7279365062713623, + "num_tokens": 222052475.0, + "step": 8579 + }, + { + "epoch": 0.9422358884252141, + "grad_norm": 1.979092001914978, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6965579986572266, + "num_tokens": 222080988.0, + "step": 8580 + }, + { + "epoch": 0.9423457061278278, + "grad_norm": 2.104353189468384, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.7044423818588257, + "num_tokens": 222108697.0, + "step": 8581 + }, + { + "epoch": 0.9424555238304415, + "grad_norm": 2.746922016143799, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7235864400863647, + "num_tokens": 222126981.0, + "step": 8582 + }, + { + "epoch": 0.9425653415330552, + "grad_norm": 2.081387758255005, + "learning_rate": 1e-06, + "loss": 1.1074, + "mean_token_accuracy": 0.6717855930328369, + "num_tokens": 222155767.0, + "step": 8583 + }, + { + "epoch": 0.9426751592356688, + "grad_norm": 2.4802355766296387, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6999019384384155, + "num_tokens": 222177533.0, + "step": 8584 + }, + { + "epoch": 0.9427849769382824, + "grad_norm": 2.094733238220215, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7152692079544067, + "num_tokens": 222201844.0, + "step": 8585 + }, + { + "epoch": 0.9428947946408961, + "grad_norm": 2.3285210132598877, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7158432006835938, + "num_tokens": 222224495.0, + "step": 8586 + }, + { + "epoch": 0.9430046123435097, + "grad_norm": 2.6137709617614746, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7055234909057617, + "num_tokens": 222243560.0, + "step": 8587 + }, + { + "epoch": 0.9431144300461234, + "grad_norm": 1.98154878616333, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7206130027770996, + "num_tokens": 222274129.0, + "step": 8588 + }, + { + "epoch": 0.9432242477487371, + "grad_norm": 2.02502179145813, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7199180722236633, + "num_tokens": 222305237.0, + "step": 8589 + }, + { + "epoch": 0.9433340654513508, + "grad_norm": 2.3873345851898193, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7035508155822754, + "num_tokens": 222328774.0, + "step": 8590 + }, + { + "epoch": 0.9434438831539644, + "grad_norm": 2.136713981628418, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7256489396095276, + "num_tokens": 222352974.0, + "step": 8591 + }, + { + "epoch": 0.9435537008565781, + "grad_norm": 2.3105218410491943, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7043288350105286, + "num_tokens": 222375481.0, + "step": 8592 + }, + { + "epoch": 0.9436635185591917, + "grad_norm": 2.213099479675293, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7047399282455444, + "num_tokens": 222399146.0, + "step": 8593 + }, + { + "epoch": 0.9437733362618054, + "grad_norm": 1.8542143106460571, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6955441236495972, + "num_tokens": 222432584.0, + "step": 8594 + }, + { + "epoch": 0.943883153964419, + "grad_norm": 2.4430365562438965, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7047404050827026, + "num_tokens": 222454366.0, + "step": 8595 + }, + { + "epoch": 0.9439929716670328, + "grad_norm": 2.280728816986084, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7089459896087646, + "num_tokens": 222479174.0, + "step": 8596 + }, + { + "epoch": 0.9441027893696464, + "grad_norm": 2.2201318740844727, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7254217267036438, + "num_tokens": 222503908.0, + "step": 8597 + }, + { + "epoch": 0.9442126070722601, + "grad_norm": 2.644712209701538, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7424653768539429, + "num_tokens": 222520848.0, + "step": 8598 + }, + { + "epoch": 0.9443224247748737, + "grad_norm": 2.495854139328003, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7310322523117065, + "num_tokens": 222538364.0, + "step": 8599 + }, + { + "epoch": 0.9444322424774874, + "grad_norm": 2.085496425628662, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6945454478263855, + "num_tokens": 222565875.0, + "step": 8600 + }, + { + "epoch": 0.944542060180101, + "grad_norm": 2.2406578063964844, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6853674054145813, + "num_tokens": 222591050.0, + "step": 8601 + }, + { + "epoch": 0.9446518778827147, + "grad_norm": 2.147858142852783, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7159780859947205, + "num_tokens": 222616289.0, + "step": 8602 + }, + { + "epoch": 0.9447616955853284, + "grad_norm": 2.0469443798065186, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7055253982543945, + "num_tokens": 222642934.0, + "step": 8603 + }, + { + "epoch": 0.944871513287942, + "grad_norm": 2.084921360015869, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7328053116798401, + "num_tokens": 222669971.0, + "step": 8604 + }, + { + "epoch": 0.9449813309905557, + "grad_norm": 2.334529399871826, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7178956866264343, + "num_tokens": 222693886.0, + "step": 8605 + }, + { + "epoch": 0.9450911486931693, + "grad_norm": 2.223313331604004, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7025123238563538, + "num_tokens": 222718751.0, + "step": 8606 + }, + { + "epoch": 0.945200966395783, + "grad_norm": 2.412233591079712, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.728847324848175, + "num_tokens": 222739723.0, + "step": 8607 + }, + { + "epoch": 0.9453107840983966, + "grad_norm": 2.2465202808380127, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7157086133956909, + "num_tokens": 222764431.0, + "step": 8608 + }, + { + "epoch": 0.9454206018010103, + "grad_norm": 2.3610262870788574, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7212724089622498, + "num_tokens": 222786067.0, + "step": 8609 + }, + { + "epoch": 0.9455304195036239, + "grad_norm": 2.186126947402954, + "learning_rate": 1e-06, + "loss": 1.0317, + "mean_token_accuracy": 0.6895071268081665, + "num_tokens": 222811849.0, + "step": 8610 + }, + { + "epoch": 0.9456402372062377, + "grad_norm": 2.2407331466674805, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6918964385986328, + "num_tokens": 222837098.0, + "step": 8611 + }, + { + "epoch": 0.9457500549088513, + "grad_norm": 2.1114697456359863, + "learning_rate": 1e-06, + "loss": 1.0557, + "mean_token_accuracy": 0.6929957270622253, + "num_tokens": 222867209.0, + "step": 8612 + }, + { + "epoch": 0.945859872611465, + "grad_norm": 2.236302137374878, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6847971677780151, + "num_tokens": 222891024.0, + "step": 8613 + }, + { + "epoch": 0.9459696903140786, + "grad_norm": 2.195984125137329, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7165873646736145, + "num_tokens": 222914495.0, + "step": 8614 + }, + { + "epoch": 0.9460795080166923, + "grad_norm": 2.4908273220062256, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7229329943656921, + "num_tokens": 222933415.0, + "step": 8615 + }, + { + "epoch": 0.9461893257193059, + "grad_norm": 2.2871153354644775, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7086745500564575, + "num_tokens": 222956258.0, + "step": 8616 + }, + { + "epoch": 0.9462991434219196, + "grad_norm": 2.429168462753296, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7184537649154663, + "num_tokens": 222976109.0, + "step": 8617 + }, + { + "epoch": 0.9464089611245333, + "grad_norm": 2.1771538257598877, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7018693685531616, + "num_tokens": 223002419.0, + "step": 8618 + }, + { + "epoch": 0.946518778827147, + "grad_norm": 2.106415033340454, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7028719186782837, + "num_tokens": 223029426.0, + "step": 8619 + }, + { + "epoch": 0.9466285965297606, + "grad_norm": 2.121270179748535, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7238799333572388, + "num_tokens": 223055601.0, + "step": 8620 + }, + { + "epoch": 0.9467384142323743, + "grad_norm": 2.021775007247925, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6905733942985535, + "num_tokens": 223086401.0, + "step": 8621 + }, + { + "epoch": 0.9468482319349879, + "grad_norm": 2.179861307144165, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7072705626487732, + "num_tokens": 223112709.0, + "step": 8622 + }, + { + "epoch": 0.9469580496376016, + "grad_norm": 2.3635263442993164, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7143843173980713, + "num_tokens": 223132726.0, + "step": 8623 + }, + { + "epoch": 0.9470678673402152, + "grad_norm": 2.607815742492676, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7257894277572632, + "num_tokens": 223151442.0, + "step": 8624 + }, + { + "epoch": 0.947177685042829, + "grad_norm": 2.2429232597351074, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7070617079734802, + "num_tokens": 223174241.0, + "step": 8625 + }, + { + "epoch": 0.9472875027454426, + "grad_norm": 2.3504676818847656, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7216507792472839, + "num_tokens": 223196993.0, + "step": 8626 + }, + { + "epoch": 0.9473973204480562, + "grad_norm": 1.9158655405044556, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7008430361747742, + "num_tokens": 223226771.0, + "step": 8627 + }, + { + "epoch": 0.9475071381506699, + "grad_norm": 2.2733993530273438, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7162479162216187, + "num_tokens": 223249487.0, + "step": 8628 + }, + { + "epoch": 0.9476169558532835, + "grad_norm": 2.1138758659362793, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6993734240531921, + "num_tokens": 223275629.0, + "step": 8629 + }, + { + "epoch": 0.9477267735558972, + "grad_norm": 2.0793070793151855, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6905149221420288, + "num_tokens": 223304025.0, + "step": 8630 + }, + { + "epoch": 0.9478365912585108, + "grad_norm": 2.0726144313812256, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7066577076911926, + "num_tokens": 223333077.0, + "step": 8631 + }, + { + "epoch": 0.9479464089611246, + "grad_norm": 2.3654680252075195, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.701543927192688, + "num_tokens": 223356647.0, + "step": 8632 + }, + { + "epoch": 0.9480562266637382, + "grad_norm": 2.7899415493011475, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7286726236343384, + "num_tokens": 223374548.0, + "step": 8633 + }, + { + "epoch": 0.9481660443663519, + "grad_norm": 2.244452714920044, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7199058532714844, + "num_tokens": 223400800.0, + "step": 8634 + }, + { + "epoch": 0.9482758620689655, + "grad_norm": 2.682227849960327, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7286479473114014, + "num_tokens": 223419768.0, + "step": 8635 + }, + { + "epoch": 0.9483856797715792, + "grad_norm": 2.054537773132324, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7153764963150024, + "num_tokens": 223446443.0, + "step": 8636 + }, + { + "epoch": 0.9484954974741928, + "grad_norm": 2.110743761062622, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7197064161300659, + "num_tokens": 223473510.0, + "step": 8637 + }, + { + "epoch": 0.9486053151768065, + "grad_norm": 2.1850883960723877, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7379543781280518, + "num_tokens": 223503095.0, + "step": 8638 + }, + { + "epoch": 0.9487151328794201, + "grad_norm": 2.089820623397827, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6997573375701904, + "num_tokens": 223532652.0, + "step": 8639 + }, + { + "epoch": 0.9488249505820339, + "grad_norm": 2.0913443565368652, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7164651155471802, + "num_tokens": 223560980.0, + "step": 8640 + }, + { + "epoch": 0.9489347682846475, + "grad_norm": 2.012768030166626, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7287713289260864, + "num_tokens": 223586348.0, + "step": 8641 + }, + { + "epoch": 0.9490445859872612, + "grad_norm": 2.4823806285858154, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7113763093948364, + "num_tokens": 223607708.0, + "step": 8642 + }, + { + "epoch": 0.9491544036898748, + "grad_norm": 2.4027087688446045, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7324993014335632, + "num_tokens": 223628312.0, + "step": 8643 + }, + { + "epoch": 0.9492642213924884, + "grad_norm": 2.117818593978882, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7085320949554443, + "num_tokens": 223656073.0, + "step": 8644 + }, + { + "epoch": 0.9493740390951021, + "grad_norm": 2.168837308883667, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7136386632919312, + "num_tokens": 223681909.0, + "step": 8645 + }, + { + "epoch": 0.9494838567977157, + "grad_norm": 2.2472288608551025, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.694516658782959, + "num_tokens": 223706150.0, + "step": 8646 + }, + { + "epoch": 0.9495936745003295, + "grad_norm": 2.316126585006714, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7047691941261292, + "num_tokens": 223730524.0, + "step": 8647 + }, + { + "epoch": 0.9497034922029431, + "grad_norm": 2.633516550064087, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7384577989578247, + "num_tokens": 223748705.0, + "step": 8648 + }, + { + "epoch": 0.9498133099055568, + "grad_norm": 1.8993695974349976, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.7012335062026978, + "num_tokens": 223784072.0, + "step": 8649 + }, + { + "epoch": 0.9499231276081704, + "grad_norm": 2.1366209983825684, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7381876707077026, + "num_tokens": 223811086.0, + "step": 8650 + }, + { + "epoch": 0.9500329453107841, + "grad_norm": 2.306915044784546, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7159358263015747, + "num_tokens": 223835135.0, + "step": 8651 + }, + { + "epoch": 0.9501427630133977, + "grad_norm": 2.206106662750244, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6900193691253662, + "num_tokens": 223861787.0, + "step": 8652 + }, + { + "epoch": 0.9502525807160114, + "grad_norm": 2.062669277191162, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6886683702468872, + "num_tokens": 223891162.0, + "step": 8653 + }, + { + "epoch": 0.9503623984186251, + "grad_norm": 2.43449068069458, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7200046181678772, + "num_tokens": 223912433.0, + "step": 8654 + }, + { + "epoch": 0.9504722161212388, + "grad_norm": 2.2694735527038574, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.688498854637146, + "num_tokens": 223938108.0, + "step": 8655 + }, + { + "epoch": 0.9505820338238524, + "grad_norm": 2.050888776779175, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6876763105392456, + "num_tokens": 223965637.0, + "step": 8656 + }, + { + "epoch": 0.9506918515264661, + "grad_norm": 2.5650699138641357, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7140176892280579, + "num_tokens": 223985296.0, + "step": 8657 + }, + { + "epoch": 0.9508016692290797, + "grad_norm": 2.0910699367523193, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6984502077102661, + "num_tokens": 224013684.0, + "step": 8658 + }, + { + "epoch": 0.9509114869316934, + "grad_norm": 2.0437734127044678, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7016102075576782, + "num_tokens": 224041925.0, + "step": 8659 + }, + { + "epoch": 0.951021304634307, + "grad_norm": 2.4077553749084473, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7041328549385071, + "num_tokens": 224062807.0, + "step": 8660 + }, + { + "epoch": 0.9511311223369208, + "grad_norm": 2.0994579792022705, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.706937313079834, + "num_tokens": 224090444.0, + "step": 8661 + }, + { + "epoch": 0.9512409400395344, + "grad_norm": 2.329746723175049, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6937453150749207, + "num_tokens": 224115116.0, + "step": 8662 + }, + { + "epoch": 0.9513507577421481, + "grad_norm": 2.0252511501312256, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.691148042678833, + "num_tokens": 224144631.0, + "step": 8663 + }, + { + "epoch": 0.9514605754447617, + "grad_norm": 2.031346321105957, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7080013751983643, + "num_tokens": 224173130.0, + "step": 8664 + }, + { + "epoch": 0.9515703931473753, + "grad_norm": 2.2928965091705322, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7301050424575806, + "num_tokens": 224196040.0, + "step": 8665 + }, + { + "epoch": 0.951680210849989, + "grad_norm": 2.0469143390655518, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.735934853553772, + "num_tokens": 224221596.0, + "step": 8666 + }, + { + "epoch": 0.9517900285526026, + "grad_norm": 2.2164359092712402, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7160745859146118, + "num_tokens": 224247333.0, + "step": 8667 + }, + { + "epoch": 0.9518998462552163, + "grad_norm": 2.1232433319091797, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7194661498069763, + "num_tokens": 224274178.0, + "step": 8668 + }, + { + "epoch": 0.95200966395783, + "grad_norm": 1.9860529899597168, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.69633549451828, + "num_tokens": 224304474.0, + "step": 8669 + }, + { + "epoch": 0.9521194816604437, + "grad_norm": 2.321812391281128, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.687511682510376, + "num_tokens": 224329214.0, + "step": 8670 + }, + { + "epoch": 0.9522292993630573, + "grad_norm": 2.1810076236724854, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7074249982833862, + "num_tokens": 224353752.0, + "step": 8671 + }, + { + "epoch": 0.952339117065671, + "grad_norm": 2.703721761703491, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7379115223884583, + "num_tokens": 224369261.0, + "step": 8672 + }, + { + "epoch": 0.9524489347682846, + "grad_norm": 2.093202829360962, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7088891863822937, + "num_tokens": 224397273.0, + "step": 8673 + }, + { + "epoch": 0.9525587524708983, + "grad_norm": 2.365975856781006, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7114777565002441, + "num_tokens": 224419111.0, + "step": 8674 + }, + { + "epoch": 0.9526685701735119, + "grad_norm": 2.1218748092651367, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7139300107955933, + "num_tokens": 224445100.0, + "step": 8675 + }, + { + "epoch": 0.9527783878761257, + "grad_norm": 2.199826717376709, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7299922704696655, + "num_tokens": 224469047.0, + "step": 8676 + }, + { + "epoch": 0.9528882055787393, + "grad_norm": 2.3006789684295654, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7208015322685242, + "num_tokens": 224491089.0, + "step": 8677 + }, + { + "epoch": 0.952998023281353, + "grad_norm": 2.18898868560791, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7050521373748779, + "num_tokens": 224515134.0, + "step": 8678 + }, + { + "epoch": 0.9531078409839666, + "grad_norm": 1.9862033128738403, + "learning_rate": 1e-06, + "loss": 1.0685, + "mean_token_accuracy": 0.6785104274749756, + "num_tokens": 224548089.0, + "step": 8679 + }, + { + "epoch": 0.9532176586865803, + "grad_norm": 2.1988422870635986, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7103455066680908, + "num_tokens": 224571705.0, + "step": 8680 + }, + { + "epoch": 0.9533274763891939, + "grad_norm": 2.1806087493896484, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.689497709274292, + "num_tokens": 224596889.0, + "step": 8681 + }, + { + "epoch": 0.9534372940918076, + "grad_norm": 1.797894835472107, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7155963182449341, + "num_tokens": 224632269.0, + "step": 8682 + }, + { + "epoch": 0.9535471117944213, + "grad_norm": 2.05155348777771, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.6998200416564941, + "num_tokens": 224659968.0, + "step": 8683 + }, + { + "epoch": 0.953656929497035, + "grad_norm": 1.9751781225204468, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7166832685470581, + "num_tokens": 224689653.0, + "step": 8684 + }, + { + "epoch": 0.9537667471996486, + "grad_norm": 1.9880869388580322, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.697393000125885, + "num_tokens": 224717652.0, + "step": 8685 + }, + { + "epoch": 0.9538765649022622, + "grad_norm": 1.9053471088409424, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6921350955963135, + "num_tokens": 224751602.0, + "step": 8686 + }, + { + "epoch": 0.9539863826048759, + "grad_norm": 2.2485008239746094, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6959552764892578, + "num_tokens": 224775812.0, + "step": 8687 + }, + { + "epoch": 0.9540962003074895, + "grad_norm": 2.2374465465545654, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7240782976150513, + "num_tokens": 224798433.0, + "step": 8688 + }, + { + "epoch": 0.9542060180101032, + "grad_norm": 2.097660779953003, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6957653164863586, + "num_tokens": 224826605.0, + "step": 8689 + }, + { + "epoch": 0.9543158357127169, + "grad_norm": 2.1716341972351074, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7266197204589844, + "num_tokens": 224850608.0, + "step": 8690 + }, + { + "epoch": 0.9544256534153306, + "grad_norm": 2.4779622554779053, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7078555822372437, + "num_tokens": 224872229.0, + "step": 8691 + }, + { + "epoch": 0.9545354711179442, + "grad_norm": 2.120056629180908, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7225399017333984, + "num_tokens": 224899062.0, + "step": 8692 + }, + { + "epoch": 0.9546452888205579, + "grad_norm": 2.4211297035217285, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7150610685348511, + "num_tokens": 224920953.0, + "step": 8693 + }, + { + "epoch": 0.9547551065231715, + "grad_norm": 2.282092332839966, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7015599012374878, + "num_tokens": 224945827.0, + "step": 8694 + }, + { + "epoch": 0.9548649242257852, + "grad_norm": 2.100257635116577, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7040650248527527, + "num_tokens": 224975422.0, + "step": 8695 + }, + { + "epoch": 0.9549747419283988, + "grad_norm": 2.2138993740081787, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6975754499435425, + "num_tokens": 225001592.0, + "step": 8696 + }, + { + "epoch": 0.9550845596310125, + "grad_norm": 2.562476873397827, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.720703125, + "num_tokens": 225022802.0, + "step": 8697 + }, + { + "epoch": 0.9551943773336262, + "grad_norm": 2.2507617473602295, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.68651282787323, + "num_tokens": 225048842.0, + "step": 8698 + }, + { + "epoch": 0.9553041950362399, + "grad_norm": 1.942553162574768, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7091706991195679, + "num_tokens": 225080619.0, + "step": 8699 + }, + { + "epoch": 0.9554140127388535, + "grad_norm": 1.9519593715667725, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7114722728729248, + "num_tokens": 225110529.0, + "step": 8700 + }, + { + "epoch": 0.9555238304414672, + "grad_norm": 2.126788377761841, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7365648746490479, + "num_tokens": 225136463.0, + "step": 8701 + }, + { + "epoch": 0.9556336481440808, + "grad_norm": 2.2018954753875732, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.687857985496521, + "num_tokens": 225161023.0, + "step": 8702 + }, + { + "epoch": 0.9557434658466945, + "grad_norm": 2.1170637607574463, + "learning_rate": 1e-06, + "loss": 1.0978, + "mean_token_accuracy": 0.6729874610900879, + "num_tokens": 225188822.0, + "step": 8703 + }, + { + "epoch": 0.9558532835493081, + "grad_norm": 1.9656143188476562, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7053869962692261, + "num_tokens": 225219426.0, + "step": 8704 + }, + { + "epoch": 0.9559631012519219, + "grad_norm": 2.332970142364502, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7128016352653503, + "num_tokens": 225241096.0, + "step": 8705 + }, + { + "epoch": 0.9560729189545355, + "grad_norm": 2.2304866313934326, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.706928014755249, + "num_tokens": 225264747.0, + "step": 8706 + }, + { + "epoch": 0.9561827366571491, + "grad_norm": 2.285517930984497, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7069856524467468, + "num_tokens": 225287683.0, + "step": 8707 + }, + { + "epoch": 0.9562925543597628, + "grad_norm": 2.056626796722412, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6922423839569092, + "num_tokens": 225316574.0, + "step": 8708 + }, + { + "epoch": 0.9564023720623764, + "grad_norm": 2.1169352531433105, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7095324993133545, + "num_tokens": 225343257.0, + "step": 8709 + }, + { + "epoch": 0.9565121897649901, + "grad_norm": 2.2761447429656982, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7406374216079712, + "num_tokens": 225366372.0, + "step": 8710 + }, + { + "epoch": 0.9566220074676037, + "grad_norm": 2.0054678916931152, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6901059150695801, + "num_tokens": 225397796.0, + "step": 8711 + }, + { + "epoch": 0.9567318251702175, + "grad_norm": 1.9963933229446411, + "learning_rate": 1e-06, + "loss": 1.1132, + "mean_token_accuracy": 0.6786406636238098, + "num_tokens": 225431624.0, + "step": 8712 + }, + { + "epoch": 0.9568416428728311, + "grad_norm": 2.0771255493164062, + "learning_rate": 1e-06, + "loss": 1.0779, + "mean_token_accuracy": 0.6795639991760254, + "num_tokens": 225459819.0, + "step": 8713 + }, + { + "epoch": 0.9569514605754448, + "grad_norm": 2.0958478450775146, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7080879807472229, + "num_tokens": 225487993.0, + "step": 8714 + }, + { + "epoch": 0.9570612782780584, + "grad_norm": 2.1011569499969482, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7200557589530945, + "num_tokens": 225513334.0, + "step": 8715 + }, + { + "epoch": 0.9571710959806721, + "grad_norm": 2.2351608276367188, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.711930513381958, + "num_tokens": 225537623.0, + "step": 8716 + }, + { + "epoch": 0.9572809136832857, + "grad_norm": 2.6064820289611816, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7081858515739441, + "num_tokens": 225556885.0, + "step": 8717 + }, + { + "epoch": 0.9573907313858994, + "grad_norm": 2.3062851428985596, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7012633085250854, + "num_tokens": 225582847.0, + "step": 8718 + }, + { + "epoch": 0.9575005490885131, + "grad_norm": 2.666294574737549, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7366624474525452, + "num_tokens": 225598872.0, + "step": 8719 + }, + { + "epoch": 0.9576103667911268, + "grad_norm": 2.2139439582824707, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6985218524932861, + "num_tokens": 225624823.0, + "step": 8720 + }, + { + "epoch": 0.9577201844937404, + "grad_norm": 2.215324878692627, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.7011568546295166, + "num_tokens": 225649755.0, + "step": 8721 + }, + { + "epoch": 0.9578300021963541, + "grad_norm": 1.8354058265686035, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6886063814163208, + "num_tokens": 225683143.0, + "step": 8722 + }, + { + "epoch": 0.9579398198989677, + "grad_norm": 1.8534846305847168, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7222840189933777, + "num_tokens": 225714652.0, + "step": 8723 + }, + { + "epoch": 0.9580496376015813, + "grad_norm": 2.2524309158325195, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7197207808494568, + "num_tokens": 225737649.0, + "step": 8724 + }, + { + "epoch": 0.958159455304195, + "grad_norm": 2.157158374786377, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7175583839416504, + "num_tokens": 225762807.0, + "step": 8725 + }, + { + "epoch": 0.9582692730068088, + "grad_norm": 2.0628631114959717, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6895163655281067, + "num_tokens": 225790806.0, + "step": 8726 + }, + { + "epoch": 0.9583790907094224, + "grad_norm": 2.048076868057251, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7154503464698792, + "num_tokens": 225818475.0, + "step": 8727 + }, + { + "epoch": 0.958488908412036, + "grad_norm": 2.4219579696655273, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7122162580490112, + "num_tokens": 225838758.0, + "step": 8728 + }, + { + "epoch": 0.9585987261146497, + "grad_norm": 1.9934988021850586, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7095905542373657, + "num_tokens": 225869813.0, + "step": 8729 + }, + { + "epoch": 0.9587085438172633, + "grad_norm": 2.1941213607788086, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7197762727737427, + "num_tokens": 225893794.0, + "step": 8730 + }, + { + "epoch": 0.958818361519877, + "grad_norm": 2.0779483318328857, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6991890668869019, + "num_tokens": 225920531.0, + "step": 8731 + }, + { + "epoch": 0.9589281792224906, + "grad_norm": 1.9356292486190796, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7123621702194214, + "num_tokens": 225951936.0, + "step": 8732 + }, + { + "epoch": 0.9590379969251043, + "grad_norm": 2.3319766521453857, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7154060006141663, + "num_tokens": 225974222.0, + "step": 8733 + }, + { + "epoch": 0.959147814627718, + "grad_norm": 2.2713675498962402, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7257025241851807, + "num_tokens": 225997213.0, + "step": 8734 + }, + { + "epoch": 0.9592576323303317, + "grad_norm": 2.41339373588562, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7048059701919556, + "num_tokens": 226017402.0, + "step": 8735 + }, + { + "epoch": 0.9593674500329453, + "grad_norm": 2.110492467880249, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6864479184150696, + "num_tokens": 226046124.0, + "step": 8736 + }, + { + "epoch": 0.959477267735559, + "grad_norm": 2.2020342350006104, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7008552551269531, + "num_tokens": 226072832.0, + "step": 8737 + }, + { + "epoch": 0.9595870854381726, + "grad_norm": 2.0187644958496094, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7144889831542969, + "num_tokens": 226099350.0, + "step": 8738 + }, + { + "epoch": 0.9596969031407863, + "grad_norm": 2.1552274227142334, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7104952335357666, + "num_tokens": 226123899.0, + "step": 8739 + }, + { + "epoch": 0.9598067208433999, + "grad_norm": 2.06105375289917, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7123773694038391, + "num_tokens": 226151450.0, + "step": 8740 + }, + { + "epoch": 0.9599165385460137, + "grad_norm": 1.990661382675171, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7181053757667542, + "num_tokens": 226180002.0, + "step": 8741 + }, + { + "epoch": 0.9600263562486273, + "grad_norm": 1.9849510192871094, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6963159441947937, + "num_tokens": 226207642.0, + "step": 8742 + }, + { + "epoch": 0.960136173951241, + "grad_norm": 2.2798120975494385, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7065946459770203, + "num_tokens": 226233048.0, + "step": 8743 + }, + { + "epoch": 0.9602459916538546, + "grad_norm": 2.2807209491729736, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7250397205352783, + "num_tokens": 226254190.0, + "step": 8744 + }, + { + "epoch": 0.9603558093564682, + "grad_norm": 2.1208324432373047, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.70374596118927, + "num_tokens": 226280390.0, + "step": 8745 + }, + { + "epoch": 0.9604656270590819, + "grad_norm": 2.0317463874816895, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7055679559707642, + "num_tokens": 226308489.0, + "step": 8746 + }, + { + "epoch": 0.9605754447616955, + "grad_norm": 2.1657469272613525, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7160865068435669, + "num_tokens": 226334100.0, + "step": 8747 + }, + { + "epoch": 0.9606852624643093, + "grad_norm": 2.4023468494415283, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7223423719406128, + "num_tokens": 226355150.0, + "step": 8748 + }, + { + "epoch": 0.9607950801669229, + "grad_norm": 2.2986133098602295, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6845311522483826, + "num_tokens": 226385231.0, + "step": 8749 + }, + { + "epoch": 0.9609048978695366, + "grad_norm": 1.9219884872436523, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.712067723274231, + "num_tokens": 226416209.0, + "step": 8750 + }, + { + "epoch": 0.9610147155721502, + "grad_norm": 2.4710166454315186, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7138376832008362, + "num_tokens": 226435635.0, + "step": 8751 + }, + { + "epoch": 0.9611245332747639, + "grad_norm": 1.8863261938095093, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7023619413375854, + "num_tokens": 226468132.0, + "step": 8752 + }, + { + "epoch": 0.9612343509773775, + "grad_norm": 2.1639416217803955, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.725874662399292, + "num_tokens": 226492763.0, + "step": 8753 + }, + { + "epoch": 0.9613441686799912, + "grad_norm": 2.2088706493377686, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7085816860198975, + "num_tokens": 226518240.0, + "step": 8754 + }, + { + "epoch": 0.9614539863826049, + "grad_norm": 2.4472429752349854, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7138009667396545, + "num_tokens": 226538071.0, + "step": 8755 + }, + { + "epoch": 0.9615638040852186, + "grad_norm": 1.8402020931243896, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7072449326515198, + "num_tokens": 226567951.0, + "step": 8756 + }, + { + "epoch": 0.9616736217878322, + "grad_norm": 2.6285195350646973, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7142918109893799, + "num_tokens": 226585283.0, + "step": 8757 + }, + { + "epoch": 0.9617834394904459, + "grad_norm": 1.988303303718567, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7023481726646423, + "num_tokens": 226614720.0, + "step": 8758 + }, + { + "epoch": 0.9618932571930595, + "grad_norm": 2.486539602279663, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7110652923583984, + "num_tokens": 226636638.0, + "step": 8759 + }, + { + "epoch": 0.9620030748956732, + "grad_norm": 2.0599818229675293, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6961507201194763, + "num_tokens": 226664693.0, + "step": 8760 + }, + { + "epoch": 0.9621128925982868, + "grad_norm": 2.194192409515381, + "learning_rate": 1e-06, + "loss": 1.0492, + "mean_token_accuracy": 0.6784325242042542, + "num_tokens": 226691569.0, + "step": 8761 + }, + { + "epoch": 0.9622227103009005, + "grad_norm": 2.6042709350585938, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7256774306297302, + "num_tokens": 226708968.0, + "step": 8762 + }, + { + "epoch": 0.9623325280035142, + "grad_norm": 1.9650455713272095, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6947453022003174, + "num_tokens": 226739371.0, + "step": 8763 + }, + { + "epoch": 0.9624423457061279, + "grad_norm": 2.139364719390869, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6979552507400513, + "num_tokens": 226765112.0, + "step": 8764 + }, + { + "epoch": 0.9625521634087415, + "grad_norm": 2.201993942260742, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7049954533576965, + "num_tokens": 226790665.0, + "step": 8765 + }, + { + "epoch": 0.9626619811113551, + "grad_norm": 1.9798805713653564, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.707202672958374, + "num_tokens": 226820430.0, + "step": 8766 + }, + { + "epoch": 0.9627717988139688, + "grad_norm": 1.94273042678833, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6944236755371094, + "num_tokens": 226852716.0, + "step": 8767 + }, + { + "epoch": 0.9628816165165824, + "grad_norm": 2.0337510108947754, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7175363302230835, + "num_tokens": 226880918.0, + "step": 8768 + }, + { + "epoch": 0.9629914342191961, + "grad_norm": 2.2280991077423096, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7036436796188354, + "num_tokens": 226904758.0, + "step": 8769 + }, + { + "epoch": 0.9631012519218098, + "grad_norm": 2.1275196075439453, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6936272978782654, + "num_tokens": 226933464.0, + "step": 8770 + }, + { + "epoch": 0.9632110696244235, + "grad_norm": 2.071026563644409, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7159810662269592, + "num_tokens": 226961442.0, + "step": 8771 + }, + { + "epoch": 0.9633208873270371, + "grad_norm": 2.192359447479248, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.7013939023017883, + "num_tokens": 226987718.0, + "step": 8772 + }, + { + "epoch": 0.9634307050296508, + "grad_norm": 2.1961374282836914, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7012268304824829, + "num_tokens": 227014707.0, + "step": 8773 + }, + { + "epoch": 0.9635405227322644, + "grad_norm": 2.0894615650177, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6971467733383179, + "num_tokens": 227042658.0, + "step": 8774 + }, + { + "epoch": 0.9636503404348781, + "grad_norm": 2.2188198566436768, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7191940546035767, + "num_tokens": 227065980.0, + "step": 8775 + }, + { + "epoch": 0.9637601581374917, + "grad_norm": 2.1524367332458496, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6882205009460449, + "num_tokens": 227092798.0, + "step": 8776 + }, + { + "epoch": 0.9638699758401055, + "grad_norm": 2.0953898429870605, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7001748085021973, + "num_tokens": 227119037.0, + "step": 8777 + }, + { + "epoch": 0.9639797935427191, + "grad_norm": 1.9965280294418335, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7136620283126831, + "num_tokens": 227147385.0, + "step": 8778 + }, + { + "epoch": 0.9640896112453328, + "grad_norm": 2.039477586746216, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7007153034210205, + "num_tokens": 227175846.0, + "step": 8779 + }, + { + "epoch": 0.9641994289479464, + "grad_norm": 2.089012861251831, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6944746971130371, + "num_tokens": 227203360.0, + "step": 8780 + }, + { + "epoch": 0.9643092466505601, + "grad_norm": 2.1842305660247803, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7004407048225403, + "num_tokens": 227228358.0, + "step": 8781 + }, + { + "epoch": 0.9644190643531737, + "grad_norm": 2.102451801300049, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7251108884811401, + "num_tokens": 227255492.0, + "step": 8782 + }, + { + "epoch": 0.9645288820557874, + "grad_norm": 2.412266254425049, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7134582996368408, + "num_tokens": 227276939.0, + "step": 8783 + }, + { + "epoch": 0.9646386997584011, + "grad_norm": 1.9998518228530884, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6983716487884521, + "num_tokens": 227305598.0, + "step": 8784 + }, + { + "epoch": 0.9647485174610148, + "grad_norm": 2.2408864498138428, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7007783651351929, + "num_tokens": 227329581.0, + "step": 8785 + }, + { + "epoch": 0.9648583351636284, + "grad_norm": 2.210937976837158, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7105677127838135, + "num_tokens": 227354844.0, + "step": 8786 + }, + { + "epoch": 0.964968152866242, + "grad_norm": 2.0454213619232178, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7002049684524536, + "num_tokens": 227383559.0, + "step": 8787 + }, + { + "epoch": 0.9650779705688557, + "grad_norm": 2.2157511711120605, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7178695201873779, + "num_tokens": 227407240.0, + "step": 8788 + }, + { + "epoch": 0.9651877882714693, + "grad_norm": 1.9335274696350098, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7078319787979126, + "num_tokens": 227438972.0, + "step": 8789 + }, + { + "epoch": 0.965297605974083, + "grad_norm": 2.061230421066284, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.7055423259735107, + "num_tokens": 227466755.0, + "step": 8790 + }, + { + "epoch": 0.9654074236766966, + "grad_norm": 2.48276686668396, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7361611127853394, + "num_tokens": 227486042.0, + "step": 8791 + }, + { + "epoch": 0.9655172413793104, + "grad_norm": 2.0124881267547607, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7326536178588867, + "num_tokens": 227513569.0, + "step": 8792 + }, + { + "epoch": 0.965627059081924, + "grad_norm": 2.440523147583008, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7034275531768799, + "num_tokens": 227536632.0, + "step": 8793 + }, + { + "epoch": 0.9657368767845377, + "grad_norm": 2.038580894470215, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6916295289993286, + "num_tokens": 227565842.0, + "step": 8794 + }, + { + "epoch": 0.9658466944871513, + "grad_norm": 1.9377278089523315, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7038046717643738, + "num_tokens": 227595085.0, + "step": 8795 + }, + { + "epoch": 0.965956512189765, + "grad_norm": 2.1608574390411377, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7120479941368103, + "num_tokens": 227619531.0, + "step": 8796 + }, + { + "epoch": 0.9660663298923786, + "grad_norm": 2.3635096549987793, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.6943324208259583, + "num_tokens": 227643118.0, + "step": 8797 + }, + { + "epoch": 0.9661761475949923, + "grad_norm": 2.1357016563415527, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7134100198745728, + "num_tokens": 227668000.0, + "step": 8798 + }, + { + "epoch": 0.966285965297606, + "grad_norm": 2.2368595600128174, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7293124198913574, + "num_tokens": 227690524.0, + "step": 8799 + }, + { + "epoch": 0.9663957830002197, + "grad_norm": 2.201951742172241, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7172385454177856, + "num_tokens": 227714620.0, + "step": 8800 + }, + { + "epoch": 0.9665056007028333, + "grad_norm": 2.111229658126831, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7254936695098877, + "num_tokens": 227739464.0, + "step": 8801 + }, + { + "epoch": 0.966615418405447, + "grad_norm": 2.1907522678375244, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.7029633522033691, + "num_tokens": 227764732.0, + "step": 8802 + }, + { + "epoch": 0.9667252361080606, + "grad_norm": 1.9428340196609497, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6841340661048889, + "num_tokens": 227795769.0, + "step": 8803 + }, + { + "epoch": 0.9668350538106742, + "grad_norm": 2.262857675552368, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.716418445110321, + "num_tokens": 227820704.0, + "step": 8804 + }, + { + "epoch": 0.9669448715132879, + "grad_norm": 1.867931842803955, + "learning_rate": 1e-06, + "loss": 1.0577, + "mean_token_accuracy": 0.691571831703186, + "num_tokens": 227857518.0, + "step": 8805 + }, + { + "epoch": 0.9670546892159017, + "grad_norm": 2.194821357727051, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7208775281906128, + "num_tokens": 227881946.0, + "step": 8806 + }, + { + "epoch": 0.9671645069185153, + "grad_norm": 2.044542074203491, + "learning_rate": 1e-06, + "loss": 1.0407, + "mean_token_accuracy": 0.687629759311676, + "num_tokens": 227912468.0, + "step": 8807 + }, + { + "epoch": 0.967274324621129, + "grad_norm": 2.3077406883239746, + "learning_rate": 1e-06, + "loss": 1.0622, + "mean_token_accuracy": 0.6817540526390076, + "num_tokens": 227938208.0, + "step": 8808 + }, + { + "epoch": 0.9673841423237426, + "grad_norm": 2.2141075134277344, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6954655051231384, + "num_tokens": 227964581.0, + "step": 8809 + }, + { + "epoch": 0.9674939600263562, + "grad_norm": 2.070091485977173, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7067541480064392, + "num_tokens": 227995051.0, + "step": 8810 + }, + { + "epoch": 0.9676037777289699, + "grad_norm": 2.7605996131896973, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.705817699432373, + "num_tokens": 228011822.0, + "step": 8811 + }, + { + "epoch": 0.9677135954315835, + "grad_norm": 2.0991623401641846, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6733262538909912, + "num_tokens": 228041097.0, + "step": 8812 + }, + { + "epoch": 0.9678234131341973, + "grad_norm": 1.8918378353118896, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7082697749137878, + "num_tokens": 228073745.0, + "step": 8813 + }, + { + "epoch": 0.9679332308368109, + "grad_norm": 2.343266010284424, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7156325578689575, + "num_tokens": 228096488.0, + "step": 8814 + }, + { + "epoch": 0.9680430485394246, + "grad_norm": 2.033461093902588, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6962020993232727, + "num_tokens": 228126843.0, + "step": 8815 + }, + { + "epoch": 0.9681528662420382, + "grad_norm": 2.4483859539031982, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.725459098815918, + "num_tokens": 228148818.0, + "step": 8816 + }, + { + "epoch": 0.9682626839446519, + "grad_norm": 2.252310037612915, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7286649942398071, + "num_tokens": 228171522.0, + "step": 8817 + }, + { + "epoch": 0.9683725016472655, + "grad_norm": 2.2809176445007324, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6906176209449768, + "num_tokens": 228197254.0, + "step": 8818 + }, + { + "epoch": 0.9684823193498792, + "grad_norm": 2.313551425933838, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7031527757644653, + "num_tokens": 228219818.0, + "step": 8819 + }, + { + "epoch": 0.9685921370524928, + "grad_norm": 2.388638496398926, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7182390689849854, + "num_tokens": 228239719.0, + "step": 8820 + }, + { + "epoch": 0.9687019547551066, + "grad_norm": 2.230346441268921, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7263935208320618, + "num_tokens": 228262713.0, + "step": 8821 + }, + { + "epoch": 0.9688117724577202, + "grad_norm": 2.134963035583496, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7081603407859802, + "num_tokens": 228289415.0, + "step": 8822 + }, + { + "epoch": 0.9689215901603339, + "grad_norm": 2.102105140686035, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.707001805305481, + "num_tokens": 228315429.0, + "step": 8823 + }, + { + "epoch": 0.9690314078629475, + "grad_norm": 2.172185182571411, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7415789365768433, + "num_tokens": 228339247.0, + "step": 8824 + }, + { + "epoch": 0.9691412255655611, + "grad_norm": 2.1307835578918457, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7242871522903442, + "num_tokens": 228364715.0, + "step": 8825 + }, + { + "epoch": 0.9692510432681748, + "grad_norm": 2.449120044708252, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.727812647819519, + "num_tokens": 228386350.0, + "step": 8826 + }, + { + "epoch": 0.9693608609707884, + "grad_norm": 1.9489197731018066, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7072961330413818, + "num_tokens": 228418504.0, + "step": 8827 + }, + { + "epoch": 0.9694706786734022, + "grad_norm": 2.2321219444274902, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7167487144470215, + "num_tokens": 228443214.0, + "step": 8828 + }, + { + "epoch": 0.9695804963760158, + "grad_norm": 2.3312621116638184, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7185554504394531, + "num_tokens": 228465758.0, + "step": 8829 + }, + { + "epoch": 0.9696903140786295, + "grad_norm": 2.0307581424713135, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.717254638671875, + "num_tokens": 228492918.0, + "step": 8830 + }, + { + "epoch": 0.9698001317812431, + "grad_norm": 2.237720251083374, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.6961846947669983, + "num_tokens": 228517881.0, + "step": 8831 + }, + { + "epoch": 0.9699099494838568, + "grad_norm": 1.9302692413330078, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7131345272064209, + "num_tokens": 228548073.0, + "step": 8832 + }, + { + "epoch": 0.9700197671864704, + "grad_norm": 2.425729274749756, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7180565595626831, + "num_tokens": 228568379.0, + "step": 8833 + }, + { + "epoch": 0.9701295848890841, + "grad_norm": 2.131470203399658, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7085669636726379, + "num_tokens": 228595878.0, + "step": 8834 + }, + { + "epoch": 0.9702394025916978, + "grad_norm": 2.112832546234131, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.7012643218040466, + "num_tokens": 228626734.0, + "step": 8835 + }, + { + "epoch": 0.9703492202943115, + "grad_norm": 2.1271848678588867, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.690166711807251, + "num_tokens": 228657659.0, + "step": 8836 + }, + { + "epoch": 0.9704590379969251, + "grad_norm": 2.359762191772461, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7090080976486206, + "num_tokens": 228679738.0, + "step": 8837 + }, + { + "epoch": 0.9705688556995388, + "grad_norm": 2.1595492362976074, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7108691930770874, + "num_tokens": 228706159.0, + "step": 8838 + }, + { + "epoch": 0.9706786734021524, + "grad_norm": 2.441688060760498, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7120251655578613, + "num_tokens": 228727542.0, + "step": 8839 + }, + { + "epoch": 0.9707884911047661, + "grad_norm": 2.233806610107422, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.6959421634674072, + "num_tokens": 228751295.0, + "step": 8840 + }, + { + "epoch": 0.9708983088073797, + "grad_norm": 2.257735252380371, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.7004804611206055, + "num_tokens": 228778417.0, + "step": 8841 + }, + { + "epoch": 0.9710081265099935, + "grad_norm": 1.834384799003601, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6972647309303284, + "num_tokens": 228812038.0, + "step": 8842 + }, + { + "epoch": 0.9711179442126071, + "grad_norm": 2.0850281715393066, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7080587148666382, + "num_tokens": 228839443.0, + "step": 8843 + }, + { + "epoch": 0.9712277619152208, + "grad_norm": 2.159102439880371, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7207074165344238, + "num_tokens": 228863670.0, + "step": 8844 + }, + { + "epoch": 0.9713375796178344, + "grad_norm": 1.8031364679336548, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7059434652328491, + "num_tokens": 228899048.0, + "step": 8845 + }, + { + "epoch": 0.971447397320448, + "grad_norm": 2.164296865463257, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6843922138214111, + "num_tokens": 228925711.0, + "step": 8846 + }, + { + "epoch": 0.9715572150230617, + "grad_norm": 2.3371450901031494, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7112323045730591, + "num_tokens": 228948934.0, + "step": 8847 + }, + { + "epoch": 0.9716670327256753, + "grad_norm": 2.111525297164917, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7120687961578369, + "num_tokens": 228975185.0, + "step": 8848 + }, + { + "epoch": 0.971776850428289, + "grad_norm": 2.127671480178833, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7288051843643188, + "num_tokens": 229001724.0, + "step": 8849 + }, + { + "epoch": 0.9718866681309027, + "grad_norm": 2.5577852725982666, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7280830144882202, + "num_tokens": 229023802.0, + "step": 8850 + }, + { + "epoch": 0.9719964858335164, + "grad_norm": 2.006272315979004, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7224875688552856, + "num_tokens": 229053214.0, + "step": 8851 + }, + { + "epoch": 0.97210630353613, + "grad_norm": 1.982175350189209, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7127031683921814, + "num_tokens": 229082556.0, + "step": 8852 + }, + { + "epoch": 0.9722161212387437, + "grad_norm": 2.416574478149414, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7017650604248047, + "num_tokens": 229102678.0, + "step": 8853 + }, + { + "epoch": 0.9723259389413573, + "grad_norm": 2.314056396484375, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.709606409072876, + "num_tokens": 229127059.0, + "step": 8854 + }, + { + "epoch": 0.972435756643971, + "grad_norm": 2.2496652603149414, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.7024569511413574, + "num_tokens": 229151902.0, + "step": 8855 + }, + { + "epoch": 0.9725455743465846, + "grad_norm": 2.275627613067627, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7212190628051758, + "num_tokens": 229174950.0, + "step": 8856 + }, + { + "epoch": 0.9726553920491984, + "grad_norm": 2.343923568725586, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7191827893257141, + "num_tokens": 229196120.0, + "step": 8857 + }, + { + "epoch": 0.972765209751812, + "grad_norm": 2.1582837104797363, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7000195980072021, + "num_tokens": 229223840.0, + "step": 8858 + }, + { + "epoch": 0.9728750274544257, + "grad_norm": 2.4975273609161377, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7137390971183777, + "num_tokens": 229243153.0, + "step": 8859 + }, + { + "epoch": 0.9729848451570393, + "grad_norm": 2.1482558250427246, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7008475065231323, + "num_tokens": 229270777.0, + "step": 8860 + }, + { + "epoch": 0.973094662859653, + "grad_norm": 2.355988025665283, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7135611772537231, + "num_tokens": 229292593.0, + "step": 8861 + }, + { + "epoch": 0.9732044805622666, + "grad_norm": 2.2354609966278076, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7129979133605957, + "num_tokens": 229317275.0, + "step": 8862 + }, + { + "epoch": 0.9733142982648803, + "grad_norm": 2.1555776596069336, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6934407949447632, + "num_tokens": 229343841.0, + "step": 8863 + }, + { + "epoch": 0.973424115967494, + "grad_norm": 2.326582431793213, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6955260038375854, + "num_tokens": 229368220.0, + "step": 8864 + }, + { + "epoch": 0.9735339336701077, + "grad_norm": 2.4167144298553467, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6932541131973267, + "num_tokens": 229389388.0, + "step": 8865 + }, + { + "epoch": 0.9736437513727213, + "grad_norm": 2.031240940093994, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6987926363945007, + "num_tokens": 229419230.0, + "step": 8866 + }, + { + "epoch": 0.973753569075335, + "grad_norm": 1.9973223209381104, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6930023431777954, + "num_tokens": 229448049.0, + "step": 8867 + }, + { + "epoch": 0.9738633867779486, + "grad_norm": 2.2594995498657227, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6961721777915955, + "num_tokens": 229472943.0, + "step": 8868 + }, + { + "epoch": 0.9739732044805622, + "grad_norm": 2.447313070297241, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7245275974273682, + "num_tokens": 229492958.0, + "step": 8869 + }, + { + "epoch": 0.9740830221831759, + "grad_norm": 1.980911374092102, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7538028955459595, + "num_tokens": 229518904.0, + "step": 8870 + }, + { + "epoch": 0.9741928398857896, + "grad_norm": 2.054494857788086, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7069560289382935, + "num_tokens": 229546570.0, + "step": 8871 + }, + { + "epoch": 0.9743026575884033, + "grad_norm": 2.3483164310455322, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7168996334075928, + "num_tokens": 229567965.0, + "step": 8872 + }, + { + "epoch": 0.9744124752910169, + "grad_norm": 2.0271098613739014, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7116844058036804, + "num_tokens": 229596198.0, + "step": 8873 + }, + { + "epoch": 0.9745222929936306, + "grad_norm": 2.0157101154327393, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7059588432312012, + "num_tokens": 229626224.0, + "step": 8874 + }, + { + "epoch": 0.9746321106962442, + "grad_norm": 2.241525650024414, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6906684637069702, + "num_tokens": 229652706.0, + "step": 8875 + }, + { + "epoch": 0.9747419283988579, + "grad_norm": 2.105837821960449, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7084885835647583, + "num_tokens": 229679203.0, + "step": 8876 + }, + { + "epoch": 0.9748517461014715, + "grad_norm": 2.3398871421813965, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7009701132774353, + "num_tokens": 229702096.0, + "step": 8877 + }, + { + "epoch": 0.9749615638040853, + "grad_norm": 2.4798216819763184, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6997808218002319, + "num_tokens": 229721871.0, + "step": 8878 + }, + { + "epoch": 0.9750713815066989, + "grad_norm": 2.3309085369110107, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6870365142822266, + "num_tokens": 229745519.0, + "step": 8879 + }, + { + "epoch": 0.9751811992093126, + "grad_norm": 2.1495237350463867, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7176632881164551, + "num_tokens": 229771220.0, + "step": 8880 + }, + { + "epoch": 0.9752910169119262, + "grad_norm": 2.301351308822632, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6930457949638367, + "num_tokens": 229795634.0, + "step": 8881 + }, + { + "epoch": 0.9754008346145399, + "grad_norm": 2.402055501937866, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7364294528961182, + "num_tokens": 229815447.0, + "step": 8882 + }, + { + "epoch": 0.9755106523171535, + "grad_norm": 1.9153428077697754, + "learning_rate": 1e-06, + "loss": 1.1156, + "mean_token_accuracy": 0.6634235978126526, + "num_tokens": 229846624.0, + "step": 8883 + }, + { + "epoch": 0.9756204700197671, + "grad_norm": 2.244542360305786, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7114705443382263, + "num_tokens": 229870063.0, + "step": 8884 + }, + { + "epoch": 0.9757302877223808, + "grad_norm": 1.957413911819458, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7074620127677917, + "num_tokens": 229898350.0, + "step": 8885 + }, + { + "epoch": 0.9758401054249946, + "grad_norm": 2.3742311000823975, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6935729384422302, + "num_tokens": 229921538.0, + "step": 8886 + }, + { + "epoch": 0.9759499231276082, + "grad_norm": 2.2160611152648926, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6979142427444458, + "num_tokens": 229945119.0, + "step": 8887 + }, + { + "epoch": 0.9760597408302218, + "grad_norm": 2.324293613433838, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6906126737594604, + "num_tokens": 229968920.0, + "step": 8888 + }, + { + "epoch": 0.9761695585328355, + "grad_norm": 1.864622950553894, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7104986906051636, + "num_tokens": 230003802.0, + "step": 8889 + }, + { + "epoch": 0.9762793762354491, + "grad_norm": 2.2035279273986816, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.699897050857544, + "num_tokens": 230030227.0, + "step": 8890 + }, + { + "epoch": 0.9763891939380628, + "grad_norm": 2.4351143836975098, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7157377600669861, + "num_tokens": 230053413.0, + "step": 8891 + }, + { + "epoch": 0.9764990116406764, + "grad_norm": 2.3979177474975586, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7036339044570923, + "num_tokens": 230076540.0, + "step": 8892 + }, + { + "epoch": 0.9766088293432902, + "grad_norm": 2.1409668922424316, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7107151746749878, + "num_tokens": 230102079.0, + "step": 8893 + }, + { + "epoch": 0.9767186470459038, + "grad_norm": 2.0945920944213867, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6961635947227478, + "num_tokens": 230129033.0, + "step": 8894 + }, + { + "epoch": 0.9768284647485175, + "grad_norm": 2.033064603805542, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6915854215621948, + "num_tokens": 230158669.0, + "step": 8895 + }, + { + "epoch": 0.9769382824511311, + "grad_norm": 2.484764575958252, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7246226072311401, + "num_tokens": 230178979.0, + "step": 8896 + }, + { + "epoch": 0.9770481001537448, + "grad_norm": 2.168241500854492, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.713221549987793, + "num_tokens": 230203677.0, + "step": 8897 + }, + { + "epoch": 0.9771579178563584, + "grad_norm": 2.4595725536346436, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7411656379699707, + "num_tokens": 230223685.0, + "step": 8898 + }, + { + "epoch": 0.9772677355589721, + "grad_norm": 2.380764961242676, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7281558513641357, + "num_tokens": 230243836.0, + "step": 8899 + }, + { + "epoch": 0.9773775532615858, + "grad_norm": 2.2080748081207275, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6952451467514038, + "num_tokens": 230267068.0, + "step": 8900 + }, + { + "epoch": 0.9774873709641995, + "grad_norm": 2.1391007900238037, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6979520320892334, + "num_tokens": 230293163.0, + "step": 8901 + }, + { + "epoch": 0.9775971886668131, + "grad_norm": 1.9462330341339111, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6831783652305603, + "num_tokens": 230325348.0, + "step": 8902 + }, + { + "epoch": 0.9777070063694268, + "grad_norm": 2.539276361465454, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7332649230957031, + "num_tokens": 230345616.0, + "step": 8903 + }, + { + "epoch": 0.9778168240720404, + "grad_norm": 2.3912031650543213, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7165240049362183, + "num_tokens": 230366751.0, + "step": 8904 + }, + { + "epoch": 0.977926641774654, + "grad_norm": 2.126131772994995, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6851130723953247, + "num_tokens": 230395863.0, + "step": 8905 + }, + { + "epoch": 0.9780364594772677, + "grad_norm": 2.6675961017608643, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.726560115814209, + "num_tokens": 230414203.0, + "step": 8906 + }, + { + "epoch": 0.9781462771798815, + "grad_norm": 2.1637990474700928, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7129949331283569, + "num_tokens": 230440005.0, + "step": 8907 + }, + { + "epoch": 0.9782560948824951, + "grad_norm": 2.309581756591797, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7020605802536011, + "num_tokens": 230464148.0, + "step": 8908 + }, + { + "epoch": 0.9783659125851087, + "grad_norm": 2.3857479095458984, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7117407321929932, + "num_tokens": 230484594.0, + "step": 8909 + }, + { + "epoch": 0.9784757302877224, + "grad_norm": 2.083432197570801, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7089845538139343, + "num_tokens": 230510260.0, + "step": 8910 + }, + { + "epoch": 0.978585547990336, + "grad_norm": 2.0321733951568604, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.696458101272583, + "num_tokens": 230539461.0, + "step": 8911 + }, + { + "epoch": 0.9786953656929497, + "grad_norm": 2.1621406078338623, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7081055045127869, + "num_tokens": 230566495.0, + "step": 8912 + }, + { + "epoch": 0.9788051833955633, + "grad_norm": 1.8101826906204224, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6942764520645142, + "num_tokens": 230601093.0, + "step": 8913 + }, + { + "epoch": 0.978915001098177, + "grad_norm": 1.9222147464752197, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7090550661087036, + "num_tokens": 230635869.0, + "step": 8914 + }, + { + "epoch": 0.9790248188007907, + "grad_norm": 2.039726495742798, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6946674585342407, + "num_tokens": 230663962.0, + "step": 8915 + }, + { + "epoch": 0.9791346365034044, + "grad_norm": 1.9255179166793823, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6896015405654907, + "num_tokens": 230696912.0, + "step": 8916 + }, + { + "epoch": 0.979244454206018, + "grad_norm": 2.0843005180358887, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7033597826957703, + "num_tokens": 230725370.0, + "step": 8917 + }, + { + "epoch": 0.9793542719086317, + "grad_norm": 2.1935296058654785, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7072075605392456, + "num_tokens": 230749997.0, + "step": 8918 + }, + { + "epoch": 0.9794640896112453, + "grad_norm": 2.156723976135254, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7375432848930359, + "num_tokens": 230774837.0, + "step": 8919 + }, + { + "epoch": 0.979573907313859, + "grad_norm": 2.285130023956299, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7232644557952881, + "num_tokens": 230799299.0, + "step": 8920 + }, + { + "epoch": 0.9796837250164726, + "grad_norm": 2.3384366035461426, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7043215036392212, + "num_tokens": 230822987.0, + "step": 8921 + }, + { + "epoch": 0.9797935427190864, + "grad_norm": 2.5012476444244385, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7071592807769775, + "num_tokens": 230843790.0, + "step": 8922 + }, + { + "epoch": 0.9799033604217, + "grad_norm": 2.253601551055908, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7213159203529358, + "num_tokens": 230866674.0, + "step": 8923 + }, + { + "epoch": 0.9800131781243137, + "grad_norm": 2.1131033897399902, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6954590082168579, + "num_tokens": 230896171.0, + "step": 8924 + }, + { + "epoch": 0.9801229958269273, + "grad_norm": 1.9561570882797241, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6918904781341553, + "num_tokens": 230927449.0, + "step": 8925 + }, + { + "epoch": 0.980232813529541, + "grad_norm": 2.3742079734802246, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7137773036956787, + "num_tokens": 230949584.0, + "step": 8926 + }, + { + "epoch": 0.9803426312321546, + "grad_norm": 2.3803458213806152, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7224849462509155, + "num_tokens": 230971160.0, + "step": 8927 + }, + { + "epoch": 0.9804524489347682, + "grad_norm": 2.068507671356201, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7133706212043762, + "num_tokens": 230998463.0, + "step": 8928 + }, + { + "epoch": 0.980562266637382, + "grad_norm": 2.0122616291046143, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7319122552871704, + "num_tokens": 231027874.0, + "step": 8929 + }, + { + "epoch": 0.9806720843399956, + "grad_norm": 2.075482130050659, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.703467845916748, + "num_tokens": 231054479.0, + "step": 8930 + }, + { + "epoch": 0.9807819020426093, + "grad_norm": 2.0248093605041504, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7016782164573669, + "num_tokens": 231084304.0, + "step": 8931 + }, + { + "epoch": 0.9808917197452229, + "grad_norm": 2.208599805831909, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7109329104423523, + "num_tokens": 231109727.0, + "step": 8932 + }, + { + "epoch": 0.9810015374478366, + "grad_norm": 2.1876800060272217, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6962749361991882, + "num_tokens": 231138001.0, + "step": 8933 + }, + { + "epoch": 0.9811113551504502, + "grad_norm": 2.4346063137054443, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7058711647987366, + "num_tokens": 231161019.0, + "step": 8934 + }, + { + "epoch": 0.9812211728530639, + "grad_norm": 2.393815040588379, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7204797267913818, + "num_tokens": 231182904.0, + "step": 8935 + }, + { + "epoch": 0.9813309905556776, + "grad_norm": 2.375523567199707, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7139734029769897, + "num_tokens": 231205834.0, + "step": 8936 + }, + { + "epoch": 0.9814408082582913, + "grad_norm": 2.3460536003112793, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7172720432281494, + "num_tokens": 231229297.0, + "step": 8937 + }, + { + "epoch": 0.9815506259609049, + "grad_norm": 2.0884017944335938, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7170842885971069, + "num_tokens": 231253732.0, + "step": 8938 + }, + { + "epoch": 0.9816604436635186, + "grad_norm": 2.161318063735962, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.70149165391922, + "num_tokens": 231281803.0, + "step": 8939 + }, + { + "epoch": 0.9817702613661322, + "grad_norm": 1.964751958847046, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6894325017929077, + "num_tokens": 231312159.0, + "step": 8940 + }, + { + "epoch": 0.9818800790687459, + "grad_norm": 2.0895156860351562, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7184234857559204, + "num_tokens": 231339780.0, + "step": 8941 + }, + { + "epoch": 0.9819898967713595, + "grad_norm": 2.162757158279419, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7272993326187134, + "num_tokens": 231365522.0, + "step": 8942 + }, + { + "epoch": 0.9820997144739732, + "grad_norm": 2.13730525970459, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6966068744659424, + "num_tokens": 231392658.0, + "step": 8943 + }, + { + "epoch": 0.9822095321765869, + "grad_norm": 1.9755305051803589, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6993058919906616, + "num_tokens": 231421138.0, + "step": 8944 + }, + { + "epoch": 0.9823193498792006, + "grad_norm": 2.0050206184387207, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7007153034210205, + "num_tokens": 231449858.0, + "step": 8945 + }, + { + "epoch": 0.9824291675818142, + "grad_norm": 2.3853588104248047, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7015953063964844, + "num_tokens": 231471215.0, + "step": 8946 + }, + { + "epoch": 0.9825389852844278, + "grad_norm": 2.22489595413208, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7526549100875854, + "num_tokens": 231493172.0, + "step": 8947 + }, + { + "epoch": 0.9826488029870415, + "grad_norm": 2.291151285171509, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.712624192237854, + "num_tokens": 231517220.0, + "step": 8948 + }, + { + "epoch": 0.9827586206896551, + "grad_norm": 2.0085036754608154, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6954801678657532, + "num_tokens": 231546201.0, + "step": 8949 + }, + { + "epoch": 0.9828684383922688, + "grad_norm": 2.2475783824920654, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.719059944152832, + "num_tokens": 231569750.0, + "step": 8950 + }, + { + "epoch": 0.9829782560948825, + "grad_norm": 2.4227945804595947, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7026969194412231, + "num_tokens": 231590161.0, + "step": 8951 + }, + { + "epoch": 0.9830880737974962, + "grad_norm": 1.9757146835327148, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7036770582199097, + "num_tokens": 231621120.0, + "step": 8952 + }, + { + "epoch": 0.9831978915001098, + "grad_norm": 1.9821481704711914, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7052721381187439, + "num_tokens": 231648905.0, + "step": 8953 + }, + { + "epoch": 0.9833077092027235, + "grad_norm": 2.2511777877807617, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6888414621353149, + "num_tokens": 231676121.0, + "step": 8954 + }, + { + "epoch": 0.9834175269053371, + "grad_norm": 2.070240020751953, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7114570736885071, + "num_tokens": 231703666.0, + "step": 8955 + }, + { + "epoch": 0.9835273446079508, + "grad_norm": 2.235034227371216, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6796503067016602, + "num_tokens": 231729719.0, + "step": 8956 + }, + { + "epoch": 0.9836371623105644, + "grad_norm": 2.1136419773101807, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6973885893821716, + "num_tokens": 231756062.0, + "step": 8957 + }, + { + "epoch": 0.9837469800131782, + "grad_norm": 1.9862865209579468, + "learning_rate": 1e-06, + "loss": 1.0938, + "mean_token_accuracy": 0.6678333282470703, + "num_tokens": 231787721.0, + "step": 8958 + }, + { + "epoch": 0.9838567977157918, + "grad_norm": 2.1451663970947266, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6998732089996338, + "num_tokens": 231813870.0, + "step": 8959 + }, + { + "epoch": 0.9839666154184055, + "grad_norm": 2.107179641723633, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7125727534294128, + "num_tokens": 231840917.0, + "step": 8960 + }, + { + "epoch": 0.9840764331210191, + "grad_norm": 2.322704792022705, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.718001663684845, + "num_tokens": 231862843.0, + "step": 8961 + }, + { + "epoch": 0.9841862508236328, + "grad_norm": 2.046205759048462, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7155227661132812, + "num_tokens": 231889043.0, + "step": 8962 + }, + { + "epoch": 0.9842960685262464, + "grad_norm": 2.107079029083252, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7066669464111328, + "num_tokens": 231916725.0, + "step": 8963 + }, + { + "epoch": 0.98440588622886, + "grad_norm": 1.9942961931228638, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7197235226631165, + "num_tokens": 231946104.0, + "step": 8964 + }, + { + "epoch": 0.9845157039314738, + "grad_norm": 2.4525654315948486, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7053781151771545, + "num_tokens": 231967785.0, + "step": 8965 + }, + { + "epoch": 0.9846255216340875, + "grad_norm": 2.0217065811157227, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7042614817619324, + "num_tokens": 231994741.0, + "step": 8966 + }, + { + "epoch": 0.9847353393367011, + "grad_norm": 2.4422974586486816, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6966559290885925, + "num_tokens": 232016422.0, + "step": 8967 + }, + { + "epoch": 0.9848451570393147, + "grad_norm": 2.242192268371582, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7042242884635925, + "num_tokens": 232040771.0, + "step": 8968 + }, + { + "epoch": 0.9849549747419284, + "grad_norm": 2.075822114944458, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7182116508483887, + "num_tokens": 232066750.0, + "step": 8969 + }, + { + "epoch": 0.985064792444542, + "grad_norm": 2.1996123790740967, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7105424404144287, + "num_tokens": 232090449.0, + "step": 8970 + }, + { + "epoch": 0.9851746101471557, + "grad_norm": 1.8877911567687988, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6907906532287598, + "num_tokens": 232122063.0, + "step": 8971 + }, + { + "epoch": 0.9852844278497693, + "grad_norm": 2.051964282989502, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7158393859863281, + "num_tokens": 232149959.0, + "step": 8972 + }, + { + "epoch": 0.9853942455523831, + "grad_norm": 2.4986917972564697, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7285233736038208, + "num_tokens": 232169056.0, + "step": 8973 + }, + { + "epoch": 0.9855040632549967, + "grad_norm": 2.190398931503296, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7083905935287476, + "num_tokens": 232194796.0, + "step": 8974 + }, + { + "epoch": 0.9856138809576104, + "grad_norm": 2.0455143451690674, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7106775045394897, + "num_tokens": 232224152.0, + "step": 8975 + }, + { + "epoch": 0.985723698660224, + "grad_norm": 2.2400152683258057, + "learning_rate": 1e-06, + "loss": 1.0371, + "mean_token_accuracy": 0.689965546131134, + "num_tokens": 232248823.0, + "step": 8976 + }, + { + "epoch": 0.9858335163628377, + "grad_norm": 1.9604514837265015, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6966267228126526, + "num_tokens": 232278960.0, + "step": 8977 + }, + { + "epoch": 0.9859433340654513, + "grad_norm": 2.471527576446533, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7293713092803955, + "num_tokens": 232299002.0, + "step": 8978 + }, + { + "epoch": 0.986053151768065, + "grad_norm": 2.0659196376800537, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7284017205238342, + "num_tokens": 232323783.0, + "step": 8979 + }, + { + "epoch": 0.9861629694706787, + "grad_norm": 2.2989983558654785, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7300081849098206, + "num_tokens": 232345943.0, + "step": 8980 + }, + { + "epoch": 0.9862727871732924, + "grad_norm": 2.236384630203247, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7006051540374756, + "num_tokens": 232370383.0, + "step": 8981 + }, + { + "epoch": 0.986382604875906, + "grad_norm": 2.044276475906372, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6914401650428772, + "num_tokens": 232399776.0, + "step": 8982 + }, + { + "epoch": 0.9864924225785197, + "grad_norm": 1.9434021711349487, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.704825758934021, + "num_tokens": 232429891.0, + "step": 8983 + }, + { + "epoch": 0.9866022402811333, + "grad_norm": 2.2798614501953125, + "learning_rate": 1e-06, + "loss": 1.0695, + "mean_token_accuracy": 0.6785402297973633, + "num_tokens": 232456867.0, + "step": 8984 + }, + { + "epoch": 0.986712057983747, + "grad_norm": 2.196658134460449, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7323284149169922, + "num_tokens": 232481328.0, + "step": 8985 + }, + { + "epoch": 0.9868218756863606, + "grad_norm": 2.1436474323272705, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7164998650550842, + "num_tokens": 232504646.0, + "step": 8986 + }, + { + "epoch": 0.9869316933889744, + "grad_norm": 2.0267958641052246, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6995077729225159, + "num_tokens": 232535282.0, + "step": 8987 + }, + { + "epoch": 0.987041511091588, + "grad_norm": 2.183016538619995, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7106986045837402, + "num_tokens": 232561180.0, + "step": 8988 + }, + { + "epoch": 0.9871513287942016, + "grad_norm": 2.1726911067962646, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7091637253761292, + "num_tokens": 232588832.0, + "step": 8989 + }, + { + "epoch": 0.9872611464968153, + "grad_norm": 2.0947985649108887, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.697585940361023, + "num_tokens": 232616528.0, + "step": 8990 + }, + { + "epoch": 0.9873709641994289, + "grad_norm": 2.1970584392547607, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.7019871473312378, + "num_tokens": 232643620.0, + "step": 8991 + }, + { + "epoch": 0.9874807819020426, + "grad_norm": 2.029867172241211, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7084503173828125, + "num_tokens": 232672758.0, + "step": 8992 + }, + { + "epoch": 0.9875905996046562, + "grad_norm": 2.2080063819885254, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7027883529663086, + "num_tokens": 232698818.0, + "step": 8993 + }, + { + "epoch": 0.98770041730727, + "grad_norm": 2.1467325687408447, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7034116983413696, + "num_tokens": 232723684.0, + "step": 8994 + }, + { + "epoch": 0.9878102350098836, + "grad_norm": 2.1345975399017334, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7198423147201538, + "num_tokens": 232748578.0, + "step": 8995 + }, + { + "epoch": 0.9879200527124973, + "grad_norm": 2.0345304012298584, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7114897966384888, + "num_tokens": 232777194.0, + "step": 8996 + }, + { + "epoch": 0.9880298704151109, + "grad_norm": 1.9573478698730469, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.697454571723938, + "num_tokens": 232806509.0, + "step": 8997 + }, + { + "epoch": 0.9881396881177246, + "grad_norm": 2.38926362991333, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6960694789886475, + "num_tokens": 232829602.0, + "step": 8998 + }, + { + "epoch": 0.9882495058203382, + "grad_norm": 2.3561158180236816, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7109352350234985, + "num_tokens": 232853101.0, + "step": 8999 + }, + { + "epoch": 0.9883593235229519, + "grad_norm": 2.2709438800811768, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7154562473297119, + "num_tokens": 232876847.0, + "step": 9000 + }, + { + "epoch": 0.9884691412255655, + "grad_norm": 2.3227877616882324, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7298420667648315, + "num_tokens": 232897886.0, + "step": 9001 + }, + { + "epoch": 0.9885789589281793, + "grad_norm": 2.1694796085357666, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6947770118713379, + "num_tokens": 232923874.0, + "step": 9002 + }, + { + "epoch": 0.9886887766307929, + "grad_norm": 2.2009615898132324, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7109636664390564, + "num_tokens": 232948175.0, + "step": 9003 + }, + { + "epoch": 0.9887985943334066, + "grad_norm": 2.3916568756103516, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.6976265907287598, + "num_tokens": 232970566.0, + "step": 9004 + }, + { + "epoch": 0.9889084120360202, + "grad_norm": 2.3453638553619385, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6945451498031616, + "num_tokens": 232993237.0, + "step": 9005 + }, + { + "epoch": 0.9890182297386338, + "grad_norm": 1.8778395652770996, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6929558515548706, + "num_tokens": 233027525.0, + "step": 9006 + }, + { + "epoch": 0.9891280474412475, + "grad_norm": 1.9262839555740356, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.703781008720398, + "num_tokens": 233061174.0, + "step": 9007 + }, + { + "epoch": 0.9892378651438611, + "grad_norm": 2.2442262172698975, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.696922242641449, + "num_tokens": 233087169.0, + "step": 9008 + }, + { + "epoch": 0.9893476828464749, + "grad_norm": 2.1584889888763428, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6999965310096741, + "num_tokens": 233113169.0, + "step": 9009 + }, + { + "epoch": 0.9894575005490885, + "grad_norm": 1.9412939548492432, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7049592137336731, + "num_tokens": 233145830.0, + "step": 9010 + }, + { + "epoch": 0.9895673182517022, + "grad_norm": 2.109571933746338, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6965621709823608, + "num_tokens": 233172642.0, + "step": 9011 + }, + { + "epoch": 0.9896771359543158, + "grad_norm": 2.2849040031433105, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7261166572570801, + "num_tokens": 233194751.0, + "step": 9012 + }, + { + "epoch": 0.9897869536569295, + "grad_norm": 1.9117944240570068, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.70473712682724, + "num_tokens": 233225997.0, + "step": 9013 + }, + { + "epoch": 0.9898967713595431, + "grad_norm": 2.0965828895568848, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7060525417327881, + "num_tokens": 233252009.0, + "step": 9014 + }, + { + "epoch": 0.9900065890621568, + "grad_norm": 2.076988935470581, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7234891653060913, + "num_tokens": 233278741.0, + "step": 9015 + }, + { + "epoch": 0.9901164067647705, + "grad_norm": 2.350085735321045, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7192861437797546, + "num_tokens": 233298654.0, + "step": 9016 + }, + { + "epoch": 0.9902262244673842, + "grad_norm": 2.0370354652404785, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.711394190788269, + "num_tokens": 233327394.0, + "step": 9017 + }, + { + "epoch": 0.9903360421699978, + "grad_norm": 2.154745578765869, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.711053729057312, + "num_tokens": 233352462.0, + "step": 9018 + }, + { + "epoch": 0.9904458598726115, + "grad_norm": 2.3895676136016846, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.71718829870224, + "num_tokens": 233374329.0, + "step": 9019 + }, + { + "epoch": 0.9905556775752251, + "grad_norm": 2.0569825172424316, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.71912682056427, + "num_tokens": 233399704.0, + "step": 9020 + }, + { + "epoch": 0.9906654952778388, + "grad_norm": 2.326234817504883, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7036881446838379, + "num_tokens": 233422103.0, + "step": 9021 + }, + { + "epoch": 0.9907753129804524, + "grad_norm": 2.365506887435913, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7102713584899902, + "num_tokens": 233444041.0, + "step": 9022 + }, + { + "epoch": 0.9908851306830662, + "grad_norm": 2.0360641479492188, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.71577388048172, + "num_tokens": 233472497.0, + "step": 9023 + }, + { + "epoch": 0.9909949483856798, + "grad_norm": 2.2711212635040283, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7087007164955139, + "num_tokens": 233495761.0, + "step": 9024 + }, + { + "epoch": 0.9911047660882935, + "grad_norm": 2.0807507038116455, + "learning_rate": 1e-06, + "loss": 1.0567, + "mean_token_accuracy": 0.6789767146110535, + "num_tokens": 233526528.0, + "step": 9025 + }, + { + "epoch": 0.9912145837909071, + "grad_norm": 2.2198057174682617, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7118045687675476, + "num_tokens": 233549552.0, + "step": 9026 + }, + { + "epoch": 0.9913244014935207, + "grad_norm": 2.1705920696258545, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7082191109657288, + "num_tokens": 233575314.0, + "step": 9027 + }, + { + "epoch": 0.9914342191961344, + "grad_norm": 2.109761953353882, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7047441601753235, + "num_tokens": 233600910.0, + "step": 9028 + }, + { + "epoch": 0.991544036898748, + "grad_norm": 2.218628406524658, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7015993595123291, + "num_tokens": 233624970.0, + "step": 9029 + }, + { + "epoch": 0.9916538546013618, + "grad_norm": 1.9598321914672852, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.7000463604927063, + "num_tokens": 233656091.0, + "step": 9030 + }, + { + "epoch": 0.9917636723039754, + "grad_norm": 2.2838656902313232, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.713380753993988, + "num_tokens": 233678512.0, + "step": 9031 + }, + { + "epoch": 0.9918734900065891, + "grad_norm": 2.1170482635498047, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7102112770080566, + "num_tokens": 233704187.0, + "step": 9032 + }, + { + "epoch": 0.9919833077092027, + "grad_norm": 2.1080336570739746, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7039780020713806, + "num_tokens": 233731448.0, + "step": 9033 + }, + { + "epoch": 0.9920931254118164, + "grad_norm": 2.2384636402130127, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.713783860206604, + "num_tokens": 233755151.0, + "step": 9034 + }, + { + "epoch": 0.99220294311443, + "grad_norm": 1.7966532707214355, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7124114036560059, + "num_tokens": 233789917.0, + "step": 9035 + }, + { + "epoch": 0.9923127608170437, + "grad_norm": 2.3270716667175293, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7147377133369446, + "num_tokens": 233812965.0, + "step": 9036 + }, + { + "epoch": 0.9924225785196573, + "grad_norm": 2.037666082382202, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.708314836025238, + "num_tokens": 233842025.0, + "step": 9037 + }, + { + "epoch": 0.9925323962222711, + "grad_norm": 2.0226809978485107, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7197690010070801, + "num_tokens": 233870113.0, + "step": 9038 + }, + { + "epoch": 0.9926422139248847, + "grad_norm": 2.0660314559936523, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7101849317550659, + "num_tokens": 233897948.0, + "step": 9039 + }, + { + "epoch": 0.9927520316274984, + "grad_norm": 2.196126699447632, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6956677436828613, + "num_tokens": 233922903.0, + "step": 9040 + }, + { + "epoch": 0.992861849330112, + "grad_norm": 2.094461679458618, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7352463603019714, + "num_tokens": 233949903.0, + "step": 9041 + }, + { + "epoch": 0.9929716670327257, + "grad_norm": 2.02840256690979, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7015671133995056, + "num_tokens": 233980714.0, + "step": 9042 + }, + { + "epoch": 0.9930814847353393, + "grad_norm": 1.9151933193206787, + "learning_rate": 1e-06, + "loss": 1.0539, + "mean_token_accuracy": 0.686366081237793, + "num_tokens": 234011634.0, + "step": 9043 + }, + { + "epoch": 0.993191302437953, + "grad_norm": 2.303865671157837, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6888790130615234, + "num_tokens": 234035125.0, + "step": 9044 + }, + { + "epoch": 0.9933011201405667, + "grad_norm": 2.365696430206299, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6999591588973999, + "num_tokens": 234057417.0, + "step": 9045 + }, + { + "epoch": 0.9934109378431804, + "grad_norm": 2.5193912982940674, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6979681849479675, + "num_tokens": 234078743.0, + "step": 9046 + }, + { + "epoch": 0.993520755545794, + "grad_norm": 2.0854694843292236, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7307698726654053, + "num_tokens": 234105817.0, + "step": 9047 + }, + { + "epoch": 0.9936305732484076, + "grad_norm": 2.04846453666687, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6939390301704407, + "num_tokens": 234132511.0, + "step": 9048 + }, + { + "epoch": 0.9937403909510213, + "grad_norm": 2.1994845867156982, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7222528457641602, + "num_tokens": 234155713.0, + "step": 9049 + }, + { + "epoch": 0.9938502086536349, + "grad_norm": 1.9430791139602661, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.711823582649231, + "num_tokens": 234184946.0, + "step": 9050 + }, + { + "epoch": 0.9939600263562486, + "grad_norm": 2.322908401489258, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7142124176025391, + "num_tokens": 234206926.0, + "step": 9051 + }, + { + "epoch": 0.9940698440588623, + "grad_norm": 2.1327641010284424, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.715906023979187, + "num_tokens": 234231917.0, + "step": 9052 + }, + { + "epoch": 0.994179661761476, + "grad_norm": 2.153679132461548, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6977586150169373, + "num_tokens": 234257913.0, + "step": 9053 + }, + { + "epoch": 0.9942894794640896, + "grad_norm": 2.2763445377349854, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7241954803466797, + "num_tokens": 234281030.0, + "step": 9054 + }, + { + "epoch": 0.9943992971667033, + "grad_norm": 2.2185606956481934, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7129449844360352, + "num_tokens": 234304409.0, + "step": 9055 + }, + { + "epoch": 0.9945091148693169, + "grad_norm": 1.9722163677215576, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.67314612865448, + "num_tokens": 234336413.0, + "step": 9056 + }, + { + "epoch": 0.9946189325719306, + "grad_norm": 2.197709321975708, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7180114984512329, + "num_tokens": 234360815.0, + "step": 9057 + }, + { + "epoch": 0.9947287502745442, + "grad_norm": 2.1636245250701904, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.716128945350647, + "num_tokens": 234389038.0, + "step": 9058 + }, + { + "epoch": 0.994838567977158, + "grad_norm": 2.3796546459198, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7000503540039062, + "num_tokens": 234410875.0, + "step": 9059 + }, + { + "epoch": 0.9949483856797716, + "grad_norm": 2.1878790855407715, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7234934568405151, + "num_tokens": 234436468.0, + "step": 9060 + }, + { + "epoch": 0.9950582033823853, + "grad_norm": 2.1166064739227295, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7105624675750732, + "num_tokens": 234462935.0, + "step": 9061 + }, + { + "epoch": 0.9951680210849989, + "grad_norm": 1.8943477869033813, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7112811803817749, + "num_tokens": 234496078.0, + "step": 9062 + }, + { + "epoch": 0.9952778387876126, + "grad_norm": 2.11907958984375, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6912485361099243, + "num_tokens": 234524273.0, + "step": 9063 + }, + { + "epoch": 0.9953876564902262, + "grad_norm": 2.1836118698120117, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7207492589950562, + "num_tokens": 234549620.0, + "step": 9064 + }, + { + "epoch": 0.9954974741928398, + "grad_norm": 2.03007173538208, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.718630313873291, + "num_tokens": 234576957.0, + "step": 9065 + }, + { + "epoch": 0.9956072918954535, + "grad_norm": 2.196441888809204, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7020962834358215, + "num_tokens": 234601466.0, + "step": 9066 + }, + { + "epoch": 0.9957171095980673, + "grad_norm": 1.9342567920684814, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7283895015716553, + "num_tokens": 234632198.0, + "step": 9067 + }, + { + "epoch": 0.9958269273006809, + "grad_norm": 1.9089138507843018, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7122102379798889, + "num_tokens": 234662027.0, + "step": 9068 + }, + { + "epoch": 0.9959367450032945, + "grad_norm": 2.049497127532959, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7120180130004883, + "num_tokens": 234688572.0, + "step": 9069 + }, + { + "epoch": 0.9960465627059082, + "grad_norm": 2.167786121368408, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6884040832519531, + "num_tokens": 234716877.0, + "step": 9070 + }, + { + "epoch": 0.9961563804085218, + "grad_norm": 1.9690096378326416, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7025144100189209, + "num_tokens": 234746332.0, + "step": 9071 + }, + { + "epoch": 0.9962661981111355, + "grad_norm": 2.261497735977173, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.713992178440094, + "num_tokens": 234769472.0, + "step": 9072 + }, + { + "epoch": 0.9963760158137491, + "grad_norm": 2.1655097007751465, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7031580209732056, + "num_tokens": 234794215.0, + "step": 9073 + }, + { + "epoch": 0.9964858335163629, + "grad_norm": 1.9712008237838745, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7148257493972778, + "num_tokens": 234821761.0, + "step": 9074 + }, + { + "epoch": 0.9965956512189765, + "grad_norm": 2.177321195602417, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7162361145019531, + "num_tokens": 234848284.0, + "step": 9075 + }, + { + "epoch": 0.9967054689215902, + "grad_norm": 2.586437702178955, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7154775857925415, + "num_tokens": 234866900.0, + "step": 9076 + }, + { + "epoch": 0.9968152866242038, + "grad_norm": 2.2674877643585205, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7199048399925232, + "num_tokens": 234890696.0, + "step": 9077 + }, + { + "epoch": 0.9969251043268175, + "grad_norm": 2.5952820777893066, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7167125940322876, + "num_tokens": 234909194.0, + "step": 9078 + }, + { + "epoch": 0.9970349220294311, + "grad_norm": 2.0115058422088623, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7006077766418457, + "num_tokens": 234938499.0, + "step": 9079 + }, + { + "epoch": 0.9971447397320448, + "grad_norm": 2.622026205062866, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7048603296279907, + "num_tokens": 234956871.0, + "step": 9080 + }, + { + "epoch": 0.9972545574346585, + "grad_norm": 2.131337881088257, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7234045267105103, + "num_tokens": 234983142.0, + "step": 9081 + }, + { + "epoch": 0.9973643751372722, + "grad_norm": 1.919262409210205, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7019524574279785, + "num_tokens": 235015256.0, + "step": 9082 + }, + { + "epoch": 0.9974741928398858, + "grad_norm": 2.1231424808502197, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7279346585273743, + "num_tokens": 235039088.0, + "step": 9083 + }, + { + "epoch": 0.9975840105424995, + "grad_norm": 2.489854574203491, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7044910788536072, + "num_tokens": 235059458.0, + "step": 9084 + }, + { + "epoch": 0.9976938282451131, + "grad_norm": 2.3725783824920654, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7222231030464172, + "num_tokens": 235080777.0, + "step": 9085 + }, + { + "epoch": 0.9978036459477267, + "grad_norm": 2.0332517623901367, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7303047180175781, + "num_tokens": 235106971.0, + "step": 9086 + }, + { + "epoch": 0.9979134636503404, + "grad_norm": 2.6110846996307373, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7178975939750671, + "num_tokens": 235126459.0, + "step": 9087 + }, + { + "epoch": 0.9980232813529542, + "grad_norm": 1.9965277910232544, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7061291933059692, + "num_tokens": 235154795.0, + "step": 9088 + }, + { + "epoch": 0.9981330990555678, + "grad_norm": 2.393467664718628, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7007760405540466, + "num_tokens": 235177941.0, + "step": 9089 + }, + { + "epoch": 0.9982429167581814, + "grad_norm": 2.0757596492767334, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.6995136141777039, + "num_tokens": 235205466.0, + "step": 9090 + }, + { + "epoch": 0.9983527344607951, + "grad_norm": 2.3060302734375, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7044067978858948, + "num_tokens": 235228890.0, + "step": 9091 + }, + { + "epoch": 0.9984625521634087, + "grad_norm": 2.3704893589019775, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7218214273452759, + "num_tokens": 235250592.0, + "step": 9092 + }, + { + "epoch": 0.9985723698660224, + "grad_norm": 1.987789273262024, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7204352617263794, + "num_tokens": 235278580.0, + "step": 9093 + }, + { + "epoch": 0.998682187568636, + "grad_norm": 2.255857229232788, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7234362363815308, + "num_tokens": 235301987.0, + "step": 9094 + }, + { + "epoch": 0.9987920052712497, + "grad_norm": 2.2785327434539795, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6995434761047363, + "num_tokens": 235328062.0, + "step": 9095 + }, + { + "epoch": 0.9989018229738634, + "grad_norm": 2.292578935623169, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7282930612564087, + "num_tokens": 235350562.0, + "step": 9096 + }, + { + "epoch": 0.9990116406764771, + "grad_norm": 1.9981977939605713, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7081596851348877, + "num_tokens": 235379626.0, + "step": 9097 + }, + { + "epoch": 0.9991214583790907, + "grad_norm": 2.237398862838745, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7116422653198242, + "num_tokens": 235405941.0, + "step": 9098 + }, + { + "epoch": 0.9992312760817044, + "grad_norm": 2.077228546142578, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7092306613922119, + "num_tokens": 235432729.0, + "step": 9099 + }, + { + "epoch": 0.999341093784318, + "grad_norm": 1.9632527828216553, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7030826807022095, + "num_tokens": 235463599.0, + "step": 9100 + }, + { + "epoch": 0.9994509114869317, + "grad_norm": 2.3155338764190674, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7080028653144836, + "num_tokens": 235486921.0, + "step": 9101 + }, + { + "epoch": 0.9995607291895453, + "grad_norm": 1.8640384674072266, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6939171552658081, + "num_tokens": 235519812.0, + "step": 9102 + }, + { + "epoch": 0.9996705468921591, + "grad_norm": 2.2341675758361816, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7269951105117798, + "num_tokens": 235543456.0, + "step": 9103 + }, + { + "epoch": 0.9997803645947727, + "grad_norm": 2.060300827026367, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6954749822616577, + "num_tokens": 235572016.0, + "step": 9104 + }, + { + "epoch": 0.9998901822973864, + "grad_norm": 2.3901915550231934, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7190898656845093, + "num_tokens": 235594912.0, + "step": 9105 + }, + { + "epoch": 1.0, + "grad_norm": 2.313845157623291, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7081006169319153, + "num_tokens": 235617245.0, + "step": 9106 + }, + { + "epoch": 1.0001098177026138, + "grad_norm": 2.080559253692627, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7027031183242798, + "num_tokens": 235647042.0, + "step": 9107 + }, + { + "epoch": 1.0002196354052273, + "grad_norm": 2.0119216442108154, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7184215784072876, + "num_tokens": 235675620.0, + "step": 9108 + }, + { + "epoch": 1.000329453107841, + "grad_norm": 2.373976230621338, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7137978672981262, + "num_tokens": 235699190.0, + "step": 9109 + }, + { + "epoch": 1.0004392708104546, + "grad_norm": 1.8445011377334595, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7156139016151428, + "num_tokens": 235729568.0, + "step": 9110 + }, + { + "epoch": 1.0005490885130683, + "grad_norm": 2.116572856903076, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.726284384727478, + "num_tokens": 235753975.0, + "step": 9111 + }, + { + "epoch": 1.0006589062156819, + "grad_norm": 2.2909176349639893, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7290475368499756, + "num_tokens": 235774114.0, + "step": 9112 + }, + { + "epoch": 1.0007687239182956, + "grad_norm": 2.1121511459350586, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6955420970916748, + "num_tokens": 235799505.0, + "step": 9113 + }, + { + "epoch": 1.0008785416209094, + "grad_norm": 2.0859994888305664, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7189939022064209, + "num_tokens": 235828000.0, + "step": 9114 + }, + { + "epoch": 1.000988359323523, + "grad_norm": 2.4187612533569336, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7460169792175293, + "num_tokens": 235851026.0, + "step": 9115 + }, + { + "epoch": 1.0010981770261367, + "grad_norm": 2.2439403533935547, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6853697299957275, + "num_tokens": 235880806.0, + "step": 9116 + }, + { + "epoch": 1.0012079947287502, + "grad_norm": 2.1308085918426514, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7040032148361206, + "num_tokens": 235913442.0, + "step": 9117 + }, + { + "epoch": 1.001317812431364, + "grad_norm": 2.3287529945373535, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7306263446807861, + "num_tokens": 235936145.0, + "step": 9118 + }, + { + "epoch": 1.0014276301339775, + "grad_norm": 2.385913610458374, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7217416167259216, + "num_tokens": 235958222.0, + "step": 9119 + }, + { + "epoch": 1.0015374478365913, + "grad_norm": 2.297664165496826, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7293086051940918, + "num_tokens": 235980934.0, + "step": 9120 + }, + { + "epoch": 1.001647265539205, + "grad_norm": 2.2600677013397217, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7081035375595093, + "num_tokens": 236005418.0, + "step": 9121 + }, + { + "epoch": 1.0017570832418186, + "grad_norm": 2.202376127243042, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7242656946182251, + "num_tokens": 236031486.0, + "step": 9122 + }, + { + "epoch": 1.0018669009444323, + "grad_norm": 2.2609457969665527, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6958346962928772, + "num_tokens": 236058222.0, + "step": 9123 + }, + { + "epoch": 1.0019767186470458, + "grad_norm": 2.200467348098755, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7343692183494568, + "num_tokens": 236082865.0, + "step": 9124 + }, + { + "epoch": 1.0020865363496596, + "grad_norm": 1.9202433824539185, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6951667666435242, + "num_tokens": 236118019.0, + "step": 9125 + }, + { + "epoch": 1.0021963540522731, + "grad_norm": 2.311880111694336, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7184363603591919, + "num_tokens": 236140259.0, + "step": 9126 + }, + { + "epoch": 1.002306171754887, + "grad_norm": 1.9484565258026123, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7055187821388245, + "num_tokens": 236171193.0, + "step": 9127 + }, + { + "epoch": 1.0024159894575007, + "grad_norm": 2.217775344848633, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7156350612640381, + "num_tokens": 236198553.0, + "step": 9128 + }, + { + "epoch": 1.0025258071601142, + "grad_norm": 2.284173011779785, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7001692056655884, + "num_tokens": 236225340.0, + "step": 9129 + }, + { + "epoch": 1.002635624862728, + "grad_norm": 2.317898750305176, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7220383882522583, + "num_tokens": 236249433.0, + "step": 9130 + }, + { + "epoch": 1.0027454425653415, + "grad_norm": 2.052058219909668, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6888449192047119, + "num_tokens": 236278592.0, + "step": 9131 + }, + { + "epoch": 1.0028552602679552, + "grad_norm": 2.589111328125, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.73764967918396, + "num_tokens": 236296550.0, + "step": 9132 + }, + { + "epoch": 1.0029650779705688, + "grad_norm": 2.053187131881714, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7141216993331909, + "num_tokens": 236325169.0, + "step": 9133 + }, + { + "epoch": 1.0030748956731825, + "grad_norm": 2.0136196613311768, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.696129560470581, + "num_tokens": 236355195.0, + "step": 9134 + }, + { + "epoch": 1.0031847133757963, + "grad_norm": 1.8721306324005127, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6972174644470215, + "num_tokens": 236392651.0, + "step": 9135 + }, + { + "epoch": 1.0032945310784098, + "grad_norm": 1.861150860786438, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.709038257598877, + "num_tokens": 236429123.0, + "step": 9136 + }, + { + "epoch": 1.0034043487810236, + "grad_norm": 2.207845687866211, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7243807315826416, + "num_tokens": 236458505.0, + "step": 9137 + }, + { + "epoch": 1.0035141664836371, + "grad_norm": 2.332094430923462, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.724310040473938, + "num_tokens": 236483365.0, + "step": 9138 + }, + { + "epoch": 1.0036239841862509, + "grad_norm": 2.1764798164367676, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7192240953445435, + "num_tokens": 236507918.0, + "step": 9139 + }, + { + "epoch": 1.0037338018888644, + "grad_norm": 1.7785180807113647, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7063738107681274, + "num_tokens": 236542471.0, + "step": 9140 + }, + { + "epoch": 1.0038436195914782, + "grad_norm": 2.2771220207214355, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7289812564849854, + "num_tokens": 236566275.0, + "step": 9141 + }, + { + "epoch": 1.0039534372940917, + "grad_norm": 2.359480857849121, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7234690189361572, + "num_tokens": 236588724.0, + "step": 9142 + }, + { + "epoch": 1.0040632549967055, + "grad_norm": 1.9935328960418701, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.694151759147644, + "num_tokens": 236619468.0, + "step": 9143 + }, + { + "epoch": 1.0041730726993192, + "grad_norm": 2.46529221534729, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7169647812843323, + "num_tokens": 236641105.0, + "step": 9144 + }, + { + "epoch": 1.0042828904019327, + "grad_norm": 2.4406216144561768, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7226927280426025, + "num_tokens": 236664417.0, + "step": 9145 + }, + { + "epoch": 1.0043927081045465, + "grad_norm": 2.42718505859375, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7235402464866638, + "num_tokens": 236685394.0, + "step": 9146 + }, + { + "epoch": 1.00450252580716, + "grad_norm": 2.388639211654663, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.736420750617981, + "num_tokens": 236709099.0, + "step": 9147 + }, + { + "epoch": 1.0046123435097738, + "grad_norm": 2.066364049911499, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.7032777070999146, + "num_tokens": 236736328.0, + "step": 9148 + }, + { + "epoch": 1.0047221612123873, + "grad_norm": 2.1615488529205322, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7240126132965088, + "num_tokens": 236762359.0, + "step": 9149 + }, + { + "epoch": 1.004831978915001, + "grad_norm": 2.1008846759796143, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7032686471939087, + "num_tokens": 236788664.0, + "step": 9150 + }, + { + "epoch": 1.0049417966176148, + "grad_norm": 2.2604305744171143, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.726787805557251, + "num_tokens": 236811440.0, + "step": 9151 + }, + { + "epoch": 1.0050516143202284, + "grad_norm": 1.9627442359924316, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7014522552490234, + "num_tokens": 236844805.0, + "step": 9152 + }, + { + "epoch": 1.0051614320228421, + "grad_norm": 2.249079942703247, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7291417717933655, + "num_tokens": 236869318.0, + "step": 9153 + }, + { + "epoch": 1.0052712497254557, + "grad_norm": 1.854759693145752, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7395275831222534, + "num_tokens": 236901039.0, + "step": 9154 + }, + { + "epoch": 1.0053810674280694, + "grad_norm": 2.168384075164795, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7004817724227905, + "num_tokens": 236928639.0, + "step": 9155 + }, + { + "epoch": 1.005490885130683, + "grad_norm": 2.2411141395568848, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.698119580745697, + "num_tokens": 236955654.0, + "step": 9156 + }, + { + "epoch": 1.0056007028332967, + "grad_norm": 2.2256991863250732, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.72980135679245, + "num_tokens": 236980378.0, + "step": 9157 + }, + { + "epoch": 1.0057105205359105, + "grad_norm": 2.169083595275879, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7043648958206177, + "num_tokens": 237008385.0, + "step": 9158 + }, + { + "epoch": 1.005820338238524, + "grad_norm": 2.155427932739258, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7104777097702026, + "num_tokens": 237034555.0, + "step": 9159 + }, + { + "epoch": 1.0059301559411378, + "grad_norm": 2.090829610824585, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7063988447189331, + "num_tokens": 237066324.0, + "step": 9160 + }, + { + "epoch": 1.0060399736437513, + "grad_norm": 2.016016721725464, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7077646851539612, + "num_tokens": 237094989.0, + "step": 9161 + }, + { + "epoch": 1.006149791346365, + "grad_norm": 2.279526472091675, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.6961310505867004, + "num_tokens": 237120765.0, + "step": 9162 + }, + { + "epoch": 1.0062596090489786, + "grad_norm": 1.9972106218338013, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.708777129650116, + "num_tokens": 237151511.0, + "step": 9163 + }, + { + "epoch": 1.0063694267515924, + "grad_norm": 2.336488723754883, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7121378779411316, + "num_tokens": 237176336.0, + "step": 9164 + }, + { + "epoch": 1.0064792444542061, + "grad_norm": 2.1081387996673584, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6935406923294067, + "num_tokens": 237207576.0, + "step": 9165 + }, + { + "epoch": 1.0065890621568196, + "grad_norm": 2.1378839015960693, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.712071418762207, + "num_tokens": 237235137.0, + "step": 9166 + }, + { + "epoch": 1.0066988798594334, + "grad_norm": 2.4654128551483154, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7244730591773987, + "num_tokens": 237255124.0, + "step": 9167 + }, + { + "epoch": 1.006808697562047, + "grad_norm": 2.455298662185669, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6976273655891418, + "num_tokens": 237278364.0, + "step": 9168 + }, + { + "epoch": 1.0069185152646607, + "grad_norm": 2.3240480422973633, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.731937050819397, + "num_tokens": 237302500.0, + "step": 9169 + }, + { + "epoch": 1.0070283329672742, + "grad_norm": 2.185685157775879, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.731078565120697, + "num_tokens": 237327908.0, + "step": 9170 + }, + { + "epoch": 1.007138150669888, + "grad_norm": 2.161069869995117, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.706926167011261, + "num_tokens": 237355523.0, + "step": 9171 + }, + { + "epoch": 1.0072479683725017, + "grad_norm": 2.1399667263031006, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7112860679626465, + "num_tokens": 237381925.0, + "step": 9172 + }, + { + "epoch": 1.0073577860751153, + "grad_norm": 2.1520578861236572, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7153280973434448, + "num_tokens": 237408436.0, + "step": 9173 + }, + { + "epoch": 1.007467603777729, + "grad_norm": 2.201688528060913, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7196547985076904, + "num_tokens": 237434340.0, + "step": 9174 + }, + { + "epoch": 1.0075774214803426, + "grad_norm": 2.1380505561828613, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7422587871551514, + "num_tokens": 237459680.0, + "step": 9175 + }, + { + "epoch": 1.0076872391829563, + "grad_norm": 1.9734721183776855, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6923011541366577, + "num_tokens": 237492224.0, + "step": 9176 + }, + { + "epoch": 1.0077970568855699, + "grad_norm": 2.142381191253662, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7453403472900391, + "num_tokens": 237516302.0, + "step": 9177 + }, + { + "epoch": 1.0079068745881836, + "grad_norm": 2.270970106124878, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7138132452964783, + "num_tokens": 237540954.0, + "step": 9178 + }, + { + "epoch": 1.0080166922907974, + "grad_norm": 2.226444721221924, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7143071889877319, + "num_tokens": 237565679.0, + "step": 9179 + }, + { + "epoch": 1.008126509993411, + "grad_norm": 2.0221798419952393, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7040481567382812, + "num_tokens": 237592971.0, + "step": 9180 + }, + { + "epoch": 1.0082363276960247, + "grad_norm": 2.160282850265503, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.720318615436554, + "num_tokens": 237618973.0, + "step": 9181 + }, + { + "epoch": 1.0083461453986382, + "grad_norm": 2.072126865386963, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7101174592971802, + "num_tokens": 237649304.0, + "step": 9182 + }, + { + "epoch": 1.008455963101252, + "grad_norm": 2.3838951587677, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.679947555065155, + "num_tokens": 237676775.0, + "step": 9183 + }, + { + "epoch": 1.0085657808038655, + "grad_norm": 2.0092132091522217, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7047326564788818, + "num_tokens": 237707669.0, + "step": 9184 + }, + { + "epoch": 1.0086755985064793, + "grad_norm": 2.2892885208129883, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7319058179855347, + "num_tokens": 237732335.0, + "step": 9185 + }, + { + "epoch": 1.008785416209093, + "grad_norm": 2.431797504425049, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7158072590827942, + "num_tokens": 237755742.0, + "step": 9186 + }, + { + "epoch": 1.0088952339117065, + "grad_norm": 2.323197603225708, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.718778669834137, + "num_tokens": 237779404.0, + "step": 9187 + }, + { + "epoch": 1.0090050516143203, + "grad_norm": 2.452561140060425, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7272933721542358, + "num_tokens": 237801424.0, + "step": 9188 + }, + { + "epoch": 1.0091148693169338, + "grad_norm": 2.4159975051879883, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7139390707015991, + "num_tokens": 237824471.0, + "step": 9189 + }, + { + "epoch": 1.0092246870195476, + "grad_norm": 1.9831194877624512, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6918437480926514, + "num_tokens": 237856973.0, + "step": 9190 + }, + { + "epoch": 1.0093345047221611, + "grad_norm": 2.11064076423645, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7082405090332031, + "num_tokens": 237886112.0, + "step": 9191 + }, + { + "epoch": 1.0094443224247749, + "grad_norm": 2.3294079303741455, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6871299743652344, + "num_tokens": 237912563.0, + "step": 9192 + }, + { + "epoch": 1.0095541401273886, + "grad_norm": 2.5034914016723633, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.728458821773529, + "num_tokens": 237931329.0, + "step": 9193 + }, + { + "epoch": 1.0096639578300022, + "grad_norm": 2.8374390602111816, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7252172827720642, + "num_tokens": 237949637.0, + "step": 9194 + }, + { + "epoch": 1.009773775532616, + "grad_norm": 2.678968906402588, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7425899505615234, + "num_tokens": 237968165.0, + "step": 9195 + }, + { + "epoch": 1.0098835932352295, + "grad_norm": 2.2822935581207275, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7097787261009216, + "num_tokens": 237993937.0, + "step": 9196 + }, + { + "epoch": 1.0099934109378432, + "grad_norm": 2.288254976272583, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7216792702674866, + "num_tokens": 238017976.0, + "step": 9197 + }, + { + "epoch": 1.0101032286404568, + "grad_norm": 1.9354960918426514, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6960128545761108, + "num_tokens": 238049863.0, + "step": 9198 + }, + { + "epoch": 1.0102130463430705, + "grad_norm": 2.0830321311950684, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7289515733718872, + "num_tokens": 238079151.0, + "step": 9199 + }, + { + "epoch": 1.010322864045684, + "grad_norm": 2.2432031631469727, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.728462815284729, + "num_tokens": 238103439.0, + "step": 9200 + }, + { + "epoch": 1.0104326817482978, + "grad_norm": 2.188572645187378, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.705109715461731, + "num_tokens": 238132132.0, + "step": 9201 + }, + { + "epoch": 1.0105424994509116, + "grad_norm": 2.555715799331665, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7328441143035889, + "num_tokens": 238152749.0, + "step": 9202 + }, + { + "epoch": 1.010652317153525, + "grad_norm": 2.0305488109588623, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7282913327217102, + "num_tokens": 238180616.0, + "step": 9203 + }, + { + "epoch": 1.0107621348561389, + "grad_norm": 2.406477928161621, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7101500630378723, + "num_tokens": 238205218.0, + "step": 9204 + }, + { + "epoch": 1.0108719525587524, + "grad_norm": 2.378314971923828, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7185100317001343, + "num_tokens": 238228805.0, + "step": 9205 + }, + { + "epoch": 1.0109817702613662, + "grad_norm": 2.0596728324890137, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7051290273666382, + "num_tokens": 238258713.0, + "step": 9206 + }, + { + "epoch": 1.0110915879639797, + "grad_norm": 2.460906505584717, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7554349899291992, + "num_tokens": 238277651.0, + "step": 9207 + }, + { + "epoch": 1.0112014056665934, + "grad_norm": 2.1154799461364746, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.6994153261184692, + "num_tokens": 238306098.0, + "step": 9208 + }, + { + "epoch": 1.0113112233692072, + "grad_norm": 2.3931615352630615, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.733527660369873, + "num_tokens": 238327071.0, + "step": 9209 + }, + { + "epoch": 1.0114210410718207, + "grad_norm": 2.178710460662842, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7055003643035889, + "num_tokens": 238353322.0, + "step": 9210 + }, + { + "epoch": 1.0115308587744345, + "grad_norm": 2.0386483669281006, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7224392294883728, + "num_tokens": 238382018.0, + "step": 9211 + }, + { + "epoch": 1.011640676477048, + "grad_norm": 2.4820830821990967, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.733576774597168, + "num_tokens": 238401017.0, + "step": 9212 + }, + { + "epoch": 1.0117504941796618, + "grad_norm": 2.2038841247558594, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7196943759918213, + "num_tokens": 238428954.0, + "step": 9213 + }, + { + "epoch": 1.0118603118822753, + "grad_norm": 2.1343343257904053, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7364618182182312, + "num_tokens": 238455499.0, + "step": 9214 + }, + { + "epoch": 1.011970129584889, + "grad_norm": 2.1581838130950928, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7180660963058472, + "num_tokens": 238481219.0, + "step": 9215 + }, + { + "epoch": 1.0120799472875028, + "grad_norm": 2.1276795864105225, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7382275462150574, + "num_tokens": 238506894.0, + "step": 9216 + }, + { + "epoch": 1.0121897649901164, + "grad_norm": 2.2694528102874756, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7231478691101074, + "num_tokens": 238534255.0, + "step": 9217 + }, + { + "epoch": 1.0122995826927301, + "grad_norm": 2.5148420333862305, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7246156930923462, + "num_tokens": 238555269.0, + "step": 9218 + }, + { + "epoch": 1.0124094003953437, + "grad_norm": 1.945743203163147, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7068991661071777, + "num_tokens": 238588256.0, + "step": 9219 + }, + { + "epoch": 1.0125192180979574, + "grad_norm": 1.8588429689407349, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7049185633659363, + "num_tokens": 238622800.0, + "step": 9220 + }, + { + "epoch": 1.012629035800571, + "grad_norm": 2.4028234481811523, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7265914082527161, + "num_tokens": 238645351.0, + "step": 9221 + }, + { + "epoch": 1.0127388535031847, + "grad_norm": 2.3441505432128906, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7050844430923462, + "num_tokens": 238668786.0, + "step": 9222 + }, + { + "epoch": 1.0128486712057985, + "grad_norm": 2.1481986045837402, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7275850176811218, + "num_tokens": 238694597.0, + "step": 9223 + }, + { + "epoch": 1.012958488908412, + "grad_norm": 2.2609071731567383, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7285329103469849, + "num_tokens": 238719112.0, + "step": 9224 + }, + { + "epoch": 1.0130683066110258, + "grad_norm": 1.837916374206543, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7227622270584106, + "num_tokens": 238751808.0, + "step": 9225 + }, + { + "epoch": 1.0131781243136393, + "grad_norm": 2.310696601867676, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.72409987449646, + "num_tokens": 238776031.0, + "step": 9226 + }, + { + "epoch": 1.013287942016253, + "grad_norm": 2.484853744506836, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7157019376754761, + "num_tokens": 238796867.0, + "step": 9227 + }, + { + "epoch": 1.0133977597188666, + "grad_norm": 2.2882213592529297, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.709301769733429, + "num_tokens": 238822315.0, + "step": 9228 + }, + { + "epoch": 1.0135075774214803, + "grad_norm": 2.2427616119384766, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7105748653411865, + "num_tokens": 238847142.0, + "step": 9229 + }, + { + "epoch": 1.013617395124094, + "grad_norm": 2.079960823059082, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7057231664657593, + "num_tokens": 238877845.0, + "step": 9230 + }, + { + "epoch": 1.0137272128267076, + "grad_norm": 2.1583094596862793, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7008311152458191, + "num_tokens": 238904482.0, + "step": 9231 + }, + { + "epoch": 1.0138370305293214, + "grad_norm": 2.3299918174743652, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7039280533790588, + "num_tokens": 238928874.0, + "step": 9232 + }, + { + "epoch": 1.013946848231935, + "grad_norm": 2.343174695968628, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7090967893600464, + "num_tokens": 238955467.0, + "step": 9233 + }, + { + "epoch": 1.0140566659345487, + "grad_norm": 2.433777332305908, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7385174632072449, + "num_tokens": 238977613.0, + "step": 9234 + }, + { + "epoch": 1.0141664836371622, + "grad_norm": 2.394618511199951, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.706475555896759, + "num_tokens": 239002498.0, + "step": 9235 + }, + { + "epoch": 1.014276301339776, + "grad_norm": 1.9719740152359009, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.703443169593811, + "num_tokens": 239038191.0, + "step": 9236 + }, + { + "epoch": 1.0143861190423897, + "grad_norm": 2.1052892208099365, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6891170740127563, + "num_tokens": 239067476.0, + "step": 9237 + }, + { + "epoch": 1.0144959367450033, + "grad_norm": 2.223734140396118, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7186720967292786, + "num_tokens": 239094249.0, + "step": 9238 + }, + { + "epoch": 1.014605754447617, + "grad_norm": 2.2806079387664795, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.715404748916626, + "num_tokens": 239119765.0, + "step": 9239 + }, + { + "epoch": 1.0147155721502306, + "grad_norm": 2.05727481842041, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7143669724464417, + "num_tokens": 239147445.0, + "step": 9240 + }, + { + "epoch": 1.0148253898528443, + "grad_norm": 2.348731756210327, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7285264730453491, + "num_tokens": 239169710.0, + "step": 9241 + }, + { + "epoch": 1.0149352075554579, + "grad_norm": 2.4102349281311035, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.731956958770752, + "num_tokens": 239191689.0, + "step": 9242 + }, + { + "epoch": 1.0150450252580716, + "grad_norm": 2.162654161453247, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7000619769096375, + "num_tokens": 239220203.0, + "step": 9243 + }, + { + "epoch": 1.0151548429606854, + "grad_norm": 2.232417106628418, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7401146292686462, + "num_tokens": 239247003.0, + "step": 9244 + }, + { + "epoch": 1.015264660663299, + "grad_norm": 2.4752347469329834, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7356300950050354, + "num_tokens": 239266673.0, + "step": 9245 + }, + { + "epoch": 1.0153744783659127, + "grad_norm": 2.25911021232605, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.709621787071228, + "num_tokens": 239291722.0, + "step": 9246 + }, + { + "epoch": 1.0154842960685262, + "grad_norm": 2.2968709468841553, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.72282874584198, + "num_tokens": 239315226.0, + "step": 9247 + }, + { + "epoch": 1.01559411377114, + "grad_norm": 2.2492010593414307, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7285372614860535, + "num_tokens": 239341129.0, + "step": 9248 + }, + { + "epoch": 1.0157039314737535, + "grad_norm": 2.1391825675964355, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7166605591773987, + "num_tokens": 239366717.0, + "step": 9249 + }, + { + "epoch": 1.0158137491763672, + "grad_norm": 2.221987724304199, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7154321670532227, + "num_tokens": 239392989.0, + "step": 9250 + }, + { + "epoch": 1.015923566878981, + "grad_norm": 2.1118171215057373, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7034804224967957, + "num_tokens": 239424703.0, + "step": 9251 + }, + { + "epoch": 1.0160333845815945, + "grad_norm": 2.1998026371002197, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6930168867111206, + "num_tokens": 239454364.0, + "step": 9252 + }, + { + "epoch": 1.0161432022842083, + "grad_norm": 2.204113721847534, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6950455904006958, + "num_tokens": 239482975.0, + "step": 9253 + }, + { + "epoch": 1.0162530199868218, + "grad_norm": 2.093168258666992, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6953632235527039, + "num_tokens": 239515619.0, + "step": 9254 + }, + { + "epoch": 1.0163628376894356, + "grad_norm": 2.4497718811035156, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7077398896217346, + "num_tokens": 239537698.0, + "step": 9255 + }, + { + "epoch": 1.0164726553920491, + "grad_norm": 2.311167001724243, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7233409881591797, + "num_tokens": 239562114.0, + "step": 9256 + }, + { + "epoch": 1.0165824730946629, + "grad_norm": 2.0352566242218018, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7001251578330994, + "num_tokens": 239590906.0, + "step": 9257 + }, + { + "epoch": 1.0166922907972764, + "grad_norm": 2.1975321769714355, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7352752685546875, + "num_tokens": 239615166.0, + "step": 9258 + }, + { + "epoch": 1.0168021084998902, + "grad_norm": 2.10933518409729, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7110641002655029, + "num_tokens": 239641273.0, + "step": 9259 + }, + { + "epoch": 1.016911926202504, + "grad_norm": 2.0948667526245117, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7102752327919006, + "num_tokens": 239669932.0, + "step": 9260 + }, + { + "epoch": 1.0170217439051175, + "grad_norm": 2.407099723815918, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7229841947555542, + "num_tokens": 239691431.0, + "step": 9261 + }, + { + "epoch": 1.0171315616077312, + "grad_norm": 2.560074806213379, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7121567726135254, + "num_tokens": 239714438.0, + "step": 9262 + }, + { + "epoch": 1.0172413793103448, + "grad_norm": 2.0263805389404297, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7197051644325256, + "num_tokens": 239741908.0, + "step": 9263 + }, + { + "epoch": 1.0173511970129585, + "grad_norm": 2.1185033321380615, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7127358913421631, + "num_tokens": 239772040.0, + "step": 9264 + }, + { + "epoch": 1.017461014715572, + "grad_norm": 2.123570203781128, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7341845631599426, + "num_tokens": 239798973.0, + "step": 9265 + }, + { + "epoch": 1.0175708324181858, + "grad_norm": 2.1563236713409424, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7276705503463745, + "num_tokens": 239826141.0, + "step": 9266 + }, + { + "epoch": 1.0176806501207996, + "grad_norm": 2.214958667755127, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7042537927627563, + "num_tokens": 239852736.0, + "step": 9267 + }, + { + "epoch": 1.017790467823413, + "grad_norm": 2.521697759628296, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.738950252532959, + "num_tokens": 239872701.0, + "step": 9268 + }, + { + "epoch": 1.0179002855260268, + "grad_norm": 2.03031325340271, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7023676633834839, + "num_tokens": 239904145.0, + "step": 9269 + }, + { + "epoch": 1.0180101032286404, + "grad_norm": 2.199475049972534, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.747910737991333, + "num_tokens": 239928765.0, + "step": 9270 + }, + { + "epoch": 1.0181199209312541, + "grad_norm": 2.2609457969665527, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6981090307235718, + "num_tokens": 239953182.0, + "step": 9271 + }, + { + "epoch": 1.0182297386338677, + "grad_norm": 2.442864418029785, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7214663028717041, + "num_tokens": 239975609.0, + "step": 9272 + }, + { + "epoch": 1.0183395563364814, + "grad_norm": 1.969957709312439, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7238661050796509, + "num_tokens": 240006532.0, + "step": 9273 + }, + { + "epoch": 1.0184493740390952, + "grad_norm": 1.8659182786941528, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7297511100769043, + "num_tokens": 240037463.0, + "step": 9274 + }, + { + "epoch": 1.0185591917417087, + "grad_norm": 2.047257423400879, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.6986315846443176, + "num_tokens": 240068775.0, + "step": 9275 + }, + { + "epoch": 1.0186690094443225, + "grad_norm": 2.054873466491699, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.705612301826477, + "num_tokens": 240098457.0, + "step": 9276 + }, + { + "epoch": 1.018778827146936, + "grad_norm": 2.230626344680786, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7052817344665527, + "num_tokens": 240126114.0, + "step": 9277 + }, + { + "epoch": 1.0188886448495498, + "grad_norm": 2.33150315284729, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7269693613052368, + "num_tokens": 240149000.0, + "step": 9278 + }, + { + "epoch": 1.0189984625521633, + "grad_norm": 1.8605544567108154, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6909183263778687, + "num_tokens": 240184417.0, + "step": 9279 + }, + { + "epoch": 1.019108280254777, + "grad_norm": 2.1316442489624023, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7110995054244995, + "num_tokens": 240213251.0, + "step": 9280 + }, + { + "epoch": 1.0192180979573908, + "grad_norm": 2.2163896560668945, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6912274360656738, + "num_tokens": 240243902.0, + "step": 9281 + }, + { + "epoch": 1.0193279156600044, + "grad_norm": 2.1234493255615234, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7169647216796875, + "num_tokens": 240271686.0, + "step": 9282 + }, + { + "epoch": 1.0194377333626181, + "grad_norm": 2.2607686519622803, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7001309394836426, + "num_tokens": 240298105.0, + "step": 9283 + }, + { + "epoch": 1.0195475510652316, + "grad_norm": 2.155130624771118, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7222289443016052, + "num_tokens": 240323329.0, + "step": 9284 + }, + { + "epoch": 1.0196573687678454, + "grad_norm": 2.2337515354156494, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7135770320892334, + "num_tokens": 240348312.0, + "step": 9285 + }, + { + "epoch": 1.019767186470459, + "grad_norm": 2.135126829147339, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7133820652961731, + "num_tokens": 240373960.0, + "step": 9286 + }, + { + "epoch": 1.0198770041730727, + "grad_norm": 2.442748546600342, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7239537239074707, + "num_tokens": 240395472.0, + "step": 9287 + }, + { + "epoch": 1.0199868218756865, + "grad_norm": 1.989802598953247, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7011772394180298, + "num_tokens": 240427402.0, + "step": 9288 + }, + { + "epoch": 1.0200966395783, + "grad_norm": 1.9054864645004272, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7258021831512451, + "num_tokens": 240461485.0, + "step": 9289 + }, + { + "epoch": 1.0202064572809137, + "grad_norm": 2.240046977996826, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7198541164398193, + "num_tokens": 240488154.0, + "step": 9290 + }, + { + "epoch": 1.0203162749835273, + "grad_norm": 2.303611993789673, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7179257869720459, + "num_tokens": 240515041.0, + "step": 9291 + }, + { + "epoch": 1.020426092686141, + "grad_norm": 2.0176842212677, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7232080698013306, + "num_tokens": 240543212.0, + "step": 9292 + }, + { + "epoch": 1.0205359103887546, + "grad_norm": 1.9204264879226685, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7252865433692932, + "num_tokens": 240577429.0, + "step": 9293 + }, + { + "epoch": 1.0206457280913683, + "grad_norm": 2.131817102432251, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7146588563919067, + "num_tokens": 240606022.0, + "step": 9294 + }, + { + "epoch": 1.020755545793982, + "grad_norm": 2.4622151851654053, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7112382650375366, + "num_tokens": 240627360.0, + "step": 9295 + }, + { + "epoch": 1.0208653634965956, + "grad_norm": 2.068605661392212, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7081412076950073, + "num_tokens": 240655953.0, + "step": 9296 + }, + { + "epoch": 1.0209751811992094, + "grad_norm": 2.143343448638916, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7180027961730957, + "num_tokens": 240681901.0, + "step": 9297 + }, + { + "epoch": 1.021084998901823, + "grad_norm": 2.1199870109558105, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7136807441711426, + "num_tokens": 240708896.0, + "step": 9298 + }, + { + "epoch": 1.0211948166044367, + "grad_norm": 2.1555628776550293, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7309677004814148, + "num_tokens": 240734080.0, + "step": 9299 + }, + { + "epoch": 1.0213046343070502, + "grad_norm": 2.4627890586853027, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7074218988418579, + "num_tokens": 240757776.0, + "step": 9300 + }, + { + "epoch": 1.021414452009664, + "grad_norm": 2.484121799468994, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7228753566741943, + "num_tokens": 240779967.0, + "step": 9301 + }, + { + "epoch": 1.0215242697122777, + "grad_norm": 2.1145012378692627, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7088674306869507, + "num_tokens": 240811222.0, + "step": 9302 + }, + { + "epoch": 1.0216340874148913, + "grad_norm": 2.1541695594787598, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.717064619064331, + "num_tokens": 240839514.0, + "step": 9303 + }, + { + "epoch": 1.021743905117505, + "grad_norm": 2.076446294784546, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.6999716758728027, + "num_tokens": 240869492.0, + "step": 9304 + }, + { + "epoch": 1.0218537228201185, + "grad_norm": 2.011518716812134, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6977418661117554, + "num_tokens": 240903330.0, + "step": 9305 + }, + { + "epoch": 1.0219635405227323, + "grad_norm": 2.0678606033325195, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7514100670814514, + "num_tokens": 240931670.0, + "step": 9306 + }, + { + "epoch": 1.0220733582253458, + "grad_norm": 2.245297908782959, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.731677770614624, + "num_tokens": 240955765.0, + "step": 9307 + }, + { + "epoch": 1.0221831759279596, + "grad_norm": 2.0858871936798096, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7187851071357727, + "num_tokens": 240984698.0, + "step": 9308 + }, + { + "epoch": 1.0222929936305734, + "grad_norm": 2.2881405353546143, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7148385047912598, + "num_tokens": 241011486.0, + "step": 9309 + }, + { + "epoch": 1.022402811333187, + "grad_norm": 2.2164623737335205, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7066288590431213, + "num_tokens": 241038036.0, + "step": 9310 + }, + { + "epoch": 1.0225126290358006, + "grad_norm": 2.3574976921081543, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7198374271392822, + "num_tokens": 241062846.0, + "step": 9311 + }, + { + "epoch": 1.0226224467384142, + "grad_norm": 2.2393598556518555, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7280911803245544, + "num_tokens": 241089714.0, + "step": 9312 + }, + { + "epoch": 1.022732264441028, + "grad_norm": 2.0152382850646973, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7013289928436279, + "num_tokens": 241122272.0, + "step": 9313 + }, + { + "epoch": 1.0228420821436415, + "grad_norm": 2.5694146156311035, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7233426570892334, + "num_tokens": 241143105.0, + "step": 9314 + }, + { + "epoch": 1.0229518998462552, + "grad_norm": 2.1837756633758545, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7051586508750916, + "num_tokens": 241170861.0, + "step": 9315 + }, + { + "epoch": 1.023061717548869, + "grad_norm": 2.3781321048736572, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7261529564857483, + "num_tokens": 241193185.0, + "step": 9316 + }, + { + "epoch": 1.0231715352514825, + "grad_norm": 2.0106265544891357, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7224492430686951, + "num_tokens": 241222262.0, + "step": 9317 + }, + { + "epoch": 1.0232813529540963, + "grad_norm": 2.400571823120117, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7108500003814697, + "num_tokens": 241245067.0, + "step": 9318 + }, + { + "epoch": 1.0233911706567098, + "grad_norm": 2.1835477352142334, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7139290571212769, + "num_tokens": 241272467.0, + "step": 9319 + }, + { + "epoch": 1.0235009883593236, + "grad_norm": 2.4337902069091797, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7145218849182129, + "num_tokens": 241297534.0, + "step": 9320 + }, + { + "epoch": 1.023610806061937, + "grad_norm": 2.0570781230926514, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7124038934707642, + "num_tokens": 241327611.0, + "step": 9321 + }, + { + "epoch": 1.0237206237645509, + "grad_norm": 2.1325228214263916, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7126678228378296, + "num_tokens": 241356755.0, + "step": 9322 + }, + { + "epoch": 1.0238304414671644, + "grad_norm": 2.44813871383667, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7195569276809692, + "num_tokens": 241378369.0, + "step": 9323 + }, + { + "epoch": 1.0239402591697782, + "grad_norm": 2.2404563426971436, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7166145443916321, + "num_tokens": 241404176.0, + "step": 9324 + }, + { + "epoch": 1.024050076872392, + "grad_norm": 2.2718353271484375, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7235943078994751, + "num_tokens": 241429475.0, + "step": 9325 + }, + { + "epoch": 1.0241598945750054, + "grad_norm": 2.481462240219116, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7177622318267822, + "num_tokens": 241450642.0, + "step": 9326 + }, + { + "epoch": 1.0242697122776192, + "grad_norm": 1.9882097244262695, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7345885038375854, + "num_tokens": 241480796.0, + "step": 9327 + }, + { + "epoch": 1.0243795299802327, + "grad_norm": 3.037085771560669, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7310169339179993, + "num_tokens": 241498114.0, + "step": 9328 + }, + { + "epoch": 1.0244893476828465, + "grad_norm": 2.5822439193725586, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7314285039901733, + "num_tokens": 241519122.0, + "step": 9329 + }, + { + "epoch": 1.02459916538546, + "grad_norm": 2.011556386947632, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7078057527542114, + "num_tokens": 241550097.0, + "step": 9330 + }, + { + "epoch": 1.0247089830880738, + "grad_norm": 2.1331560611724854, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6945705413818359, + "num_tokens": 241578237.0, + "step": 9331 + }, + { + "epoch": 1.0248188007906875, + "grad_norm": 2.3555049896240234, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7384529709815979, + "num_tokens": 241599538.0, + "step": 9332 + }, + { + "epoch": 1.024928618493301, + "grad_norm": 2.3153624534606934, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7241501808166504, + "num_tokens": 241623433.0, + "step": 9333 + }, + { + "epoch": 1.0250384361959148, + "grad_norm": 2.2699625492095947, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7233468294143677, + "num_tokens": 241647005.0, + "step": 9334 + }, + { + "epoch": 1.0251482538985284, + "grad_norm": 2.084534168243408, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7057767510414124, + "num_tokens": 241676177.0, + "step": 9335 + }, + { + "epoch": 1.0252580716011421, + "grad_norm": 2.1054859161376953, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7261958718299866, + "num_tokens": 241705183.0, + "step": 9336 + }, + { + "epoch": 1.0253678893037557, + "grad_norm": 2.5378432273864746, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7077829837799072, + "num_tokens": 241728831.0, + "step": 9337 + }, + { + "epoch": 1.0254777070063694, + "grad_norm": 1.9633290767669678, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7304754257202148, + "num_tokens": 241757350.0, + "step": 9338 + }, + { + "epoch": 1.0255875247089832, + "grad_norm": 1.9470784664154053, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7089716196060181, + "num_tokens": 241792208.0, + "step": 9339 + }, + { + "epoch": 1.0256973424115967, + "grad_norm": 2.112698554992676, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7228511571884155, + "num_tokens": 241820190.0, + "step": 9340 + }, + { + "epoch": 1.0258071601142105, + "grad_norm": 2.2767810821533203, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7215660810470581, + "num_tokens": 241844640.0, + "step": 9341 + }, + { + "epoch": 1.025916977816824, + "grad_norm": 2.4584178924560547, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.733333170413971, + "num_tokens": 241866014.0, + "step": 9342 + }, + { + "epoch": 1.0260267955194378, + "grad_norm": 2.068077802658081, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6946752071380615, + "num_tokens": 241896370.0, + "step": 9343 + }, + { + "epoch": 1.0261366132220513, + "grad_norm": 2.207252264022827, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7075818777084351, + "num_tokens": 241921445.0, + "step": 9344 + }, + { + "epoch": 1.026246430924665, + "grad_norm": 2.4212589263916016, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7277531623840332, + "num_tokens": 241942105.0, + "step": 9345 + }, + { + "epoch": 1.0263562486272788, + "grad_norm": 2.0575132369995117, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7004437446594238, + "num_tokens": 241974155.0, + "step": 9346 + }, + { + "epoch": 1.0264660663298923, + "grad_norm": 2.1858267784118652, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7142723798751831, + "num_tokens": 242000788.0, + "step": 9347 + }, + { + "epoch": 1.026575884032506, + "grad_norm": 2.114436626434326, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7265915274620056, + "num_tokens": 242028371.0, + "step": 9348 + }, + { + "epoch": 1.0266857017351196, + "grad_norm": 2.1250903606414795, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7264652252197266, + "num_tokens": 242056593.0, + "step": 9349 + }, + { + "epoch": 1.0267955194377334, + "grad_norm": 2.126988410949707, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7391639351844788, + "num_tokens": 242083641.0, + "step": 9350 + }, + { + "epoch": 1.026905337140347, + "grad_norm": 2.1300432682037354, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7317649126052856, + "num_tokens": 242109187.0, + "step": 9351 + }, + { + "epoch": 1.0270151548429607, + "grad_norm": 2.359697103500366, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7167903780937195, + "num_tokens": 242131308.0, + "step": 9352 + }, + { + "epoch": 1.0271249725455744, + "grad_norm": 2.2984673976898193, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7163784503936768, + "num_tokens": 242158139.0, + "step": 9353 + }, + { + "epoch": 1.027234790248188, + "grad_norm": 2.164287567138672, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7071057558059692, + "num_tokens": 242185517.0, + "step": 9354 + }, + { + "epoch": 1.0273446079508017, + "grad_norm": 2.1651041507720947, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7163728475570679, + "num_tokens": 242212944.0, + "step": 9355 + }, + { + "epoch": 1.0274544256534153, + "grad_norm": 2.2742035388946533, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7166472673416138, + "num_tokens": 242237921.0, + "step": 9356 + }, + { + "epoch": 1.027564243356029, + "grad_norm": 2.2188615798950195, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7070578932762146, + "num_tokens": 242264084.0, + "step": 9357 + }, + { + "epoch": 1.0276740610586426, + "grad_norm": 2.223517894744873, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6965138912200928, + "num_tokens": 242289515.0, + "step": 9358 + }, + { + "epoch": 1.0277838787612563, + "grad_norm": 2.3779616355895996, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7315747737884521, + "num_tokens": 242312676.0, + "step": 9359 + }, + { + "epoch": 1.02789369646387, + "grad_norm": 2.1225533485412598, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.709011435508728, + "num_tokens": 242341108.0, + "step": 9360 + }, + { + "epoch": 1.0280035141664836, + "grad_norm": 2.3053345680236816, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7166552543640137, + "num_tokens": 242364261.0, + "step": 9361 + }, + { + "epoch": 1.0281133318690974, + "grad_norm": 2.5229082107543945, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7207375764846802, + "num_tokens": 242385200.0, + "step": 9362 + }, + { + "epoch": 1.028223149571711, + "grad_norm": 1.9152899980545044, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7017073631286621, + "num_tokens": 242421340.0, + "step": 9363 + }, + { + "epoch": 1.0283329672743247, + "grad_norm": 2.129467487335205, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7044779062271118, + "num_tokens": 242449861.0, + "step": 9364 + }, + { + "epoch": 1.0284427849769382, + "grad_norm": 2.174567937850952, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7362539768218994, + "num_tokens": 242476627.0, + "step": 9365 + }, + { + "epoch": 1.028552602679552, + "grad_norm": 2.324970006942749, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7084279656410217, + "num_tokens": 242502244.0, + "step": 9366 + }, + { + "epoch": 1.0286624203821657, + "grad_norm": 2.2821240425109863, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7160640954971313, + "num_tokens": 242528568.0, + "step": 9367 + }, + { + "epoch": 1.0287722380847792, + "grad_norm": 2.1454734802246094, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7324005365371704, + "num_tokens": 242554618.0, + "step": 9368 + }, + { + "epoch": 1.028882055787393, + "grad_norm": 2.281926393508911, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7149515748023987, + "num_tokens": 242576525.0, + "step": 9369 + }, + { + "epoch": 1.0289918734900065, + "grad_norm": 2.0669384002685547, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6985528469085693, + "num_tokens": 242604868.0, + "step": 9370 + }, + { + "epoch": 1.0291016911926203, + "grad_norm": 2.2961440086364746, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7274521589279175, + "num_tokens": 242627934.0, + "step": 9371 + }, + { + "epoch": 1.0292115088952338, + "grad_norm": 2.4965949058532715, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7295730710029602, + "num_tokens": 242648332.0, + "step": 9372 + }, + { + "epoch": 1.0293213265978476, + "grad_norm": 2.1648287773132324, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7254046201705933, + "num_tokens": 242675591.0, + "step": 9373 + }, + { + "epoch": 1.0294311443004613, + "grad_norm": 2.3110013008117676, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7123488187789917, + "num_tokens": 242699493.0, + "step": 9374 + }, + { + "epoch": 1.0295409620030749, + "grad_norm": 2.1181442737579346, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6845858693122864, + "num_tokens": 242726766.0, + "step": 9375 + }, + { + "epoch": 1.0296507797056886, + "grad_norm": 1.9014015197753906, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6925451159477234, + "num_tokens": 242761048.0, + "step": 9376 + }, + { + "epoch": 1.0297605974083022, + "grad_norm": 2.512692928314209, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7002419233322144, + "num_tokens": 242785961.0, + "step": 9377 + }, + { + "epoch": 1.029870415110916, + "grad_norm": 2.178178310394287, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7109530568122864, + "num_tokens": 242813003.0, + "step": 9378 + }, + { + "epoch": 1.0299802328135295, + "grad_norm": 2.08227801322937, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6969007253646851, + "num_tokens": 242844550.0, + "step": 9379 + }, + { + "epoch": 1.0300900505161432, + "grad_norm": 2.5046303272247314, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7126003503799438, + "num_tokens": 242868020.0, + "step": 9380 + }, + { + "epoch": 1.030199868218757, + "grad_norm": 2.4498770236968994, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7075365781784058, + "num_tokens": 242890243.0, + "step": 9381 + }, + { + "epoch": 1.0303096859213705, + "grad_norm": 2.206000566482544, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7200583219528198, + "num_tokens": 242915586.0, + "step": 9382 + }, + { + "epoch": 1.0304195036239843, + "grad_norm": 2.1122982501983643, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7081894874572754, + "num_tokens": 242943042.0, + "step": 9383 + }, + { + "epoch": 1.0305293213265978, + "grad_norm": 2.2015931606292725, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7219136953353882, + "num_tokens": 242968983.0, + "step": 9384 + }, + { + "epoch": 1.0306391390292116, + "grad_norm": 2.1996593475341797, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7183792591094971, + "num_tokens": 242999537.0, + "step": 9385 + }, + { + "epoch": 1.030748956731825, + "grad_norm": 2.0724637508392334, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7082575559616089, + "num_tokens": 243029698.0, + "step": 9386 + }, + { + "epoch": 1.0308587744344389, + "grad_norm": 2.1952807903289795, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7222479581832886, + "num_tokens": 243057058.0, + "step": 9387 + }, + { + "epoch": 1.0309685921370524, + "grad_norm": 2.1302523612976074, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7239537835121155, + "num_tokens": 243083376.0, + "step": 9388 + }, + { + "epoch": 1.0310784098396661, + "grad_norm": 2.293914556503296, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7170538902282715, + "num_tokens": 243106844.0, + "step": 9389 + }, + { + "epoch": 1.03118822754228, + "grad_norm": 1.9554023742675781, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7122633457183838, + "num_tokens": 243139308.0, + "step": 9390 + }, + { + "epoch": 1.0312980452448934, + "grad_norm": 2.1495285034179688, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7087187767028809, + "num_tokens": 243165509.0, + "step": 9391 + }, + { + "epoch": 1.0314078629475072, + "grad_norm": 2.3119091987609863, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7289894819259644, + "num_tokens": 243187638.0, + "step": 9392 + }, + { + "epoch": 1.0315176806501207, + "grad_norm": 2.5613772869110107, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7332842946052551, + "num_tokens": 243207999.0, + "step": 9393 + }, + { + "epoch": 1.0316274983527345, + "grad_norm": 2.1795480251312256, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7066690921783447, + "num_tokens": 243236555.0, + "step": 9394 + }, + { + "epoch": 1.031737316055348, + "grad_norm": 2.126786470413208, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7357100248336792, + "num_tokens": 243262821.0, + "step": 9395 + }, + { + "epoch": 1.0318471337579618, + "grad_norm": 2.0852725505828857, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7112419605255127, + "num_tokens": 243292156.0, + "step": 9396 + }, + { + "epoch": 1.0319569514605755, + "grad_norm": 2.2870075702667236, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7142099142074585, + "num_tokens": 243317821.0, + "step": 9397 + }, + { + "epoch": 1.032066769163189, + "grad_norm": 2.1899466514587402, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7051597833633423, + "num_tokens": 243345859.0, + "step": 9398 + }, + { + "epoch": 1.0321765868658028, + "grad_norm": 2.2467384338378906, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7053482532501221, + "num_tokens": 243372271.0, + "step": 9399 + }, + { + "epoch": 1.0322864045684164, + "grad_norm": 2.2693498134613037, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.6969010829925537, + "num_tokens": 243399628.0, + "step": 9400 + }, + { + "epoch": 1.0323962222710301, + "grad_norm": 2.686105489730835, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7391897439956665, + "num_tokens": 243419002.0, + "step": 9401 + }, + { + "epoch": 1.0325060399736437, + "grad_norm": 2.0796055793762207, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7092803716659546, + "num_tokens": 243449824.0, + "step": 9402 + }, + { + "epoch": 1.0326158576762574, + "grad_norm": 2.302783727645874, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7152785062789917, + "num_tokens": 243474777.0, + "step": 9403 + }, + { + "epoch": 1.0327256753788712, + "grad_norm": 1.983114242553711, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7083435654640198, + "num_tokens": 243508762.0, + "step": 9404 + }, + { + "epoch": 1.0328354930814847, + "grad_norm": 2.6008760929107666, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7362476587295532, + "num_tokens": 243528461.0, + "step": 9405 + }, + { + "epoch": 1.0329453107840985, + "grad_norm": 2.250288248062134, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7146101593971252, + "num_tokens": 243556317.0, + "step": 9406 + }, + { + "epoch": 1.033055128486712, + "grad_norm": 2.120835542678833, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7188445329666138, + "num_tokens": 243585250.0, + "step": 9407 + }, + { + "epoch": 1.0331649461893258, + "grad_norm": 2.4607512950897217, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7373747825622559, + "num_tokens": 243606557.0, + "step": 9408 + }, + { + "epoch": 1.0332747638919393, + "grad_norm": 2.153022050857544, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7104834914207458, + "num_tokens": 243632472.0, + "step": 9409 + }, + { + "epoch": 1.033384581594553, + "grad_norm": 2.079082727432251, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7275202870368958, + "num_tokens": 243660567.0, + "step": 9410 + }, + { + "epoch": 1.0334943992971668, + "grad_norm": 2.5353336334228516, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7239693403244019, + "num_tokens": 243681101.0, + "step": 9411 + }, + { + "epoch": 1.0336042169997803, + "grad_norm": 2.0904314517974854, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7112585306167603, + "num_tokens": 243710120.0, + "step": 9412 + }, + { + "epoch": 1.033714034702394, + "grad_norm": 2.492222785949707, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7358291149139404, + "num_tokens": 243731564.0, + "step": 9413 + }, + { + "epoch": 1.0338238524050076, + "grad_norm": 2.5146753787994385, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7520872950553894, + "num_tokens": 243752083.0, + "step": 9414 + }, + { + "epoch": 1.0339336701076214, + "grad_norm": 2.31551456451416, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7532663345336914, + "num_tokens": 243776098.0, + "step": 9415 + }, + { + "epoch": 1.034043487810235, + "grad_norm": 2.420069932937622, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7369465827941895, + "num_tokens": 243798831.0, + "step": 9416 + }, + { + "epoch": 1.0341533055128487, + "grad_norm": 2.0666067600250244, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7106742858886719, + "num_tokens": 243826736.0, + "step": 9417 + }, + { + "epoch": 1.0342631232154624, + "grad_norm": 2.323911190032959, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7240839004516602, + "num_tokens": 243850406.0, + "step": 9418 + }, + { + "epoch": 1.034372940918076, + "grad_norm": 5.724987506866455, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7474541068077087, + "num_tokens": 243866665.0, + "step": 9419 + }, + { + "epoch": 1.0344827586206897, + "grad_norm": 2.291367292404175, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7247308492660522, + "num_tokens": 243891260.0, + "step": 9420 + }, + { + "epoch": 1.0345925763233033, + "grad_norm": 2.4183313846588135, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7256446480751038, + "num_tokens": 243912863.0, + "step": 9421 + }, + { + "epoch": 1.034702394025917, + "grad_norm": 2.083991289138794, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7214914560317993, + "num_tokens": 243941500.0, + "step": 9422 + }, + { + "epoch": 1.0348122117285306, + "grad_norm": 2.354883909225464, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7022498250007629, + "num_tokens": 243965604.0, + "step": 9423 + }, + { + "epoch": 1.0349220294311443, + "grad_norm": 2.271286725997925, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7268435955047607, + "num_tokens": 243989569.0, + "step": 9424 + }, + { + "epoch": 1.035031847133758, + "grad_norm": 2.1390092372894287, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7292705774307251, + "num_tokens": 244018389.0, + "step": 9425 + }, + { + "epoch": 1.0351416648363716, + "grad_norm": 2.2885122299194336, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7176109552383423, + "num_tokens": 244042053.0, + "step": 9426 + }, + { + "epoch": 1.0352514825389854, + "grad_norm": 2.042856216430664, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7165687084197998, + "num_tokens": 244072302.0, + "step": 9427 + }, + { + "epoch": 1.035361300241599, + "grad_norm": 2.4307634830474854, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7074978947639465, + "num_tokens": 244095758.0, + "step": 9428 + }, + { + "epoch": 1.0354711179442126, + "grad_norm": 2.0577597618103027, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7206387519836426, + "num_tokens": 244124923.0, + "step": 9429 + }, + { + "epoch": 1.0355809356468262, + "grad_norm": 2.334096908569336, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7147384285926819, + "num_tokens": 244148357.0, + "step": 9430 + }, + { + "epoch": 1.03569075334944, + "grad_norm": 2.1765592098236084, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7187432050704956, + "num_tokens": 244175301.0, + "step": 9431 + }, + { + "epoch": 1.0358005710520537, + "grad_norm": 2.201780319213867, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7219381332397461, + "num_tokens": 244201386.0, + "step": 9432 + }, + { + "epoch": 1.0359103887546672, + "grad_norm": 2.182297945022583, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7103020548820496, + "num_tokens": 244228633.0, + "step": 9433 + }, + { + "epoch": 1.036020206457281, + "grad_norm": 2.1913700103759766, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7146521806716919, + "num_tokens": 244256930.0, + "step": 9434 + }, + { + "epoch": 1.0361300241598945, + "grad_norm": 2.3542401790618896, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7494873404502869, + "num_tokens": 244278992.0, + "step": 9435 + }, + { + "epoch": 1.0362398418625083, + "grad_norm": 2.095668315887451, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7180162668228149, + "num_tokens": 244307073.0, + "step": 9436 + }, + { + "epoch": 1.0363496595651218, + "grad_norm": 2.244600772857666, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7158032059669495, + "num_tokens": 244333029.0, + "step": 9437 + }, + { + "epoch": 1.0364594772677356, + "grad_norm": 2.421593189239502, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7390671968460083, + "num_tokens": 244354581.0, + "step": 9438 + }, + { + "epoch": 1.036569294970349, + "grad_norm": 2.574918270111084, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7158796191215515, + "num_tokens": 244376640.0, + "step": 9439 + }, + { + "epoch": 1.0366791126729629, + "grad_norm": 2.2025668621063232, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6985713243484497, + "num_tokens": 244403385.0, + "step": 9440 + }, + { + "epoch": 1.0367889303755766, + "grad_norm": 2.271134614944458, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7059862613677979, + "num_tokens": 244431391.0, + "step": 9441 + }, + { + "epoch": 1.0368987480781902, + "grad_norm": 2.140636682510376, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7089951038360596, + "num_tokens": 244458610.0, + "step": 9442 + }, + { + "epoch": 1.037008565780804, + "grad_norm": 2.0015711784362793, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7075810432434082, + "num_tokens": 244490252.0, + "step": 9443 + }, + { + "epoch": 1.0371183834834174, + "grad_norm": 2.1880290508270264, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7272505760192871, + "num_tokens": 244516220.0, + "step": 9444 + }, + { + "epoch": 1.0372282011860312, + "grad_norm": 2.154562473297119, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7189503312110901, + "num_tokens": 244544253.0, + "step": 9445 + }, + { + "epoch": 1.037338018888645, + "grad_norm": 2.3334269523620605, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7215056419372559, + "num_tokens": 244567126.0, + "step": 9446 + }, + { + "epoch": 1.0374478365912585, + "grad_norm": 2.112393379211426, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7214161157608032, + "num_tokens": 244595282.0, + "step": 9447 + }, + { + "epoch": 1.0375576542938723, + "grad_norm": 2.5919294357299805, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7039021253585815, + "num_tokens": 244617865.0, + "step": 9448 + }, + { + "epoch": 1.0376674719964858, + "grad_norm": 2.095212697982788, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7163410186767578, + "num_tokens": 244645296.0, + "step": 9449 + }, + { + "epoch": 1.0377772896990995, + "grad_norm": 2.6846981048583984, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7389440536499023, + "num_tokens": 244663788.0, + "step": 9450 + }, + { + "epoch": 1.037887107401713, + "grad_norm": 2.480142831802368, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7199281454086304, + "num_tokens": 244685999.0, + "step": 9451 + }, + { + "epoch": 1.0379969251043268, + "grad_norm": 2.2104063034057617, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7041269540786743, + "num_tokens": 244713331.0, + "step": 9452 + }, + { + "epoch": 1.0381067428069404, + "grad_norm": 2.1265311241149902, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.728289008140564, + "num_tokens": 244740153.0, + "step": 9453 + }, + { + "epoch": 1.0382165605095541, + "grad_norm": 2.378706932067871, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7219541072845459, + "num_tokens": 244763602.0, + "step": 9454 + }, + { + "epoch": 1.0383263782121679, + "grad_norm": 2.0061521530151367, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7272972464561462, + "num_tokens": 244793724.0, + "step": 9455 + }, + { + "epoch": 1.0384361959147814, + "grad_norm": 2.190676689147949, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7376561164855957, + "num_tokens": 244818177.0, + "step": 9456 + }, + { + "epoch": 1.0385460136173952, + "grad_norm": 1.952265739440918, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7284093499183655, + "num_tokens": 244848557.0, + "step": 9457 + }, + { + "epoch": 1.0386558313200087, + "grad_norm": 2.215259552001953, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7283228635787964, + "num_tokens": 244874269.0, + "step": 9458 + }, + { + "epoch": 1.0387656490226225, + "grad_norm": 2.4648263454437256, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7283003330230713, + "num_tokens": 244894552.0, + "step": 9459 + }, + { + "epoch": 1.038875466725236, + "grad_norm": 2.2097084522247314, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7345016002655029, + "num_tokens": 244919365.0, + "step": 9460 + }, + { + "epoch": 1.0389852844278498, + "grad_norm": 2.2514026165008545, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7128371596336365, + "num_tokens": 244943450.0, + "step": 9461 + }, + { + "epoch": 1.0390951021304635, + "grad_norm": 2.350222110748291, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7164008021354675, + "num_tokens": 244966856.0, + "step": 9462 + }, + { + "epoch": 1.039204919833077, + "grad_norm": 2.051600933074951, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6986855864524841, + "num_tokens": 244998059.0, + "step": 9463 + }, + { + "epoch": 1.0393147375356908, + "grad_norm": 2.2715227603912354, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7083427906036377, + "num_tokens": 245022171.0, + "step": 9464 + }, + { + "epoch": 1.0394245552383043, + "grad_norm": 2.2241051197052, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7273685336112976, + "num_tokens": 245049263.0, + "step": 9465 + }, + { + "epoch": 1.039534372940918, + "grad_norm": 2.3060381412506104, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7393315434455872, + "num_tokens": 245073793.0, + "step": 9466 + }, + { + "epoch": 1.0396441906435316, + "grad_norm": 1.9983199834823608, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7189823985099792, + "num_tokens": 245103029.0, + "step": 9467 + }, + { + "epoch": 1.0397540083461454, + "grad_norm": 2.225226879119873, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7164635062217712, + "num_tokens": 245128495.0, + "step": 9468 + }, + { + "epoch": 1.0398638260487592, + "grad_norm": 2.0903995037078857, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.725069522857666, + "num_tokens": 245156033.0, + "step": 9469 + }, + { + "epoch": 1.0399736437513727, + "grad_norm": 2.67055344581604, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7387497425079346, + "num_tokens": 245174744.0, + "step": 9470 + }, + { + "epoch": 1.0400834614539864, + "grad_norm": 2.3552191257476807, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7103534936904907, + "num_tokens": 245200011.0, + "step": 9471 + }, + { + "epoch": 1.0401932791566, + "grad_norm": 2.144068717956543, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7285206317901611, + "num_tokens": 245227165.0, + "step": 9472 + }, + { + "epoch": 1.0403030968592137, + "grad_norm": 2.0070273876190186, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7114663124084473, + "num_tokens": 245257985.0, + "step": 9473 + }, + { + "epoch": 1.0404129145618273, + "grad_norm": 2.4007468223571777, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7506425380706787, + "num_tokens": 245279876.0, + "step": 9474 + }, + { + "epoch": 1.040522732264441, + "grad_norm": 2.303030252456665, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7038720846176147, + "num_tokens": 245305272.0, + "step": 9475 + }, + { + "epoch": 1.0406325499670548, + "grad_norm": 2.3593862056732178, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7153993844985962, + "num_tokens": 245328051.0, + "step": 9476 + }, + { + "epoch": 1.0407423676696683, + "grad_norm": 1.9620451927185059, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7141146659851074, + "num_tokens": 245358157.0, + "step": 9477 + }, + { + "epoch": 1.040852185372282, + "grad_norm": 2.1063737869262695, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7075888514518738, + "num_tokens": 245384741.0, + "step": 9478 + }, + { + "epoch": 1.0409620030748956, + "grad_norm": 2.2590718269348145, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7076666951179504, + "num_tokens": 245409463.0, + "step": 9479 + }, + { + "epoch": 1.0410718207775094, + "grad_norm": 2.176236867904663, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.733325183391571, + "num_tokens": 245438723.0, + "step": 9480 + }, + { + "epoch": 1.041181638480123, + "grad_norm": 2.50406813621521, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7314006090164185, + "num_tokens": 245458902.0, + "step": 9481 + }, + { + "epoch": 1.0412914561827367, + "grad_norm": 2.116814613342285, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7055248022079468, + "num_tokens": 245485163.0, + "step": 9482 + }, + { + "epoch": 1.0414012738853504, + "grad_norm": 2.095442056655884, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7300736308097839, + "num_tokens": 245511526.0, + "step": 9483 + }, + { + "epoch": 1.041511091587964, + "grad_norm": 2.4249143600463867, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7250645160675049, + "num_tokens": 245533940.0, + "step": 9484 + }, + { + "epoch": 1.0416209092905777, + "grad_norm": 2.4550728797912598, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7504945397377014, + "num_tokens": 245553366.0, + "step": 9485 + }, + { + "epoch": 1.0417307269931912, + "grad_norm": 2.520965576171875, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7358089685440063, + "num_tokens": 245573331.0, + "step": 9486 + }, + { + "epoch": 1.041840544695805, + "grad_norm": 2.4743008613586426, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7433381080627441, + "num_tokens": 245595441.0, + "step": 9487 + }, + { + "epoch": 1.0419503623984185, + "grad_norm": 2.5094454288482666, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7283312082290649, + "num_tokens": 245616215.0, + "step": 9488 + }, + { + "epoch": 1.0420601801010323, + "grad_norm": 2.2017273902893066, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7150632739067078, + "num_tokens": 245641474.0, + "step": 9489 + }, + { + "epoch": 1.042169997803646, + "grad_norm": 2.283200979232788, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7136317491531372, + "num_tokens": 245666626.0, + "step": 9490 + }, + { + "epoch": 1.0422798155062596, + "grad_norm": 2.0924859046936035, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7224778532981873, + "num_tokens": 245693609.0, + "step": 9491 + }, + { + "epoch": 1.0423896332088733, + "grad_norm": 2.4171407222747803, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7190089821815491, + "num_tokens": 245715909.0, + "step": 9492 + }, + { + "epoch": 1.0424994509114869, + "grad_norm": 2.273069381713867, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7266401052474976, + "num_tokens": 245743609.0, + "step": 9493 + }, + { + "epoch": 1.0426092686141006, + "grad_norm": 2.5106444358825684, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7207004427909851, + "num_tokens": 245765807.0, + "step": 9494 + }, + { + "epoch": 1.0427190863167142, + "grad_norm": 2.0362987518310547, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7025826573371887, + "num_tokens": 245798404.0, + "step": 9495 + }, + { + "epoch": 1.042828904019328, + "grad_norm": 2.4028329849243164, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7151307463645935, + "num_tokens": 245820725.0, + "step": 9496 + }, + { + "epoch": 1.0429387217219417, + "grad_norm": 2.281956911087036, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7093287706375122, + "num_tokens": 245844551.0, + "step": 9497 + }, + { + "epoch": 1.0430485394245552, + "grad_norm": 2.143730640411377, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7334127426147461, + "num_tokens": 245870890.0, + "step": 9498 + }, + { + "epoch": 1.043158357127169, + "grad_norm": 2.1868910789489746, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7105704545974731, + "num_tokens": 245896044.0, + "step": 9499 + }, + { + "epoch": 1.0432681748297825, + "grad_norm": 2.253142833709717, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7006897330284119, + "num_tokens": 245923223.0, + "step": 9500 + }, + { + "epoch": 1.0433779925323963, + "grad_norm": 2.170745611190796, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7071751952171326, + "num_tokens": 245950175.0, + "step": 9501 + }, + { + "epoch": 1.0434878102350098, + "grad_norm": 2.337568521499634, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7284698486328125, + "num_tokens": 245974500.0, + "step": 9502 + }, + { + "epoch": 1.0435976279376236, + "grad_norm": 1.9771709442138672, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7021178007125854, + "num_tokens": 246005891.0, + "step": 9503 + }, + { + "epoch": 1.043707445640237, + "grad_norm": 2.295152187347412, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6879363656044006, + "num_tokens": 246030961.0, + "step": 9504 + }, + { + "epoch": 1.0438172633428509, + "grad_norm": 2.3028199672698975, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7103891968727112, + "num_tokens": 246054915.0, + "step": 9505 + }, + { + "epoch": 1.0439270810454646, + "grad_norm": 2.193671703338623, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7265781164169312, + "num_tokens": 246081890.0, + "step": 9506 + }, + { + "epoch": 1.0440368987480781, + "grad_norm": 2.100219249725342, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7224931716918945, + "num_tokens": 246110456.0, + "step": 9507 + }, + { + "epoch": 1.044146716450692, + "grad_norm": 2.0449025630950928, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6973137855529785, + "num_tokens": 246139574.0, + "step": 9508 + }, + { + "epoch": 1.0442565341533054, + "grad_norm": 2.26877498626709, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.705131471157074, + "num_tokens": 246165445.0, + "step": 9509 + }, + { + "epoch": 1.0443663518559192, + "grad_norm": 2.479870557785034, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7335301637649536, + "num_tokens": 246185870.0, + "step": 9510 + }, + { + "epoch": 1.0444761695585327, + "grad_norm": 1.9804171323776245, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.6991922855377197, + "num_tokens": 246216822.0, + "step": 9511 + }, + { + "epoch": 1.0445859872611465, + "grad_norm": 2.451728582382202, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7334033250808716, + "num_tokens": 246239351.0, + "step": 9512 + }, + { + "epoch": 1.0446958049637602, + "grad_norm": 2.3923449516296387, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7347837090492249, + "num_tokens": 246262271.0, + "step": 9513 + }, + { + "epoch": 1.0448056226663738, + "grad_norm": 2.394443988800049, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.716167151927948, + "num_tokens": 246286562.0, + "step": 9514 + }, + { + "epoch": 1.0449154403689875, + "grad_norm": 2.4527225494384766, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7125715017318726, + "num_tokens": 246310265.0, + "step": 9515 + }, + { + "epoch": 1.045025258071601, + "grad_norm": 2.365293025970459, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7024250030517578, + "num_tokens": 246335195.0, + "step": 9516 + }, + { + "epoch": 1.0451350757742148, + "grad_norm": 2.3487210273742676, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7267391085624695, + "num_tokens": 246360099.0, + "step": 9517 + }, + { + "epoch": 1.0452448934768284, + "grad_norm": 2.282809019088745, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6859724521636963, + "num_tokens": 246385857.0, + "step": 9518 + }, + { + "epoch": 1.0453547111794421, + "grad_norm": 2.206109046936035, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7386741638183594, + "num_tokens": 246411918.0, + "step": 9519 + }, + { + "epoch": 1.0454645288820559, + "grad_norm": 2.3792850971221924, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7427259683609009, + "num_tokens": 246433859.0, + "step": 9520 + }, + { + "epoch": 1.0455743465846694, + "grad_norm": 2.47082257270813, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7340124845504761, + "num_tokens": 246455155.0, + "step": 9521 + }, + { + "epoch": 1.0456841642872832, + "grad_norm": 2.08064341545105, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7182443737983704, + "num_tokens": 246482223.0, + "step": 9522 + }, + { + "epoch": 1.0457939819898967, + "grad_norm": 2.2803189754486084, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.70548015832901, + "num_tokens": 246507535.0, + "step": 9523 + }, + { + "epoch": 1.0459037996925105, + "grad_norm": 2.400294065475464, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7170747518539429, + "num_tokens": 246530389.0, + "step": 9524 + }, + { + "epoch": 1.046013617395124, + "grad_norm": 2.125910758972168, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7101457715034485, + "num_tokens": 246558410.0, + "step": 9525 + }, + { + "epoch": 1.0461234350977378, + "grad_norm": 2.0515198707580566, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7316612005233765, + "num_tokens": 246586971.0, + "step": 9526 + }, + { + "epoch": 1.0462332528003515, + "grad_norm": 2.224313735961914, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7213879227638245, + "num_tokens": 246611921.0, + "step": 9527 + }, + { + "epoch": 1.046343070502965, + "grad_norm": 2.287370443344116, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7231743931770325, + "num_tokens": 246637183.0, + "step": 9528 + }, + { + "epoch": 1.0464528882055788, + "grad_norm": 2.2175443172454834, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7113785147666931, + "num_tokens": 246661123.0, + "step": 9529 + }, + { + "epoch": 1.0465627059081923, + "grad_norm": 2.250997304916382, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7305523157119751, + "num_tokens": 246684376.0, + "step": 9530 + }, + { + "epoch": 1.046672523610806, + "grad_norm": 2.6065802574157715, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7259560227394104, + "num_tokens": 246703163.0, + "step": 9531 + }, + { + "epoch": 1.0467823413134196, + "grad_norm": 2.2240865230560303, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7309809923171997, + "num_tokens": 246727436.0, + "step": 9532 + }, + { + "epoch": 1.0468921590160334, + "grad_norm": 2.1257247924804688, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7237449288368225, + "num_tokens": 246754963.0, + "step": 9533 + }, + { + "epoch": 1.0470019767186471, + "grad_norm": 2.409899950027466, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7333709597587585, + "num_tokens": 246777203.0, + "step": 9534 + }, + { + "epoch": 1.0471117944212607, + "grad_norm": 2.3774945735931396, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7474758625030518, + "num_tokens": 246800178.0, + "step": 9535 + }, + { + "epoch": 1.0472216121238744, + "grad_norm": 2.1728286743164062, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7081544995307922, + "num_tokens": 246827037.0, + "step": 9536 + }, + { + "epoch": 1.047331429826488, + "grad_norm": 2.4594779014587402, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7493654489517212, + "num_tokens": 246848866.0, + "step": 9537 + }, + { + "epoch": 1.0474412475291017, + "grad_norm": 2.5199060440063477, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7240512371063232, + "num_tokens": 246869487.0, + "step": 9538 + }, + { + "epoch": 1.0475510652317153, + "grad_norm": 2.2861838340759277, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7001718878746033, + "num_tokens": 246897217.0, + "step": 9539 + }, + { + "epoch": 1.047660882934329, + "grad_norm": 2.0598039627075195, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7037033438682556, + "num_tokens": 246927339.0, + "step": 9540 + }, + { + "epoch": 1.0477707006369428, + "grad_norm": 2.0163416862487793, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7195941209793091, + "num_tokens": 246957855.0, + "step": 9541 + }, + { + "epoch": 1.0478805183395563, + "grad_norm": 2.3725595474243164, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7327219247817993, + "num_tokens": 246979698.0, + "step": 9542 + }, + { + "epoch": 1.04799033604217, + "grad_norm": 2.122727870941162, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7412768602371216, + "num_tokens": 247007272.0, + "step": 9543 + }, + { + "epoch": 1.0481001537447836, + "grad_norm": 2.4020307064056396, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7379138469696045, + "num_tokens": 247030032.0, + "step": 9544 + }, + { + "epoch": 1.0482099714473974, + "grad_norm": 2.287970542907715, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7258522510528564, + "num_tokens": 247053901.0, + "step": 9545 + }, + { + "epoch": 1.048319789150011, + "grad_norm": 2.422614574432373, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.714474081993103, + "num_tokens": 247076976.0, + "step": 9546 + }, + { + "epoch": 1.0484296068526247, + "grad_norm": 2.542941093444824, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7371387481689453, + "num_tokens": 247097690.0, + "step": 9547 + }, + { + "epoch": 1.0485394245552384, + "grad_norm": 2.2935523986816406, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7373467087745667, + "num_tokens": 247121076.0, + "step": 9548 + }, + { + "epoch": 1.048649242257852, + "grad_norm": 2.222928047180176, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7318155765533447, + "num_tokens": 247145963.0, + "step": 9549 + }, + { + "epoch": 1.0487590599604657, + "grad_norm": 2.1792426109313965, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7105987071990967, + "num_tokens": 247172368.0, + "step": 9550 + }, + { + "epoch": 1.0488688776630792, + "grad_norm": 2.164456605911255, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.6985138058662415, + "num_tokens": 247201180.0, + "step": 9551 + }, + { + "epoch": 1.048978695365693, + "grad_norm": 2.0425682067871094, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7048625946044922, + "num_tokens": 247232563.0, + "step": 9552 + }, + { + "epoch": 1.0490885130683065, + "grad_norm": 2.289836883544922, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6984504461288452, + "num_tokens": 247257790.0, + "step": 9553 + }, + { + "epoch": 1.0491983307709203, + "grad_norm": 2.358640193939209, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7210758924484253, + "num_tokens": 247279700.0, + "step": 9554 + }, + { + "epoch": 1.0493081484735338, + "grad_norm": 2.3391177654266357, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7383090853691101, + "num_tokens": 247303222.0, + "step": 9555 + }, + { + "epoch": 1.0494179661761476, + "grad_norm": 2.3346593379974365, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7390426397323608, + "num_tokens": 247325782.0, + "step": 9556 + }, + { + "epoch": 1.0495277838787613, + "grad_norm": 2.0750763416290283, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6973470449447632, + "num_tokens": 247356404.0, + "step": 9557 + }, + { + "epoch": 1.0496376015813749, + "grad_norm": 1.9342108964920044, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7026644945144653, + "num_tokens": 247389537.0, + "step": 9558 + }, + { + "epoch": 1.0497474192839886, + "grad_norm": 2.40276837348938, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7135554552078247, + "num_tokens": 247412214.0, + "step": 9559 + }, + { + "epoch": 1.0498572369866022, + "grad_norm": 2.449660539627075, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7495704889297485, + "num_tokens": 247432745.0, + "step": 9560 + }, + { + "epoch": 1.049967054689216, + "grad_norm": 2.238708257675171, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7196576595306396, + "num_tokens": 247457197.0, + "step": 9561 + }, + { + "epoch": 1.0500768723918297, + "grad_norm": 2.578103542327881, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7012076377868652, + "num_tokens": 247479184.0, + "step": 9562 + }, + { + "epoch": 1.0501866900944432, + "grad_norm": 2.411846160888672, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7374162673950195, + "num_tokens": 247502682.0, + "step": 9563 + }, + { + "epoch": 1.050296507797057, + "grad_norm": 2.2443082332611084, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.706405758857727, + "num_tokens": 247531199.0, + "step": 9564 + }, + { + "epoch": 1.0504063254996705, + "grad_norm": 2.1391215324401855, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6996220350265503, + "num_tokens": 247560239.0, + "step": 9565 + }, + { + "epoch": 1.0505161432022843, + "grad_norm": 2.3373851776123047, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7033069133758545, + "num_tokens": 247584598.0, + "step": 9566 + }, + { + "epoch": 1.0506259609048978, + "grad_norm": 2.1538870334625244, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7274954319000244, + "num_tokens": 247610185.0, + "step": 9567 + }, + { + "epoch": 1.0507357786075116, + "grad_norm": 2.3608736991882324, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.751922070980072, + "num_tokens": 247632298.0, + "step": 9568 + }, + { + "epoch": 1.050845596310125, + "grad_norm": 2.050166368484497, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7304350137710571, + "num_tokens": 247659916.0, + "step": 9569 + }, + { + "epoch": 1.0509554140127388, + "grad_norm": 2.080735921859741, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7249835133552551, + "num_tokens": 247688480.0, + "step": 9570 + }, + { + "epoch": 1.0510652317153526, + "grad_norm": 2.164189338684082, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7170629501342773, + "num_tokens": 247716386.0, + "step": 9571 + }, + { + "epoch": 1.0511750494179661, + "grad_norm": 2.2697184085845947, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7142858505249023, + "num_tokens": 247741248.0, + "step": 9572 + }, + { + "epoch": 1.05128486712058, + "grad_norm": 2.2116308212280273, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7209838628768921, + "num_tokens": 247767135.0, + "step": 9573 + }, + { + "epoch": 1.0513946848231934, + "grad_norm": 2.200298309326172, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7187384366989136, + "num_tokens": 247793866.0, + "step": 9574 + }, + { + "epoch": 1.0515045025258072, + "grad_norm": 1.9790784120559692, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7107871770858765, + "num_tokens": 247826488.0, + "step": 9575 + }, + { + "epoch": 1.0516143202284207, + "grad_norm": 2.427279472351074, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7047704458236694, + "num_tokens": 247848265.0, + "step": 9576 + }, + { + "epoch": 1.0517241379310345, + "grad_norm": 2.0600671768188477, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7130360007286072, + "num_tokens": 247878027.0, + "step": 9577 + }, + { + "epoch": 1.0518339556336482, + "grad_norm": 2.5098958015441895, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7317934632301331, + "num_tokens": 247899160.0, + "step": 9578 + }, + { + "epoch": 1.0519437733362618, + "grad_norm": 2.299187660217285, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6856330037117004, + "num_tokens": 247924094.0, + "step": 9579 + }, + { + "epoch": 1.0520535910388755, + "grad_norm": 1.8356155157089233, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.702592670917511, + "num_tokens": 247958200.0, + "step": 9580 + }, + { + "epoch": 1.052163408741489, + "grad_norm": 2.26425838470459, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7177440524101257, + "num_tokens": 247982276.0, + "step": 9581 + }, + { + "epoch": 1.0522732264441028, + "grad_norm": 2.1467366218566895, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7247474789619446, + "num_tokens": 248010082.0, + "step": 9582 + }, + { + "epoch": 1.0523830441467164, + "grad_norm": 2.2747645378112793, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7351266145706177, + "num_tokens": 248035005.0, + "step": 9583 + }, + { + "epoch": 1.05249286184933, + "grad_norm": 2.183824300765991, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7091786861419678, + "num_tokens": 248061281.0, + "step": 9584 + }, + { + "epoch": 1.0526026795519439, + "grad_norm": 2.250694751739502, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7061895132064819, + "num_tokens": 248089291.0, + "step": 9585 + }, + { + "epoch": 1.0527124972545574, + "grad_norm": 2.0912575721740723, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7179539203643799, + "num_tokens": 248118085.0, + "step": 9586 + }, + { + "epoch": 1.0528223149571712, + "grad_norm": 2.368147611618042, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7146711349487305, + "num_tokens": 248141100.0, + "step": 9587 + }, + { + "epoch": 1.0529321326597847, + "grad_norm": 2.2931149005889893, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7235976457595825, + "num_tokens": 248165329.0, + "step": 9588 + }, + { + "epoch": 1.0530419503623984, + "grad_norm": 2.0296459197998047, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7008914947509766, + "num_tokens": 248196463.0, + "step": 9589 + }, + { + "epoch": 1.053151768065012, + "grad_norm": 2.3874716758728027, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7388497591018677, + "num_tokens": 248219419.0, + "step": 9590 + }, + { + "epoch": 1.0532615857676257, + "grad_norm": 2.557813882827759, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7005935907363892, + "num_tokens": 248242837.0, + "step": 9591 + }, + { + "epoch": 1.0533714034702395, + "grad_norm": 2.3692822456359863, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7148422002792358, + "num_tokens": 248267252.0, + "step": 9592 + }, + { + "epoch": 1.053481221172853, + "grad_norm": 1.8591079711914062, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.6993658542633057, + "num_tokens": 248303356.0, + "step": 9593 + }, + { + "epoch": 1.0535910388754668, + "grad_norm": 2.177053213119507, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7056192755699158, + "num_tokens": 248331023.0, + "step": 9594 + }, + { + "epoch": 1.0537008565780803, + "grad_norm": 1.9562976360321045, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7229084968566895, + "num_tokens": 248360940.0, + "step": 9595 + }, + { + "epoch": 1.053810674280694, + "grad_norm": 2.1322052478790283, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7193204164505005, + "num_tokens": 248390650.0, + "step": 9596 + }, + { + "epoch": 1.0539204919833076, + "grad_norm": 2.1050305366516113, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.719636857509613, + "num_tokens": 248417494.0, + "step": 9597 + }, + { + "epoch": 1.0540303096859214, + "grad_norm": 2.0215628147125244, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7369635105133057, + "num_tokens": 248444723.0, + "step": 9598 + }, + { + "epoch": 1.0541401273885351, + "grad_norm": 2.3926098346710205, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7121171951293945, + "num_tokens": 248466685.0, + "step": 9599 + }, + { + "epoch": 1.0542499450911487, + "grad_norm": 2.5109405517578125, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7279736995697021, + "num_tokens": 248488630.0, + "step": 9600 + }, + { + "epoch": 1.0543597627937624, + "grad_norm": 2.0648109912872314, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6856846809387207, + "num_tokens": 248518374.0, + "step": 9601 + }, + { + "epoch": 1.054469580496376, + "grad_norm": 2.258315086364746, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7145071029663086, + "num_tokens": 248542665.0, + "step": 9602 + }, + { + "epoch": 1.0545793981989897, + "grad_norm": 1.8397773504257202, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7260907888412476, + "num_tokens": 248576368.0, + "step": 9603 + }, + { + "epoch": 1.0546892159016032, + "grad_norm": 1.9565982818603516, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6942380666732788, + "num_tokens": 248606520.0, + "step": 9604 + }, + { + "epoch": 1.054799033604217, + "grad_norm": 2.0643234252929688, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6886059045791626, + "num_tokens": 248634950.0, + "step": 9605 + }, + { + "epoch": 1.0549088513068308, + "grad_norm": 2.0736446380615234, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7195097804069519, + "num_tokens": 248663549.0, + "step": 9606 + }, + { + "epoch": 1.0550186690094443, + "grad_norm": 2.3163647651672363, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7097358703613281, + "num_tokens": 248688326.0, + "step": 9607 + }, + { + "epoch": 1.055128486712058, + "grad_norm": 2.1895499229431152, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7116742134094238, + "num_tokens": 248714511.0, + "step": 9608 + }, + { + "epoch": 1.0552383044146716, + "grad_norm": 2.17254900932312, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7212448120117188, + "num_tokens": 248742304.0, + "step": 9609 + }, + { + "epoch": 1.0553481221172853, + "grad_norm": 2.1188220977783203, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7390178442001343, + "num_tokens": 248768175.0, + "step": 9610 + }, + { + "epoch": 1.0554579398198989, + "grad_norm": 2.087876319885254, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7117090821266174, + "num_tokens": 248798423.0, + "step": 9611 + }, + { + "epoch": 1.0555677575225126, + "grad_norm": 2.2555439472198486, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7137880325317383, + "num_tokens": 248823545.0, + "step": 9612 + }, + { + "epoch": 1.0556775752251264, + "grad_norm": 2.575920820236206, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7381613850593567, + "num_tokens": 248842430.0, + "step": 9613 + }, + { + "epoch": 1.05578739292774, + "grad_norm": 2.105632781982422, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7121748328208923, + "num_tokens": 248870096.0, + "step": 9614 + }, + { + "epoch": 1.0558972106303537, + "grad_norm": 2.346005916595459, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7116400599479675, + "num_tokens": 248893581.0, + "step": 9615 + }, + { + "epoch": 1.0560070283329672, + "grad_norm": 2.186910629272461, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7101800441741943, + "num_tokens": 248919001.0, + "step": 9616 + }, + { + "epoch": 1.056116846035581, + "grad_norm": 2.1280088424682617, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7094061374664307, + "num_tokens": 248946510.0, + "step": 9617 + }, + { + "epoch": 1.0562266637381945, + "grad_norm": 2.0028843879699707, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7113164663314819, + "num_tokens": 248978612.0, + "step": 9618 + }, + { + "epoch": 1.0563364814408083, + "grad_norm": 2.0845065116882324, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6900066137313843, + "num_tokens": 249008011.0, + "step": 9619 + }, + { + "epoch": 1.0564462991434218, + "grad_norm": 2.248380184173584, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.731050431728363, + "num_tokens": 249031806.0, + "step": 9620 + }, + { + "epoch": 1.0565561168460356, + "grad_norm": 2.529676675796509, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.719550371170044, + "num_tokens": 249052702.0, + "step": 9621 + }, + { + "epoch": 1.0566659345486493, + "grad_norm": 1.9063535928726196, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.687928318977356, + "num_tokens": 249087582.0, + "step": 9622 + }, + { + "epoch": 1.0567757522512629, + "grad_norm": 2.3813631534576416, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7108055353164673, + "num_tokens": 249111026.0, + "step": 9623 + }, + { + "epoch": 1.0568855699538766, + "grad_norm": 2.3449513912200928, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7142761945724487, + "num_tokens": 249133904.0, + "step": 9624 + }, + { + "epoch": 1.0569953876564901, + "grad_norm": 2.708991050720215, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.722597599029541, + "num_tokens": 249153840.0, + "step": 9625 + }, + { + "epoch": 1.057105205359104, + "grad_norm": 2.3730955123901367, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7119425535202026, + "num_tokens": 249177898.0, + "step": 9626 + }, + { + "epoch": 1.0572150230617177, + "grad_norm": 2.4829261302948, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7233523726463318, + "num_tokens": 249199392.0, + "step": 9627 + }, + { + "epoch": 1.0573248407643312, + "grad_norm": 2.1313669681549072, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7153352499008179, + "num_tokens": 249227251.0, + "step": 9628 + }, + { + "epoch": 1.057434658466945, + "grad_norm": 2.475271463394165, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.741242527961731, + "num_tokens": 249248156.0, + "step": 9629 + }, + { + "epoch": 1.0575444761695585, + "grad_norm": 2.506127119064331, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6976068019866943, + "num_tokens": 249271245.0, + "step": 9630 + }, + { + "epoch": 1.0576542938721722, + "grad_norm": 2.380977153778076, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6963350176811218, + "num_tokens": 249297210.0, + "step": 9631 + }, + { + "epoch": 1.0577641115747858, + "grad_norm": 2.5044877529144287, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7305580377578735, + "num_tokens": 249317807.0, + "step": 9632 + }, + { + "epoch": 1.0578739292773995, + "grad_norm": 2.1446051597595215, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.719323992729187, + "num_tokens": 249347204.0, + "step": 9633 + }, + { + "epoch": 1.057983746980013, + "grad_norm": 2.0379867553710938, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7154185771942139, + "num_tokens": 249377018.0, + "step": 9634 + }, + { + "epoch": 1.0580935646826268, + "grad_norm": 2.148038148880005, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.727348268032074, + "num_tokens": 249404706.0, + "step": 9635 + }, + { + "epoch": 1.0582033823852406, + "grad_norm": 2.405194044113159, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7243237495422363, + "num_tokens": 249428576.0, + "step": 9636 + }, + { + "epoch": 1.0583132000878541, + "grad_norm": 2.2991578578948975, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7338145971298218, + "num_tokens": 249455495.0, + "step": 9637 + }, + { + "epoch": 1.0584230177904679, + "grad_norm": 2.6041088104248047, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7303965091705322, + "num_tokens": 249474722.0, + "step": 9638 + }, + { + "epoch": 1.0585328354930814, + "grad_norm": 2.2673532962799072, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7198100090026855, + "num_tokens": 249501296.0, + "step": 9639 + }, + { + "epoch": 1.0586426531956952, + "grad_norm": 2.5589921474456787, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7344483137130737, + "num_tokens": 249521551.0, + "step": 9640 + }, + { + "epoch": 1.0587524708983087, + "grad_norm": 2.303590774536133, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.725935697555542, + "num_tokens": 249546353.0, + "step": 9641 + }, + { + "epoch": 1.0588622886009225, + "grad_norm": 2.3563525676727295, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7088268995285034, + "num_tokens": 249572895.0, + "step": 9642 + }, + { + "epoch": 1.0589721063035362, + "grad_norm": 2.513172149658203, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7401149868965149, + "num_tokens": 249596783.0, + "step": 9643 + }, + { + "epoch": 1.0590819240061498, + "grad_norm": 2.230841636657715, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7165254354476929, + "num_tokens": 249623279.0, + "step": 9644 + }, + { + "epoch": 1.0591917417087635, + "grad_norm": 2.2814793586730957, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7150742411613464, + "num_tokens": 249651574.0, + "step": 9645 + }, + { + "epoch": 1.059301559411377, + "grad_norm": 2.260258197784424, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7091858386993408, + "num_tokens": 249677631.0, + "step": 9646 + }, + { + "epoch": 1.0594113771139908, + "grad_norm": 2.3703105449676514, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7068729400634766, + "num_tokens": 249703995.0, + "step": 9647 + }, + { + "epoch": 1.0595211948166043, + "grad_norm": 2.2707364559173584, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7226895093917847, + "num_tokens": 249728629.0, + "step": 9648 + }, + { + "epoch": 1.059631012519218, + "grad_norm": 2.1388602256774902, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.7001293897628784, + "num_tokens": 249757537.0, + "step": 9649 + }, + { + "epoch": 1.0597408302218319, + "grad_norm": 2.4733479022979736, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.716562032699585, + "num_tokens": 249779905.0, + "step": 9650 + }, + { + "epoch": 1.0598506479244454, + "grad_norm": 2.283409595489502, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7256098985671997, + "num_tokens": 249806155.0, + "step": 9651 + }, + { + "epoch": 1.0599604656270591, + "grad_norm": 2.1258270740509033, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.706992506980896, + "num_tokens": 249836082.0, + "step": 9652 + }, + { + "epoch": 1.0600702833296727, + "grad_norm": 2.356839418411255, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7203545570373535, + "num_tokens": 249860640.0, + "step": 9653 + }, + { + "epoch": 1.0601801010322864, + "grad_norm": 2.3296022415161133, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7122195959091187, + "num_tokens": 249884934.0, + "step": 9654 + }, + { + "epoch": 1.0602899187349, + "grad_norm": 2.2439815998077393, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7142921686172485, + "num_tokens": 249910064.0, + "step": 9655 + }, + { + "epoch": 1.0603997364375137, + "grad_norm": 2.1837189197540283, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7040926218032837, + "num_tokens": 249937239.0, + "step": 9656 + }, + { + "epoch": 1.0605095541401275, + "grad_norm": 2.7785449028015137, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7418348789215088, + "num_tokens": 249955880.0, + "step": 9657 + }, + { + "epoch": 1.060619371842741, + "grad_norm": 1.9459562301635742, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7020037174224854, + "num_tokens": 249988295.0, + "step": 9658 + }, + { + "epoch": 1.0607291895453548, + "grad_norm": 2.3934144973754883, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7267142534255981, + "num_tokens": 250014371.0, + "step": 9659 + }, + { + "epoch": 1.0608390072479683, + "grad_norm": 2.4582080841064453, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7321320176124573, + "num_tokens": 250035108.0, + "step": 9660 + }, + { + "epoch": 1.060948824950582, + "grad_norm": 2.386603593826294, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7248353362083435, + "num_tokens": 250057924.0, + "step": 9661 + }, + { + "epoch": 1.0610586426531956, + "grad_norm": 2.0417590141296387, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7121615409851074, + "num_tokens": 250088872.0, + "step": 9662 + }, + { + "epoch": 1.0611684603558094, + "grad_norm": 2.5937135219573975, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6974944472312927, + "num_tokens": 250110129.0, + "step": 9663 + }, + { + "epoch": 1.0612782780584231, + "grad_norm": 2.4284868240356445, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7180049419403076, + "num_tokens": 250132915.0, + "step": 9664 + }, + { + "epoch": 1.0613880957610367, + "grad_norm": 2.2272181510925293, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7113835215568542, + "num_tokens": 250158548.0, + "step": 9665 + }, + { + "epoch": 1.0614979134636504, + "grad_norm": 2.2934458255767822, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7095609903335571, + "num_tokens": 250183701.0, + "step": 9666 + }, + { + "epoch": 1.061607731166264, + "grad_norm": 2.1162595748901367, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7133468985557556, + "num_tokens": 250212131.0, + "step": 9667 + }, + { + "epoch": 1.0617175488688777, + "grad_norm": 2.423173427581787, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.736240029335022, + "num_tokens": 250235131.0, + "step": 9668 + }, + { + "epoch": 1.0618273665714912, + "grad_norm": 2.39827299118042, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7178318500518799, + "num_tokens": 250258718.0, + "step": 9669 + }, + { + "epoch": 1.061937184274105, + "grad_norm": 2.178731918334961, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7424681186676025, + "num_tokens": 250283341.0, + "step": 9670 + }, + { + "epoch": 1.0620470019767188, + "grad_norm": 2.2811050415039062, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7400686740875244, + "num_tokens": 250304962.0, + "step": 9671 + }, + { + "epoch": 1.0621568196793323, + "grad_norm": 2.2953758239746094, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7014039754867554, + "num_tokens": 250332787.0, + "step": 9672 + }, + { + "epoch": 1.062266637381946, + "grad_norm": 1.9039020538330078, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6930782198905945, + "num_tokens": 250368623.0, + "step": 9673 + }, + { + "epoch": 1.0623764550845596, + "grad_norm": 2.3607680797576904, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7238284349441528, + "num_tokens": 250392124.0, + "step": 9674 + }, + { + "epoch": 1.0624862727871733, + "grad_norm": 2.181185483932495, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7194716930389404, + "num_tokens": 250417204.0, + "step": 9675 + }, + { + "epoch": 1.0625960904897869, + "grad_norm": 2.212623119354248, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7056695818901062, + "num_tokens": 250444684.0, + "step": 9676 + }, + { + "epoch": 1.0627059081924006, + "grad_norm": 2.319669723510742, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7039515972137451, + "num_tokens": 250469887.0, + "step": 9677 + }, + { + "epoch": 1.0628157258950144, + "grad_norm": 2.1247191429138184, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7241151332855225, + "num_tokens": 250498534.0, + "step": 9678 + }, + { + "epoch": 1.062925543597628, + "grad_norm": 2.1706085205078125, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7418889999389648, + "num_tokens": 250525802.0, + "step": 9679 + }, + { + "epoch": 1.0630353613002417, + "grad_norm": 2.1264989376068115, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7260890007019043, + "num_tokens": 250553585.0, + "step": 9680 + }, + { + "epoch": 1.0631451790028552, + "grad_norm": 2.4068825244903564, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7332367897033691, + "num_tokens": 250576137.0, + "step": 9681 + }, + { + "epoch": 1.063254996705469, + "grad_norm": 2.008195400238037, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6937743425369263, + "num_tokens": 250606474.0, + "step": 9682 + }, + { + "epoch": 1.0633648144080825, + "grad_norm": 1.9983488321304321, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7214163541793823, + "num_tokens": 250636724.0, + "step": 9683 + }, + { + "epoch": 1.0634746321106963, + "grad_norm": 2.0965094566345215, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7118950486183167, + "num_tokens": 250665159.0, + "step": 9684 + }, + { + "epoch": 1.0635844498133098, + "grad_norm": 2.321898937225342, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7369388341903687, + "num_tokens": 250687852.0, + "step": 9685 + }, + { + "epoch": 1.0636942675159236, + "grad_norm": 2.0960142612457275, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7327055335044861, + "num_tokens": 250717529.0, + "step": 9686 + }, + { + "epoch": 1.0638040852185373, + "grad_norm": 2.4320781230926514, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7161920070648193, + "num_tokens": 250738915.0, + "step": 9687 + }, + { + "epoch": 1.0639139029211508, + "grad_norm": 2.415848731994629, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.730125904083252, + "num_tokens": 250761103.0, + "step": 9688 + }, + { + "epoch": 1.0640237206237646, + "grad_norm": 2.4188759326934814, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7347104549407959, + "num_tokens": 250784246.0, + "step": 9689 + }, + { + "epoch": 1.0641335383263781, + "grad_norm": 2.2422053813934326, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7251613736152649, + "num_tokens": 250808922.0, + "step": 9690 + }, + { + "epoch": 1.064243356028992, + "grad_norm": 2.0555126667022705, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.701326847076416, + "num_tokens": 250837921.0, + "step": 9691 + }, + { + "epoch": 1.0643531737316057, + "grad_norm": 2.223388195037842, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7225147485733032, + "num_tokens": 250864218.0, + "step": 9692 + }, + { + "epoch": 1.0644629914342192, + "grad_norm": 2.210200309753418, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7334221601486206, + "num_tokens": 250889667.0, + "step": 9693 + }, + { + "epoch": 1.064572809136833, + "grad_norm": 2.7400805950164795, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.750619649887085, + "num_tokens": 250906367.0, + "step": 9694 + }, + { + "epoch": 1.0646826268394465, + "grad_norm": 2.0988268852233887, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7187790274620056, + "num_tokens": 250935459.0, + "step": 9695 + }, + { + "epoch": 1.0647924445420602, + "grad_norm": 2.331737518310547, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7243958115577698, + "num_tokens": 250960205.0, + "step": 9696 + }, + { + "epoch": 1.0649022622446738, + "grad_norm": 2.156170129776001, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7164461612701416, + "num_tokens": 250986395.0, + "step": 9697 + }, + { + "epoch": 1.0650120799472875, + "grad_norm": 2.439885377883911, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7112951278686523, + "num_tokens": 251010206.0, + "step": 9698 + }, + { + "epoch": 1.065121897649901, + "grad_norm": 2.137026071548462, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.710624098777771, + "num_tokens": 251037764.0, + "step": 9699 + }, + { + "epoch": 1.0652317153525148, + "grad_norm": 2.16793155670166, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7204651832580566, + "num_tokens": 251065632.0, + "step": 9700 + }, + { + "epoch": 1.0653415330551286, + "grad_norm": 2.5335748195648193, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7077692747116089, + "num_tokens": 251087567.0, + "step": 9701 + }, + { + "epoch": 1.065451350757742, + "grad_norm": 2.736405372619629, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7395469546318054, + "num_tokens": 251105169.0, + "step": 9702 + }, + { + "epoch": 1.0655611684603559, + "grad_norm": 2.1837832927703857, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7600914239883423, + "num_tokens": 251129897.0, + "step": 9703 + }, + { + "epoch": 1.0656709861629694, + "grad_norm": 2.4754974842071533, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7284583449363708, + "num_tokens": 251152836.0, + "step": 9704 + }, + { + "epoch": 1.0657808038655832, + "grad_norm": 2.3687567710876465, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7175314426422119, + "num_tokens": 251178610.0, + "step": 9705 + }, + { + "epoch": 1.0658906215681967, + "grad_norm": 2.1864430904388428, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7152851819992065, + "num_tokens": 251203642.0, + "step": 9706 + }, + { + "epoch": 1.0660004392708105, + "grad_norm": 2.248349666595459, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7195072174072266, + "num_tokens": 251228970.0, + "step": 9707 + }, + { + "epoch": 1.0661102569734242, + "grad_norm": 2.1083602905273438, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.721000075340271, + "num_tokens": 251256481.0, + "step": 9708 + }, + { + "epoch": 1.0662200746760377, + "grad_norm": 2.225212574005127, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6988308429718018, + "num_tokens": 251284419.0, + "step": 9709 + }, + { + "epoch": 1.0663298923786515, + "grad_norm": 2.225555896759033, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7194821834564209, + "num_tokens": 251308932.0, + "step": 9710 + }, + { + "epoch": 1.066439710081265, + "grad_norm": 2.061553955078125, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7046226859092712, + "num_tokens": 251339253.0, + "step": 9711 + }, + { + "epoch": 1.0665495277838788, + "grad_norm": 2.614657163619995, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7261419296264648, + "num_tokens": 251360860.0, + "step": 9712 + }, + { + "epoch": 1.0666593454864923, + "grad_norm": 2.2402472496032715, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7277998328208923, + "num_tokens": 251385661.0, + "step": 9713 + }, + { + "epoch": 1.066769163189106, + "grad_norm": 2.2525970935821533, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7179053425788879, + "num_tokens": 251411174.0, + "step": 9714 + }, + { + "epoch": 1.0668789808917198, + "grad_norm": 2.5057408809661865, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7221900224685669, + "num_tokens": 251433780.0, + "step": 9715 + }, + { + "epoch": 1.0669887985943334, + "grad_norm": 2.468837261199951, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7162420749664307, + "num_tokens": 251457958.0, + "step": 9716 + }, + { + "epoch": 1.0670986162969471, + "grad_norm": 2.219099283218384, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7200121283531189, + "num_tokens": 251483937.0, + "step": 9717 + }, + { + "epoch": 1.0672084339995607, + "grad_norm": 2.363952159881592, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7248139977455139, + "num_tokens": 251506303.0, + "step": 9718 + }, + { + "epoch": 1.0673182517021744, + "grad_norm": 2.3392300605773926, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7084265351295471, + "num_tokens": 251531605.0, + "step": 9719 + }, + { + "epoch": 1.067428069404788, + "grad_norm": 2.199967622756958, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6920654773712158, + "num_tokens": 251559470.0, + "step": 9720 + }, + { + "epoch": 1.0675378871074017, + "grad_norm": 2.467639207839966, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7437242269515991, + "num_tokens": 251580254.0, + "step": 9721 + }, + { + "epoch": 1.0676477048100155, + "grad_norm": 2.172194004058838, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7161966562271118, + "num_tokens": 251606712.0, + "step": 9722 + }, + { + "epoch": 1.067757522512629, + "grad_norm": 2.20603609085083, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7079316973686218, + "num_tokens": 251632873.0, + "step": 9723 + }, + { + "epoch": 1.0678673402152428, + "grad_norm": 2.142564058303833, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.705303430557251, + "num_tokens": 251662345.0, + "step": 9724 + }, + { + "epoch": 1.0679771579178563, + "grad_norm": 1.9699044227600098, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7220058441162109, + "num_tokens": 251694926.0, + "step": 9725 + }, + { + "epoch": 1.06808697562047, + "grad_norm": 2.4032294750213623, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.721481442451477, + "num_tokens": 251717795.0, + "step": 9726 + }, + { + "epoch": 1.0681967933230836, + "grad_norm": 2.123626708984375, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7249813675880432, + "num_tokens": 251744985.0, + "step": 9727 + }, + { + "epoch": 1.0683066110256974, + "grad_norm": 2.2733941078186035, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7118072509765625, + "num_tokens": 251769990.0, + "step": 9728 + }, + { + "epoch": 1.068416428728311, + "grad_norm": 2.524014472961426, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7061725854873657, + "num_tokens": 251793256.0, + "step": 9729 + }, + { + "epoch": 1.0685262464309246, + "grad_norm": 2.040163993835449, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6915267705917358, + "num_tokens": 251824616.0, + "step": 9730 + }, + { + "epoch": 1.0686360641335384, + "grad_norm": 2.359032392501831, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7042489051818848, + "num_tokens": 251849164.0, + "step": 9731 + }, + { + "epoch": 1.068745881836152, + "grad_norm": 2.0492429733276367, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7327576875686646, + "num_tokens": 251879989.0, + "step": 9732 + }, + { + "epoch": 1.0688556995387657, + "grad_norm": 2.1521878242492676, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7153365612030029, + "num_tokens": 251906896.0, + "step": 9733 + }, + { + "epoch": 1.0689655172413792, + "grad_norm": 2.113344430923462, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7067140936851501, + "num_tokens": 251935270.0, + "step": 9734 + }, + { + "epoch": 1.069075334943993, + "grad_norm": 2.459084987640381, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7261662483215332, + "num_tokens": 251958585.0, + "step": 9735 + }, + { + "epoch": 1.0691851526466065, + "grad_norm": 2.044502019882202, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.684434175491333, + "num_tokens": 251989265.0, + "step": 9736 + }, + { + "epoch": 1.0692949703492203, + "grad_norm": 2.126755714416504, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7379149198532104, + "num_tokens": 252015010.0, + "step": 9737 + }, + { + "epoch": 1.069404788051834, + "grad_norm": 2.3651113510131836, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.727996289730072, + "num_tokens": 252038183.0, + "step": 9738 + }, + { + "epoch": 1.0695146057544476, + "grad_norm": 2.489196300506592, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7064069509506226, + "num_tokens": 252061193.0, + "step": 9739 + }, + { + "epoch": 1.0696244234570613, + "grad_norm": 2.072690725326538, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7164819836616516, + "num_tokens": 252091708.0, + "step": 9740 + }, + { + "epoch": 1.0697342411596749, + "grad_norm": 2.5762739181518555, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.737769603729248, + "num_tokens": 252110662.0, + "step": 9741 + }, + { + "epoch": 1.0698440588622886, + "grad_norm": 2.09450364112854, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7212095260620117, + "num_tokens": 252138537.0, + "step": 9742 + }, + { + "epoch": 1.0699538765649024, + "grad_norm": 2.217461109161377, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7170148491859436, + "num_tokens": 252164974.0, + "step": 9743 + }, + { + "epoch": 1.070063694267516, + "grad_norm": 2.177253246307373, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7216030955314636, + "num_tokens": 252189644.0, + "step": 9744 + }, + { + "epoch": 1.0701735119701297, + "grad_norm": 2.287612199783325, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6969804167747498, + "num_tokens": 252215816.0, + "step": 9745 + }, + { + "epoch": 1.0702833296727432, + "grad_norm": 2.1474015712738037, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7172764539718628, + "num_tokens": 252242546.0, + "step": 9746 + }, + { + "epoch": 1.070393147375357, + "grad_norm": 2.6094624996185303, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7207549214363098, + "num_tokens": 252261287.0, + "step": 9747 + }, + { + "epoch": 1.0705029650779705, + "grad_norm": 2.205510139465332, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7254545092582703, + "num_tokens": 252288336.0, + "step": 9748 + }, + { + "epoch": 1.0706127827805842, + "grad_norm": 2.1173431873321533, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7290682196617126, + "num_tokens": 252315179.0, + "step": 9749 + }, + { + "epoch": 1.0707226004831978, + "grad_norm": 2.2192864418029785, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7082829475402832, + "num_tokens": 252341554.0, + "step": 9750 + }, + { + "epoch": 1.0708324181858115, + "grad_norm": 2.17447566986084, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.707802951335907, + "num_tokens": 252369884.0, + "step": 9751 + }, + { + "epoch": 1.0709422358884253, + "grad_norm": 2.221956968307495, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7237985134124756, + "num_tokens": 252395945.0, + "step": 9752 + }, + { + "epoch": 1.0710520535910388, + "grad_norm": 2.2468152046203613, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7210301160812378, + "num_tokens": 252421213.0, + "step": 9753 + }, + { + "epoch": 1.0711618712936526, + "grad_norm": 2.3752174377441406, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7181852459907532, + "num_tokens": 252444980.0, + "step": 9754 + }, + { + "epoch": 1.0712716889962661, + "grad_norm": 2.285613536834717, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7114431858062744, + "num_tokens": 252472709.0, + "step": 9755 + }, + { + "epoch": 1.0713815066988799, + "grad_norm": 1.9740935564041138, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7171251177787781, + "num_tokens": 252502923.0, + "step": 9756 + }, + { + "epoch": 1.0714913244014934, + "grad_norm": 2.122797966003418, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7113094329833984, + "num_tokens": 252530288.0, + "step": 9757 + }, + { + "epoch": 1.0716011421041072, + "grad_norm": 2.0623879432678223, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6951739192008972, + "num_tokens": 252561438.0, + "step": 9758 + }, + { + "epoch": 1.071710959806721, + "grad_norm": 2.0377299785614014, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6942497491836548, + "num_tokens": 252591279.0, + "step": 9759 + }, + { + "epoch": 1.0718207775093345, + "grad_norm": 2.126847743988037, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.722077488899231, + "num_tokens": 252621008.0, + "step": 9760 + }, + { + "epoch": 1.0719305952119482, + "grad_norm": 2.2321300506591797, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7127336859703064, + "num_tokens": 252647364.0, + "step": 9761 + }, + { + "epoch": 1.0720404129145618, + "grad_norm": 2.3630855083465576, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7245326042175293, + "num_tokens": 252670834.0, + "step": 9762 + }, + { + "epoch": 1.0721502306171755, + "grad_norm": 2.293231248855591, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7068241834640503, + "num_tokens": 252695647.0, + "step": 9763 + }, + { + "epoch": 1.072260048319789, + "grad_norm": 2.0493454933166504, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7206020355224609, + "num_tokens": 252725425.0, + "step": 9764 + }, + { + "epoch": 1.0723698660224028, + "grad_norm": 2.0224695205688477, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7028914093971252, + "num_tokens": 252755765.0, + "step": 9765 + }, + { + "epoch": 1.0724796837250166, + "grad_norm": 2.0562021732330322, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7199084758758545, + "num_tokens": 252784490.0, + "step": 9766 + }, + { + "epoch": 1.07258950142763, + "grad_norm": 2.253753900527954, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7232012152671814, + "num_tokens": 252812843.0, + "step": 9767 + }, + { + "epoch": 1.0726993191302439, + "grad_norm": 2.540407180786133, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7196657061576843, + "num_tokens": 252834022.0, + "step": 9768 + }, + { + "epoch": 1.0728091368328574, + "grad_norm": 2.285550594329834, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7204516530036926, + "num_tokens": 252860473.0, + "step": 9769 + }, + { + "epoch": 1.0729189545354711, + "grad_norm": 2.0495309829711914, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6906739473342896, + "num_tokens": 252892540.0, + "step": 9770 + }, + { + "epoch": 1.0730287722380847, + "grad_norm": 1.9496314525604248, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7207595109939575, + "num_tokens": 252925170.0, + "step": 9771 + }, + { + "epoch": 1.0731385899406984, + "grad_norm": 2.492532253265381, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7250767946243286, + "num_tokens": 252946526.0, + "step": 9772 + }, + { + "epoch": 1.0732484076433122, + "grad_norm": 2.4285333156585693, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7007405757904053, + "num_tokens": 252974024.0, + "step": 9773 + }, + { + "epoch": 1.0733582253459257, + "grad_norm": 2.4473540782928467, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7209939360618591, + "num_tokens": 252997552.0, + "step": 9774 + }, + { + "epoch": 1.0734680430485395, + "grad_norm": 2.2292754650115967, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7179397940635681, + "num_tokens": 253024018.0, + "step": 9775 + }, + { + "epoch": 1.073577860751153, + "grad_norm": 1.8046153783798218, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7455651760101318, + "num_tokens": 253059439.0, + "step": 9776 + }, + { + "epoch": 1.0736876784537668, + "grad_norm": 1.8845481872558594, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7087377905845642, + "num_tokens": 253092060.0, + "step": 9777 + }, + { + "epoch": 1.0737974961563803, + "grad_norm": 2.241624593734741, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7218294739723206, + "num_tokens": 253116732.0, + "step": 9778 + }, + { + "epoch": 1.073907313858994, + "grad_norm": 2.4823994636535645, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7121235728263855, + "num_tokens": 253137444.0, + "step": 9779 + }, + { + "epoch": 1.0740171315616078, + "grad_norm": 2.3319950103759766, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7087134122848511, + "num_tokens": 253163728.0, + "step": 9780 + }, + { + "epoch": 1.0741269492642214, + "grad_norm": 2.5292067527770996, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7154013514518738, + "num_tokens": 253185007.0, + "step": 9781 + }, + { + "epoch": 1.0742367669668351, + "grad_norm": 2.2288479804992676, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7173612117767334, + "num_tokens": 253210039.0, + "step": 9782 + }, + { + "epoch": 1.0743465846694487, + "grad_norm": 2.3043577671051025, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7125217914581299, + "num_tokens": 253233730.0, + "step": 9783 + }, + { + "epoch": 1.0744564023720624, + "grad_norm": 2.5120394229888916, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7456890344619751, + "num_tokens": 253253245.0, + "step": 9784 + }, + { + "epoch": 1.074566220074676, + "grad_norm": 2.31257700920105, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7396291494369507, + "num_tokens": 253276720.0, + "step": 9785 + }, + { + "epoch": 1.0746760377772897, + "grad_norm": 2.065673828125, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7122551202774048, + "num_tokens": 253304939.0, + "step": 9786 + }, + { + "epoch": 1.0747858554799035, + "grad_norm": 2.035611391067505, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7018615007400513, + "num_tokens": 253337224.0, + "step": 9787 + }, + { + "epoch": 1.074895673182517, + "grad_norm": 2.4017651081085205, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7175663113594055, + "num_tokens": 253361449.0, + "step": 9788 + }, + { + "epoch": 1.0750054908851308, + "grad_norm": 2.5286600589752197, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7330918908119202, + "num_tokens": 253382157.0, + "step": 9789 + }, + { + "epoch": 1.0751153085877443, + "grad_norm": 1.9536266326904297, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.707423746585846, + "num_tokens": 253416618.0, + "step": 9790 + }, + { + "epoch": 1.075225126290358, + "grad_norm": 2.432968854904175, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7308349609375, + "num_tokens": 253437656.0, + "step": 9791 + }, + { + "epoch": 1.0753349439929716, + "grad_norm": 2.0752737522125244, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7087646126747131, + "num_tokens": 253465676.0, + "step": 9792 + }, + { + "epoch": 1.0754447616955853, + "grad_norm": 2.373460531234741, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7187269926071167, + "num_tokens": 253488247.0, + "step": 9793 + }, + { + "epoch": 1.075554579398199, + "grad_norm": 2.5218632221221924, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7158762812614441, + "num_tokens": 253509582.0, + "step": 9794 + }, + { + "epoch": 1.0756643971008126, + "grad_norm": 2.3198678493499756, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7113161087036133, + "num_tokens": 253533499.0, + "step": 9795 + }, + { + "epoch": 1.0757742148034264, + "grad_norm": 2.153822183609009, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7202103734016418, + "num_tokens": 253559942.0, + "step": 9796 + }, + { + "epoch": 1.07588403250604, + "grad_norm": 2.216478109359741, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7244507074356079, + "num_tokens": 253587048.0, + "step": 9797 + }, + { + "epoch": 1.0759938502086537, + "grad_norm": 2.3265504837036133, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7130458354949951, + "num_tokens": 253614210.0, + "step": 9798 + }, + { + "epoch": 1.0761036679112672, + "grad_norm": 1.9824200868606567, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7166416645050049, + "num_tokens": 253645388.0, + "step": 9799 + }, + { + "epoch": 1.076213485613881, + "grad_norm": 2.499002695083618, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7293386459350586, + "num_tokens": 253666976.0, + "step": 9800 + }, + { + "epoch": 1.0763233033164945, + "grad_norm": 2.095102310180664, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6957114934921265, + "num_tokens": 253695884.0, + "step": 9801 + }, + { + "epoch": 1.0764331210191083, + "grad_norm": 1.8935532569885254, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.6977428793907166, + "num_tokens": 253730963.0, + "step": 9802 + }, + { + "epoch": 1.076542938721722, + "grad_norm": 2.3824758529663086, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7252044081687927, + "num_tokens": 253754933.0, + "step": 9803 + }, + { + "epoch": 1.0766527564243356, + "grad_norm": 2.265901803970337, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7068610191345215, + "num_tokens": 253782586.0, + "step": 9804 + }, + { + "epoch": 1.0767625741269493, + "grad_norm": 1.9990854263305664, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7241669297218323, + "num_tokens": 253816135.0, + "step": 9805 + }, + { + "epoch": 1.0768723918295628, + "grad_norm": 2.279036045074463, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7365232706069946, + "num_tokens": 253840466.0, + "step": 9806 + }, + { + "epoch": 1.0769822095321766, + "grad_norm": 2.1472721099853516, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7232218980789185, + "num_tokens": 253866512.0, + "step": 9807 + }, + { + "epoch": 1.0770920272347904, + "grad_norm": 2.279595375061035, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7399766445159912, + "num_tokens": 253890883.0, + "step": 9808 + }, + { + "epoch": 1.077201844937404, + "grad_norm": 2.4731180667877197, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7245358228683472, + "num_tokens": 253913297.0, + "step": 9809 + }, + { + "epoch": 1.0773116626400177, + "grad_norm": 2.138097047805786, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7144104242324829, + "num_tokens": 253941360.0, + "step": 9810 + }, + { + "epoch": 1.0774214803426312, + "grad_norm": 2.0174379348754883, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6931254863739014, + "num_tokens": 253971636.0, + "step": 9811 + }, + { + "epoch": 1.077531298045245, + "grad_norm": 2.321749448776245, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7200354337692261, + "num_tokens": 253996545.0, + "step": 9812 + }, + { + "epoch": 1.0776411157478585, + "grad_norm": 1.9379076957702637, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7004662752151489, + "num_tokens": 254030517.0, + "step": 9813 + }, + { + "epoch": 1.0777509334504722, + "grad_norm": 2.5959556102752686, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7137413024902344, + "num_tokens": 254052332.0, + "step": 9814 + }, + { + "epoch": 1.0778607511530858, + "grad_norm": 2.395397186279297, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7195224761962891, + "num_tokens": 254075027.0, + "step": 9815 + }, + { + "epoch": 1.0779705688556995, + "grad_norm": 2.157426118850708, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7062486410140991, + "num_tokens": 254105596.0, + "step": 9816 + }, + { + "epoch": 1.0780803865583133, + "grad_norm": 2.3224575519561768, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7167961597442627, + "num_tokens": 254129912.0, + "step": 9817 + }, + { + "epoch": 1.0781902042609268, + "grad_norm": 2.2990458011627197, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7391011118888855, + "num_tokens": 254152853.0, + "step": 9818 + }, + { + "epoch": 1.0783000219635406, + "grad_norm": 2.234931707382202, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7163200378417969, + "num_tokens": 254179367.0, + "step": 9819 + }, + { + "epoch": 1.0784098396661541, + "grad_norm": 2.196110725402832, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7024433612823486, + "num_tokens": 254205615.0, + "step": 9820 + }, + { + "epoch": 1.0785196573687679, + "grad_norm": 2.165038824081421, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7301221489906311, + "num_tokens": 254232185.0, + "step": 9821 + }, + { + "epoch": 1.0786294750713814, + "grad_norm": 2.198162078857422, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7070795893669128, + "num_tokens": 254258918.0, + "step": 9822 + }, + { + "epoch": 1.0787392927739952, + "grad_norm": 2.521728992462158, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7313891649246216, + "num_tokens": 254281886.0, + "step": 9823 + }, + { + "epoch": 1.078849110476609, + "grad_norm": 2.0671374797821045, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7066450119018555, + "num_tokens": 254312053.0, + "step": 9824 + }, + { + "epoch": 1.0789589281792225, + "grad_norm": 2.3463714122772217, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7296772003173828, + "num_tokens": 254335024.0, + "step": 9825 + }, + { + "epoch": 1.0790687458818362, + "grad_norm": 2.152606248855591, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7128036618232727, + "num_tokens": 254361962.0, + "step": 9826 + }, + { + "epoch": 1.0791785635844497, + "grad_norm": 1.8505096435546875, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7148518562316895, + "num_tokens": 254396842.0, + "step": 9827 + }, + { + "epoch": 1.0792883812870635, + "grad_norm": 2.2619361877441406, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7143895626068115, + "num_tokens": 254422731.0, + "step": 9828 + }, + { + "epoch": 1.079398198989677, + "grad_norm": 2.1660423278808594, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7335392832756042, + "num_tokens": 254448617.0, + "step": 9829 + }, + { + "epoch": 1.0795080166922908, + "grad_norm": 2.5804355144500732, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7063708305358887, + "num_tokens": 254468811.0, + "step": 9830 + }, + { + "epoch": 1.0796178343949046, + "grad_norm": 2.3553855419158936, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7136908769607544, + "num_tokens": 254492675.0, + "step": 9831 + }, + { + "epoch": 1.079727652097518, + "grad_norm": 2.6410470008850098, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.72182297706604, + "num_tokens": 254512410.0, + "step": 9832 + }, + { + "epoch": 1.0798374698001318, + "grad_norm": 2.4231200218200684, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7174064517021179, + "num_tokens": 254535716.0, + "step": 9833 + }, + { + "epoch": 1.0799472875027454, + "grad_norm": 2.509948492050171, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.728224515914917, + "num_tokens": 254557037.0, + "step": 9834 + }, + { + "epoch": 1.0800571052053591, + "grad_norm": 2.253422260284424, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6937733888626099, + "num_tokens": 254584259.0, + "step": 9835 + }, + { + "epoch": 1.0801669229079727, + "grad_norm": 2.2413454055786133, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6814707517623901, + "num_tokens": 254614845.0, + "step": 9836 + }, + { + "epoch": 1.0802767406105864, + "grad_norm": 2.1817657947540283, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7272409200668335, + "num_tokens": 254640490.0, + "step": 9837 + }, + { + "epoch": 1.0803865583132002, + "grad_norm": 2.1581664085388184, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7205033302307129, + "num_tokens": 254667390.0, + "step": 9838 + }, + { + "epoch": 1.0804963760158137, + "grad_norm": 2.4132134914398193, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7037444114685059, + "num_tokens": 254690296.0, + "step": 9839 + }, + { + "epoch": 1.0806061937184275, + "grad_norm": 2.381432056427002, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7047991752624512, + "num_tokens": 254715809.0, + "step": 9840 + }, + { + "epoch": 1.080716011421041, + "grad_norm": 2.254579782485962, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7500613331794739, + "num_tokens": 254741448.0, + "step": 9841 + }, + { + "epoch": 1.0808258291236548, + "grad_norm": 2.1877756118774414, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7070251703262329, + "num_tokens": 254767857.0, + "step": 9842 + }, + { + "epoch": 1.0809356468262683, + "grad_norm": 2.352543592453003, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7229490280151367, + "num_tokens": 254792749.0, + "step": 9843 + }, + { + "epoch": 1.081045464528882, + "grad_norm": 2.0025901794433594, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6954479217529297, + "num_tokens": 254827509.0, + "step": 9844 + }, + { + "epoch": 1.0811552822314958, + "grad_norm": 2.2624030113220215, + "learning_rate": 1e-06, + "loss": 0.7954, + "mean_token_accuracy": 0.7456446290016174, + "num_tokens": 254849998.0, + "step": 9845 + }, + { + "epoch": 1.0812650999341094, + "grad_norm": 2.4308252334594727, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7075303792953491, + "num_tokens": 254873692.0, + "step": 9846 + }, + { + "epoch": 1.081374917636723, + "grad_norm": 2.14469575881958, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7280226349830627, + "num_tokens": 254901458.0, + "step": 9847 + }, + { + "epoch": 1.0814847353393366, + "grad_norm": 2.4472124576568604, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7214261293411255, + "num_tokens": 254922737.0, + "step": 9848 + }, + { + "epoch": 1.0815945530419504, + "grad_norm": 2.1937313079833984, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7076294422149658, + "num_tokens": 254951216.0, + "step": 9849 + }, + { + "epoch": 1.081704370744564, + "grad_norm": 2.372755289077759, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7159327864646912, + "num_tokens": 254975725.0, + "step": 9850 + }, + { + "epoch": 1.0818141884471777, + "grad_norm": 1.8764777183532715, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6999534368515015, + "num_tokens": 255012158.0, + "step": 9851 + }, + { + "epoch": 1.0819240061497912, + "grad_norm": 2.815472364425659, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7392444610595703, + "num_tokens": 255028102.0, + "step": 9852 + }, + { + "epoch": 1.082033823852405, + "grad_norm": 2.1874263286590576, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7287498712539673, + "num_tokens": 255053098.0, + "step": 9853 + }, + { + "epoch": 1.0821436415550187, + "grad_norm": 2.134087085723877, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.750593364238739, + "num_tokens": 255079204.0, + "step": 9854 + }, + { + "epoch": 1.0822534592576323, + "grad_norm": 1.9503949880599976, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.726264476776123, + "num_tokens": 255109320.0, + "step": 9855 + }, + { + "epoch": 1.082363276960246, + "grad_norm": 1.9119691848754883, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6963566541671753, + "num_tokens": 255142634.0, + "step": 9856 + }, + { + "epoch": 1.0824730946628596, + "grad_norm": 2.287055492401123, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7443210482597351, + "num_tokens": 255166313.0, + "step": 9857 + }, + { + "epoch": 1.0825829123654733, + "grad_norm": 2.267788887023926, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7143626809120178, + "num_tokens": 255194625.0, + "step": 9858 + }, + { + "epoch": 1.082692730068087, + "grad_norm": 2.183924913406372, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7320830821990967, + "num_tokens": 255220850.0, + "step": 9859 + }, + { + "epoch": 1.0828025477707006, + "grad_norm": 2.4908535480499268, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7265769243240356, + "num_tokens": 255243028.0, + "step": 9860 + }, + { + "epoch": 1.0829123654733144, + "grad_norm": 2.2313690185546875, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.695029616355896, + "num_tokens": 255272468.0, + "step": 9861 + }, + { + "epoch": 1.083022183175928, + "grad_norm": 2.4356446266174316, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7184958457946777, + "num_tokens": 255296927.0, + "step": 9862 + }, + { + "epoch": 1.0831320008785417, + "grad_norm": 2.1243674755096436, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7413519620895386, + "num_tokens": 255323354.0, + "step": 9863 + }, + { + "epoch": 1.0832418185811552, + "grad_norm": 2.0051448345184326, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.6975758075714111, + "num_tokens": 255356309.0, + "step": 9864 + }, + { + "epoch": 1.083351636283769, + "grad_norm": 2.167752504348755, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7159155607223511, + "num_tokens": 255383162.0, + "step": 9865 + }, + { + "epoch": 1.0834614539863825, + "grad_norm": 2.3011209964752197, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7048923969268799, + "num_tokens": 255408460.0, + "step": 9866 + }, + { + "epoch": 1.0835712716889963, + "grad_norm": 2.2463302612304688, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7189232110977173, + "num_tokens": 255434169.0, + "step": 9867 + }, + { + "epoch": 1.08368108939161, + "grad_norm": 2.5312108993530273, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.708519697189331, + "num_tokens": 255454970.0, + "step": 9868 + }, + { + "epoch": 1.0837909070942235, + "grad_norm": 2.47257661819458, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7409515380859375, + "num_tokens": 255475839.0, + "step": 9869 + }, + { + "epoch": 1.0839007247968373, + "grad_norm": 2.0996010303497314, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6975756287574768, + "num_tokens": 255505850.0, + "step": 9870 + }, + { + "epoch": 1.0840105424994508, + "grad_norm": 2.1145596504211426, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.6997494101524353, + "num_tokens": 255536149.0, + "step": 9871 + }, + { + "epoch": 1.0841203602020646, + "grad_norm": 2.2628090381622314, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7242308855056763, + "num_tokens": 255561334.0, + "step": 9872 + }, + { + "epoch": 1.0842301779046783, + "grad_norm": 2.157226085662842, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7152437567710876, + "num_tokens": 255589566.0, + "step": 9873 + }, + { + "epoch": 1.0843399956072919, + "grad_norm": 2.3751914501190186, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7418642044067383, + "num_tokens": 255613867.0, + "step": 9874 + }, + { + "epoch": 1.0844498133099056, + "grad_norm": 2.5581514835357666, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7277703881263733, + "num_tokens": 255634796.0, + "step": 9875 + }, + { + "epoch": 1.0845596310125192, + "grad_norm": 1.9585484266281128, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.6993044018745422, + "num_tokens": 255666912.0, + "step": 9876 + }, + { + "epoch": 1.084669448715133, + "grad_norm": 2.2227160930633545, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7076730132102966, + "num_tokens": 255693497.0, + "step": 9877 + }, + { + "epoch": 1.0847792664177465, + "grad_norm": 2.207376003265381, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7079471349716187, + "num_tokens": 255719485.0, + "step": 9878 + }, + { + "epoch": 1.0848890841203602, + "grad_norm": 2.257645606994629, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7188265323638916, + "num_tokens": 255743656.0, + "step": 9879 + }, + { + "epoch": 1.0849989018229738, + "grad_norm": 2.0564379692077637, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.711617112159729, + "num_tokens": 255774053.0, + "step": 9880 + }, + { + "epoch": 1.0851087195255875, + "grad_norm": 2.621093988418579, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7022993564605713, + "num_tokens": 255794035.0, + "step": 9881 + }, + { + "epoch": 1.0852185372282013, + "grad_norm": 2.167637348175049, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7490999698638916, + "num_tokens": 255819485.0, + "step": 9882 + }, + { + "epoch": 1.0853283549308148, + "grad_norm": 2.3951127529144287, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6908539533615112, + "num_tokens": 255844570.0, + "step": 9883 + }, + { + "epoch": 1.0854381726334286, + "grad_norm": 2.1983871459960938, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6936492323875427, + "num_tokens": 255873762.0, + "step": 9884 + }, + { + "epoch": 1.085547990336042, + "grad_norm": 2.4593281745910645, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.717979907989502, + "num_tokens": 255896096.0, + "step": 9885 + }, + { + "epoch": 1.0856578080386559, + "grad_norm": 2.0478498935699463, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6884002685546875, + "num_tokens": 255927815.0, + "step": 9886 + }, + { + "epoch": 1.0857676257412694, + "grad_norm": 2.3825812339782715, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7314918041229248, + "num_tokens": 255949988.0, + "step": 9887 + }, + { + "epoch": 1.0858774434438832, + "grad_norm": 2.652841329574585, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7497774362564087, + "num_tokens": 255968301.0, + "step": 9888 + }, + { + "epoch": 1.085987261146497, + "grad_norm": 2.4582362174987793, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7199826240539551, + "num_tokens": 255991431.0, + "step": 9889 + }, + { + "epoch": 1.0860970788491104, + "grad_norm": 2.0224878787994385, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6943212747573853, + "num_tokens": 256023317.0, + "step": 9890 + }, + { + "epoch": 1.0862068965517242, + "grad_norm": 2.7985613346099854, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7173942923545837, + "num_tokens": 256042239.0, + "step": 9891 + }, + { + "epoch": 1.0863167142543377, + "grad_norm": 2.1871423721313477, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7420117259025574, + "num_tokens": 256069235.0, + "step": 9892 + }, + { + "epoch": 1.0864265319569515, + "grad_norm": 2.2773547172546387, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7227353453636169, + "num_tokens": 256094727.0, + "step": 9893 + }, + { + "epoch": 1.086536349659565, + "grad_norm": 2.4406144618988037, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7321165204048157, + "num_tokens": 256116410.0, + "step": 9894 + }, + { + "epoch": 1.0866461673621788, + "grad_norm": 2.2928287982940674, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7323527932167053, + "num_tokens": 256140523.0, + "step": 9895 + }, + { + "epoch": 1.0867559850647925, + "grad_norm": 2.0404319763183594, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.729397177696228, + "num_tokens": 256173084.0, + "step": 9896 + }, + { + "epoch": 1.086865802767406, + "grad_norm": 2.4201722145080566, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7093828916549683, + "num_tokens": 256197211.0, + "step": 9897 + }, + { + "epoch": 1.0869756204700198, + "grad_norm": 2.340242624282837, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7271497845649719, + "num_tokens": 256219446.0, + "step": 9898 + }, + { + "epoch": 1.0870854381726334, + "grad_norm": 2.5080583095550537, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7412993907928467, + "num_tokens": 256240081.0, + "step": 9899 + }, + { + "epoch": 1.0871952558752471, + "grad_norm": 2.0887656211853027, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.731412410736084, + "num_tokens": 256268252.0, + "step": 9900 + }, + { + "epoch": 1.0873050735778607, + "grad_norm": 2.572256326675415, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7459062337875366, + "num_tokens": 256288236.0, + "step": 9901 + }, + { + "epoch": 1.0874148912804744, + "grad_norm": 2.608487367630005, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7355498671531677, + "num_tokens": 256307306.0, + "step": 9902 + }, + { + "epoch": 1.0875247089830882, + "grad_norm": 2.084930181503296, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6931004524230957, + "num_tokens": 256336305.0, + "step": 9903 + }, + { + "epoch": 1.0876345266857017, + "grad_norm": 2.3312370777130127, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7383488416671753, + "num_tokens": 256360289.0, + "step": 9904 + }, + { + "epoch": 1.0877443443883155, + "grad_norm": 1.9981931447982788, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7226344347000122, + "num_tokens": 256390078.0, + "step": 9905 + }, + { + "epoch": 1.087854162090929, + "grad_norm": 2.0720841884613037, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7120128273963928, + "num_tokens": 256420136.0, + "step": 9906 + }, + { + "epoch": 1.0879639797935428, + "grad_norm": 2.115999460220337, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7216955423355103, + "num_tokens": 256448256.0, + "step": 9907 + }, + { + "epoch": 1.0880737974961563, + "grad_norm": 2.0831401348114014, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7049286365509033, + "num_tokens": 256477398.0, + "step": 9908 + }, + { + "epoch": 1.08818361519877, + "grad_norm": 2.3968935012817383, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7368851900100708, + "num_tokens": 256499772.0, + "step": 9909 + }, + { + "epoch": 1.0882934329013838, + "grad_norm": 2.1263375282287598, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7156683206558228, + "num_tokens": 256528746.0, + "step": 9910 + }, + { + "epoch": 1.0884032506039973, + "grad_norm": 2.4617044925689697, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6837483644485474, + "num_tokens": 256553700.0, + "step": 9911 + }, + { + "epoch": 1.088513068306611, + "grad_norm": 2.1337409019470215, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7258796691894531, + "num_tokens": 256581267.0, + "step": 9912 + }, + { + "epoch": 1.0886228860092246, + "grad_norm": 2.1875522136688232, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.702075719833374, + "num_tokens": 256606458.0, + "step": 9913 + }, + { + "epoch": 1.0887327037118384, + "grad_norm": 2.3327066898345947, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7243639826774597, + "num_tokens": 256631026.0, + "step": 9914 + }, + { + "epoch": 1.088842521414452, + "grad_norm": 2.0653388500213623, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7088845372200012, + "num_tokens": 256659438.0, + "step": 9915 + }, + { + "epoch": 1.0889523391170657, + "grad_norm": 2.175644874572754, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6881594061851501, + "num_tokens": 256690266.0, + "step": 9916 + }, + { + "epoch": 1.0890621568196792, + "grad_norm": 2.105112314224243, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7304226160049438, + "num_tokens": 256719995.0, + "step": 9917 + }, + { + "epoch": 1.089171974522293, + "grad_norm": 2.514824867248535, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7135389447212219, + "num_tokens": 256741027.0, + "step": 9918 + }, + { + "epoch": 1.0892817922249067, + "grad_norm": 2.423190116882324, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6878044605255127, + "num_tokens": 256764856.0, + "step": 9919 + }, + { + "epoch": 1.0893916099275203, + "grad_norm": 2.1654837131500244, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7044853568077087, + "num_tokens": 256791611.0, + "step": 9920 + }, + { + "epoch": 1.089501427630134, + "grad_norm": 2.372258424758911, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7135518789291382, + "num_tokens": 256816799.0, + "step": 9921 + }, + { + "epoch": 1.0896112453327476, + "grad_norm": 2.1249942779541016, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6951048374176025, + "num_tokens": 256844587.0, + "step": 9922 + }, + { + "epoch": 1.0897210630353613, + "grad_norm": 2.2534725666046143, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7175322771072388, + "num_tokens": 256869769.0, + "step": 9923 + }, + { + "epoch": 1.089830880737975, + "grad_norm": 2.064610481262207, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6943708658218384, + "num_tokens": 256902427.0, + "step": 9924 + }, + { + "epoch": 1.0899406984405886, + "grad_norm": 2.304065465927124, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7145822048187256, + "num_tokens": 256928302.0, + "step": 9925 + }, + { + "epoch": 1.0900505161432024, + "grad_norm": 2.524829387664795, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7102209329605103, + "num_tokens": 256949129.0, + "step": 9926 + }, + { + "epoch": 1.090160333845816, + "grad_norm": 2.17812442779541, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7167621850967407, + "num_tokens": 256979446.0, + "step": 9927 + }, + { + "epoch": 1.0902701515484297, + "grad_norm": 2.2421138286590576, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7195222973823547, + "num_tokens": 257006538.0, + "step": 9928 + }, + { + "epoch": 1.0903799692510432, + "grad_norm": 2.2441415786743164, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7377469539642334, + "num_tokens": 257030709.0, + "step": 9929 + }, + { + "epoch": 1.090489786953657, + "grad_norm": 2.31303071975708, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7271790504455566, + "num_tokens": 257054215.0, + "step": 9930 + }, + { + "epoch": 1.0905996046562705, + "grad_norm": 2.1579980850219727, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7207401990890503, + "num_tokens": 257080815.0, + "step": 9931 + }, + { + "epoch": 1.0907094223588842, + "grad_norm": 2.1913485527038574, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7262890338897705, + "num_tokens": 257105877.0, + "step": 9932 + }, + { + "epoch": 1.090819240061498, + "grad_norm": 2.471203088760376, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7226133942604065, + "num_tokens": 257128601.0, + "step": 9933 + }, + { + "epoch": 1.0909290577641115, + "grad_norm": 2.471217632293701, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.730016827583313, + "num_tokens": 257149597.0, + "step": 9934 + }, + { + "epoch": 1.0910388754667253, + "grad_norm": 2.457531690597534, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.712844729423523, + "num_tokens": 257171312.0, + "step": 9935 + }, + { + "epoch": 1.0911486931693388, + "grad_norm": 2.0116071701049805, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7058049440383911, + "num_tokens": 257204631.0, + "step": 9936 + }, + { + "epoch": 1.0912585108719526, + "grad_norm": 2.3436167240142822, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7049247622489929, + "num_tokens": 257228960.0, + "step": 9937 + }, + { + "epoch": 1.0913683285745663, + "grad_norm": 2.418318748474121, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7269074320793152, + "num_tokens": 257251455.0, + "step": 9938 + }, + { + "epoch": 1.0914781462771799, + "grad_norm": 2.339916706085205, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7184348106384277, + "num_tokens": 257277132.0, + "step": 9939 + }, + { + "epoch": 1.0915879639797936, + "grad_norm": 2.292426824569702, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.714851438999176, + "num_tokens": 257301807.0, + "step": 9940 + }, + { + "epoch": 1.0916977816824072, + "grad_norm": 1.9720417261123657, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7098899483680725, + "num_tokens": 257333641.0, + "step": 9941 + }, + { + "epoch": 1.091807599385021, + "grad_norm": 2.0897765159606934, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7294732928276062, + "num_tokens": 257361424.0, + "step": 9942 + }, + { + "epoch": 1.0919174170876345, + "grad_norm": 2.443575620651245, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7131339311599731, + "num_tokens": 257383212.0, + "step": 9943 + }, + { + "epoch": 1.0920272347902482, + "grad_norm": 2.077120780944824, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7111291885375977, + "num_tokens": 257412798.0, + "step": 9944 + }, + { + "epoch": 1.0921370524928617, + "grad_norm": 2.2134740352630615, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7131309509277344, + "num_tokens": 257440412.0, + "step": 9945 + }, + { + "epoch": 1.0922468701954755, + "grad_norm": 2.1352286338806152, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.721933126449585, + "num_tokens": 257468699.0, + "step": 9946 + }, + { + "epoch": 1.0923566878980893, + "grad_norm": 2.042848587036133, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7055968046188354, + "num_tokens": 257497606.0, + "step": 9947 + }, + { + "epoch": 1.0924665056007028, + "grad_norm": 2.3552703857421875, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7319973707199097, + "num_tokens": 257520184.0, + "step": 9948 + }, + { + "epoch": 1.0925763233033166, + "grad_norm": 2.169172525405884, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6969338655471802, + "num_tokens": 257550058.0, + "step": 9949 + }, + { + "epoch": 1.09268614100593, + "grad_norm": 2.17073655128479, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7109353542327881, + "num_tokens": 257577836.0, + "step": 9950 + }, + { + "epoch": 1.0927959587085438, + "grad_norm": 1.8275690078735352, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7043871283531189, + "num_tokens": 257614681.0, + "step": 9951 + }, + { + "epoch": 1.0929057764111574, + "grad_norm": 2.284345865249634, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7198789715766907, + "num_tokens": 257638113.0, + "step": 9952 + }, + { + "epoch": 1.0930155941137711, + "grad_norm": 2.2761504650115967, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7318875193595886, + "num_tokens": 257661459.0, + "step": 9953 + }, + { + "epoch": 1.093125411816385, + "grad_norm": 1.9636237621307373, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6993638277053833, + "num_tokens": 257694517.0, + "step": 9954 + }, + { + "epoch": 1.0932352295189984, + "grad_norm": 2.108828067779541, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7292251586914062, + "num_tokens": 257721198.0, + "step": 9955 + }, + { + "epoch": 1.0933450472216122, + "grad_norm": 2.3615591526031494, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7416056394577026, + "num_tokens": 257741074.0, + "step": 9956 + }, + { + "epoch": 1.0934548649242257, + "grad_norm": 2.23075008392334, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7084426879882812, + "num_tokens": 257767076.0, + "step": 9957 + }, + { + "epoch": 1.0935646826268395, + "grad_norm": 2.0366694927215576, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6931141018867493, + "num_tokens": 257798867.0, + "step": 9958 + }, + { + "epoch": 1.093674500329453, + "grad_norm": 2.544956922531128, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7296712398529053, + "num_tokens": 257821000.0, + "step": 9959 + }, + { + "epoch": 1.0937843180320668, + "grad_norm": 2.217042922973633, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7372923493385315, + "num_tokens": 257845751.0, + "step": 9960 + }, + { + "epoch": 1.0938941357346805, + "grad_norm": 2.2370071411132812, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7228406667709351, + "num_tokens": 257870734.0, + "step": 9961 + }, + { + "epoch": 1.094003953437294, + "grad_norm": 2.46529483795166, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7106971740722656, + "num_tokens": 257892104.0, + "step": 9962 + }, + { + "epoch": 1.0941137711399078, + "grad_norm": 2.259042978286743, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7065930366516113, + "num_tokens": 257918169.0, + "step": 9963 + }, + { + "epoch": 1.0942235888425214, + "grad_norm": 2.610107421875, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7258638739585876, + "num_tokens": 257939361.0, + "step": 9964 + }, + { + "epoch": 1.0943334065451351, + "grad_norm": 2.427142858505249, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7205959558486938, + "num_tokens": 257962990.0, + "step": 9965 + }, + { + "epoch": 1.0944432242477486, + "grad_norm": 2.872987747192383, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7512495517730713, + "num_tokens": 257980096.0, + "step": 9966 + }, + { + "epoch": 1.0945530419503624, + "grad_norm": 2.1349165439605713, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7403242588043213, + "num_tokens": 258007207.0, + "step": 9967 + }, + { + "epoch": 1.0946628596529762, + "grad_norm": 1.8251467943191528, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7002274990081787, + "num_tokens": 258043075.0, + "step": 9968 + }, + { + "epoch": 1.0947726773555897, + "grad_norm": 1.9886034727096558, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.739258885383606, + "num_tokens": 258073828.0, + "step": 9969 + }, + { + "epoch": 1.0948824950582035, + "grad_norm": 2.3168954849243164, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7048194408416748, + "num_tokens": 258100626.0, + "step": 9970 + }, + { + "epoch": 1.094992312760817, + "grad_norm": 2.1603519916534424, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7550767660140991, + "num_tokens": 258125951.0, + "step": 9971 + }, + { + "epoch": 1.0951021304634307, + "grad_norm": 2.0935239791870117, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7361016869544983, + "num_tokens": 258151984.0, + "step": 9972 + }, + { + "epoch": 1.0952119481660443, + "grad_norm": 2.2524821758270264, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7011724710464478, + "num_tokens": 258177525.0, + "step": 9973 + }, + { + "epoch": 1.095321765868658, + "grad_norm": 2.2766144275665283, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7136480212211609, + "num_tokens": 258203254.0, + "step": 9974 + }, + { + "epoch": 1.0954315835712718, + "grad_norm": 2.259199619293213, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7106635570526123, + "num_tokens": 258230192.0, + "step": 9975 + }, + { + "epoch": 1.0955414012738853, + "grad_norm": 2.3962414264678955, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.701350212097168, + "num_tokens": 258253607.0, + "step": 9976 + }, + { + "epoch": 1.095651218976499, + "grad_norm": 2.4572973251342773, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7193810939788818, + "num_tokens": 258276578.0, + "step": 9977 + }, + { + "epoch": 1.0957610366791126, + "grad_norm": 2.3025553226470947, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7149016261100769, + "num_tokens": 258302613.0, + "step": 9978 + }, + { + "epoch": 1.0958708543817264, + "grad_norm": 1.8613693714141846, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7066155076026917, + "num_tokens": 258337946.0, + "step": 9979 + }, + { + "epoch": 1.09598067208434, + "grad_norm": 2.023005723953247, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7026922702789307, + "num_tokens": 258369426.0, + "step": 9980 + }, + { + "epoch": 1.0960904897869537, + "grad_norm": 2.455209970474243, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7238049507141113, + "num_tokens": 258391493.0, + "step": 9981 + }, + { + "epoch": 1.0962003074895672, + "grad_norm": 2.1924102306365967, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7145709991455078, + "num_tokens": 258417199.0, + "step": 9982 + }, + { + "epoch": 1.096310125192181, + "grad_norm": 2.5009541511535645, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7147424221038818, + "num_tokens": 258438100.0, + "step": 9983 + }, + { + "epoch": 1.0964199428947947, + "grad_norm": 2.213676929473877, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7427833676338196, + "num_tokens": 258463474.0, + "step": 9984 + }, + { + "epoch": 1.0965297605974083, + "grad_norm": 2.2948739528656006, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7064394354820251, + "num_tokens": 258488155.0, + "step": 9985 + }, + { + "epoch": 1.096639578300022, + "grad_norm": 2.5294055938720703, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7244458794593811, + "num_tokens": 258511403.0, + "step": 9986 + }, + { + "epoch": 1.0967493960026355, + "grad_norm": 2.470231294631958, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.705288290977478, + "num_tokens": 258533843.0, + "step": 9987 + }, + { + "epoch": 1.0968592137052493, + "grad_norm": 2.205390691757202, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.722175121307373, + "num_tokens": 258559380.0, + "step": 9988 + }, + { + "epoch": 1.096969031407863, + "grad_norm": 2.224846839904785, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6811031103134155, + "num_tokens": 258589868.0, + "step": 9989 + }, + { + "epoch": 1.0970788491104766, + "grad_norm": 2.3707845211029053, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7374409437179565, + "num_tokens": 258611281.0, + "step": 9990 + }, + { + "epoch": 1.0971886668130904, + "grad_norm": 2.2101237773895264, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7013736367225647, + "num_tokens": 258639167.0, + "step": 9991 + }, + { + "epoch": 1.0972984845157039, + "grad_norm": 2.2131831645965576, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.718535304069519, + "num_tokens": 258665202.0, + "step": 9992 + }, + { + "epoch": 1.0974083022183176, + "grad_norm": 2.5007050037384033, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7318329811096191, + "num_tokens": 258686204.0, + "step": 9993 + }, + { + "epoch": 1.0975181199209312, + "grad_norm": 2.1358864307403564, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7299720048904419, + "num_tokens": 258712260.0, + "step": 9994 + }, + { + "epoch": 1.097627937623545, + "grad_norm": 2.62050461769104, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7271803617477417, + "num_tokens": 258732364.0, + "step": 9995 + }, + { + "epoch": 1.0977377553261585, + "grad_norm": 2.273847818374634, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7499983310699463, + "num_tokens": 258758137.0, + "step": 9996 + }, + { + "epoch": 1.0978475730287722, + "grad_norm": 2.3885152339935303, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7184540629386902, + "num_tokens": 258781894.0, + "step": 9997 + }, + { + "epoch": 1.097957390731386, + "grad_norm": 1.9022319316864014, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.726943850517273, + "num_tokens": 258816433.0, + "step": 9998 + }, + { + "epoch": 1.0980672084339995, + "grad_norm": 2.277127504348755, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7036228179931641, + "num_tokens": 258842293.0, + "step": 9999 + }, + { + "epoch": 1.0981770261366133, + "grad_norm": 2.3220224380493164, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7116084694862366, + "num_tokens": 258868498.0, + "step": 10000 + }, + { + "epoch": 1.0982868438392268, + "grad_norm": 2.298854112625122, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7132164239883423, + "num_tokens": 258894056.0, + "step": 10001 + }, + { + "epoch": 1.0983966615418406, + "grad_norm": 2.6456289291381836, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7428440451622009, + "num_tokens": 258913204.0, + "step": 10002 + }, + { + "epoch": 1.098506479244454, + "grad_norm": 2.5111639499664307, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7485849261283875, + "num_tokens": 258934560.0, + "step": 10003 + }, + { + "epoch": 1.0986162969470679, + "grad_norm": 2.5431647300720215, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7383071184158325, + "num_tokens": 258957202.0, + "step": 10004 + }, + { + "epoch": 1.0987261146496816, + "grad_norm": 2.5821077823638916, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7248056530952454, + "num_tokens": 258978611.0, + "step": 10005 + }, + { + "epoch": 1.0988359323522952, + "grad_norm": 2.3071846961975098, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7138767242431641, + "num_tokens": 259004896.0, + "step": 10006 + }, + { + "epoch": 1.098945750054909, + "grad_norm": 2.390357494354248, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7133631706237793, + "num_tokens": 259029899.0, + "step": 10007 + }, + { + "epoch": 1.0990555677575224, + "grad_norm": 1.992169976234436, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7262868881225586, + "num_tokens": 259058605.0, + "step": 10008 + }, + { + "epoch": 1.0991653854601362, + "grad_norm": 2.256112575531006, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7179819941520691, + "num_tokens": 259082524.0, + "step": 10009 + }, + { + "epoch": 1.0992752031627497, + "grad_norm": 2.304692268371582, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7227120995521545, + "num_tokens": 259107190.0, + "step": 10010 + }, + { + "epoch": 1.0993850208653635, + "grad_norm": 1.949533224105835, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7183636426925659, + "num_tokens": 259140725.0, + "step": 10011 + }, + { + "epoch": 1.0994948385679773, + "grad_norm": 2.2931482791900635, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7010030150413513, + "num_tokens": 259166114.0, + "step": 10012 + }, + { + "epoch": 1.0996046562705908, + "grad_norm": 2.3363144397735596, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7221572995185852, + "num_tokens": 259188512.0, + "step": 10013 + }, + { + "epoch": 1.0997144739732045, + "grad_norm": 2.176788806915283, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7074248790740967, + "num_tokens": 259213322.0, + "step": 10014 + }, + { + "epoch": 1.099824291675818, + "grad_norm": 1.944883942604065, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.711940348148346, + "num_tokens": 259244247.0, + "step": 10015 + }, + { + "epoch": 1.0999341093784318, + "grad_norm": 2.1149182319641113, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7268177270889282, + "num_tokens": 259272335.0, + "step": 10016 + }, + { + "epoch": 1.1000439270810454, + "grad_norm": 2.121299982070923, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7239419221878052, + "num_tokens": 259299824.0, + "step": 10017 + }, + { + "epoch": 1.1001537447836591, + "grad_norm": 2.2468981742858887, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7297548055648804, + "num_tokens": 259322970.0, + "step": 10018 + }, + { + "epoch": 1.1002635624862729, + "grad_norm": 2.157810926437378, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.714093029499054, + "num_tokens": 259352116.0, + "step": 10019 + }, + { + "epoch": 1.1003733801888864, + "grad_norm": 2.049607038497925, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6977213621139526, + "num_tokens": 259383382.0, + "step": 10020 + }, + { + "epoch": 1.1004831978915002, + "grad_norm": 1.9764801263809204, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.718242883682251, + "num_tokens": 259412079.0, + "step": 10021 + }, + { + "epoch": 1.1005930155941137, + "grad_norm": 1.984932541847229, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7141163349151611, + "num_tokens": 259445070.0, + "step": 10022 + }, + { + "epoch": 1.1007028332967275, + "grad_norm": 2.2485198974609375, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7186768054962158, + "num_tokens": 259473069.0, + "step": 10023 + }, + { + "epoch": 1.100812650999341, + "grad_norm": 2.241055965423584, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7148809432983398, + "num_tokens": 259498640.0, + "step": 10024 + }, + { + "epoch": 1.1009224687019548, + "grad_norm": 2.2026779651641846, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.694164514541626, + "num_tokens": 259525249.0, + "step": 10025 + }, + { + "epoch": 1.1010322864045685, + "grad_norm": 1.997334599494934, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.6944637298583984, + "num_tokens": 259558002.0, + "step": 10026 + }, + { + "epoch": 1.101142104107182, + "grad_norm": 2.054187297821045, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7262780070304871, + "num_tokens": 259587005.0, + "step": 10027 + }, + { + "epoch": 1.1012519218097958, + "grad_norm": 2.486375093460083, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.703104555606842, + "num_tokens": 259608555.0, + "step": 10028 + }, + { + "epoch": 1.1013617395124093, + "grad_norm": 2.2873289585113525, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.715022623538971, + "num_tokens": 259632527.0, + "step": 10029 + }, + { + "epoch": 1.101471557215023, + "grad_norm": 2.1939027309417725, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7375034093856812, + "num_tokens": 259657114.0, + "step": 10030 + }, + { + "epoch": 1.1015813749176366, + "grad_norm": 2.0913028717041016, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7181510925292969, + "num_tokens": 259687056.0, + "step": 10031 + }, + { + "epoch": 1.1016911926202504, + "grad_norm": 2.284268379211426, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7318142652511597, + "num_tokens": 259711041.0, + "step": 10032 + }, + { + "epoch": 1.1018010103228641, + "grad_norm": 2.2152798175811768, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7077627182006836, + "num_tokens": 259735799.0, + "step": 10033 + }, + { + "epoch": 1.1019108280254777, + "grad_norm": 2.387956380844116, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7214414477348328, + "num_tokens": 259759245.0, + "step": 10034 + }, + { + "epoch": 1.1020206457280914, + "grad_norm": 2.232182264328003, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7106064558029175, + "num_tokens": 259786430.0, + "step": 10035 + }, + { + "epoch": 1.102130463430705, + "grad_norm": 2.2979226112365723, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7049275636672974, + "num_tokens": 259812205.0, + "step": 10036 + }, + { + "epoch": 1.1022402811333187, + "grad_norm": 2.4719910621643066, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7192515134811401, + "num_tokens": 259834090.0, + "step": 10037 + }, + { + "epoch": 1.1023500988359323, + "grad_norm": 2.273505210876465, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7183209657669067, + "num_tokens": 259859684.0, + "step": 10038 + }, + { + "epoch": 1.102459916538546, + "grad_norm": 2.1126441955566406, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7302389144897461, + "num_tokens": 259887694.0, + "step": 10039 + }, + { + "epoch": 1.1025697342411598, + "grad_norm": 2.1585607528686523, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7072985172271729, + "num_tokens": 259915361.0, + "step": 10040 + }, + { + "epoch": 1.1026795519437733, + "grad_norm": 2.380460500717163, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7096251249313354, + "num_tokens": 259939604.0, + "step": 10041 + }, + { + "epoch": 1.102789369646387, + "grad_norm": 2.2373814582824707, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7043529748916626, + "num_tokens": 259964570.0, + "step": 10042 + }, + { + "epoch": 1.1028991873490006, + "grad_norm": 2.324476957321167, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7286567091941833, + "num_tokens": 259988233.0, + "step": 10043 + }, + { + "epoch": 1.1030090050516144, + "grad_norm": 2.236560106277466, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7021260261535645, + "num_tokens": 260016806.0, + "step": 10044 + }, + { + "epoch": 1.103118822754228, + "grad_norm": 2.1879141330718994, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7100061178207397, + "num_tokens": 260044011.0, + "step": 10045 + }, + { + "epoch": 1.1032286404568417, + "grad_norm": 2.3551316261291504, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7238396406173706, + "num_tokens": 260068734.0, + "step": 10046 + }, + { + "epoch": 1.1033384581594552, + "grad_norm": 2.1954314708709717, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7261130809783936, + "num_tokens": 260093825.0, + "step": 10047 + }, + { + "epoch": 1.103448275862069, + "grad_norm": 2.0856876373291016, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6929441690444946, + "num_tokens": 260122935.0, + "step": 10048 + }, + { + "epoch": 1.1035580935646827, + "grad_norm": 2.485562324523926, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7329424619674683, + "num_tokens": 260148211.0, + "step": 10049 + }, + { + "epoch": 1.1036679112672962, + "grad_norm": 2.4114885330200195, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7286041975021362, + "num_tokens": 260171463.0, + "step": 10050 + }, + { + "epoch": 1.10377772896991, + "grad_norm": 2.295048713684082, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7346376180648804, + "num_tokens": 260194896.0, + "step": 10051 + }, + { + "epoch": 1.1038875466725235, + "grad_norm": 2.461366653442383, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7000764608383179, + "num_tokens": 260218383.0, + "step": 10052 + }, + { + "epoch": 1.1039973643751373, + "grad_norm": 2.2771596908569336, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7084143757820129, + "num_tokens": 260244991.0, + "step": 10053 + }, + { + "epoch": 1.104107182077751, + "grad_norm": 2.2509124279022217, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7229811549186707, + "num_tokens": 260271284.0, + "step": 10054 + }, + { + "epoch": 1.1042169997803646, + "grad_norm": 2.4957287311553955, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.7020756006240845, + "num_tokens": 260294880.0, + "step": 10055 + }, + { + "epoch": 1.1043268174829783, + "grad_norm": 2.227189064025879, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6873317956924438, + "num_tokens": 260322375.0, + "step": 10056 + }, + { + "epoch": 1.1044366351855919, + "grad_norm": 2.118194341659546, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6964763402938843, + "num_tokens": 260352339.0, + "step": 10057 + }, + { + "epoch": 1.1045464528882056, + "grad_norm": 2.4441447257995605, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.715747058391571, + "num_tokens": 260376081.0, + "step": 10058 + }, + { + "epoch": 1.1046562705908192, + "grad_norm": 2.303300142288208, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7110379338264465, + "num_tokens": 260400360.0, + "step": 10059 + }, + { + "epoch": 1.104766088293433, + "grad_norm": 2.234229326248169, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7206286191940308, + "num_tokens": 260427736.0, + "step": 10060 + }, + { + "epoch": 1.1048759059960465, + "grad_norm": 2.284945487976074, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7113546133041382, + "num_tokens": 260451174.0, + "step": 10061 + }, + { + "epoch": 1.1049857236986602, + "grad_norm": 2.3749940395355225, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7168412208557129, + "num_tokens": 260474631.0, + "step": 10062 + }, + { + "epoch": 1.105095541401274, + "grad_norm": 2.248812675476074, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7334393262863159, + "num_tokens": 260499111.0, + "step": 10063 + }, + { + "epoch": 1.1052053591038875, + "grad_norm": 2.438453435897827, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7269044518470764, + "num_tokens": 260520362.0, + "step": 10064 + }, + { + "epoch": 1.1053151768065013, + "grad_norm": 2.151395082473755, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7292956113815308, + "num_tokens": 260547511.0, + "step": 10065 + }, + { + "epoch": 1.1054249945091148, + "grad_norm": 2.406121253967285, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.744587779045105, + "num_tokens": 260569000.0, + "step": 10066 + }, + { + "epoch": 1.1055348122117286, + "grad_norm": 1.812235713005066, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.722514271736145, + "num_tokens": 260602601.0, + "step": 10067 + }, + { + "epoch": 1.105644629914342, + "grad_norm": 2.162209987640381, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7038832902908325, + "num_tokens": 260631014.0, + "step": 10068 + }, + { + "epoch": 1.1057544476169558, + "grad_norm": 2.2233731746673584, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7237017750740051, + "num_tokens": 260656564.0, + "step": 10069 + }, + { + "epoch": 1.1058642653195696, + "grad_norm": 2.2978243827819824, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7264128923416138, + "num_tokens": 260678529.0, + "step": 10070 + }, + { + "epoch": 1.1059740830221831, + "grad_norm": 2.157240152359009, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7304423451423645, + "num_tokens": 260703650.0, + "step": 10071 + }, + { + "epoch": 1.106083900724797, + "grad_norm": 2.4302866458892822, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7198551893234253, + "num_tokens": 260725587.0, + "step": 10072 + }, + { + "epoch": 1.1061937184274104, + "grad_norm": 2.3041465282440186, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7379035353660583, + "num_tokens": 260749706.0, + "step": 10073 + }, + { + "epoch": 1.1063035361300242, + "grad_norm": 2.2021732330322266, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7365016937255859, + "num_tokens": 260775255.0, + "step": 10074 + }, + { + "epoch": 1.1064133538326377, + "grad_norm": 2.1290338039398193, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7211297154426575, + "num_tokens": 260803395.0, + "step": 10075 + }, + { + "epoch": 1.1065231715352515, + "grad_norm": 2.3523621559143066, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.717578113079071, + "num_tokens": 260827511.0, + "step": 10076 + }, + { + "epoch": 1.1066329892378652, + "grad_norm": 2.0569450855255127, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7197373509407043, + "num_tokens": 260857517.0, + "step": 10077 + }, + { + "epoch": 1.1067428069404788, + "grad_norm": 2.393442392349243, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7253769040107727, + "num_tokens": 260881444.0, + "step": 10078 + }, + { + "epoch": 1.1068526246430925, + "grad_norm": 2.287505865097046, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7242940068244934, + "num_tokens": 260907718.0, + "step": 10079 + }, + { + "epoch": 1.106962442345706, + "grad_norm": 2.2438416481018066, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7137260437011719, + "num_tokens": 260934543.0, + "step": 10080 + }, + { + "epoch": 1.1070722600483198, + "grad_norm": 2.358067750930786, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7076184153556824, + "num_tokens": 260960660.0, + "step": 10081 + }, + { + "epoch": 1.1071820777509334, + "grad_norm": 2.171082019805908, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.721089243888855, + "num_tokens": 260988653.0, + "step": 10082 + }, + { + "epoch": 1.1072918954535471, + "grad_norm": 2.1444921493530273, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7100588083267212, + "num_tokens": 261017363.0, + "step": 10083 + }, + { + "epoch": 1.1074017131561609, + "grad_norm": 2.2091808319091797, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7104378938674927, + "num_tokens": 261044320.0, + "step": 10084 + }, + { + "epoch": 1.1075115308587744, + "grad_norm": 2.1481683254241943, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7416287660598755, + "num_tokens": 261069403.0, + "step": 10085 + }, + { + "epoch": 1.1076213485613882, + "grad_norm": 2.6678872108459473, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7287255525588989, + "num_tokens": 261088395.0, + "step": 10086 + }, + { + "epoch": 1.1077311662640017, + "grad_norm": 2.2379658222198486, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.734292209148407, + "num_tokens": 261113142.0, + "step": 10087 + }, + { + "epoch": 1.1078409839666155, + "grad_norm": 2.4199256896972656, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6932662725448608, + "num_tokens": 261135076.0, + "step": 10088 + }, + { + "epoch": 1.107950801669229, + "grad_norm": 2.1855878829956055, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7115657925605774, + "num_tokens": 261159556.0, + "step": 10089 + }, + { + "epoch": 1.1080606193718427, + "grad_norm": 2.3930485248565674, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7257214784622192, + "num_tokens": 261185178.0, + "step": 10090 + }, + { + "epoch": 1.1081704370744565, + "grad_norm": 2.2022526264190674, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7433252334594727, + "num_tokens": 261210986.0, + "step": 10091 + }, + { + "epoch": 1.10828025477707, + "grad_norm": 1.9927524328231812, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7041597366333008, + "num_tokens": 261242510.0, + "step": 10092 + }, + { + "epoch": 1.1083900724796838, + "grad_norm": 2.2182130813598633, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7189315557479858, + "num_tokens": 261267792.0, + "step": 10093 + }, + { + "epoch": 1.1084998901822973, + "grad_norm": 2.295466423034668, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7274202108383179, + "num_tokens": 261293488.0, + "step": 10094 + }, + { + "epoch": 1.108609707884911, + "grad_norm": 2.5836479663848877, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7251443266868591, + "num_tokens": 261313720.0, + "step": 10095 + }, + { + "epoch": 1.1087195255875246, + "grad_norm": 2.7066681385040283, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7384788990020752, + "num_tokens": 261331486.0, + "step": 10096 + }, + { + "epoch": 1.1088293432901384, + "grad_norm": 2.4089066982269287, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7190805673599243, + "num_tokens": 261353896.0, + "step": 10097 + }, + { + "epoch": 1.108939160992752, + "grad_norm": 2.5915181636810303, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7122319936752319, + "num_tokens": 261376889.0, + "step": 10098 + }, + { + "epoch": 1.1090489786953657, + "grad_norm": 2.675232172012329, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7212092876434326, + "num_tokens": 261395160.0, + "step": 10099 + }, + { + "epoch": 1.1091587963979794, + "grad_norm": 2.2477335929870605, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7225161194801331, + "num_tokens": 261421634.0, + "step": 10100 + }, + { + "epoch": 1.109268614100593, + "grad_norm": 2.13358998298645, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7177591919898987, + "num_tokens": 261450939.0, + "step": 10101 + }, + { + "epoch": 1.1093784318032067, + "grad_norm": 2.357215404510498, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7231265902519226, + "num_tokens": 261474246.0, + "step": 10102 + }, + { + "epoch": 1.1094882495058203, + "grad_norm": 2.218085527420044, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7391557693481445, + "num_tokens": 261499381.0, + "step": 10103 + }, + { + "epoch": 1.109598067208434, + "grad_norm": 2.0047783851623535, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7244805693626404, + "num_tokens": 261529173.0, + "step": 10104 + }, + { + "epoch": 1.1097078849110478, + "grad_norm": 1.9603791236877441, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.714061439037323, + "num_tokens": 261562491.0, + "step": 10105 + }, + { + "epoch": 1.1098177026136613, + "grad_norm": 2.294914484024048, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7237612009048462, + "num_tokens": 261587022.0, + "step": 10106 + }, + { + "epoch": 1.109927520316275, + "grad_norm": 2.2070369720458984, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7094947099685669, + "num_tokens": 261613628.0, + "step": 10107 + }, + { + "epoch": 1.1100373380188886, + "grad_norm": 2.334847927093506, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.72679603099823, + "num_tokens": 261634952.0, + "step": 10108 + }, + { + "epoch": 1.1101471557215024, + "grad_norm": 2.3482742309570312, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7276880741119385, + "num_tokens": 261657788.0, + "step": 10109 + }, + { + "epoch": 1.110256973424116, + "grad_norm": 1.98099684715271, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6725927591323853, + "num_tokens": 261691464.0, + "step": 10110 + }, + { + "epoch": 1.1103667911267296, + "grad_norm": 2.3580873012542725, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7334524393081665, + "num_tokens": 261713940.0, + "step": 10111 + }, + { + "epoch": 1.1104766088293432, + "grad_norm": 2.2306129932403564, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7303809523582458, + "num_tokens": 261739358.0, + "step": 10112 + }, + { + "epoch": 1.110586426531957, + "grad_norm": 2.2466237545013428, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7342710494995117, + "num_tokens": 261765350.0, + "step": 10113 + }, + { + "epoch": 1.1106962442345707, + "grad_norm": 2.403982162475586, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7184864282608032, + "num_tokens": 261790069.0, + "step": 10114 + }, + { + "epoch": 1.1108060619371842, + "grad_norm": 5.401348114013672, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7239766120910645, + "num_tokens": 261815604.0, + "step": 10115 + }, + { + "epoch": 1.110915879639798, + "grad_norm": 2.1859567165374756, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7058403491973877, + "num_tokens": 261841623.0, + "step": 10116 + }, + { + "epoch": 1.1110256973424115, + "grad_norm": 2.5510995388031006, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.6996286511421204, + "num_tokens": 261862501.0, + "step": 10117 + }, + { + "epoch": 1.1111355150450253, + "grad_norm": 2.1246564388275146, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7006103992462158, + "num_tokens": 261890576.0, + "step": 10118 + }, + { + "epoch": 1.111245332747639, + "grad_norm": 2.2718002796173096, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7341806292533875, + "num_tokens": 261915171.0, + "step": 10119 + }, + { + "epoch": 1.1113551504502526, + "grad_norm": 2.2005412578582764, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7337729334831238, + "num_tokens": 261942007.0, + "step": 10120 + }, + { + "epoch": 1.1114649681528663, + "grad_norm": 2.244084596633911, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7084163427352905, + "num_tokens": 261966752.0, + "step": 10121 + }, + { + "epoch": 1.1115747858554799, + "grad_norm": 2.453779458999634, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7157235145568848, + "num_tokens": 261990830.0, + "step": 10122 + }, + { + "epoch": 1.1116846035580936, + "grad_norm": 2.2009119987487793, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7058938145637512, + "num_tokens": 262015820.0, + "step": 10123 + }, + { + "epoch": 1.1117944212607072, + "grad_norm": 2.6621055603027344, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7385460138320923, + "num_tokens": 262034653.0, + "step": 10124 + }, + { + "epoch": 1.111904238963321, + "grad_norm": 2.151585102081299, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6858161091804504, + "num_tokens": 262065968.0, + "step": 10125 + }, + { + "epoch": 1.1120140566659344, + "grad_norm": 1.8526190519332886, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.705396294593811, + "num_tokens": 262102326.0, + "step": 10126 + }, + { + "epoch": 1.1121238743685482, + "grad_norm": 2.1979689598083496, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7049570083618164, + "num_tokens": 262129269.0, + "step": 10127 + }, + { + "epoch": 1.112233692071162, + "grad_norm": 2.139023780822754, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6814696192741394, + "num_tokens": 262159214.0, + "step": 10128 + }, + { + "epoch": 1.1123435097737755, + "grad_norm": 2.061741352081299, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.700861930847168, + "num_tokens": 262187605.0, + "step": 10129 + }, + { + "epoch": 1.1124533274763893, + "grad_norm": 2.2427761554718018, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.701803982257843, + "num_tokens": 262215300.0, + "step": 10130 + }, + { + "epoch": 1.1125631451790028, + "grad_norm": 2.311659097671509, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.725011944770813, + "num_tokens": 262237667.0, + "step": 10131 + }, + { + "epoch": 1.1126729628816165, + "grad_norm": 2.3847758769989014, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7208945751190186, + "num_tokens": 262260847.0, + "step": 10132 + }, + { + "epoch": 1.11278278058423, + "grad_norm": 2.1300060749053955, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7355951070785522, + "num_tokens": 262288164.0, + "step": 10133 + }, + { + "epoch": 1.1128925982868438, + "grad_norm": 2.115436315536499, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.717942476272583, + "num_tokens": 262315357.0, + "step": 10134 + }, + { + "epoch": 1.1130024159894576, + "grad_norm": 2.1725575923919678, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7505838871002197, + "num_tokens": 262340817.0, + "step": 10135 + }, + { + "epoch": 1.1131122336920711, + "grad_norm": 2.0623159408569336, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7228398323059082, + "num_tokens": 262369774.0, + "step": 10136 + }, + { + "epoch": 1.1132220513946849, + "grad_norm": 2.182476282119751, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7071799039840698, + "num_tokens": 262395648.0, + "step": 10137 + }, + { + "epoch": 1.1133318690972984, + "grad_norm": 2.151135206222534, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7193256616592407, + "num_tokens": 262421124.0, + "step": 10138 + }, + { + "epoch": 1.1134416867999122, + "grad_norm": 2.200115919113159, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7389525771141052, + "num_tokens": 262446550.0, + "step": 10139 + }, + { + "epoch": 1.1135515045025257, + "grad_norm": 2.1710848808288574, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7347147464752197, + "num_tokens": 262471616.0, + "step": 10140 + }, + { + "epoch": 1.1136613222051395, + "grad_norm": 2.8375158309936523, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7224017381668091, + "num_tokens": 262489414.0, + "step": 10141 + }, + { + "epoch": 1.1137711399077532, + "grad_norm": 1.8192193508148193, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.703460693359375, + "num_tokens": 262526085.0, + "step": 10142 + }, + { + "epoch": 1.1138809576103668, + "grad_norm": 2.2653326988220215, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7163070440292358, + "num_tokens": 262550686.0, + "step": 10143 + }, + { + "epoch": 1.1139907753129805, + "grad_norm": 2.2450664043426514, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.703453540802002, + "num_tokens": 262575657.0, + "step": 10144 + }, + { + "epoch": 1.114100593015594, + "grad_norm": 2.2405080795288086, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6909852623939514, + "num_tokens": 262603127.0, + "step": 10145 + }, + { + "epoch": 1.1142104107182078, + "grad_norm": 2.2705092430114746, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.6996275186538696, + "num_tokens": 262628529.0, + "step": 10146 + }, + { + "epoch": 1.1143202284208213, + "grad_norm": 2.1563832759857178, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6856464147567749, + "num_tokens": 262656454.0, + "step": 10147 + }, + { + "epoch": 1.114430046123435, + "grad_norm": 2.090219497680664, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7381099462509155, + "num_tokens": 262685153.0, + "step": 10148 + }, + { + "epoch": 1.1145398638260489, + "grad_norm": 2.1082005500793457, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7026445865631104, + "num_tokens": 262713813.0, + "step": 10149 + }, + { + "epoch": 1.1146496815286624, + "grad_norm": 1.9972163438796997, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7263093590736389, + "num_tokens": 262743376.0, + "step": 10150 + }, + { + "epoch": 1.1147594992312762, + "grad_norm": 2.137935161590576, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7393653392791748, + "num_tokens": 262770450.0, + "step": 10151 + }, + { + "epoch": 1.1148693169338897, + "grad_norm": 2.3090386390686035, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6921570897102356, + "num_tokens": 262795402.0, + "step": 10152 + }, + { + "epoch": 1.1149791346365034, + "grad_norm": 2.4523673057556152, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7005618810653687, + "num_tokens": 262818722.0, + "step": 10153 + }, + { + "epoch": 1.115088952339117, + "grad_norm": 2.3002519607543945, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7400716543197632, + "num_tokens": 262841738.0, + "step": 10154 + }, + { + "epoch": 1.1151987700417307, + "grad_norm": 2.2221975326538086, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.744248628616333, + "num_tokens": 262865964.0, + "step": 10155 + }, + { + "epoch": 1.1153085877443445, + "grad_norm": 2.741230010986328, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7044878005981445, + "num_tokens": 262885839.0, + "step": 10156 + }, + { + "epoch": 1.115418405446958, + "grad_norm": 2.3697140216827393, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7233223915100098, + "num_tokens": 262908507.0, + "step": 10157 + }, + { + "epoch": 1.1155282231495718, + "grad_norm": 2.448225975036621, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7219899892807007, + "num_tokens": 262931525.0, + "step": 10158 + }, + { + "epoch": 1.1156380408521853, + "grad_norm": 2.50120210647583, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7087640762329102, + "num_tokens": 262954414.0, + "step": 10159 + }, + { + "epoch": 1.115747858554799, + "grad_norm": 2.3517184257507324, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7074346542358398, + "num_tokens": 262976802.0, + "step": 10160 + }, + { + "epoch": 1.1158576762574126, + "grad_norm": 2.090057849884033, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.719470739364624, + "num_tokens": 263006436.0, + "step": 10161 + }, + { + "epoch": 1.1159674939600264, + "grad_norm": 2.1694507598876953, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7033496499061584, + "num_tokens": 263033693.0, + "step": 10162 + }, + { + "epoch": 1.11607731166264, + "grad_norm": 2.5002286434173584, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.720422625541687, + "num_tokens": 263055731.0, + "step": 10163 + }, + { + "epoch": 1.1161871293652537, + "grad_norm": 2.2337422370910645, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7099449634552002, + "num_tokens": 263081564.0, + "step": 10164 + }, + { + "epoch": 1.1162969470678674, + "grad_norm": 2.172868490219116, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7215806841850281, + "num_tokens": 263110244.0, + "step": 10165 + }, + { + "epoch": 1.116406764770481, + "grad_norm": 2.3008034229278564, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7038072347640991, + "num_tokens": 263135014.0, + "step": 10166 + }, + { + "epoch": 1.1165165824730947, + "grad_norm": 2.520838975906372, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7020677924156189, + "num_tokens": 263156727.0, + "step": 10167 + }, + { + "epoch": 1.1166264001757082, + "grad_norm": 2.3102805614471436, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.722148060798645, + "num_tokens": 263183522.0, + "step": 10168 + }, + { + "epoch": 1.116736217878322, + "grad_norm": 2.272956371307373, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7345752120018005, + "num_tokens": 263208367.0, + "step": 10169 + }, + { + "epoch": 1.1168460355809358, + "grad_norm": 2.540581703186035, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7531890869140625, + "num_tokens": 263227569.0, + "step": 10170 + }, + { + "epoch": 1.1169558532835493, + "grad_norm": 2.087439775466919, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7227866649627686, + "num_tokens": 263255747.0, + "step": 10171 + }, + { + "epoch": 1.117065670986163, + "grad_norm": 2.4834377765655518, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7216212153434753, + "num_tokens": 263277067.0, + "step": 10172 + }, + { + "epoch": 1.1171754886887766, + "grad_norm": 2.2986252307891846, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7257153987884521, + "num_tokens": 263300135.0, + "step": 10173 + }, + { + "epoch": 1.1172853063913903, + "grad_norm": 2.7308385372161865, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7341112494468689, + "num_tokens": 263318790.0, + "step": 10174 + }, + { + "epoch": 1.1173951240940039, + "grad_norm": 2.1021647453308105, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7149213552474976, + "num_tokens": 263348357.0, + "step": 10175 + }, + { + "epoch": 1.1175049417966176, + "grad_norm": 2.241973876953125, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7355460524559021, + "num_tokens": 263372010.0, + "step": 10176 + }, + { + "epoch": 1.1176147594992312, + "grad_norm": 2.268376111984253, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7137550711631775, + "num_tokens": 263397378.0, + "step": 10177 + }, + { + "epoch": 1.117724577201845, + "grad_norm": 2.162148952484131, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7223672866821289, + "num_tokens": 263422840.0, + "step": 10178 + }, + { + "epoch": 1.1178343949044587, + "grad_norm": 2.3000471591949463, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7243975400924683, + "num_tokens": 263447140.0, + "step": 10179 + }, + { + "epoch": 1.1179442126070722, + "grad_norm": 2.0483412742614746, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.703561544418335, + "num_tokens": 263477950.0, + "step": 10180 + }, + { + "epoch": 1.118054030309686, + "grad_norm": 1.959225058555603, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6984901428222656, + "num_tokens": 263511333.0, + "step": 10181 + }, + { + "epoch": 1.1181638480122995, + "grad_norm": 2.0815975666046143, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7120670676231384, + "num_tokens": 263539008.0, + "step": 10182 + }, + { + "epoch": 1.1182736657149133, + "grad_norm": 2.4676918983459473, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7290410399436951, + "num_tokens": 263559854.0, + "step": 10183 + }, + { + "epoch": 1.1183834834175268, + "grad_norm": 2.3240647315979004, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7222795486450195, + "num_tokens": 263583069.0, + "step": 10184 + }, + { + "epoch": 1.1184933011201406, + "grad_norm": 2.586982488632202, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7549542188644409, + "num_tokens": 263603230.0, + "step": 10185 + }, + { + "epoch": 1.1186031188227543, + "grad_norm": 2.070483922958374, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7106202244758606, + "num_tokens": 263632745.0, + "step": 10186 + }, + { + "epoch": 1.1187129365253679, + "grad_norm": 2.367839813232422, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7097936868667603, + "num_tokens": 263656588.0, + "step": 10187 + }, + { + "epoch": 1.1188227542279816, + "grad_norm": 2.433198928833008, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7295822501182556, + "num_tokens": 263677098.0, + "step": 10188 + }, + { + "epoch": 1.1189325719305951, + "grad_norm": 2.3252804279327393, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7044470310211182, + "num_tokens": 263700576.0, + "step": 10189 + }, + { + "epoch": 1.119042389633209, + "grad_norm": 2.0728695392608643, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7151301503181458, + "num_tokens": 263728753.0, + "step": 10190 + }, + { + "epoch": 1.1191522073358224, + "grad_norm": 2.071225881576538, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7129033803939819, + "num_tokens": 263754501.0, + "step": 10191 + }, + { + "epoch": 1.1192620250384362, + "grad_norm": 2.114406108856201, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7081880569458008, + "num_tokens": 263781638.0, + "step": 10192 + }, + { + "epoch": 1.11937184274105, + "grad_norm": 2.3201403617858887, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7038102149963379, + "num_tokens": 263806214.0, + "step": 10193 + }, + { + "epoch": 1.1194816604436635, + "grad_norm": 2.1278791427612305, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.711806058883667, + "num_tokens": 263834287.0, + "step": 10194 + }, + { + "epoch": 1.1195914781462772, + "grad_norm": 2.196974277496338, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7161250114440918, + "num_tokens": 263861180.0, + "step": 10195 + }, + { + "epoch": 1.1197012958488908, + "grad_norm": 2.449678659439087, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.730466365814209, + "num_tokens": 263883046.0, + "step": 10196 + }, + { + "epoch": 1.1198111135515045, + "grad_norm": 2.173362970352173, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7247157096862793, + "num_tokens": 263910091.0, + "step": 10197 + }, + { + "epoch": 1.119920931254118, + "grad_norm": 2.130629777908325, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7290767431259155, + "num_tokens": 263937454.0, + "step": 10198 + }, + { + "epoch": 1.1200307489567318, + "grad_norm": 2.1962342262268066, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7212602496147156, + "num_tokens": 263963185.0, + "step": 10199 + }, + { + "epoch": 1.1201405666593456, + "grad_norm": 2.4990386962890625, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7294849157333374, + "num_tokens": 263985334.0, + "step": 10200 + }, + { + "epoch": 1.1202503843619591, + "grad_norm": 2.182060718536377, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7017897963523865, + "num_tokens": 264013050.0, + "step": 10201 + }, + { + "epoch": 1.1203602020645729, + "grad_norm": 2.2538669109344482, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7061853408813477, + "num_tokens": 264040065.0, + "step": 10202 + }, + { + "epoch": 1.1204700197671864, + "grad_norm": 2.159237861633301, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7068054676055908, + "num_tokens": 264067824.0, + "step": 10203 + }, + { + "epoch": 1.1205798374698002, + "grad_norm": 2.1817996501922607, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7029151916503906, + "num_tokens": 264094078.0, + "step": 10204 + }, + { + "epoch": 1.1206896551724137, + "grad_norm": 2.070535898208618, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7074844241142273, + "num_tokens": 264122206.0, + "step": 10205 + }, + { + "epoch": 1.1207994728750275, + "grad_norm": 2.3333799839019775, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7039964199066162, + "num_tokens": 264146490.0, + "step": 10206 + }, + { + "epoch": 1.1209092905776412, + "grad_norm": 2.9315764904022217, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7439562082290649, + "num_tokens": 264162769.0, + "step": 10207 + }, + { + "epoch": 1.1210191082802548, + "grad_norm": 2.316267490386963, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7242705821990967, + "num_tokens": 264185118.0, + "step": 10208 + }, + { + "epoch": 1.1211289259828685, + "grad_norm": 2.054398775100708, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7128883600234985, + "num_tokens": 264217125.0, + "step": 10209 + }, + { + "epoch": 1.121238743685482, + "grad_norm": 2.1503591537475586, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7302101850509644, + "num_tokens": 264243982.0, + "step": 10210 + }, + { + "epoch": 1.1213485613880958, + "grad_norm": 2.347566843032837, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7041328549385071, + "num_tokens": 264268132.0, + "step": 10211 + }, + { + "epoch": 1.1214583790907093, + "grad_norm": 2.3347067832946777, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7221957445144653, + "num_tokens": 264290913.0, + "step": 10212 + }, + { + "epoch": 1.121568196793323, + "grad_norm": 2.043261766433716, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7331609725952148, + "num_tokens": 264317797.0, + "step": 10213 + }, + { + "epoch": 1.1216780144959368, + "grad_norm": 2.3358073234558105, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7123082876205444, + "num_tokens": 264343062.0, + "step": 10214 + }, + { + "epoch": 1.1217878321985504, + "grad_norm": 2.3678395748138428, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7137077450752258, + "num_tokens": 264365747.0, + "step": 10215 + }, + { + "epoch": 1.1218976499011641, + "grad_norm": 2.3386240005493164, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7079948782920837, + "num_tokens": 264389155.0, + "step": 10216 + }, + { + "epoch": 1.1220074676037777, + "grad_norm": 2.2018113136291504, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7632421851158142, + "num_tokens": 264412724.0, + "step": 10217 + }, + { + "epoch": 1.1221172853063914, + "grad_norm": 2.294116735458374, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7039262652397156, + "num_tokens": 264437737.0, + "step": 10218 + }, + { + "epoch": 1.122227103009005, + "grad_norm": 2.3980064392089844, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.6988006234169006, + "num_tokens": 264464141.0, + "step": 10219 + }, + { + "epoch": 1.1223369207116187, + "grad_norm": 2.2726433277130127, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7108798027038574, + "num_tokens": 264489951.0, + "step": 10220 + }, + { + "epoch": 1.1224467384142325, + "grad_norm": 1.9510080814361572, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7018482685089111, + "num_tokens": 264522420.0, + "step": 10221 + }, + { + "epoch": 1.122556556116846, + "grad_norm": 2.143568992614746, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7111645936965942, + "num_tokens": 264550110.0, + "step": 10222 + }, + { + "epoch": 1.1226663738194598, + "grad_norm": 2.16908860206604, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7039086222648621, + "num_tokens": 264577054.0, + "step": 10223 + }, + { + "epoch": 1.1227761915220733, + "grad_norm": 2.283179998397827, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7151280641555786, + "num_tokens": 264602938.0, + "step": 10224 + }, + { + "epoch": 1.122886009224687, + "grad_norm": 2.2228760719299316, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7178559303283691, + "num_tokens": 264629059.0, + "step": 10225 + }, + { + "epoch": 1.1229958269273006, + "grad_norm": 2.245809555053711, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7283837199211121, + "num_tokens": 264655928.0, + "step": 10226 + }, + { + "epoch": 1.1231056446299144, + "grad_norm": 2.2299230098724365, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7294970154762268, + "num_tokens": 264681093.0, + "step": 10227 + }, + { + "epoch": 1.123215462332528, + "grad_norm": 2.2523739337921143, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7285726070404053, + "num_tokens": 264705936.0, + "step": 10228 + }, + { + "epoch": 1.1233252800351416, + "grad_norm": 2.1900830268859863, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7130327224731445, + "num_tokens": 264731489.0, + "step": 10229 + }, + { + "epoch": 1.1234350977377554, + "grad_norm": 2.3422751426696777, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7390931248664856, + "num_tokens": 264753490.0, + "step": 10230 + }, + { + "epoch": 1.123544915440369, + "grad_norm": 2.3139302730560303, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7147316932678223, + "num_tokens": 264779633.0, + "step": 10231 + }, + { + "epoch": 1.1236547331429827, + "grad_norm": 2.2351813316345215, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7067369222640991, + "num_tokens": 264804496.0, + "step": 10232 + }, + { + "epoch": 1.1237645508455962, + "grad_norm": 2.2080917358398438, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6914132237434387, + "num_tokens": 264832883.0, + "step": 10233 + }, + { + "epoch": 1.12387436854821, + "grad_norm": 2.225234031677246, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7319012880325317, + "num_tokens": 264858624.0, + "step": 10234 + }, + { + "epoch": 1.1239841862508237, + "grad_norm": 2.619898796081543, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7127476930618286, + "num_tokens": 264880895.0, + "step": 10235 + }, + { + "epoch": 1.1240940039534373, + "grad_norm": 2.2567458152770996, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7004515528678894, + "num_tokens": 264907135.0, + "step": 10236 + }, + { + "epoch": 1.124203821656051, + "grad_norm": 2.1951780319213867, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7122640609741211, + "num_tokens": 264934646.0, + "step": 10237 + }, + { + "epoch": 1.1243136393586646, + "grad_norm": 2.009899854660034, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7206279039382935, + "num_tokens": 264963994.0, + "step": 10238 + }, + { + "epoch": 1.1244234570612783, + "grad_norm": 2.3259549140930176, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7238938808441162, + "num_tokens": 264988809.0, + "step": 10239 + }, + { + "epoch": 1.1245332747638919, + "grad_norm": 2.306330919265747, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7274060845375061, + "num_tokens": 265011918.0, + "step": 10240 + }, + { + "epoch": 1.1246430924665056, + "grad_norm": 2.1146273612976074, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7135096788406372, + "num_tokens": 265040220.0, + "step": 10241 + }, + { + "epoch": 1.1247529101691192, + "grad_norm": 2.1764492988586426, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7332921028137207, + "num_tokens": 265066840.0, + "step": 10242 + }, + { + "epoch": 1.124862727871733, + "grad_norm": 2.41662335395813, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7294766902923584, + "num_tokens": 265090591.0, + "step": 10243 + }, + { + "epoch": 1.1249725455743467, + "grad_norm": 2.2210123538970947, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7094568610191345, + "num_tokens": 265117912.0, + "step": 10244 + }, + { + "epoch": 1.1250823632769602, + "grad_norm": 2.2857866287231445, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7136526703834534, + "num_tokens": 265143335.0, + "step": 10245 + }, + { + "epoch": 1.125192180979574, + "grad_norm": 2.346130132675171, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6954278945922852, + "num_tokens": 265168875.0, + "step": 10246 + }, + { + "epoch": 1.1253019986821875, + "grad_norm": 2.2499027252197266, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6963300704956055, + "num_tokens": 265195136.0, + "step": 10247 + }, + { + "epoch": 1.1254118163848013, + "grad_norm": 2.061022996902466, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7124061584472656, + "num_tokens": 265222875.0, + "step": 10248 + }, + { + "epoch": 1.125521634087415, + "grad_norm": 2.57285737991333, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7231087684631348, + "num_tokens": 265242173.0, + "step": 10249 + }, + { + "epoch": 1.1256314517900285, + "grad_norm": 2.310101270675659, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7367822527885437, + "num_tokens": 265265667.0, + "step": 10250 + }, + { + "epoch": 1.1257412694926423, + "grad_norm": 2.196812152862549, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6939005255699158, + "num_tokens": 265295514.0, + "step": 10251 + }, + { + "epoch": 1.1258510871952558, + "grad_norm": 2.4364442825317383, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7386287450790405, + "num_tokens": 265317261.0, + "step": 10252 + }, + { + "epoch": 1.1259609048978696, + "grad_norm": 2.309248924255371, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.704008162021637, + "num_tokens": 265342770.0, + "step": 10253 + }, + { + "epoch": 1.1260707226004831, + "grad_norm": 2.546297550201416, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7193231582641602, + "num_tokens": 265363346.0, + "step": 10254 + }, + { + "epoch": 1.1261805403030969, + "grad_norm": 2.132610321044922, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7141042351722717, + "num_tokens": 265391417.0, + "step": 10255 + }, + { + "epoch": 1.1262903580057104, + "grad_norm": 2.54797625541687, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7445304989814758, + "num_tokens": 265412658.0, + "step": 10256 + }, + { + "epoch": 1.1264001757083242, + "grad_norm": 2.112443208694458, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7240896224975586, + "num_tokens": 265439582.0, + "step": 10257 + }, + { + "epoch": 1.126509993410938, + "grad_norm": 2.222771167755127, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7170095443725586, + "num_tokens": 265466946.0, + "step": 10258 + }, + { + "epoch": 1.1266198111135515, + "grad_norm": 2.0187771320343018, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7140031456947327, + "num_tokens": 265499854.0, + "step": 10259 + }, + { + "epoch": 1.1267296288161652, + "grad_norm": 2.446727991104126, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7304908037185669, + "num_tokens": 265523938.0, + "step": 10260 + }, + { + "epoch": 1.1268394465187788, + "grad_norm": 1.8506436347961426, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.718721866607666, + "num_tokens": 265560468.0, + "step": 10261 + }, + { + "epoch": 1.1269492642213925, + "grad_norm": 2.438189744949341, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7087571620941162, + "num_tokens": 265583667.0, + "step": 10262 + }, + { + "epoch": 1.127059081924006, + "grad_norm": 2.2627246379852295, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7382957935333252, + "num_tokens": 265606595.0, + "step": 10263 + }, + { + "epoch": 1.1271688996266198, + "grad_norm": 2.1969187259674072, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7092442512512207, + "num_tokens": 265633660.0, + "step": 10264 + }, + { + "epoch": 1.1272787173292333, + "grad_norm": 2.1644692420959473, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7102212309837341, + "num_tokens": 265659487.0, + "step": 10265 + }, + { + "epoch": 1.127388535031847, + "grad_norm": 2.1488797664642334, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7089242935180664, + "num_tokens": 265688273.0, + "step": 10266 + }, + { + "epoch": 1.1274983527344609, + "grad_norm": 2.24690842628479, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.698430061340332, + "num_tokens": 265713972.0, + "step": 10267 + }, + { + "epoch": 1.1276081704370744, + "grad_norm": 2.312028646469116, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6971737742424011, + "num_tokens": 265739755.0, + "step": 10268 + }, + { + "epoch": 1.1277179881396882, + "grad_norm": 2.427476167678833, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6970085501670837, + "num_tokens": 265764051.0, + "step": 10269 + }, + { + "epoch": 1.1278278058423017, + "grad_norm": 2.364936351776123, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7129625678062439, + "num_tokens": 265789653.0, + "step": 10270 + }, + { + "epoch": 1.1279376235449154, + "grad_norm": 2.1158394813537598, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7292002439498901, + "num_tokens": 265816462.0, + "step": 10271 + }, + { + "epoch": 1.1280474412475292, + "grad_norm": 2.41711163520813, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7362810373306274, + "num_tokens": 265840306.0, + "step": 10272 + }, + { + "epoch": 1.1281572589501427, + "grad_norm": 2.00457501411438, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7292326092720032, + "num_tokens": 265870347.0, + "step": 10273 + }, + { + "epoch": 1.1282670766527565, + "grad_norm": 2.507321357727051, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7068890333175659, + "num_tokens": 265894363.0, + "step": 10274 + }, + { + "epoch": 1.12837689435537, + "grad_norm": 2.254701852798462, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.716726541519165, + "num_tokens": 265921549.0, + "step": 10275 + }, + { + "epoch": 1.1284867120579838, + "grad_norm": 2.1839776039123535, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7246443033218384, + "num_tokens": 265948626.0, + "step": 10276 + }, + { + "epoch": 1.1285965297605973, + "grad_norm": 1.930516242980957, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7087503671646118, + "num_tokens": 265983453.0, + "step": 10277 + }, + { + "epoch": 1.128706347463211, + "grad_norm": 2.4345386028289795, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.731594443321228, + "num_tokens": 266005444.0, + "step": 10278 + }, + { + "epoch": 1.1288161651658246, + "grad_norm": 2.208975076675415, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7367634773254395, + "num_tokens": 266031241.0, + "step": 10279 + }, + { + "epoch": 1.1289259828684384, + "grad_norm": 2.493595600128174, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7266473174095154, + "num_tokens": 266053070.0, + "step": 10280 + }, + { + "epoch": 1.1290358005710521, + "grad_norm": 2.3353371620178223, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7034575939178467, + "num_tokens": 266076940.0, + "step": 10281 + }, + { + "epoch": 1.1291456182736657, + "grad_norm": 1.966942310333252, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7041100263595581, + "num_tokens": 266111542.0, + "step": 10282 + }, + { + "epoch": 1.1292554359762794, + "grad_norm": 2.0530238151550293, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7477478981018066, + "num_tokens": 266137534.0, + "step": 10283 + }, + { + "epoch": 1.129365253678893, + "grad_norm": 2.249042510986328, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7328994870185852, + "num_tokens": 266161658.0, + "step": 10284 + }, + { + "epoch": 1.1294750713815067, + "grad_norm": 2.4150516986846924, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7180458307266235, + "num_tokens": 266184514.0, + "step": 10285 + }, + { + "epoch": 1.1295848890841205, + "grad_norm": 2.2385263442993164, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6937564611434937, + "num_tokens": 266214113.0, + "step": 10286 + }, + { + "epoch": 1.129694706786734, + "grad_norm": 2.4797515869140625, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7283792495727539, + "num_tokens": 266234590.0, + "step": 10287 + }, + { + "epoch": 1.1298045244893478, + "grad_norm": 2.0921401977539062, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7358343601226807, + "num_tokens": 266262165.0, + "step": 10288 + }, + { + "epoch": 1.1299143421919613, + "grad_norm": 2.2258386611938477, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7276111245155334, + "num_tokens": 266287184.0, + "step": 10289 + }, + { + "epoch": 1.130024159894575, + "grad_norm": 2.7577364444732666, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7296836972236633, + "num_tokens": 266305758.0, + "step": 10290 + }, + { + "epoch": 1.1301339775971886, + "grad_norm": 2.114044427871704, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7219725251197815, + "num_tokens": 266335294.0, + "step": 10291 + }, + { + "epoch": 1.1302437952998023, + "grad_norm": 2.070211172103882, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7139438390731812, + "num_tokens": 266363544.0, + "step": 10292 + }, + { + "epoch": 1.1303536130024159, + "grad_norm": 2.315999746322632, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6889524459838867, + "num_tokens": 266389150.0, + "step": 10293 + }, + { + "epoch": 1.1304634307050296, + "grad_norm": 2.001849889755249, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6979342699050903, + "num_tokens": 266418930.0, + "step": 10294 + }, + { + "epoch": 1.1305732484076434, + "grad_norm": 2.2781429290771484, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6973998546600342, + "num_tokens": 266445519.0, + "step": 10295 + }, + { + "epoch": 1.130683066110257, + "grad_norm": 2.7693607807159424, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7230228185653687, + "num_tokens": 266465326.0, + "step": 10296 + }, + { + "epoch": 1.1307928838128707, + "grad_norm": 2.0066750049591064, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7095487117767334, + "num_tokens": 266498034.0, + "step": 10297 + }, + { + "epoch": 1.1309027015154842, + "grad_norm": 2.3669092655181885, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7091517448425293, + "num_tokens": 266523244.0, + "step": 10298 + }, + { + "epoch": 1.131012519218098, + "grad_norm": 2.011077880859375, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.6985452771186829, + "num_tokens": 266556348.0, + "step": 10299 + }, + { + "epoch": 1.1311223369207117, + "grad_norm": 2.1724777221679688, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7111467123031616, + "num_tokens": 266585163.0, + "step": 10300 + }, + { + "epoch": 1.1312321546233253, + "grad_norm": 2.3516972064971924, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7361367344856262, + "num_tokens": 266608597.0, + "step": 10301 + }, + { + "epoch": 1.131341972325939, + "grad_norm": 2.4946467876434326, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7267694473266602, + "num_tokens": 266631533.0, + "step": 10302 + }, + { + "epoch": 1.1314517900285526, + "grad_norm": 2.2833666801452637, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6856895685195923, + "num_tokens": 266662321.0, + "step": 10303 + }, + { + "epoch": 1.1315616077311663, + "grad_norm": 2.740029811859131, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7194368839263916, + "num_tokens": 266681531.0, + "step": 10304 + }, + { + "epoch": 1.1316714254337799, + "grad_norm": 2.098578929901123, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7291818857192993, + "num_tokens": 266709453.0, + "step": 10305 + }, + { + "epoch": 1.1317812431363936, + "grad_norm": 2.207252025604248, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7309119701385498, + "num_tokens": 266735508.0, + "step": 10306 + }, + { + "epoch": 1.1318910608390071, + "grad_norm": 2.459834337234497, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7054576873779297, + "num_tokens": 266758362.0, + "step": 10307 + }, + { + "epoch": 1.132000878541621, + "grad_norm": 2.15889835357666, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7092729210853577, + "num_tokens": 266784633.0, + "step": 10308 + }, + { + "epoch": 1.1321106962442347, + "grad_norm": 2.2208402156829834, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7148553133010864, + "num_tokens": 266811149.0, + "step": 10309 + }, + { + "epoch": 1.1322205139468482, + "grad_norm": 2.7498085498809814, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7326735258102417, + "num_tokens": 266829030.0, + "step": 10310 + }, + { + "epoch": 1.132330331649462, + "grad_norm": 2.0704703330993652, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7223692536354065, + "num_tokens": 266859845.0, + "step": 10311 + }, + { + "epoch": 1.1324401493520755, + "grad_norm": 2.2379517555236816, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7169748544692993, + "num_tokens": 266884529.0, + "step": 10312 + }, + { + "epoch": 1.1325499670546892, + "grad_norm": 2.23332142829895, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7056968808174133, + "num_tokens": 266910291.0, + "step": 10313 + }, + { + "epoch": 1.132659784757303, + "grad_norm": 2.1540334224700928, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7410200834274292, + "num_tokens": 266937059.0, + "step": 10314 + }, + { + "epoch": 1.1327696024599165, + "grad_norm": 2.2622666358947754, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.7003037929534912, + "num_tokens": 266964277.0, + "step": 10315 + }, + { + "epoch": 1.1328794201625303, + "grad_norm": 2.0056004524230957, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7073000073432922, + "num_tokens": 266996463.0, + "step": 10316 + }, + { + "epoch": 1.1329892378651438, + "grad_norm": 2.4606070518493652, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7148790955543518, + "num_tokens": 267019702.0, + "step": 10317 + }, + { + "epoch": 1.1330990555677576, + "grad_norm": 2.447369337081909, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7302724123001099, + "num_tokens": 267040614.0, + "step": 10318 + }, + { + "epoch": 1.1332088732703711, + "grad_norm": 2.2318596839904785, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7208472490310669, + "num_tokens": 267067908.0, + "step": 10319 + }, + { + "epoch": 1.1333186909729849, + "grad_norm": 2.2254538536071777, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7143808603286743, + "num_tokens": 267093389.0, + "step": 10320 + }, + { + "epoch": 1.1334285086755984, + "grad_norm": 2.0273897647857666, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7125146389007568, + "num_tokens": 267123727.0, + "step": 10321 + }, + { + "epoch": 1.1335383263782122, + "grad_norm": 2.0062968730926514, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7051522731781006, + "num_tokens": 267153359.0, + "step": 10322 + }, + { + "epoch": 1.133648144080826, + "grad_norm": 2.124466896057129, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7197659015655518, + "num_tokens": 267180260.0, + "step": 10323 + }, + { + "epoch": 1.1337579617834395, + "grad_norm": 2.029615879058838, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6875743865966797, + "num_tokens": 267214943.0, + "step": 10324 + }, + { + "epoch": 1.1338677794860532, + "grad_norm": 2.372177839279175, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7125312089920044, + "num_tokens": 267238721.0, + "step": 10325 + }, + { + "epoch": 1.1339775971886668, + "grad_norm": 2.068960428237915, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7196798324584961, + "num_tokens": 267269105.0, + "step": 10326 + }, + { + "epoch": 1.1340874148912805, + "grad_norm": 2.2387540340423584, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7110491991043091, + "num_tokens": 267295469.0, + "step": 10327 + }, + { + "epoch": 1.134197232593894, + "grad_norm": 2.0671494007110596, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7057689428329468, + "num_tokens": 267323179.0, + "step": 10328 + }, + { + "epoch": 1.1343070502965078, + "grad_norm": 2.458766460418701, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7213162779808044, + "num_tokens": 267347769.0, + "step": 10329 + }, + { + "epoch": 1.1344168679991213, + "grad_norm": 2.2024145126342773, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7162413597106934, + "num_tokens": 267373815.0, + "step": 10330 + }, + { + "epoch": 1.134526685701735, + "grad_norm": 2.3800153732299805, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7182217836380005, + "num_tokens": 267396441.0, + "step": 10331 + }, + { + "epoch": 1.1346365034043489, + "grad_norm": 2.2225821018218994, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7202695608139038, + "num_tokens": 267422653.0, + "step": 10332 + }, + { + "epoch": 1.1347463211069624, + "grad_norm": 2.460970401763916, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7266153693199158, + "num_tokens": 267445598.0, + "step": 10333 + }, + { + "epoch": 1.1348561388095761, + "grad_norm": 2.1885082721710205, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7298043966293335, + "num_tokens": 267471406.0, + "step": 10334 + }, + { + "epoch": 1.1349659565121897, + "grad_norm": 2.411898612976074, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7153692245483398, + "num_tokens": 267494400.0, + "step": 10335 + }, + { + "epoch": 1.1350757742148034, + "grad_norm": 2.1102118492126465, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7294413447380066, + "num_tokens": 267524579.0, + "step": 10336 + }, + { + "epoch": 1.1351855919174172, + "grad_norm": 1.9585833549499512, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7352038621902466, + "num_tokens": 267555767.0, + "step": 10337 + }, + { + "epoch": 1.1352954096200307, + "grad_norm": 1.8969624042510986, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7133914232254028, + "num_tokens": 267589006.0, + "step": 10338 + }, + { + "epoch": 1.1354052273226445, + "grad_norm": 2.355766534805298, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7247358560562134, + "num_tokens": 267611517.0, + "step": 10339 + }, + { + "epoch": 1.135515045025258, + "grad_norm": 2.018234968185425, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6994457840919495, + "num_tokens": 267644383.0, + "step": 10340 + }, + { + "epoch": 1.1356248627278718, + "grad_norm": 2.3143258094787598, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.704098641872406, + "num_tokens": 267667731.0, + "step": 10341 + }, + { + "epoch": 1.1357346804304853, + "grad_norm": 2.384901285171509, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7103223204612732, + "num_tokens": 267692049.0, + "step": 10342 + }, + { + "epoch": 1.135844498133099, + "grad_norm": 2.180795669555664, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7268334031105042, + "num_tokens": 267719834.0, + "step": 10343 + }, + { + "epoch": 1.1359543158357126, + "grad_norm": 2.643292188644409, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7344143390655518, + "num_tokens": 267741202.0, + "step": 10344 + }, + { + "epoch": 1.1360641335383264, + "grad_norm": 2.105876922607422, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6984347701072693, + "num_tokens": 267771467.0, + "step": 10345 + }, + { + "epoch": 1.1361739512409401, + "grad_norm": 2.4576377868652344, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7306082248687744, + "num_tokens": 267792675.0, + "step": 10346 + }, + { + "epoch": 1.1362837689435537, + "grad_norm": 2.110203504562378, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7036905288696289, + "num_tokens": 267821926.0, + "step": 10347 + }, + { + "epoch": 1.1363935866461674, + "grad_norm": 2.0423240661621094, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7122685313224792, + "num_tokens": 267852223.0, + "step": 10348 + }, + { + "epoch": 1.136503404348781, + "grad_norm": 2.2485299110412598, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7138763666152954, + "num_tokens": 267876621.0, + "step": 10349 + }, + { + "epoch": 1.1366132220513947, + "grad_norm": 1.9538503885269165, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7135422825813293, + "num_tokens": 267908585.0, + "step": 10350 + }, + { + "epoch": 1.1367230397540085, + "grad_norm": 2.0915138721466064, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7257786989212036, + "num_tokens": 267940588.0, + "step": 10351 + }, + { + "epoch": 1.136832857456622, + "grad_norm": 2.3842554092407227, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7206904888153076, + "num_tokens": 267965779.0, + "step": 10352 + }, + { + "epoch": 1.1369426751592357, + "grad_norm": 2.272757053375244, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6949429512023926, + "num_tokens": 267992571.0, + "step": 10353 + }, + { + "epoch": 1.1370524928618493, + "grad_norm": 2.2282607555389404, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7135708332061768, + "num_tokens": 268019872.0, + "step": 10354 + }, + { + "epoch": 1.137162310564463, + "grad_norm": 2.344357967376709, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7242063879966736, + "num_tokens": 268043544.0, + "step": 10355 + }, + { + "epoch": 1.1372721282670766, + "grad_norm": 2.175117015838623, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7131270170211792, + "num_tokens": 268069948.0, + "step": 10356 + }, + { + "epoch": 1.1373819459696903, + "grad_norm": 2.625809669494629, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7135469913482666, + "num_tokens": 268089960.0, + "step": 10357 + }, + { + "epoch": 1.1374917636723039, + "grad_norm": 2.4296181201934814, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7326676249504089, + "num_tokens": 268111988.0, + "step": 10358 + }, + { + "epoch": 1.1376015813749176, + "grad_norm": 2.229015827178955, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7287280559539795, + "num_tokens": 268137247.0, + "step": 10359 + }, + { + "epoch": 1.1377113990775314, + "grad_norm": 2.311368942260742, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7463390827178955, + "num_tokens": 268161393.0, + "step": 10360 + }, + { + "epoch": 1.137821216780145, + "grad_norm": 2.164036273956299, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7097817659378052, + "num_tokens": 268190923.0, + "step": 10361 + }, + { + "epoch": 1.1379310344827587, + "grad_norm": 2.4219491481781006, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7043466567993164, + "num_tokens": 268214444.0, + "step": 10362 + }, + { + "epoch": 1.1380408521853722, + "grad_norm": 2.431238889694214, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7192617654800415, + "num_tokens": 268236428.0, + "step": 10363 + }, + { + "epoch": 1.138150669887986, + "grad_norm": 2.48270845413208, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.746306300163269, + "num_tokens": 268258604.0, + "step": 10364 + }, + { + "epoch": 1.1382604875905997, + "grad_norm": 2.171294927597046, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.752408504486084, + "num_tokens": 268285344.0, + "step": 10365 + }, + { + "epoch": 1.1383703052932133, + "grad_norm": 1.8986046314239502, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7066516280174255, + "num_tokens": 268320968.0, + "step": 10366 + }, + { + "epoch": 1.138480122995827, + "grad_norm": 2.287763833999634, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7309765815734863, + "num_tokens": 268347096.0, + "step": 10367 + }, + { + "epoch": 1.1385899406984406, + "grad_norm": 2.2799811363220215, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7258486747741699, + "num_tokens": 268373691.0, + "step": 10368 + }, + { + "epoch": 1.1386997584010543, + "grad_norm": 2.2788193225860596, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7047375440597534, + "num_tokens": 268398350.0, + "step": 10369 + }, + { + "epoch": 1.1388095761036678, + "grad_norm": 2.365577459335327, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7257099151611328, + "num_tokens": 268422393.0, + "step": 10370 + }, + { + "epoch": 1.1389193938062816, + "grad_norm": 2.197817087173462, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.742997407913208, + "num_tokens": 268447765.0, + "step": 10371 + }, + { + "epoch": 1.1390292115088951, + "grad_norm": 2.0942342281341553, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7232259511947632, + "num_tokens": 268476834.0, + "step": 10372 + }, + { + "epoch": 1.139139029211509, + "grad_norm": 2.32962965965271, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.724651575088501, + "num_tokens": 268501276.0, + "step": 10373 + }, + { + "epoch": 1.1392488469141226, + "grad_norm": 1.968349575996399, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7236149907112122, + "num_tokens": 268533575.0, + "step": 10374 + }, + { + "epoch": 1.1393586646167362, + "grad_norm": 2.502039909362793, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.738206148147583, + "num_tokens": 268554506.0, + "step": 10375 + }, + { + "epoch": 1.13946848231935, + "grad_norm": 2.2709109783172607, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7218595743179321, + "num_tokens": 268579127.0, + "step": 10376 + }, + { + "epoch": 1.1395783000219635, + "grad_norm": 2.3956522941589355, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7104688882827759, + "num_tokens": 268602082.0, + "step": 10377 + }, + { + "epoch": 1.1396881177245772, + "grad_norm": 2.2511491775512695, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7014825344085693, + "num_tokens": 268629322.0, + "step": 10378 + }, + { + "epoch": 1.1397979354271908, + "grad_norm": 2.294384002685547, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7401279211044312, + "num_tokens": 268654858.0, + "step": 10379 + }, + { + "epoch": 1.1399077531298045, + "grad_norm": 2.600644588470459, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7273313999176025, + "num_tokens": 268674996.0, + "step": 10380 + }, + { + "epoch": 1.1400175708324183, + "grad_norm": 2.1859211921691895, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7080574035644531, + "num_tokens": 268701396.0, + "step": 10381 + }, + { + "epoch": 1.1401273885350318, + "grad_norm": 2.4580349922180176, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7169836759567261, + "num_tokens": 268725730.0, + "step": 10382 + }, + { + "epoch": 1.1402372062376456, + "grad_norm": 2.062556505203247, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7343618869781494, + "num_tokens": 268756266.0, + "step": 10383 + }, + { + "epoch": 1.140347023940259, + "grad_norm": 2.2704222202301025, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7353249788284302, + "num_tokens": 268779905.0, + "step": 10384 + }, + { + "epoch": 1.1404568416428729, + "grad_norm": 2.1716537475585938, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7280716896057129, + "num_tokens": 268805544.0, + "step": 10385 + }, + { + "epoch": 1.1405666593454864, + "grad_norm": 2.2761356830596924, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7416828870773315, + "num_tokens": 268828418.0, + "step": 10386 + }, + { + "epoch": 1.1406764770481002, + "grad_norm": 2.682480573654175, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7191565036773682, + "num_tokens": 268846625.0, + "step": 10387 + }, + { + "epoch": 1.140786294750714, + "grad_norm": 2.1890203952789307, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.715208888053894, + "num_tokens": 268871966.0, + "step": 10388 + }, + { + "epoch": 1.1408961124533274, + "grad_norm": 2.1558024883270264, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7416167855262756, + "num_tokens": 268897336.0, + "step": 10389 + }, + { + "epoch": 1.1410059301559412, + "grad_norm": 2.2530581951141357, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7065122127532959, + "num_tokens": 268924918.0, + "step": 10390 + }, + { + "epoch": 1.1411157478585547, + "grad_norm": 2.216348886489868, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.721082329750061, + "num_tokens": 268950848.0, + "step": 10391 + }, + { + "epoch": 1.1412255655611685, + "grad_norm": 2.300109386444092, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7096691131591797, + "num_tokens": 268976844.0, + "step": 10392 + }, + { + "epoch": 1.141335383263782, + "grad_norm": 2.148861885070801, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7253789901733398, + "num_tokens": 269004861.0, + "step": 10393 + }, + { + "epoch": 1.1414452009663958, + "grad_norm": 2.457174301147461, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7286522388458252, + "num_tokens": 269025965.0, + "step": 10394 + }, + { + "epoch": 1.1415550186690093, + "grad_norm": 2.361402750015259, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7304052114486694, + "num_tokens": 269048403.0, + "step": 10395 + }, + { + "epoch": 1.141664836371623, + "grad_norm": 2.1470589637756348, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6929066181182861, + "num_tokens": 269077585.0, + "step": 10396 + }, + { + "epoch": 1.1417746540742368, + "grad_norm": 2.043739080429077, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7403262853622437, + "num_tokens": 269108294.0, + "step": 10397 + }, + { + "epoch": 1.1418844717768504, + "grad_norm": 2.212231159210205, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7125979661941528, + "num_tokens": 269135641.0, + "step": 10398 + }, + { + "epoch": 1.1419942894794641, + "grad_norm": 1.9973773956298828, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6953682899475098, + "num_tokens": 269165374.0, + "step": 10399 + }, + { + "epoch": 1.1421041071820777, + "grad_norm": 2.4235005378723145, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7149170637130737, + "num_tokens": 269187117.0, + "step": 10400 + }, + { + "epoch": 1.1422139248846914, + "grad_norm": 2.1869430541992188, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7302358150482178, + "num_tokens": 269213528.0, + "step": 10401 + }, + { + "epoch": 1.1423237425873052, + "grad_norm": 2.1765592098236084, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7209305763244629, + "num_tokens": 269237503.0, + "step": 10402 + }, + { + "epoch": 1.1424335602899187, + "grad_norm": 2.2151458263397217, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7353829145431519, + "num_tokens": 269263747.0, + "step": 10403 + }, + { + "epoch": 1.1425433779925325, + "grad_norm": 2.4234461784362793, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7173525094985962, + "num_tokens": 269286290.0, + "step": 10404 + }, + { + "epoch": 1.142653195695146, + "grad_norm": 2.455604076385498, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7411474585533142, + "num_tokens": 269306380.0, + "step": 10405 + }, + { + "epoch": 1.1427630133977598, + "grad_norm": 2.33612060546875, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7206205129623413, + "num_tokens": 269330612.0, + "step": 10406 + }, + { + "epoch": 1.1428728311003733, + "grad_norm": 2.363609790802002, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7407975196838379, + "num_tokens": 269352725.0, + "step": 10407 + }, + { + "epoch": 1.142982648802987, + "grad_norm": 1.9112062454223633, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.713793158531189, + "num_tokens": 269385926.0, + "step": 10408 + }, + { + "epoch": 1.1430924665056006, + "grad_norm": 2.311168670654297, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6984676122665405, + "num_tokens": 269411889.0, + "step": 10409 + }, + { + "epoch": 1.1432022842082143, + "grad_norm": 2.332380771636963, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7042168378829956, + "num_tokens": 269436910.0, + "step": 10410 + }, + { + "epoch": 1.143312101910828, + "grad_norm": 2.146559476852417, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.712893009185791, + "num_tokens": 269464438.0, + "step": 10411 + }, + { + "epoch": 1.1434219196134416, + "grad_norm": 2.23850154876709, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7103917002677917, + "num_tokens": 269491963.0, + "step": 10412 + }, + { + "epoch": 1.1435317373160554, + "grad_norm": 2.1103861331939697, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7116222977638245, + "num_tokens": 269521966.0, + "step": 10413 + }, + { + "epoch": 1.143641555018669, + "grad_norm": 1.965772032737732, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7136836647987366, + "num_tokens": 269554796.0, + "step": 10414 + }, + { + "epoch": 1.1437513727212827, + "grad_norm": 2.3106517791748047, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.703071117401123, + "num_tokens": 269580105.0, + "step": 10415 + }, + { + "epoch": 1.1438611904238964, + "grad_norm": 2.0906729698181152, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6888324618339539, + "num_tokens": 269608531.0, + "step": 10416 + }, + { + "epoch": 1.14397100812651, + "grad_norm": 2.102234125137329, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7063431143760681, + "num_tokens": 269638760.0, + "step": 10417 + }, + { + "epoch": 1.1440808258291237, + "grad_norm": 2.180251121520996, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6975774765014648, + "num_tokens": 269666361.0, + "step": 10418 + }, + { + "epoch": 1.1441906435317373, + "grad_norm": 2.211632251739502, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.713741660118103, + "num_tokens": 269692763.0, + "step": 10419 + }, + { + "epoch": 1.144300461234351, + "grad_norm": 1.9583145380020142, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.6991027593612671, + "num_tokens": 269727061.0, + "step": 10420 + }, + { + "epoch": 1.1444102789369646, + "grad_norm": 2.2246639728546143, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7447428703308105, + "num_tokens": 269752838.0, + "step": 10421 + }, + { + "epoch": 1.1445200966395783, + "grad_norm": 2.0190365314483643, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7069164514541626, + "num_tokens": 269783618.0, + "step": 10422 + }, + { + "epoch": 1.1446299143421919, + "grad_norm": 2.078758478164673, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.716976523399353, + "num_tokens": 269813083.0, + "step": 10423 + }, + { + "epoch": 1.1447397320448056, + "grad_norm": 2.1661312580108643, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7190560698509216, + "num_tokens": 269841125.0, + "step": 10424 + }, + { + "epoch": 1.1448495497474194, + "grad_norm": 2.289184808731079, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7420361042022705, + "num_tokens": 269865835.0, + "step": 10425 + }, + { + "epoch": 1.144959367450033, + "grad_norm": 2.3998732566833496, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7132294774055481, + "num_tokens": 269888757.0, + "step": 10426 + }, + { + "epoch": 1.1450691851526467, + "grad_norm": 2.434568166732788, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7113696336746216, + "num_tokens": 269912248.0, + "step": 10427 + }, + { + "epoch": 1.1451790028552602, + "grad_norm": 2.3327364921569824, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7387499213218689, + "num_tokens": 269936794.0, + "step": 10428 + }, + { + "epoch": 1.145288820557874, + "grad_norm": 2.1504955291748047, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7179277539253235, + "num_tokens": 269962722.0, + "step": 10429 + }, + { + "epoch": 1.1453986382604877, + "grad_norm": 2.3543636798858643, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7231646776199341, + "num_tokens": 269986040.0, + "step": 10430 + }, + { + "epoch": 1.1455084559631012, + "grad_norm": 2.0263867378234863, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7142001390457153, + "num_tokens": 270018188.0, + "step": 10431 + }, + { + "epoch": 1.145618273665715, + "grad_norm": 2.175044536590576, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7320057153701782, + "num_tokens": 270045447.0, + "step": 10432 + }, + { + "epoch": 1.1457280913683285, + "grad_norm": 2.190539836883545, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7062535881996155, + "num_tokens": 270073881.0, + "step": 10433 + }, + { + "epoch": 1.1458379090709423, + "grad_norm": 2.5943596363067627, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7245481014251709, + "num_tokens": 270097232.0, + "step": 10434 + }, + { + "epoch": 1.1459477267735558, + "grad_norm": 2.353541135787964, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7112722396850586, + "num_tokens": 270122091.0, + "step": 10435 + }, + { + "epoch": 1.1460575444761696, + "grad_norm": 2.2619802951812744, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.6989470720291138, + "num_tokens": 270147807.0, + "step": 10436 + }, + { + "epoch": 1.1461673621787831, + "grad_norm": 2.4490227699279785, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.724664568901062, + "num_tokens": 270169830.0, + "step": 10437 + }, + { + "epoch": 1.1462771798813969, + "grad_norm": 2.372563123703003, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7284219264984131, + "num_tokens": 270194328.0, + "step": 10438 + }, + { + "epoch": 1.1463869975840106, + "grad_norm": 2.166496992111206, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7179268598556519, + "num_tokens": 270219357.0, + "step": 10439 + }, + { + "epoch": 1.1464968152866242, + "grad_norm": 2.3895373344421387, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7125526666641235, + "num_tokens": 270241779.0, + "step": 10440 + }, + { + "epoch": 1.146606632989238, + "grad_norm": 2.5873920917510986, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7288773059844971, + "num_tokens": 270261733.0, + "step": 10441 + }, + { + "epoch": 1.1467164506918515, + "grad_norm": 2.4437389373779297, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7222687005996704, + "num_tokens": 270283333.0, + "step": 10442 + }, + { + "epoch": 1.1468262683944652, + "grad_norm": 2.1725149154663086, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7126635313034058, + "num_tokens": 270315335.0, + "step": 10443 + }, + { + "epoch": 1.1469360860970788, + "grad_norm": 2.111102819442749, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7331711649894714, + "num_tokens": 270342964.0, + "step": 10444 + }, + { + "epoch": 1.1470459037996925, + "grad_norm": 2.2907848358154297, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6984184980392456, + "num_tokens": 270368653.0, + "step": 10445 + }, + { + "epoch": 1.147155721502306, + "grad_norm": 2.2998294830322266, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7135089039802551, + "num_tokens": 270393362.0, + "step": 10446 + }, + { + "epoch": 1.1472655392049198, + "grad_norm": 2.4867053031921387, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7343449592590332, + "num_tokens": 270416873.0, + "step": 10447 + }, + { + "epoch": 1.1473753569075336, + "grad_norm": 2.320571184158325, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7250985503196716, + "num_tokens": 270440103.0, + "step": 10448 + }, + { + "epoch": 1.147485174610147, + "grad_norm": 2.3661763668060303, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7231317758560181, + "num_tokens": 270462076.0, + "step": 10449 + }, + { + "epoch": 1.1475949923127609, + "grad_norm": 2.180558681488037, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7051413059234619, + "num_tokens": 270488307.0, + "step": 10450 + }, + { + "epoch": 1.1477048100153744, + "grad_norm": 2.042685031890869, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7262156009674072, + "num_tokens": 270517987.0, + "step": 10451 + }, + { + "epoch": 1.1478146277179881, + "grad_norm": 2.344630718231201, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.707301139831543, + "num_tokens": 270543828.0, + "step": 10452 + }, + { + "epoch": 1.147924445420602, + "grad_norm": 2.2490034103393555, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7051949501037598, + "num_tokens": 270570706.0, + "step": 10453 + }, + { + "epoch": 1.1480342631232154, + "grad_norm": 2.227839946746826, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7387464642524719, + "num_tokens": 270596739.0, + "step": 10454 + }, + { + "epoch": 1.1481440808258292, + "grad_norm": 2.073065757751465, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7385700941085815, + "num_tokens": 270623788.0, + "step": 10455 + }, + { + "epoch": 1.1482538985284427, + "grad_norm": 2.191528558731079, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7343185544013977, + "num_tokens": 270649744.0, + "step": 10456 + }, + { + "epoch": 1.1483637162310565, + "grad_norm": 2.410231113433838, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7306690216064453, + "num_tokens": 270671889.0, + "step": 10457 + }, + { + "epoch": 1.14847353393367, + "grad_norm": 2.473900318145752, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7588282227516174, + "num_tokens": 270692169.0, + "step": 10458 + }, + { + "epoch": 1.1485833516362838, + "grad_norm": 2.221575975418091, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7302152514457703, + "num_tokens": 270716963.0, + "step": 10459 + }, + { + "epoch": 1.1486931693388973, + "grad_norm": 2.1147546768188477, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7182814478874207, + "num_tokens": 270747232.0, + "step": 10460 + }, + { + "epoch": 1.148802987041511, + "grad_norm": 2.0223779678344727, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7092271447181702, + "num_tokens": 270777490.0, + "step": 10461 + }, + { + "epoch": 1.1489128047441248, + "grad_norm": 2.3654897212982178, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7280510663986206, + "num_tokens": 270799558.0, + "step": 10462 + }, + { + "epoch": 1.1490226224467384, + "grad_norm": 2.320812940597534, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7084331512451172, + "num_tokens": 270824945.0, + "step": 10463 + }, + { + "epoch": 1.1491324401493521, + "grad_norm": 1.9505093097686768, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.703968346118927, + "num_tokens": 270859452.0, + "step": 10464 + }, + { + "epoch": 1.1492422578519657, + "grad_norm": 2.067258834838867, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7230029106140137, + "num_tokens": 270889401.0, + "step": 10465 + }, + { + "epoch": 1.1493520755545794, + "grad_norm": 2.60821533203125, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7297060489654541, + "num_tokens": 270911780.0, + "step": 10466 + }, + { + "epoch": 1.1494618932571932, + "grad_norm": 2.225691556930542, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7167804837226868, + "num_tokens": 270939541.0, + "step": 10467 + }, + { + "epoch": 1.1495717109598067, + "grad_norm": 2.310931921005249, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7190573811531067, + "num_tokens": 270961764.0, + "step": 10468 + }, + { + "epoch": 1.1496815286624205, + "grad_norm": 2.4886839389801025, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7308664321899414, + "num_tokens": 270982189.0, + "step": 10469 + }, + { + "epoch": 1.149791346365034, + "grad_norm": 2.7925896644592285, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7229511737823486, + "num_tokens": 271000362.0, + "step": 10470 + }, + { + "epoch": 1.1499011640676478, + "grad_norm": 2.3358781337738037, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7283935546875, + "num_tokens": 271023383.0, + "step": 10471 + }, + { + "epoch": 1.1500109817702613, + "grad_norm": 2.1174111366271973, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.732935905456543, + "num_tokens": 271052828.0, + "step": 10472 + }, + { + "epoch": 1.150120799472875, + "grad_norm": 2.187563419342041, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7006937265396118, + "num_tokens": 271081617.0, + "step": 10473 + }, + { + "epoch": 1.1502306171754886, + "grad_norm": 2.2240450382232666, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6815386414527893, + "num_tokens": 271110567.0, + "step": 10474 + }, + { + "epoch": 1.1503404348781023, + "grad_norm": 2.2897086143493652, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7124347686767578, + "num_tokens": 271134094.0, + "step": 10475 + }, + { + "epoch": 1.150450252580716, + "grad_norm": 2.2494077682495117, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7380952835083008, + "num_tokens": 271159825.0, + "step": 10476 + }, + { + "epoch": 1.1505600702833296, + "grad_norm": 2.388219118118286, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7331264019012451, + "num_tokens": 271179892.0, + "step": 10477 + }, + { + "epoch": 1.1506698879859434, + "grad_norm": 2.1502020359039307, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7151234745979309, + "num_tokens": 271206416.0, + "step": 10478 + }, + { + "epoch": 1.150779705688557, + "grad_norm": 2.1197550296783447, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6925500631332397, + "num_tokens": 271235255.0, + "step": 10479 + }, + { + "epoch": 1.1508895233911707, + "grad_norm": 1.8735333681106567, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6819225549697876, + "num_tokens": 271273431.0, + "step": 10480 + }, + { + "epoch": 1.1509993410937844, + "grad_norm": 2.3901290893554688, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7032109498977661, + "num_tokens": 271296536.0, + "step": 10481 + }, + { + "epoch": 1.151109158796398, + "grad_norm": 1.9778684377670288, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6940715909004211, + "num_tokens": 271331104.0, + "step": 10482 + }, + { + "epoch": 1.1512189764990117, + "grad_norm": 2.180339813232422, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7188661694526672, + "num_tokens": 271358864.0, + "step": 10483 + }, + { + "epoch": 1.1513287942016253, + "grad_norm": 2.1241440773010254, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7088406085968018, + "num_tokens": 271385636.0, + "step": 10484 + }, + { + "epoch": 1.151438611904239, + "grad_norm": 2.4944491386413574, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7267163991928101, + "num_tokens": 271406751.0, + "step": 10485 + }, + { + "epoch": 1.1515484296068526, + "grad_norm": 2.1899962425231934, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.702556312084198, + "num_tokens": 271433759.0, + "step": 10486 + }, + { + "epoch": 1.1516582473094663, + "grad_norm": 2.538604259490967, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7342031598091125, + "num_tokens": 271453057.0, + "step": 10487 + }, + { + "epoch": 1.1517680650120798, + "grad_norm": 2.371840000152588, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7203918695449829, + "num_tokens": 271478297.0, + "step": 10488 + }, + { + "epoch": 1.1518778827146936, + "grad_norm": 2.4144411087036133, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7261037826538086, + "num_tokens": 271500563.0, + "step": 10489 + }, + { + "epoch": 1.1519877004173074, + "grad_norm": 2.0359199047088623, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7035911083221436, + "num_tokens": 271533148.0, + "step": 10490 + }, + { + "epoch": 1.152097518119921, + "grad_norm": 2.1813621520996094, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.713955283164978, + "num_tokens": 271561841.0, + "step": 10491 + }, + { + "epoch": 1.1522073358225347, + "grad_norm": 2.4564783573150635, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7330607771873474, + "num_tokens": 271583226.0, + "step": 10492 + }, + { + "epoch": 1.1523171535251482, + "grad_norm": 2.382800817489624, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7304503321647644, + "num_tokens": 271605953.0, + "step": 10493 + }, + { + "epoch": 1.152426971227762, + "grad_norm": 2.2526116371154785, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.720563530921936, + "num_tokens": 271632154.0, + "step": 10494 + }, + { + "epoch": 1.1525367889303757, + "grad_norm": 2.4792680740356445, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.724594235420227, + "num_tokens": 271653841.0, + "step": 10495 + }, + { + "epoch": 1.1526466066329892, + "grad_norm": 2.2153279781341553, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7202675342559814, + "num_tokens": 271677977.0, + "step": 10496 + }, + { + "epoch": 1.152756424335603, + "grad_norm": 2.312831401824951, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7191823720932007, + "num_tokens": 271701221.0, + "step": 10497 + }, + { + "epoch": 1.1528662420382165, + "grad_norm": 2.073387861251831, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.6975324749946594, + "num_tokens": 271732023.0, + "step": 10498 + }, + { + "epoch": 1.1529760597408303, + "grad_norm": 2.2223048210144043, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7240656018257141, + "num_tokens": 271757810.0, + "step": 10499 + }, + { + "epoch": 1.1530858774434438, + "grad_norm": 2.798187494277954, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7263858318328857, + "num_tokens": 271775593.0, + "step": 10500 + }, + { + "epoch": 1.1531956951460576, + "grad_norm": 2.189164161682129, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7297292351722717, + "num_tokens": 271802380.0, + "step": 10501 + }, + { + "epoch": 1.153305512848671, + "grad_norm": 2.1397647857666016, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7151901125907898, + "num_tokens": 271829159.0, + "step": 10502 + }, + { + "epoch": 1.1534153305512849, + "grad_norm": 2.3529562950134277, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7141118049621582, + "num_tokens": 271853961.0, + "step": 10503 + }, + { + "epoch": 1.1535251482538986, + "grad_norm": 2.1459081172943115, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7288767099380493, + "num_tokens": 271880431.0, + "step": 10504 + }, + { + "epoch": 1.1536349659565122, + "grad_norm": 1.997671365737915, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7205202579498291, + "num_tokens": 271908730.0, + "step": 10505 + }, + { + "epoch": 1.153744783659126, + "grad_norm": 2.2852227687835693, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7344812154769897, + "num_tokens": 271931465.0, + "step": 10506 + }, + { + "epoch": 1.1538546013617395, + "grad_norm": 2.3120741844177246, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.724143385887146, + "num_tokens": 271957167.0, + "step": 10507 + }, + { + "epoch": 1.1539644190643532, + "grad_norm": 2.2526445388793945, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7338681221008301, + "num_tokens": 271982253.0, + "step": 10508 + }, + { + "epoch": 1.1540742367669667, + "grad_norm": 2.0248758792877197, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6794663667678833, + "num_tokens": 272014376.0, + "step": 10509 + }, + { + "epoch": 1.1541840544695805, + "grad_norm": 2.1346118450164795, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7199041247367859, + "num_tokens": 272043291.0, + "step": 10510 + }, + { + "epoch": 1.154293872172194, + "grad_norm": 2.1111321449279785, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.6974521279335022, + "num_tokens": 272071838.0, + "step": 10511 + }, + { + "epoch": 1.1544036898748078, + "grad_norm": 2.1779329776763916, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7131561040878296, + "num_tokens": 272099074.0, + "step": 10512 + }, + { + "epoch": 1.1545135075774215, + "grad_norm": 2.347128391265869, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.730991005897522, + "num_tokens": 272121508.0, + "step": 10513 + }, + { + "epoch": 1.154623325280035, + "grad_norm": 2.2488515377044678, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7330023050308228, + "num_tokens": 272145371.0, + "step": 10514 + }, + { + "epoch": 1.1547331429826488, + "grad_norm": 2.406524896621704, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7072845697402954, + "num_tokens": 272170042.0, + "step": 10515 + }, + { + "epoch": 1.1548429606852624, + "grad_norm": 2.0989506244659424, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7362238168716431, + "num_tokens": 272197860.0, + "step": 10516 + }, + { + "epoch": 1.1549527783878761, + "grad_norm": 2.659766435623169, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7312480211257935, + "num_tokens": 272219986.0, + "step": 10517 + }, + { + "epoch": 1.15506259609049, + "grad_norm": 2.2400245666503906, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7121243476867676, + "num_tokens": 272247393.0, + "step": 10518 + }, + { + "epoch": 1.1551724137931034, + "grad_norm": 2.5707013607025146, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7414647340774536, + "num_tokens": 272266856.0, + "step": 10519 + }, + { + "epoch": 1.1552822314957172, + "grad_norm": 2.1364004611968994, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7039346098899841, + "num_tokens": 272296842.0, + "step": 10520 + }, + { + "epoch": 1.1553920491983307, + "grad_norm": 2.122687578201294, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6931979060173035, + "num_tokens": 272324349.0, + "step": 10521 + }, + { + "epoch": 1.1555018669009445, + "grad_norm": 2.1514437198638916, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7329910397529602, + "num_tokens": 272351169.0, + "step": 10522 + }, + { + "epoch": 1.155611684603558, + "grad_norm": 2.3308358192443848, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7089458703994751, + "num_tokens": 272375508.0, + "step": 10523 + }, + { + "epoch": 1.1557215023061718, + "grad_norm": 2.0006930828094482, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7193458676338196, + "num_tokens": 272405700.0, + "step": 10524 + }, + { + "epoch": 1.1558313200087853, + "grad_norm": 2.086452007293701, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6924970149993896, + "num_tokens": 272435400.0, + "step": 10525 + }, + { + "epoch": 1.155941137711399, + "grad_norm": 2.259971857070923, + "learning_rate": 1e-06, + "loss": 1.0828, + "mean_token_accuracy": 0.6802374124526978, + "num_tokens": 272462424.0, + "step": 10526 + }, + { + "epoch": 1.1560509554140128, + "grad_norm": 2.1640305519104004, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6968944072723389, + "num_tokens": 272492794.0, + "step": 10527 + }, + { + "epoch": 1.1561607731166264, + "grad_norm": 2.275542974472046, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7661483287811279, + "num_tokens": 272515654.0, + "step": 10528 + }, + { + "epoch": 1.15627059081924, + "grad_norm": 2.256598949432373, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7187797427177429, + "num_tokens": 272540293.0, + "step": 10529 + }, + { + "epoch": 1.1563804085218536, + "grad_norm": 2.076094627380371, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7237488627433777, + "num_tokens": 272570385.0, + "step": 10530 + }, + { + "epoch": 1.1564902262244674, + "grad_norm": 2.396766424179077, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7134011387825012, + "num_tokens": 272591512.0, + "step": 10531 + }, + { + "epoch": 1.1566000439270812, + "grad_norm": 2.393362522125244, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7212203741073608, + "num_tokens": 272614670.0, + "step": 10532 + }, + { + "epoch": 1.1567098616296947, + "grad_norm": 2.2205440998077393, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7165987491607666, + "num_tokens": 272640736.0, + "step": 10533 + }, + { + "epoch": 1.1568196793323084, + "grad_norm": 2.240029811859131, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7164620757102966, + "num_tokens": 272665179.0, + "step": 10534 + }, + { + "epoch": 1.156929497034922, + "grad_norm": 2.000141143798828, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7280234098434448, + "num_tokens": 272693745.0, + "step": 10535 + }, + { + "epoch": 1.1570393147375357, + "grad_norm": 2.5926666259765625, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7277812957763672, + "num_tokens": 272714068.0, + "step": 10536 + }, + { + "epoch": 1.1571491324401493, + "grad_norm": 2.33321475982666, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7042580842971802, + "num_tokens": 272737718.0, + "step": 10537 + }, + { + "epoch": 1.157258950142763, + "grad_norm": 2.858354091644287, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7251495122909546, + "num_tokens": 272760243.0, + "step": 10538 + }, + { + "epoch": 1.1573687678453766, + "grad_norm": 2.103825569152832, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7080979347229004, + "num_tokens": 272788577.0, + "step": 10539 + }, + { + "epoch": 1.1574785855479903, + "grad_norm": 2.2116901874542236, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6913784742355347, + "num_tokens": 272816076.0, + "step": 10540 + }, + { + "epoch": 1.157588403250604, + "grad_norm": 2.3726487159729004, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7205455899238586, + "num_tokens": 272838322.0, + "step": 10541 + }, + { + "epoch": 1.1576982209532176, + "grad_norm": 2.2250983715057373, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7400763034820557, + "num_tokens": 272864169.0, + "step": 10542 + }, + { + "epoch": 1.1578080386558314, + "grad_norm": 2.232537031173706, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7273079752922058, + "num_tokens": 272888136.0, + "step": 10543 + }, + { + "epoch": 1.157917856358445, + "grad_norm": 2.2806105613708496, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7087283134460449, + "num_tokens": 272912632.0, + "step": 10544 + }, + { + "epoch": 1.1580276740610587, + "grad_norm": 2.2915186882019043, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7144784331321716, + "num_tokens": 272939621.0, + "step": 10545 + }, + { + "epoch": 1.1581374917636724, + "grad_norm": 2.339646577835083, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7073062658309937, + "num_tokens": 272963177.0, + "step": 10546 + }, + { + "epoch": 1.158247309466286, + "grad_norm": 2.2664942741394043, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7367084622383118, + "num_tokens": 272987430.0, + "step": 10547 + }, + { + "epoch": 1.1583571271688997, + "grad_norm": 2.4893643856048584, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7288154363632202, + "num_tokens": 273008216.0, + "step": 10548 + }, + { + "epoch": 1.1584669448715132, + "grad_norm": 2.485139846801758, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7105933427810669, + "num_tokens": 273031222.0, + "step": 10549 + }, + { + "epoch": 1.158576762574127, + "grad_norm": 2.3806002140045166, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7293785810470581, + "num_tokens": 273053142.0, + "step": 10550 + }, + { + "epoch": 1.1586865802767405, + "grad_norm": 2.2237625122070312, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7035856246948242, + "num_tokens": 273080271.0, + "step": 10551 + }, + { + "epoch": 1.1587963979793543, + "grad_norm": 2.3084912300109863, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7263699769973755, + "num_tokens": 273103267.0, + "step": 10552 + }, + { + "epoch": 1.1589062156819678, + "grad_norm": 2.213928461074829, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7139840126037598, + "num_tokens": 273129129.0, + "step": 10553 + }, + { + "epoch": 1.1590160333845816, + "grad_norm": 2.026923418045044, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7308382391929626, + "num_tokens": 273159544.0, + "step": 10554 + }, + { + "epoch": 1.1591258510871953, + "grad_norm": 2.0971298217773438, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7197763323783875, + "num_tokens": 273186077.0, + "step": 10555 + }, + { + "epoch": 1.1592356687898089, + "grad_norm": 2.1480672359466553, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7191076874732971, + "num_tokens": 273213808.0, + "step": 10556 + }, + { + "epoch": 1.1593454864924226, + "grad_norm": 2.1569595336914062, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7256550788879395, + "num_tokens": 273242014.0, + "step": 10557 + }, + { + "epoch": 1.1594553041950362, + "grad_norm": 2.0241892337799072, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7111318111419678, + "num_tokens": 273272011.0, + "step": 10558 + }, + { + "epoch": 1.15956512189765, + "grad_norm": 2.3296163082122803, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6947227716445923, + "num_tokens": 273296193.0, + "step": 10559 + }, + { + "epoch": 1.1596749396002635, + "grad_norm": 2.1493446826934814, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7191323041915894, + "num_tokens": 273324154.0, + "step": 10560 + }, + { + "epoch": 1.1597847573028772, + "grad_norm": 1.996116042137146, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7028007507324219, + "num_tokens": 273356852.0, + "step": 10561 + }, + { + "epoch": 1.159894575005491, + "grad_norm": 2.1653904914855957, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7030478715896606, + "num_tokens": 273386090.0, + "step": 10562 + }, + { + "epoch": 1.1600043927081045, + "grad_norm": 2.0497355461120605, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7140626311302185, + "num_tokens": 273416319.0, + "step": 10563 + }, + { + "epoch": 1.1601142104107183, + "grad_norm": 2.4248292446136475, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7088207006454468, + "num_tokens": 273439730.0, + "step": 10564 + }, + { + "epoch": 1.1602240281133318, + "grad_norm": 2.296335220336914, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7010817527770996, + "num_tokens": 273465595.0, + "step": 10565 + }, + { + "epoch": 1.1603338458159456, + "grad_norm": 2.088986873626709, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6834930181503296, + "num_tokens": 273496619.0, + "step": 10566 + }, + { + "epoch": 1.160443663518559, + "grad_norm": 2.616389036178589, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.733402669429779, + "num_tokens": 273517232.0, + "step": 10567 + }, + { + "epoch": 1.1605534812211729, + "grad_norm": 2.250777244567871, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7106107473373413, + "num_tokens": 273543123.0, + "step": 10568 + }, + { + "epoch": 1.1606632989237866, + "grad_norm": 2.0211267471313477, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7178610563278198, + "num_tokens": 273573824.0, + "step": 10569 + }, + { + "epoch": 1.1607731166264001, + "grad_norm": 2.368942975997925, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7344963550567627, + "num_tokens": 273596774.0, + "step": 10570 + }, + { + "epoch": 1.160882934329014, + "grad_norm": 2.026538610458374, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7185393571853638, + "num_tokens": 273625480.0, + "step": 10571 + }, + { + "epoch": 1.1609927520316274, + "grad_norm": 2.2690088748931885, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7199863791465759, + "num_tokens": 273650325.0, + "step": 10572 + }, + { + "epoch": 1.1611025697342412, + "grad_norm": 2.325307846069336, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6990125179290771, + "num_tokens": 273673384.0, + "step": 10573 + }, + { + "epoch": 1.1612123874368547, + "grad_norm": 2.330692768096924, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7205239534378052, + "num_tokens": 273696548.0, + "step": 10574 + }, + { + "epoch": 1.1613222051394685, + "grad_norm": 2.130228042602539, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7120313048362732, + "num_tokens": 273725958.0, + "step": 10575 + }, + { + "epoch": 1.161432022842082, + "grad_norm": 1.9574635028839111, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7074129581451416, + "num_tokens": 273758568.0, + "step": 10576 + }, + { + "epoch": 1.1615418405446958, + "grad_norm": 2.1466479301452637, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7539843320846558, + "num_tokens": 273786322.0, + "step": 10577 + }, + { + "epoch": 1.1616516582473095, + "grad_norm": 2.2318618297576904, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7193315625190735, + "num_tokens": 273812749.0, + "step": 10578 + }, + { + "epoch": 1.161761475949923, + "grad_norm": 2.2594549655914307, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7287293076515198, + "num_tokens": 273836812.0, + "step": 10579 + }, + { + "epoch": 1.1618712936525368, + "grad_norm": 2.376014232635498, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6900343894958496, + "num_tokens": 273861639.0, + "step": 10580 + }, + { + "epoch": 1.1619811113551504, + "grad_norm": 2.0791072845458984, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7184558510780334, + "num_tokens": 273890785.0, + "step": 10581 + }, + { + "epoch": 1.1620909290577641, + "grad_norm": 2.0404748916625977, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7123403549194336, + "num_tokens": 273920468.0, + "step": 10582 + }, + { + "epoch": 1.1622007467603779, + "grad_norm": 2.1583120822906494, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7407457828521729, + "num_tokens": 273946197.0, + "step": 10583 + }, + { + "epoch": 1.1623105644629914, + "grad_norm": 2.090036392211914, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7114765644073486, + "num_tokens": 273974399.0, + "step": 10584 + }, + { + "epoch": 1.1624203821656052, + "grad_norm": 2.226468324661255, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7238121032714844, + "num_tokens": 274001424.0, + "step": 10585 + }, + { + "epoch": 1.1625301998682187, + "grad_norm": 2.5510196685791016, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7042855620384216, + "num_tokens": 274024276.0, + "step": 10586 + }, + { + "epoch": 1.1626400175708325, + "grad_norm": 2.2924673557281494, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7083821296691895, + "num_tokens": 274048851.0, + "step": 10587 + }, + { + "epoch": 1.162749835273446, + "grad_norm": 2.417489767074585, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7168454527854919, + "num_tokens": 274071445.0, + "step": 10588 + }, + { + "epoch": 1.1628596529760598, + "grad_norm": 2.0789215564727783, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7594693303108215, + "num_tokens": 274098455.0, + "step": 10589 + }, + { + "epoch": 1.1629694706786733, + "grad_norm": 2.202911138534546, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7434507608413696, + "num_tokens": 274124178.0, + "step": 10590 + }, + { + "epoch": 1.163079288381287, + "grad_norm": 2.1514759063720703, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7165691256523132, + "num_tokens": 274151069.0, + "step": 10591 + }, + { + "epoch": 1.1631891060839008, + "grad_norm": 2.4723763465881348, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7092145681381226, + "num_tokens": 274173561.0, + "step": 10592 + }, + { + "epoch": 1.1632989237865143, + "grad_norm": 2.0346839427948, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.6846852898597717, + "num_tokens": 274205144.0, + "step": 10593 + }, + { + "epoch": 1.163408741489128, + "grad_norm": 2.227184772491455, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7019157409667969, + "num_tokens": 274231929.0, + "step": 10594 + }, + { + "epoch": 1.1635185591917416, + "grad_norm": 2.361060619354248, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7375842928886414, + "num_tokens": 274256158.0, + "step": 10595 + }, + { + "epoch": 1.1636283768943554, + "grad_norm": 2.3543694019317627, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6986317038536072, + "num_tokens": 274283026.0, + "step": 10596 + }, + { + "epoch": 1.1637381945969691, + "grad_norm": 2.210721969604492, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7022093534469604, + "num_tokens": 274309775.0, + "step": 10597 + }, + { + "epoch": 1.1638480122995827, + "grad_norm": 2.3636112213134766, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7166280746459961, + "num_tokens": 274334540.0, + "step": 10598 + }, + { + "epoch": 1.1639578300021964, + "grad_norm": 2.4442780017852783, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7173576951026917, + "num_tokens": 274357535.0, + "step": 10599 + }, + { + "epoch": 1.16406764770481, + "grad_norm": 2.308030128479004, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7195584177970886, + "num_tokens": 274382477.0, + "step": 10600 + }, + { + "epoch": 1.1641774654074237, + "grad_norm": 2.2586240768432617, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7295551300048828, + "num_tokens": 274408585.0, + "step": 10601 + }, + { + "epoch": 1.1642872831100373, + "grad_norm": 1.9971483945846558, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.748572051525116, + "num_tokens": 274437220.0, + "step": 10602 + }, + { + "epoch": 1.164397100812651, + "grad_norm": 1.9868628978729248, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7174416780471802, + "num_tokens": 274469802.0, + "step": 10603 + }, + { + "epoch": 1.1645069185152646, + "grad_norm": 1.9750369787216187, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7144153118133545, + "num_tokens": 274499495.0, + "step": 10604 + }, + { + "epoch": 1.1646167362178783, + "grad_norm": 2.362823963165283, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.729028046131134, + "num_tokens": 274522048.0, + "step": 10605 + }, + { + "epoch": 1.164726553920492, + "grad_norm": 2.175689458847046, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6998175382614136, + "num_tokens": 274550397.0, + "step": 10606 + }, + { + "epoch": 1.1648363716231056, + "grad_norm": 2.323962688446045, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7266324162483215, + "num_tokens": 274572289.0, + "step": 10607 + }, + { + "epoch": 1.1649461893257194, + "grad_norm": 2.25870680809021, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7092670202255249, + "num_tokens": 274598271.0, + "step": 10608 + }, + { + "epoch": 1.165056007028333, + "grad_norm": 1.9887977838516235, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7209672331809998, + "num_tokens": 274626039.0, + "step": 10609 + }, + { + "epoch": 1.1651658247309467, + "grad_norm": 2.090994119644165, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7067153453826904, + "num_tokens": 274657055.0, + "step": 10610 + }, + { + "epoch": 1.1652756424335604, + "grad_norm": 2.129913091659546, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7234544157981873, + "num_tokens": 274685455.0, + "step": 10611 + }, + { + "epoch": 1.165385460136174, + "grad_norm": 2.391169786453247, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6957349181175232, + "num_tokens": 274709761.0, + "step": 10612 + }, + { + "epoch": 1.1654952778387877, + "grad_norm": 2.3860878944396973, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7245346307754517, + "num_tokens": 274734396.0, + "step": 10613 + }, + { + "epoch": 1.1656050955414012, + "grad_norm": 1.9442168474197388, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7147378921508789, + "num_tokens": 274768069.0, + "step": 10614 + }, + { + "epoch": 1.165714913244015, + "grad_norm": 1.9838283061981201, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7247568964958191, + "num_tokens": 274800029.0, + "step": 10615 + }, + { + "epoch": 1.1658247309466285, + "grad_norm": 2.6574158668518066, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7280429601669312, + "num_tokens": 274819886.0, + "step": 10616 + }, + { + "epoch": 1.1659345486492423, + "grad_norm": 2.2193377017974854, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.709125280380249, + "num_tokens": 274848395.0, + "step": 10617 + }, + { + "epoch": 1.1660443663518558, + "grad_norm": 2.240889072418213, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7192654013633728, + "num_tokens": 274872749.0, + "step": 10618 + }, + { + "epoch": 1.1661541840544696, + "grad_norm": 2.2248666286468506, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7308245897293091, + "num_tokens": 274897661.0, + "step": 10619 + }, + { + "epoch": 1.1662640017570833, + "grad_norm": 2.249424457550049, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7057764530181885, + "num_tokens": 274922900.0, + "step": 10620 + }, + { + "epoch": 1.1663738194596969, + "grad_norm": 2.0909669399261475, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.71931391954422, + "num_tokens": 274953199.0, + "step": 10621 + }, + { + "epoch": 1.1664836371623106, + "grad_norm": 2.4145965576171875, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7253430485725403, + "num_tokens": 274975312.0, + "step": 10622 + }, + { + "epoch": 1.1665934548649242, + "grad_norm": 2.114440679550171, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6969008445739746, + "num_tokens": 275005756.0, + "step": 10623 + }, + { + "epoch": 1.166703272567538, + "grad_norm": 2.277031421661377, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7313793897628784, + "num_tokens": 275029331.0, + "step": 10624 + }, + { + "epoch": 1.1668130902701515, + "grad_norm": 2.5037755966186523, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.680763304233551, + "num_tokens": 275054582.0, + "step": 10625 + }, + { + "epoch": 1.1669229079727652, + "grad_norm": 2.3189144134521484, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.738296389579773, + "num_tokens": 275079544.0, + "step": 10626 + }, + { + "epoch": 1.167032725675379, + "grad_norm": 2.1561026573181152, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7083667516708374, + "num_tokens": 275109495.0, + "step": 10627 + }, + { + "epoch": 1.1671425433779925, + "grad_norm": 2.3162941932678223, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7157213687896729, + "num_tokens": 275132812.0, + "step": 10628 + }, + { + "epoch": 1.1672523610806063, + "grad_norm": 2.069570779800415, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6944980621337891, + "num_tokens": 275161256.0, + "step": 10629 + }, + { + "epoch": 1.1673621787832198, + "grad_norm": 2.2414090633392334, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7285934686660767, + "num_tokens": 275186318.0, + "step": 10630 + }, + { + "epoch": 1.1674719964858336, + "grad_norm": 2.512162208557129, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6961891651153564, + "num_tokens": 275207258.0, + "step": 10631 + }, + { + "epoch": 1.167581814188447, + "grad_norm": 2.256838083267212, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7160354852676392, + "num_tokens": 275233345.0, + "step": 10632 + }, + { + "epoch": 1.1676916318910608, + "grad_norm": 2.2798986434936523, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7326380014419556, + "num_tokens": 275258457.0, + "step": 10633 + }, + { + "epoch": 1.1678014495936746, + "grad_norm": 2.6317028999328613, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7324748039245605, + "num_tokens": 275278010.0, + "step": 10634 + }, + { + "epoch": 1.1679112672962881, + "grad_norm": 2.3272743225097656, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7243528366088867, + "num_tokens": 275304279.0, + "step": 10635 + }, + { + "epoch": 1.168021084998902, + "grad_norm": 2.432168960571289, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7172387838363647, + "num_tokens": 275326567.0, + "step": 10636 + }, + { + "epoch": 1.1681309027015154, + "grad_norm": 1.8867703676223755, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6886533498764038, + "num_tokens": 275363466.0, + "step": 10637 + }, + { + "epoch": 1.1682407204041292, + "grad_norm": 2.511516571044922, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.73503577709198, + "num_tokens": 275383405.0, + "step": 10638 + }, + { + "epoch": 1.1683505381067427, + "grad_norm": 2.0786213874816895, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7281618118286133, + "num_tokens": 275410995.0, + "step": 10639 + }, + { + "epoch": 1.1684603558093565, + "grad_norm": 2.147458553314209, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6965832710266113, + "num_tokens": 275442348.0, + "step": 10640 + }, + { + "epoch": 1.16857017351197, + "grad_norm": 1.9653210639953613, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7082308530807495, + "num_tokens": 275473597.0, + "step": 10641 + }, + { + "epoch": 1.1686799912145838, + "grad_norm": 2.02528977394104, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6978932619094849, + "num_tokens": 275505412.0, + "step": 10642 + }, + { + "epoch": 1.1687898089171975, + "grad_norm": 2.4278602600097656, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7153489589691162, + "num_tokens": 275528121.0, + "step": 10643 + }, + { + "epoch": 1.168899626619811, + "grad_norm": 2.3563010692596436, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7400391697883606, + "num_tokens": 275549898.0, + "step": 10644 + }, + { + "epoch": 1.1690094443224248, + "grad_norm": 2.0510053634643555, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6943125128746033, + "num_tokens": 275579992.0, + "step": 10645 + }, + { + "epoch": 1.1691192620250384, + "grad_norm": 2.456479549407959, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7107876539230347, + "num_tokens": 275602864.0, + "step": 10646 + }, + { + "epoch": 1.169229079727652, + "grad_norm": 2.0040700435638428, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7302106618881226, + "num_tokens": 275631977.0, + "step": 10647 + }, + { + "epoch": 1.1693388974302659, + "grad_norm": 2.1565256118774414, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7210582494735718, + "num_tokens": 275661055.0, + "step": 10648 + }, + { + "epoch": 1.1694487151328794, + "grad_norm": 2.1541748046875, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7298961281776428, + "num_tokens": 275686776.0, + "step": 10649 + }, + { + "epoch": 1.1695585328354932, + "grad_norm": 2.0267937183380127, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.742838442325592, + "num_tokens": 275715084.0, + "step": 10650 + }, + { + "epoch": 1.1696683505381067, + "grad_norm": 1.9522837400436401, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6935911178588867, + "num_tokens": 275748446.0, + "step": 10651 + }, + { + "epoch": 1.1697781682407205, + "grad_norm": 2.47082781791687, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7074204683303833, + "num_tokens": 275770040.0, + "step": 10652 + }, + { + "epoch": 1.169887985943334, + "grad_norm": 2.452540636062622, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7233400344848633, + "num_tokens": 275793328.0, + "step": 10653 + }, + { + "epoch": 1.1699978036459477, + "grad_norm": 2.0303540229797363, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.707595705986023, + "num_tokens": 275823134.0, + "step": 10654 + }, + { + "epoch": 1.1701076213485613, + "grad_norm": 2.1757850646972656, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7026689648628235, + "num_tokens": 275851797.0, + "step": 10655 + }, + { + "epoch": 1.170217439051175, + "grad_norm": 2.464390277862549, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7152532339096069, + "num_tokens": 275874996.0, + "step": 10656 + }, + { + "epoch": 1.1703272567537888, + "grad_norm": 2.2093329429626465, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7121723294258118, + "num_tokens": 275901105.0, + "step": 10657 + }, + { + "epoch": 1.1704370744564023, + "grad_norm": 2.179771900177002, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7153776288032532, + "num_tokens": 275928391.0, + "step": 10658 + }, + { + "epoch": 1.170546892159016, + "grad_norm": 2.222557306289673, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.721503496170044, + "num_tokens": 275955151.0, + "step": 10659 + }, + { + "epoch": 1.1706567098616296, + "grad_norm": 2.4204232692718506, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7178694605827332, + "num_tokens": 275977107.0, + "step": 10660 + }, + { + "epoch": 1.1707665275642434, + "grad_norm": 2.056865930557251, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6866929531097412, + "num_tokens": 276006547.0, + "step": 10661 + }, + { + "epoch": 1.1708763452668571, + "grad_norm": 2.014486789703369, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7178332805633545, + "num_tokens": 276034604.0, + "step": 10662 + }, + { + "epoch": 1.1709861629694707, + "grad_norm": 2.2249629497528076, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7029005885124207, + "num_tokens": 276062945.0, + "step": 10663 + }, + { + "epoch": 1.1710959806720844, + "grad_norm": 2.6681838035583496, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7448140978813171, + "num_tokens": 276081219.0, + "step": 10664 + }, + { + "epoch": 1.171205798374698, + "grad_norm": 2.331946611404419, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7052897810935974, + "num_tokens": 276105780.0, + "step": 10665 + }, + { + "epoch": 1.1713156160773117, + "grad_norm": 2.389254570007324, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7355430126190186, + "num_tokens": 276127673.0, + "step": 10666 + }, + { + "epoch": 1.1714254337799253, + "grad_norm": 2.1591198444366455, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7170524597167969, + "num_tokens": 276155202.0, + "step": 10667 + }, + { + "epoch": 1.171535251482539, + "grad_norm": 2.276425361633301, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7084861993789673, + "num_tokens": 276182164.0, + "step": 10668 + }, + { + "epoch": 1.1716450691851525, + "grad_norm": 1.9911363124847412, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7323692440986633, + "num_tokens": 276210888.0, + "step": 10669 + }, + { + "epoch": 1.1717548868877663, + "grad_norm": 2.037599563598633, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7260568737983704, + "num_tokens": 276238745.0, + "step": 10670 + }, + { + "epoch": 1.17186470459038, + "grad_norm": 2.3574957847595215, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7263924479484558, + "num_tokens": 276262017.0, + "step": 10671 + }, + { + "epoch": 1.1719745222929936, + "grad_norm": 2.0942466259002686, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.71014404296875, + "num_tokens": 276291389.0, + "step": 10672 + }, + { + "epoch": 1.1720843399956073, + "grad_norm": 2.3946008682250977, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7248191237449646, + "num_tokens": 276314507.0, + "step": 10673 + }, + { + "epoch": 1.1721941576982209, + "grad_norm": 2.102933645248413, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.716168999671936, + "num_tokens": 276341334.0, + "step": 10674 + }, + { + "epoch": 1.1723039754008346, + "grad_norm": 2.209275484085083, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7130172848701477, + "num_tokens": 276366493.0, + "step": 10675 + }, + { + "epoch": 1.1724137931034484, + "grad_norm": 2.3513286113739014, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7229485511779785, + "num_tokens": 276388674.0, + "step": 10676 + }, + { + "epoch": 1.172523610806062, + "grad_norm": 2.215426445007324, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7096067667007446, + "num_tokens": 276412824.0, + "step": 10677 + }, + { + "epoch": 1.1726334285086757, + "grad_norm": 2.2451467514038086, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7252267003059387, + "num_tokens": 276440248.0, + "step": 10678 + }, + { + "epoch": 1.1727432462112892, + "grad_norm": 2.2311289310455322, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7245401740074158, + "num_tokens": 276466961.0, + "step": 10679 + }, + { + "epoch": 1.172853063913903, + "grad_norm": 2.4130895137786865, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7435330748558044, + "num_tokens": 276488568.0, + "step": 10680 + }, + { + "epoch": 1.1729628816165165, + "grad_norm": 2.290968418121338, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7247543931007385, + "num_tokens": 276514628.0, + "step": 10681 + }, + { + "epoch": 1.1730726993191303, + "grad_norm": 2.03816294670105, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7147423624992371, + "num_tokens": 276546001.0, + "step": 10682 + }, + { + "epoch": 1.1731825170217438, + "grad_norm": 2.3731918334960938, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7268660664558411, + "num_tokens": 276567144.0, + "step": 10683 + }, + { + "epoch": 1.1732923347243576, + "grad_norm": 2.0129892826080322, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7042131423950195, + "num_tokens": 276600189.0, + "step": 10684 + }, + { + "epoch": 1.1734021524269713, + "grad_norm": 2.2858355045318604, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7207010984420776, + "num_tokens": 276626814.0, + "step": 10685 + }, + { + "epoch": 1.1735119701295849, + "grad_norm": 2.0299885272979736, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6840096712112427, + "num_tokens": 276660157.0, + "step": 10686 + }, + { + "epoch": 1.1736217878321986, + "grad_norm": 2.334653854370117, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7100766897201538, + "num_tokens": 276687209.0, + "step": 10687 + }, + { + "epoch": 1.1737316055348122, + "grad_norm": 1.9716002941131592, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7235431671142578, + "num_tokens": 276719689.0, + "step": 10688 + }, + { + "epoch": 1.173841423237426, + "grad_norm": 1.9819517135620117, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7196317911148071, + "num_tokens": 276750718.0, + "step": 10689 + }, + { + "epoch": 1.1739512409400394, + "grad_norm": 2.3197381496429443, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7155513763427734, + "num_tokens": 276775344.0, + "step": 10690 + }, + { + "epoch": 1.1740610586426532, + "grad_norm": 2.169968843460083, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6947113275527954, + "num_tokens": 276802480.0, + "step": 10691 + }, + { + "epoch": 1.1741708763452667, + "grad_norm": 2.3352723121643066, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7476180791854858, + "num_tokens": 276825093.0, + "step": 10692 + }, + { + "epoch": 1.1742806940478805, + "grad_norm": 2.11964750289917, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6906293630599976, + "num_tokens": 276859244.0, + "step": 10693 + }, + { + "epoch": 1.1743905117504942, + "grad_norm": 2.217567205429077, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7199698090553284, + "num_tokens": 276883396.0, + "step": 10694 + }, + { + "epoch": 1.1745003294531078, + "grad_norm": 2.132709503173828, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7308443784713745, + "num_tokens": 276910340.0, + "step": 10695 + }, + { + "epoch": 1.1746101471557215, + "grad_norm": 2.3373398780822754, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7272094488143921, + "num_tokens": 276935607.0, + "step": 10696 + }, + { + "epoch": 1.174719964858335, + "grad_norm": 2.4543850421905518, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7186948657035828, + "num_tokens": 276958407.0, + "step": 10697 + }, + { + "epoch": 1.1748297825609488, + "grad_norm": 2.2529430389404297, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7107742428779602, + "num_tokens": 276984274.0, + "step": 10698 + }, + { + "epoch": 1.1749396002635626, + "grad_norm": 2.479928731918335, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7408662438392639, + "num_tokens": 277003924.0, + "step": 10699 + }, + { + "epoch": 1.1750494179661761, + "grad_norm": 2.2942991256713867, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7084851861000061, + "num_tokens": 277029326.0, + "step": 10700 + }, + { + "epoch": 1.1751592356687899, + "grad_norm": 2.3371267318725586, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7467716932296753, + "num_tokens": 277052982.0, + "step": 10701 + }, + { + "epoch": 1.1752690533714034, + "grad_norm": 2.0618350505828857, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7232071161270142, + "num_tokens": 277082856.0, + "step": 10702 + }, + { + "epoch": 1.1753788710740172, + "grad_norm": 2.192473888397217, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.6957187056541443, + "num_tokens": 277110580.0, + "step": 10703 + }, + { + "epoch": 1.1754886887766307, + "grad_norm": 2.3673763275146484, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7216764092445374, + "num_tokens": 277133087.0, + "step": 10704 + }, + { + "epoch": 1.1755985064792445, + "grad_norm": 2.278639316558838, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7198912501335144, + "num_tokens": 277157403.0, + "step": 10705 + }, + { + "epoch": 1.175708324181858, + "grad_norm": 2.145069122314453, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7141898274421692, + "num_tokens": 277183739.0, + "step": 10706 + }, + { + "epoch": 1.1758181418844718, + "grad_norm": 2.0798728466033936, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7072423696517944, + "num_tokens": 277210821.0, + "step": 10707 + }, + { + "epoch": 1.1759279595870855, + "grad_norm": 2.0866878032684326, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.729809582233429, + "num_tokens": 277236725.0, + "step": 10708 + }, + { + "epoch": 1.176037777289699, + "grad_norm": 1.9752070903778076, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.711690366268158, + "num_tokens": 277268483.0, + "step": 10709 + }, + { + "epoch": 1.1761475949923128, + "grad_norm": 2.0372917652130127, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7299790382385254, + "num_tokens": 277296117.0, + "step": 10710 + }, + { + "epoch": 1.1762574126949263, + "grad_norm": 2.432223081588745, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7233055830001831, + "num_tokens": 277318543.0, + "step": 10711 + }, + { + "epoch": 1.17636723039754, + "grad_norm": 2.10174298286438, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7107211351394653, + "num_tokens": 277347866.0, + "step": 10712 + }, + { + "epoch": 1.1764770481001539, + "grad_norm": 2.171682596206665, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6931648254394531, + "num_tokens": 277376341.0, + "step": 10713 + }, + { + "epoch": 1.1765868658027674, + "grad_norm": 2.164135694503784, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7303304076194763, + "num_tokens": 277403427.0, + "step": 10714 + }, + { + "epoch": 1.1766966835053811, + "grad_norm": 2.0863840579986572, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7247301340103149, + "num_tokens": 277430711.0, + "step": 10715 + }, + { + "epoch": 1.1768065012079947, + "grad_norm": 2.029284954071045, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6972740292549133, + "num_tokens": 277460092.0, + "step": 10716 + }, + { + "epoch": 1.1769163189106084, + "grad_norm": 2.138915777206421, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.6991199254989624, + "num_tokens": 277486876.0, + "step": 10717 + }, + { + "epoch": 1.177026136613222, + "grad_norm": 2.3538355827331543, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7101089954376221, + "num_tokens": 277510917.0, + "step": 10718 + }, + { + "epoch": 1.1771359543158357, + "grad_norm": 2.2206249237060547, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7068926095962524, + "num_tokens": 277537077.0, + "step": 10719 + }, + { + "epoch": 1.1772457720184493, + "grad_norm": 2.457392692565918, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7224280834197998, + "num_tokens": 277559338.0, + "step": 10720 + }, + { + "epoch": 1.177355589721063, + "grad_norm": 2.6030404567718506, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7353131175041199, + "num_tokens": 277580893.0, + "step": 10721 + }, + { + "epoch": 1.1774654074236768, + "grad_norm": 2.414109230041504, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.712114155292511, + "num_tokens": 277602487.0, + "step": 10722 + }, + { + "epoch": 1.1775752251262903, + "grad_norm": 2.0395395755767822, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.711193323135376, + "num_tokens": 277633982.0, + "step": 10723 + }, + { + "epoch": 1.177685042828904, + "grad_norm": 2.375222682952881, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7280439138412476, + "num_tokens": 277656418.0, + "step": 10724 + }, + { + "epoch": 1.1777948605315176, + "grad_norm": 2.404416084289551, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7021313905715942, + "num_tokens": 277681706.0, + "step": 10725 + }, + { + "epoch": 1.1779046782341314, + "grad_norm": 2.3874409198760986, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7310713529586792, + "num_tokens": 277704022.0, + "step": 10726 + }, + { + "epoch": 1.1780144959367451, + "grad_norm": 2.5647799968719482, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7510567903518677, + "num_tokens": 277722483.0, + "step": 10727 + }, + { + "epoch": 1.1781243136393587, + "grad_norm": 2.3902530670166016, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7247067093849182, + "num_tokens": 277744425.0, + "step": 10728 + }, + { + "epoch": 1.1782341313419724, + "grad_norm": 2.6868185997009277, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7292525768280029, + "num_tokens": 277762468.0, + "step": 10729 + }, + { + "epoch": 1.178343949044586, + "grad_norm": 2.2253122329711914, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7079020738601685, + "num_tokens": 277789603.0, + "step": 10730 + }, + { + "epoch": 1.1784537667471997, + "grad_norm": 2.205334424972534, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.6980395913124084, + "num_tokens": 277816454.0, + "step": 10731 + }, + { + "epoch": 1.1785635844498132, + "grad_norm": 2.1052370071411133, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7354726791381836, + "num_tokens": 277844616.0, + "step": 10732 + }, + { + "epoch": 1.178673402152427, + "grad_norm": 2.2749555110931396, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7114269733428955, + "num_tokens": 277870836.0, + "step": 10733 + }, + { + "epoch": 1.1787832198550405, + "grad_norm": 2.300192356109619, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7293466329574585, + "num_tokens": 277894579.0, + "step": 10734 + }, + { + "epoch": 1.1788930375576543, + "grad_norm": 2.1192212104797363, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6873807907104492, + "num_tokens": 277924162.0, + "step": 10735 + }, + { + "epoch": 1.179002855260268, + "grad_norm": 2.0751490592956543, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7200250029563904, + "num_tokens": 277952941.0, + "step": 10736 + }, + { + "epoch": 1.1791126729628816, + "grad_norm": 2.271638870239258, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.704466700553894, + "num_tokens": 277977943.0, + "step": 10737 + }, + { + "epoch": 1.1792224906654953, + "grad_norm": 2.308894157409668, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7074160575866699, + "num_tokens": 278000702.0, + "step": 10738 + }, + { + "epoch": 1.1793323083681089, + "grad_norm": 2.511384963989258, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.722488284111023, + "num_tokens": 278023195.0, + "step": 10739 + }, + { + "epoch": 1.1794421260707226, + "grad_norm": 2.13684344291687, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7272344827651978, + "num_tokens": 278051203.0, + "step": 10740 + }, + { + "epoch": 1.1795519437733364, + "grad_norm": 2.146186590194702, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7232617139816284, + "num_tokens": 278076421.0, + "step": 10741 + }, + { + "epoch": 1.17966176147595, + "grad_norm": 2.3698832988739014, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7095067501068115, + "num_tokens": 278101100.0, + "step": 10742 + }, + { + "epoch": 1.1797715791785637, + "grad_norm": 2.3792967796325684, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7039330005645752, + "num_tokens": 278126360.0, + "step": 10743 + }, + { + "epoch": 1.1798813968811772, + "grad_norm": 2.1410744190216064, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7217521667480469, + "num_tokens": 278152646.0, + "step": 10744 + }, + { + "epoch": 1.179991214583791, + "grad_norm": 2.4343090057373047, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7098835706710815, + "num_tokens": 278174769.0, + "step": 10745 + }, + { + "epoch": 1.1801010322864045, + "grad_norm": 2.110349655151367, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7246610522270203, + "num_tokens": 278203501.0, + "step": 10746 + }, + { + "epoch": 1.1802108499890183, + "grad_norm": 2.1023292541503906, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7312811613082886, + "num_tokens": 278230519.0, + "step": 10747 + }, + { + "epoch": 1.1803206676916318, + "grad_norm": 2.21671724319458, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7311744689941406, + "num_tokens": 278256524.0, + "step": 10748 + }, + { + "epoch": 1.1804304853942456, + "grad_norm": 2.466122627258301, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.733931303024292, + "num_tokens": 278278805.0, + "step": 10749 + }, + { + "epoch": 1.1805403030968593, + "grad_norm": 2.170424222946167, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7161399126052856, + "num_tokens": 278307389.0, + "step": 10750 + }, + { + "epoch": 1.1806501207994728, + "grad_norm": 2.286001205444336, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7084617018699646, + "num_tokens": 278332455.0, + "step": 10751 + }, + { + "epoch": 1.1807599385020866, + "grad_norm": 2.0711593627929688, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7143856287002563, + "num_tokens": 278362515.0, + "step": 10752 + }, + { + "epoch": 1.1808697562047001, + "grad_norm": 2.228572130203247, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7361310720443726, + "num_tokens": 278388616.0, + "step": 10753 + }, + { + "epoch": 1.180979573907314, + "grad_norm": 2.385193109512329, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7084639072418213, + "num_tokens": 278414433.0, + "step": 10754 + }, + { + "epoch": 1.1810893916099274, + "grad_norm": 2.4471354484558105, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7282992005348206, + "num_tokens": 278435896.0, + "step": 10755 + }, + { + "epoch": 1.1811992093125412, + "grad_norm": 2.4096500873565674, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7427229881286621, + "num_tokens": 278457544.0, + "step": 10756 + }, + { + "epoch": 1.1813090270151547, + "grad_norm": 2.2102015018463135, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7030618190765381, + "num_tokens": 278483248.0, + "step": 10757 + }, + { + "epoch": 1.1814188447177685, + "grad_norm": 2.3839375972747803, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7106236219406128, + "num_tokens": 278506615.0, + "step": 10758 + }, + { + "epoch": 1.1815286624203822, + "grad_norm": 2.200204372406006, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7129117250442505, + "num_tokens": 278533229.0, + "step": 10759 + }, + { + "epoch": 1.1816384801229958, + "grad_norm": 2.1504313945770264, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7247084379196167, + "num_tokens": 278559948.0, + "step": 10760 + }, + { + "epoch": 1.1817482978256095, + "grad_norm": 2.298555612564087, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7162676453590393, + "num_tokens": 278583103.0, + "step": 10761 + }, + { + "epoch": 1.181858115528223, + "grad_norm": 2.2220070362091064, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7248508334159851, + "num_tokens": 278610819.0, + "step": 10762 + }, + { + "epoch": 1.1819679332308368, + "grad_norm": 2.083482027053833, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.741741418838501, + "num_tokens": 278639297.0, + "step": 10763 + }, + { + "epoch": 1.1820777509334506, + "grad_norm": 2.147688865661621, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7019774317741394, + "num_tokens": 278668029.0, + "step": 10764 + }, + { + "epoch": 1.1821875686360641, + "grad_norm": 3.0045862197875977, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7184735536575317, + "num_tokens": 278684837.0, + "step": 10765 + }, + { + "epoch": 1.1822973863386779, + "grad_norm": 2.1435015201568604, + "learning_rate": 1e-06, + "loss": 1.0562, + "mean_token_accuracy": 0.6792972087860107, + "num_tokens": 278716488.0, + "step": 10766 + }, + { + "epoch": 1.1824072040412914, + "grad_norm": 2.5406887531280518, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7152683734893799, + "num_tokens": 278739057.0, + "step": 10767 + }, + { + "epoch": 1.1825170217439052, + "grad_norm": 2.702471971511841, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7339437007904053, + "num_tokens": 278757143.0, + "step": 10768 + }, + { + "epoch": 1.1826268394465187, + "grad_norm": 2.0098488330841064, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7224230766296387, + "num_tokens": 278788023.0, + "step": 10769 + }, + { + "epoch": 1.1827366571491325, + "grad_norm": 2.2744195461273193, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.720254123210907, + "num_tokens": 278811875.0, + "step": 10770 + }, + { + "epoch": 1.182846474851746, + "grad_norm": 2.372803211212158, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7083269357681274, + "num_tokens": 278837201.0, + "step": 10771 + }, + { + "epoch": 1.1829562925543597, + "grad_norm": 2.6802945137023926, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7369141578674316, + "num_tokens": 278856995.0, + "step": 10772 + }, + { + "epoch": 1.1830661102569735, + "grad_norm": 2.4097886085510254, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7162256240844727, + "num_tokens": 278879410.0, + "step": 10773 + }, + { + "epoch": 1.183175927959587, + "grad_norm": 2.3300271034240723, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7124324440956116, + "num_tokens": 278902806.0, + "step": 10774 + }, + { + "epoch": 1.1832857456622008, + "grad_norm": 2.3228068351745605, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7062035202980042, + "num_tokens": 278928228.0, + "step": 10775 + }, + { + "epoch": 1.1833955633648143, + "grad_norm": 2.2381365299224854, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6973899006843567, + "num_tokens": 278956107.0, + "step": 10776 + }, + { + "epoch": 1.183505381067428, + "grad_norm": 2.188786745071411, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7395907044410706, + "num_tokens": 278980976.0, + "step": 10777 + }, + { + "epoch": 1.1836151987700418, + "grad_norm": 2.3335585594177246, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7187280058860779, + "num_tokens": 279006438.0, + "step": 10778 + }, + { + "epoch": 1.1837250164726554, + "grad_norm": 2.7522244453430176, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7143396139144897, + "num_tokens": 279025046.0, + "step": 10779 + }, + { + "epoch": 1.1838348341752691, + "grad_norm": 2.0970470905303955, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7186088562011719, + "num_tokens": 279054000.0, + "step": 10780 + }, + { + "epoch": 1.1839446518778827, + "grad_norm": 1.9621268510818481, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6940219402313232, + "num_tokens": 279086191.0, + "step": 10781 + }, + { + "epoch": 1.1840544695804964, + "grad_norm": 2.5689032077789307, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7222062349319458, + "num_tokens": 279105947.0, + "step": 10782 + }, + { + "epoch": 1.18416428728311, + "grad_norm": 2.2420544624328613, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7355166673660278, + "num_tokens": 279132323.0, + "step": 10783 + }, + { + "epoch": 1.1842741049857237, + "grad_norm": 2.2804720401763916, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7097412347793579, + "num_tokens": 279157452.0, + "step": 10784 + }, + { + "epoch": 1.1843839226883373, + "grad_norm": 2.15755558013916, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7166424989700317, + "num_tokens": 279185069.0, + "step": 10785 + }, + { + "epoch": 1.184493740390951, + "grad_norm": 2.1371824741363525, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7288986444473267, + "num_tokens": 279212124.0, + "step": 10786 + }, + { + "epoch": 1.1846035580935648, + "grad_norm": 2.244668483734131, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7199643850326538, + "num_tokens": 279239591.0, + "step": 10787 + }, + { + "epoch": 1.1847133757961783, + "grad_norm": 2.0580341815948486, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7268624305725098, + "num_tokens": 279269866.0, + "step": 10788 + }, + { + "epoch": 1.184823193498792, + "grad_norm": 2.3346354961395264, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.711272120475769, + "num_tokens": 279293521.0, + "step": 10789 + }, + { + "epoch": 1.1849330112014056, + "grad_norm": 1.9080166816711426, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7288435101509094, + "num_tokens": 279326470.0, + "step": 10790 + }, + { + "epoch": 1.1850428289040194, + "grad_norm": 2.173407554626465, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7502362728118896, + "num_tokens": 279351307.0, + "step": 10791 + }, + { + "epoch": 1.185152646606633, + "grad_norm": 2.1270861625671387, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6988739967346191, + "num_tokens": 279379656.0, + "step": 10792 + }, + { + "epoch": 1.1852624643092466, + "grad_norm": 2.3354856967926025, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7309566736221313, + "num_tokens": 279401365.0, + "step": 10793 + }, + { + "epoch": 1.1853722820118604, + "grad_norm": 1.9846957921981812, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7022304534912109, + "num_tokens": 279431184.0, + "step": 10794 + }, + { + "epoch": 1.185482099714474, + "grad_norm": 2.3620009422302246, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7106595039367676, + "num_tokens": 279455386.0, + "step": 10795 + }, + { + "epoch": 1.1855919174170877, + "grad_norm": 2.27822208404541, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7174968719482422, + "num_tokens": 279480157.0, + "step": 10796 + }, + { + "epoch": 1.1857017351197012, + "grad_norm": 1.8960589170455933, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7213485240936279, + "num_tokens": 279511184.0, + "step": 10797 + }, + { + "epoch": 1.185811552822315, + "grad_norm": 2.3744969367980957, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7180852293968201, + "num_tokens": 279536340.0, + "step": 10798 + }, + { + "epoch": 1.1859213705249285, + "grad_norm": 1.9998588562011719, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7260321974754333, + "num_tokens": 279568355.0, + "step": 10799 + }, + { + "epoch": 1.1860311882275423, + "grad_norm": 2.3276023864746094, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7172247171401978, + "num_tokens": 279592058.0, + "step": 10800 + }, + { + "epoch": 1.186141005930156, + "grad_norm": 2.354504108428955, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7053723335266113, + "num_tokens": 279616851.0, + "step": 10801 + }, + { + "epoch": 1.1862508236327696, + "grad_norm": 2.2419984340667725, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7188343405723572, + "num_tokens": 279643692.0, + "step": 10802 + }, + { + "epoch": 1.1863606413353833, + "grad_norm": 2.230031967163086, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.722972571849823, + "num_tokens": 279668872.0, + "step": 10803 + }, + { + "epoch": 1.1864704590379969, + "grad_norm": 2.238152265548706, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7197407484054565, + "num_tokens": 279695404.0, + "step": 10804 + }, + { + "epoch": 1.1865802767406106, + "grad_norm": 2.3889546394348145, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7262186408042908, + "num_tokens": 279718073.0, + "step": 10805 + }, + { + "epoch": 1.1866900944432242, + "grad_norm": 2.259542942047119, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7106523513793945, + "num_tokens": 279747779.0, + "step": 10806 + }, + { + "epoch": 1.186799912145838, + "grad_norm": 1.9539611339569092, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7067769169807434, + "num_tokens": 279781480.0, + "step": 10807 + }, + { + "epoch": 1.1869097298484517, + "grad_norm": 2.1237034797668457, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.716876208782196, + "num_tokens": 279809082.0, + "step": 10808 + }, + { + "epoch": 1.1870195475510652, + "grad_norm": 2.4058172702789307, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7289553284645081, + "num_tokens": 279832111.0, + "step": 10809 + }, + { + "epoch": 1.187129365253679, + "grad_norm": 2.167031764984131, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7139227390289307, + "num_tokens": 279860701.0, + "step": 10810 + }, + { + "epoch": 1.1872391829562925, + "grad_norm": 2.0239875316619873, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7065553665161133, + "num_tokens": 279891858.0, + "step": 10811 + }, + { + "epoch": 1.1873490006589063, + "grad_norm": 2.552029848098755, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7267031669616699, + "num_tokens": 279911871.0, + "step": 10812 + }, + { + "epoch": 1.1874588183615198, + "grad_norm": 2.151033639907837, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7160303592681885, + "num_tokens": 279937887.0, + "step": 10813 + }, + { + "epoch": 1.1875686360641335, + "grad_norm": 2.1313371658325195, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7118086218833923, + "num_tokens": 279965585.0, + "step": 10814 + }, + { + "epoch": 1.1876784537667473, + "grad_norm": 2.6667253971099854, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7093547582626343, + "num_tokens": 279985822.0, + "step": 10815 + }, + { + "epoch": 1.1877882714693608, + "grad_norm": 2.0337655544281006, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.721134603023529, + "num_tokens": 280016291.0, + "step": 10816 + }, + { + "epoch": 1.1878980891719746, + "grad_norm": 2.128073215484619, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7068718075752258, + "num_tokens": 280045789.0, + "step": 10817 + }, + { + "epoch": 1.1880079068745881, + "grad_norm": 2.4371397495269775, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7379838228225708, + "num_tokens": 280068256.0, + "step": 10818 + }, + { + "epoch": 1.1881177245772019, + "grad_norm": 2.3132264614105225, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7513126134872437, + "num_tokens": 280092650.0, + "step": 10819 + }, + { + "epoch": 1.1882275422798154, + "grad_norm": 2.2913548946380615, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.71702641248703, + "num_tokens": 280118424.0, + "step": 10820 + }, + { + "epoch": 1.1883373599824292, + "grad_norm": 2.294548988342285, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7198158502578735, + "num_tokens": 280142730.0, + "step": 10821 + }, + { + "epoch": 1.1884471776850427, + "grad_norm": 2.556288003921509, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7108333110809326, + "num_tokens": 280163927.0, + "step": 10822 + }, + { + "epoch": 1.1885569953876565, + "grad_norm": 2.5213587284088135, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7221838235855103, + "num_tokens": 280184953.0, + "step": 10823 + }, + { + "epoch": 1.1886668130902702, + "grad_norm": 2.576336622238159, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7111836075782776, + "num_tokens": 280206670.0, + "step": 10824 + }, + { + "epoch": 1.1887766307928838, + "grad_norm": 2.1790883541107178, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7237974405288696, + "num_tokens": 280232882.0, + "step": 10825 + }, + { + "epoch": 1.1888864484954975, + "grad_norm": 2.3217036724090576, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7287107110023499, + "num_tokens": 280257097.0, + "step": 10826 + }, + { + "epoch": 1.188996266198111, + "grad_norm": 2.1505136489868164, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7306428551673889, + "num_tokens": 280283838.0, + "step": 10827 + }, + { + "epoch": 1.1891060839007248, + "grad_norm": 2.563153028488159, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7280915379524231, + "num_tokens": 280305715.0, + "step": 10828 + }, + { + "epoch": 1.1892159016033386, + "grad_norm": 2.675353527069092, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7345202565193176, + "num_tokens": 280323948.0, + "step": 10829 + }, + { + "epoch": 1.189325719305952, + "grad_norm": 2.209550619125366, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7161452770233154, + "num_tokens": 280352256.0, + "step": 10830 + }, + { + "epoch": 1.1894355370085659, + "grad_norm": 2.325404644012451, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7011673450469971, + "num_tokens": 280381460.0, + "step": 10831 + }, + { + "epoch": 1.1895453547111794, + "grad_norm": 2.4910805225372314, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7296428680419922, + "num_tokens": 280402295.0, + "step": 10832 + }, + { + "epoch": 1.1896551724137931, + "grad_norm": 2.344559907913208, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7534345388412476, + "num_tokens": 280424774.0, + "step": 10833 + }, + { + "epoch": 1.1897649901164067, + "grad_norm": 2.597593307495117, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.703338086605072, + "num_tokens": 280444824.0, + "step": 10834 + }, + { + "epoch": 1.1898748078190204, + "grad_norm": 2.4094173908233643, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7134697437286377, + "num_tokens": 280469134.0, + "step": 10835 + }, + { + "epoch": 1.189984625521634, + "grad_norm": 2.023897171020508, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7121865749359131, + "num_tokens": 280499041.0, + "step": 10836 + }, + { + "epoch": 1.1900944432242477, + "grad_norm": 2.113337755203247, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7100422382354736, + "num_tokens": 280528314.0, + "step": 10837 + }, + { + "epoch": 1.1902042609268615, + "grad_norm": 2.225609302520752, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7123605608940125, + "num_tokens": 280554192.0, + "step": 10838 + }, + { + "epoch": 1.190314078629475, + "grad_norm": 2.1909565925598145, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7235330939292908, + "num_tokens": 280579201.0, + "step": 10839 + }, + { + "epoch": 1.1904238963320888, + "grad_norm": 1.9407826662063599, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6948617696762085, + "num_tokens": 280613242.0, + "step": 10840 + }, + { + "epoch": 1.1905337140347023, + "grad_norm": 2.115553617477417, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7246953248977661, + "num_tokens": 280641227.0, + "step": 10841 + }, + { + "epoch": 1.190643531737316, + "grad_norm": 2.009437322616577, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7126543521881104, + "num_tokens": 280671471.0, + "step": 10842 + }, + { + "epoch": 1.1907533494399298, + "grad_norm": 2.3241653442382812, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7065644264221191, + "num_tokens": 280695950.0, + "step": 10843 + }, + { + "epoch": 1.1908631671425434, + "grad_norm": 2.6976478099823, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7452992796897888, + "num_tokens": 280714361.0, + "step": 10844 + }, + { + "epoch": 1.1909729848451571, + "grad_norm": 2.6673872470855713, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7459070682525635, + "num_tokens": 280734389.0, + "step": 10845 + }, + { + "epoch": 1.1910828025477707, + "grad_norm": 2.150629758834839, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.73409104347229, + "num_tokens": 280759798.0, + "step": 10846 + }, + { + "epoch": 1.1911926202503844, + "grad_norm": 2.450093984603882, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7295297384262085, + "num_tokens": 280782156.0, + "step": 10847 + }, + { + "epoch": 1.191302437952998, + "grad_norm": 2.276397466659546, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7138605117797852, + "num_tokens": 280807703.0, + "step": 10848 + }, + { + "epoch": 1.1914122556556117, + "grad_norm": 2.321026086807251, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7124327421188354, + "num_tokens": 280831959.0, + "step": 10849 + }, + { + "epoch": 1.1915220733582252, + "grad_norm": 1.9417835474014282, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7174147367477417, + "num_tokens": 280864614.0, + "step": 10850 + }, + { + "epoch": 1.191631891060839, + "grad_norm": 2.2514472007751465, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7052489519119263, + "num_tokens": 280891174.0, + "step": 10851 + }, + { + "epoch": 1.1917417087634528, + "grad_norm": 2.132241725921631, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7161556482315063, + "num_tokens": 280919838.0, + "step": 10852 + }, + { + "epoch": 1.1918515264660663, + "grad_norm": 2.6650071144104004, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7253434658050537, + "num_tokens": 280938891.0, + "step": 10853 + }, + { + "epoch": 1.19196134416868, + "grad_norm": 2.2933292388916016, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.719514012336731, + "num_tokens": 280962915.0, + "step": 10854 + }, + { + "epoch": 1.1920711618712936, + "grad_norm": 2.063051462173462, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7060384750366211, + "num_tokens": 280991722.0, + "step": 10855 + }, + { + "epoch": 1.1921809795739073, + "grad_norm": 2.044161319732666, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7067739963531494, + "num_tokens": 281021599.0, + "step": 10856 + }, + { + "epoch": 1.192290797276521, + "grad_norm": 1.9804179668426514, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7237630486488342, + "num_tokens": 281051685.0, + "step": 10857 + }, + { + "epoch": 1.1924006149791346, + "grad_norm": 2.3869550228118896, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7137125730514526, + "num_tokens": 281074117.0, + "step": 10858 + }, + { + "epoch": 1.1925104326817484, + "grad_norm": 2.2238008975982666, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7073785662651062, + "num_tokens": 281102277.0, + "step": 10859 + }, + { + "epoch": 1.192620250384362, + "grad_norm": 2.2362329959869385, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7113772630691528, + "num_tokens": 281128923.0, + "step": 10860 + }, + { + "epoch": 1.1927300680869757, + "grad_norm": 2.0730507373809814, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7461026310920715, + "num_tokens": 281156396.0, + "step": 10861 + }, + { + "epoch": 1.1928398857895892, + "grad_norm": 2.322744846343994, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7265623211860657, + "num_tokens": 281181795.0, + "step": 10862 + }, + { + "epoch": 1.192949703492203, + "grad_norm": 2.377173662185669, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7126300930976868, + "num_tokens": 281205956.0, + "step": 10863 + }, + { + "epoch": 1.1930595211948165, + "grad_norm": 2.199361562728882, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7104761600494385, + "num_tokens": 281231044.0, + "step": 10864 + }, + { + "epoch": 1.1931693388974303, + "grad_norm": 2.2268214225769043, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.734567403793335, + "num_tokens": 281256379.0, + "step": 10865 + }, + { + "epoch": 1.193279156600044, + "grad_norm": 2.0579452514648438, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7166761755943298, + "num_tokens": 281286506.0, + "step": 10866 + }, + { + "epoch": 1.1933889743026576, + "grad_norm": 2.2207562923431396, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7121285200119019, + "num_tokens": 281315045.0, + "step": 10867 + }, + { + "epoch": 1.1934987920052713, + "grad_norm": 2.353156566619873, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7249237895011902, + "num_tokens": 281338268.0, + "step": 10868 + }, + { + "epoch": 1.1936086097078848, + "grad_norm": 2.6131033897399902, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.731967568397522, + "num_tokens": 281358237.0, + "step": 10869 + }, + { + "epoch": 1.1937184274104986, + "grad_norm": 2.4796741008758545, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7389814853668213, + "num_tokens": 281379501.0, + "step": 10870 + }, + { + "epoch": 1.1938282451131121, + "grad_norm": 2.512359142303467, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7283164262771606, + "num_tokens": 281399907.0, + "step": 10871 + }, + { + "epoch": 1.193938062815726, + "grad_norm": 2.156519651412964, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7195816040039062, + "num_tokens": 281426171.0, + "step": 10872 + }, + { + "epoch": 1.1940478805183394, + "grad_norm": 2.2789556980133057, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7132924795150757, + "num_tokens": 281452592.0, + "step": 10873 + }, + { + "epoch": 1.1941576982209532, + "grad_norm": 2.185743808746338, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7053770422935486, + "num_tokens": 281479924.0, + "step": 10874 + }, + { + "epoch": 1.194267515923567, + "grad_norm": 2.3626835346221924, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7200157642364502, + "num_tokens": 281503398.0, + "step": 10875 + }, + { + "epoch": 1.1943773336261805, + "grad_norm": 2.3353281021118164, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7201665043830872, + "num_tokens": 281525676.0, + "step": 10876 + }, + { + "epoch": 1.1944871513287942, + "grad_norm": 2.2833995819091797, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7269631028175354, + "num_tokens": 281551631.0, + "step": 10877 + }, + { + "epoch": 1.1945969690314078, + "grad_norm": 2.576777696609497, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7193424105644226, + "num_tokens": 281572796.0, + "step": 10878 + }, + { + "epoch": 1.1947067867340215, + "grad_norm": 2.421938180923462, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7142722606658936, + "num_tokens": 281597170.0, + "step": 10879 + }, + { + "epoch": 1.1948166044366353, + "grad_norm": 2.3930859565734863, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7277618050575256, + "num_tokens": 281620614.0, + "step": 10880 + }, + { + "epoch": 1.1949264221392488, + "grad_norm": 2.224250555038452, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7171276807785034, + "num_tokens": 281646288.0, + "step": 10881 + }, + { + "epoch": 1.1950362398418626, + "grad_norm": 2.3912806510925293, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7329065799713135, + "num_tokens": 281667782.0, + "step": 10882 + }, + { + "epoch": 1.1951460575444761, + "grad_norm": 2.2033331394195557, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7244166135787964, + "num_tokens": 281694308.0, + "step": 10883 + }, + { + "epoch": 1.1952558752470899, + "grad_norm": 2.4092512130737305, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7312748432159424, + "num_tokens": 281716188.0, + "step": 10884 + }, + { + "epoch": 1.1953656929497034, + "grad_norm": 2.189816951751709, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.697365403175354, + "num_tokens": 281743927.0, + "step": 10885 + }, + { + "epoch": 1.1954755106523172, + "grad_norm": 2.0960545539855957, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7167271971702576, + "num_tokens": 281770630.0, + "step": 10886 + }, + { + "epoch": 1.1955853283549307, + "grad_norm": 2.6195859909057617, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7011017799377441, + "num_tokens": 281793140.0, + "step": 10887 + }, + { + "epoch": 1.1956951460575445, + "grad_norm": 2.4848878383636475, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7324211597442627, + "num_tokens": 281814623.0, + "step": 10888 + }, + { + "epoch": 1.1958049637601582, + "grad_norm": 2.3629791736602783, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7172210812568665, + "num_tokens": 281838998.0, + "step": 10889 + }, + { + "epoch": 1.1959147814627717, + "grad_norm": 2.0835447311401367, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.732043445110321, + "num_tokens": 281867363.0, + "step": 10890 + }, + { + "epoch": 1.1960245991653855, + "grad_norm": 2.3756794929504395, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7248536348342896, + "num_tokens": 281889231.0, + "step": 10891 + }, + { + "epoch": 1.196134416867999, + "grad_norm": 2.173356771469116, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7104349136352539, + "num_tokens": 281917109.0, + "step": 10892 + }, + { + "epoch": 1.1962442345706128, + "grad_norm": 2.365128517150879, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7095545530319214, + "num_tokens": 281942238.0, + "step": 10893 + }, + { + "epoch": 1.1963540522732266, + "grad_norm": 2.5676541328430176, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7239441871643066, + "num_tokens": 281963200.0, + "step": 10894 + }, + { + "epoch": 1.19646386997584, + "grad_norm": 2.202868700027466, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7078289985656738, + "num_tokens": 281991008.0, + "step": 10895 + }, + { + "epoch": 1.1965736876784538, + "grad_norm": 2.429971694946289, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7349822521209717, + "num_tokens": 282013040.0, + "step": 10896 + }, + { + "epoch": 1.1966835053810674, + "grad_norm": 2.6523773670196533, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7272153496742249, + "num_tokens": 282033724.0, + "step": 10897 + }, + { + "epoch": 1.1967933230836811, + "grad_norm": 2.037712574005127, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7022724151611328, + "num_tokens": 282064888.0, + "step": 10898 + }, + { + "epoch": 1.1969031407862947, + "grad_norm": 2.1842827796936035, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7197964787483215, + "num_tokens": 282091842.0, + "step": 10899 + }, + { + "epoch": 1.1970129584889084, + "grad_norm": 2.037763833999634, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7294087409973145, + "num_tokens": 282121276.0, + "step": 10900 + }, + { + "epoch": 1.197122776191522, + "grad_norm": 2.879697799682617, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7322404384613037, + "num_tokens": 282138436.0, + "step": 10901 + }, + { + "epoch": 1.1972325938941357, + "grad_norm": 2.3556838035583496, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7273300290107727, + "num_tokens": 282161974.0, + "step": 10902 + }, + { + "epoch": 1.1973424115967495, + "grad_norm": 2.2900547981262207, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.6969658136367798, + "num_tokens": 282187070.0, + "step": 10903 + }, + { + "epoch": 1.197452229299363, + "grad_norm": 2.5818910598754883, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.716166079044342, + "num_tokens": 282210391.0, + "step": 10904 + }, + { + "epoch": 1.1975620470019768, + "grad_norm": 2.180168390274048, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7317332625389099, + "num_tokens": 282236268.0, + "step": 10905 + }, + { + "epoch": 1.1976718647045903, + "grad_norm": 1.9944908618927002, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6929879188537598, + "num_tokens": 282266725.0, + "step": 10906 + }, + { + "epoch": 1.197781682407204, + "grad_norm": 2.4901938438415527, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7503145933151245, + "num_tokens": 282286293.0, + "step": 10907 + }, + { + "epoch": 1.1978915001098178, + "grad_norm": 2.1486799716949463, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7043434381484985, + "num_tokens": 282317493.0, + "step": 10908 + }, + { + "epoch": 1.1980013178124314, + "grad_norm": 2.4354639053344727, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6942827701568604, + "num_tokens": 282342822.0, + "step": 10909 + }, + { + "epoch": 1.1981111355150451, + "grad_norm": 2.3495945930480957, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7414946556091309, + "num_tokens": 282367191.0, + "step": 10910 + }, + { + "epoch": 1.1982209532176586, + "grad_norm": 2.0684096813201904, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.6974164843559265, + "num_tokens": 282397156.0, + "step": 10911 + }, + { + "epoch": 1.1983307709202724, + "grad_norm": 2.348949670791626, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.729851484298706, + "num_tokens": 282422169.0, + "step": 10912 + }, + { + "epoch": 1.198440588622886, + "grad_norm": 2.00890851020813, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7222741842269897, + "num_tokens": 282451878.0, + "step": 10913 + }, + { + "epoch": 1.1985504063254997, + "grad_norm": 2.244720935821533, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7063164710998535, + "num_tokens": 282479141.0, + "step": 10914 + }, + { + "epoch": 1.1986602240281132, + "grad_norm": 2.0795891284942627, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7075780630111694, + "num_tokens": 282508341.0, + "step": 10915 + }, + { + "epoch": 1.198770041730727, + "grad_norm": 2.2183711528778076, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7276954650878906, + "num_tokens": 282533930.0, + "step": 10916 + }, + { + "epoch": 1.1988798594333407, + "grad_norm": 2.341468095779419, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7215040922164917, + "num_tokens": 282556891.0, + "step": 10917 + }, + { + "epoch": 1.1989896771359543, + "grad_norm": 2.5517656803131104, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7264353036880493, + "num_tokens": 282578796.0, + "step": 10918 + }, + { + "epoch": 1.199099494838568, + "grad_norm": 2.3093388080596924, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7099803686141968, + "num_tokens": 282603968.0, + "step": 10919 + }, + { + "epoch": 1.1992093125411816, + "grad_norm": 2.6897494792938232, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7224955558776855, + "num_tokens": 282623436.0, + "step": 10920 + }, + { + "epoch": 1.1993191302437953, + "grad_norm": 2.4405810832977295, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7256360054016113, + "num_tokens": 282644872.0, + "step": 10921 + }, + { + "epoch": 1.199428947946409, + "grad_norm": 2.0576837062835693, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6860251426696777, + "num_tokens": 282675436.0, + "step": 10922 + }, + { + "epoch": 1.1995387656490226, + "grad_norm": 2.4034433364868164, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7417693138122559, + "num_tokens": 282698546.0, + "step": 10923 + }, + { + "epoch": 1.1996485833516364, + "grad_norm": 2.2974166870117188, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7125287055969238, + "num_tokens": 282722349.0, + "step": 10924 + }, + { + "epoch": 1.19975840105425, + "grad_norm": 2.4548230171203613, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7032593488693237, + "num_tokens": 282745554.0, + "step": 10925 + }, + { + "epoch": 1.1998682187568637, + "grad_norm": 2.374662160873413, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7234807014465332, + "num_tokens": 282770136.0, + "step": 10926 + }, + { + "epoch": 1.1999780364594772, + "grad_norm": 2.400571823120117, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7143236398696899, + "num_tokens": 282793880.0, + "step": 10927 + }, + { + "epoch": 1.200087854162091, + "grad_norm": 2.3526716232299805, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7216897010803223, + "num_tokens": 282819050.0, + "step": 10928 + }, + { + "epoch": 1.2001976718647045, + "grad_norm": 2.5770204067230225, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7299414873123169, + "num_tokens": 282838645.0, + "step": 10929 + }, + { + "epoch": 1.2003074895673183, + "grad_norm": 1.9778616428375244, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7126773595809937, + "num_tokens": 282870167.0, + "step": 10930 + }, + { + "epoch": 1.200417307269932, + "grad_norm": 2.10652756690979, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.718626856803894, + "num_tokens": 282900252.0, + "step": 10931 + }, + { + "epoch": 1.2005271249725455, + "grad_norm": 2.2422983646392822, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7116231918334961, + "num_tokens": 282924793.0, + "step": 10932 + }, + { + "epoch": 1.2006369426751593, + "grad_norm": 2.1345975399017334, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7121328115463257, + "num_tokens": 282951368.0, + "step": 10933 + }, + { + "epoch": 1.2007467603777728, + "grad_norm": 2.0187251567840576, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6895466446876526, + "num_tokens": 282982310.0, + "step": 10934 + }, + { + "epoch": 1.2008565780803866, + "grad_norm": 2.4567782878875732, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7097629904747009, + "num_tokens": 283005262.0, + "step": 10935 + }, + { + "epoch": 1.2009663957830001, + "grad_norm": 2.4221487045288086, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.736358642578125, + "num_tokens": 283029265.0, + "step": 10936 + }, + { + "epoch": 1.2010762134856139, + "grad_norm": 2.2278919219970703, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7371585369110107, + "num_tokens": 283055446.0, + "step": 10937 + }, + { + "epoch": 1.2011860311882274, + "grad_norm": 2.2686634063720703, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.749008297920227, + "num_tokens": 283080794.0, + "step": 10938 + }, + { + "epoch": 1.2012958488908412, + "grad_norm": 2.4130494594573975, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7173311710357666, + "num_tokens": 283103477.0, + "step": 10939 + }, + { + "epoch": 1.201405666593455, + "grad_norm": 2.3540689945220947, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7019171118736267, + "num_tokens": 283128921.0, + "step": 10940 + }, + { + "epoch": 1.2015154842960685, + "grad_norm": 1.954947829246521, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.714063823223114, + "num_tokens": 283159069.0, + "step": 10941 + }, + { + "epoch": 1.2016253019986822, + "grad_norm": 2.440474510192871, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7197756767272949, + "num_tokens": 283182509.0, + "step": 10942 + }, + { + "epoch": 1.2017351197012958, + "grad_norm": 2.1449484825134277, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6935226917266846, + "num_tokens": 283211447.0, + "step": 10943 + }, + { + "epoch": 1.2018449374039095, + "grad_norm": 2.1150925159454346, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7226125001907349, + "num_tokens": 283239895.0, + "step": 10944 + }, + { + "epoch": 1.2019547551065233, + "grad_norm": 2.8309686183929443, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7111287713050842, + "num_tokens": 283260397.0, + "step": 10945 + }, + { + "epoch": 1.2020645728091368, + "grad_norm": 2.2833733558654785, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7329056859016418, + "num_tokens": 283285092.0, + "step": 10946 + }, + { + "epoch": 1.2021743905117506, + "grad_norm": 2.405200481414795, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7306526899337769, + "num_tokens": 283308409.0, + "step": 10947 + }, + { + "epoch": 1.202284208214364, + "grad_norm": 2.198376178741455, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7056341171264648, + "num_tokens": 283337028.0, + "step": 10948 + }, + { + "epoch": 1.2023940259169779, + "grad_norm": 2.1753287315368652, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7235381603240967, + "num_tokens": 283362535.0, + "step": 10949 + }, + { + "epoch": 1.2025038436195914, + "grad_norm": 2.108417272567749, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6872043609619141, + "num_tokens": 283392480.0, + "step": 10950 + }, + { + "epoch": 1.2026136613222052, + "grad_norm": 2.4371540546417236, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7218213081359863, + "num_tokens": 283414290.0, + "step": 10951 + }, + { + "epoch": 1.2027234790248187, + "grad_norm": 2.515688896179199, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7399212121963501, + "num_tokens": 283434515.0, + "step": 10952 + }, + { + "epoch": 1.2028332967274324, + "grad_norm": 2.3315346240997314, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7223477363586426, + "num_tokens": 283459691.0, + "step": 10953 + }, + { + "epoch": 1.2029431144300462, + "grad_norm": 2.270786762237549, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7256869077682495, + "num_tokens": 283483621.0, + "step": 10954 + }, + { + "epoch": 1.2030529321326597, + "grad_norm": 2.2902345657348633, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7325922250747681, + "num_tokens": 283507753.0, + "step": 10955 + }, + { + "epoch": 1.2031627498352735, + "grad_norm": 2.3281664848327637, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7172260284423828, + "num_tokens": 283532007.0, + "step": 10956 + }, + { + "epoch": 1.203272567537887, + "grad_norm": 2.4227066040039062, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.7019940614700317, + "num_tokens": 283555781.0, + "step": 10957 + }, + { + "epoch": 1.2033823852405008, + "grad_norm": 2.4204463958740234, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7362365126609802, + "num_tokens": 283578730.0, + "step": 10958 + }, + { + "epoch": 1.2034922029431145, + "grad_norm": 2.2473175525665283, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7459907531738281, + "num_tokens": 283602621.0, + "step": 10959 + }, + { + "epoch": 1.203602020645728, + "grad_norm": 1.8908283710479736, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7181291580200195, + "num_tokens": 283638322.0, + "step": 10960 + }, + { + "epoch": 1.2037118383483418, + "grad_norm": 2.07055401802063, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7148018479347229, + "num_tokens": 283668495.0, + "step": 10961 + }, + { + "epoch": 1.2038216560509554, + "grad_norm": 2.5374724864959717, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7284660339355469, + "num_tokens": 283688969.0, + "step": 10962 + }, + { + "epoch": 1.2039314737535691, + "grad_norm": 2.2375826835632324, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6998540163040161, + "num_tokens": 283717810.0, + "step": 10963 + }, + { + "epoch": 1.2040412914561827, + "grad_norm": 2.458712100982666, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7066082954406738, + "num_tokens": 283740254.0, + "step": 10964 + }, + { + "epoch": 1.2041511091587964, + "grad_norm": 2.5756564140319824, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.72493577003479, + "num_tokens": 283760259.0, + "step": 10965 + }, + { + "epoch": 1.20426092686141, + "grad_norm": 2.31816029548645, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7154054045677185, + "num_tokens": 283784178.0, + "step": 10966 + }, + { + "epoch": 1.2043707445640237, + "grad_norm": 2.3677244186401367, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7428736686706543, + "num_tokens": 283806168.0, + "step": 10967 + }, + { + "epoch": 1.2044805622666375, + "grad_norm": 2.385582685470581, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7286480665206909, + "num_tokens": 283830981.0, + "step": 10968 + }, + { + "epoch": 1.204590379969251, + "grad_norm": 2.2176408767700195, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7131334543228149, + "num_tokens": 283856809.0, + "step": 10969 + }, + { + "epoch": 1.2047001976718648, + "grad_norm": 2.4356329441070557, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7215683460235596, + "num_tokens": 283880097.0, + "step": 10970 + }, + { + "epoch": 1.2048100153744783, + "grad_norm": 2.2884654998779297, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7324264049530029, + "num_tokens": 283904831.0, + "step": 10971 + }, + { + "epoch": 1.204919833077092, + "grad_norm": 2.2110352516174316, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7093175649642944, + "num_tokens": 283931547.0, + "step": 10972 + }, + { + "epoch": 1.2050296507797058, + "grad_norm": 2.5999884605407715, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7243617177009583, + "num_tokens": 283950698.0, + "step": 10973 + }, + { + "epoch": 1.2051394684823193, + "grad_norm": 2.552581310272217, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7261264324188232, + "num_tokens": 283974470.0, + "step": 10974 + }, + { + "epoch": 1.205249286184933, + "grad_norm": 2.3673579692840576, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7068840265274048, + "num_tokens": 284001087.0, + "step": 10975 + }, + { + "epoch": 1.2053591038875466, + "grad_norm": 2.6870434284210205, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7385040521621704, + "num_tokens": 284020322.0, + "step": 10976 + }, + { + "epoch": 1.2054689215901604, + "grad_norm": 2.5703954696655273, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7006020545959473, + "num_tokens": 284040858.0, + "step": 10977 + }, + { + "epoch": 1.205578739292774, + "grad_norm": 2.209864377975464, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.701898992061615, + "num_tokens": 284067636.0, + "step": 10978 + }, + { + "epoch": 1.2056885569953877, + "grad_norm": 2.1122868061065674, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.7038214802742004, + "num_tokens": 284096336.0, + "step": 10979 + }, + { + "epoch": 1.2057983746980012, + "grad_norm": 2.4235856533050537, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7103073000907898, + "num_tokens": 284120413.0, + "step": 10980 + }, + { + "epoch": 1.205908192400615, + "grad_norm": 2.161613702774048, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.724332869052887, + "num_tokens": 284145303.0, + "step": 10981 + }, + { + "epoch": 1.2060180101032287, + "grad_norm": 2.2739012241363525, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7588043212890625, + "num_tokens": 284167398.0, + "step": 10982 + }, + { + "epoch": 1.2061278278058423, + "grad_norm": 2.3313896656036377, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7227346897125244, + "num_tokens": 284191959.0, + "step": 10983 + }, + { + "epoch": 1.206237645508456, + "grad_norm": 2.2496936321258545, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7168838977813721, + "num_tokens": 284217135.0, + "step": 10984 + }, + { + "epoch": 1.2063474632110696, + "grad_norm": 1.9108428955078125, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6945136785507202, + "num_tokens": 284252636.0, + "step": 10985 + }, + { + "epoch": 1.2064572809136833, + "grad_norm": 2.265381336212158, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7286795377731323, + "num_tokens": 284277462.0, + "step": 10986 + }, + { + "epoch": 1.2065670986162969, + "grad_norm": 2.4343886375427246, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7479468584060669, + "num_tokens": 284298181.0, + "step": 10987 + }, + { + "epoch": 1.2066769163189106, + "grad_norm": 2.1901650428771973, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7103069424629211, + "num_tokens": 284326972.0, + "step": 10988 + }, + { + "epoch": 1.2067867340215244, + "grad_norm": 2.285149097442627, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7281596660614014, + "num_tokens": 284352298.0, + "step": 10989 + }, + { + "epoch": 1.206896551724138, + "grad_norm": 2.6676182746887207, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7212791442871094, + "num_tokens": 284371893.0, + "step": 10990 + }, + { + "epoch": 1.2070063694267517, + "grad_norm": 2.2441554069519043, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.706950306892395, + "num_tokens": 284398281.0, + "step": 10991 + }, + { + "epoch": 1.2071161871293652, + "grad_norm": 2.403688669204712, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7179481983184814, + "num_tokens": 284421089.0, + "step": 10992 + }, + { + "epoch": 1.207226004831979, + "grad_norm": 2.6316475868225098, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7344695925712585, + "num_tokens": 284441192.0, + "step": 10993 + }, + { + "epoch": 1.2073358225345925, + "grad_norm": 2.1117289066314697, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6977251768112183, + "num_tokens": 284471296.0, + "step": 10994 + }, + { + "epoch": 1.2074456402372062, + "grad_norm": 2.1592259407043457, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.710458517074585, + "num_tokens": 284497238.0, + "step": 10995 + }, + { + "epoch": 1.20755545793982, + "grad_norm": 2.114258050918579, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.701917290687561, + "num_tokens": 284525660.0, + "step": 10996 + }, + { + "epoch": 1.2076652756424335, + "grad_norm": 2.147357225418091, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7240159511566162, + "num_tokens": 284554370.0, + "step": 10997 + }, + { + "epoch": 1.2077750933450473, + "grad_norm": 2.36633038520813, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7278627753257751, + "num_tokens": 284577055.0, + "step": 10998 + }, + { + "epoch": 1.2078849110476608, + "grad_norm": 2.295619249343872, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7326140403747559, + "num_tokens": 284601534.0, + "step": 10999 + }, + { + "epoch": 1.2079947287502746, + "grad_norm": 2.4721343517303467, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7200760841369629, + "num_tokens": 284624533.0, + "step": 11000 + }, + { + "epoch": 1.2081045464528881, + "grad_norm": 2.000195264816284, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6803706884384155, + "num_tokens": 284655548.0, + "step": 11001 + }, + { + "epoch": 1.2082143641555019, + "grad_norm": 2.3009133338928223, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7157979011535645, + "num_tokens": 284678884.0, + "step": 11002 + }, + { + "epoch": 1.2083241818581154, + "grad_norm": 2.3783910274505615, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7156251668930054, + "num_tokens": 284703846.0, + "step": 11003 + }, + { + "epoch": 1.2084339995607292, + "grad_norm": 2.1095945835113525, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7224617600440979, + "num_tokens": 284729345.0, + "step": 11004 + }, + { + "epoch": 1.208543817263343, + "grad_norm": 2.028003215789795, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7199897170066833, + "num_tokens": 284759880.0, + "step": 11005 + }, + { + "epoch": 1.2086536349659565, + "grad_norm": 2.6739296913146973, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7387567758560181, + "num_tokens": 284778553.0, + "step": 11006 + }, + { + "epoch": 1.2087634526685702, + "grad_norm": 2.164457321166992, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7351595759391785, + "num_tokens": 284803349.0, + "step": 11007 + }, + { + "epoch": 1.2088732703711838, + "grad_norm": 2.0941789150238037, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6933295726776123, + "num_tokens": 284836162.0, + "step": 11008 + }, + { + "epoch": 1.2089830880737975, + "grad_norm": 2.544187068939209, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7289755344390869, + "num_tokens": 284855788.0, + "step": 11009 + }, + { + "epoch": 1.2090929057764113, + "grad_norm": 2.3537237644195557, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7191250324249268, + "num_tokens": 284880287.0, + "step": 11010 + }, + { + "epoch": 1.2092027234790248, + "grad_norm": 2.2556653022766113, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7271895408630371, + "num_tokens": 284904543.0, + "step": 11011 + }, + { + "epoch": 1.2093125411816386, + "grad_norm": 2.1144495010375977, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7108472585678101, + "num_tokens": 284933876.0, + "step": 11012 + }, + { + "epoch": 1.209422358884252, + "grad_norm": 2.244816541671753, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7106891870498657, + "num_tokens": 284961257.0, + "step": 11013 + }, + { + "epoch": 1.2095321765868658, + "grad_norm": 2.2101621627807617, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6951689124107361, + "num_tokens": 284988656.0, + "step": 11014 + }, + { + "epoch": 1.2096419942894794, + "grad_norm": 2.537768602371216, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7314751148223877, + "num_tokens": 285009556.0, + "step": 11015 + }, + { + "epoch": 1.2097518119920931, + "grad_norm": 2.246723175048828, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7387305498123169, + "num_tokens": 285034371.0, + "step": 11016 + }, + { + "epoch": 1.2098616296947067, + "grad_norm": 2.364565372467041, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7267913222312927, + "num_tokens": 285056245.0, + "step": 11017 + }, + { + "epoch": 1.2099714473973204, + "grad_norm": 2.235570192337036, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7193893194198608, + "num_tokens": 285081384.0, + "step": 11018 + }, + { + "epoch": 1.2100812650999342, + "grad_norm": 2.555729627609253, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7271660566329956, + "num_tokens": 285101546.0, + "step": 11019 + }, + { + "epoch": 1.2101910828025477, + "grad_norm": 2.092273473739624, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7361643314361572, + "num_tokens": 285128909.0, + "step": 11020 + }, + { + "epoch": 1.2103009005051615, + "grad_norm": 2.65210223197937, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.714233934879303, + "num_tokens": 285149217.0, + "step": 11021 + }, + { + "epoch": 1.210410718207775, + "grad_norm": 2.2917397022247314, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7423017621040344, + "num_tokens": 285173057.0, + "step": 11022 + }, + { + "epoch": 1.2105205359103888, + "grad_norm": 2.3024706840515137, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7298930883407593, + "num_tokens": 285198193.0, + "step": 11023 + }, + { + "epoch": 1.2106303536130025, + "grad_norm": 2.2055554389953613, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7094416618347168, + "num_tokens": 285226208.0, + "step": 11024 + }, + { + "epoch": 1.210740171315616, + "grad_norm": 2.351351022720337, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7155501842498779, + "num_tokens": 285249441.0, + "step": 11025 + }, + { + "epoch": 1.2108499890182298, + "grad_norm": 2.187765121459961, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.744897723197937, + "num_tokens": 285274785.0, + "step": 11026 + }, + { + "epoch": 1.2109598067208434, + "grad_norm": 1.9507076740264893, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.6924880146980286, + "num_tokens": 285311003.0, + "step": 11027 + }, + { + "epoch": 1.2110696244234571, + "grad_norm": 2.3594253063201904, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7250252366065979, + "num_tokens": 285334538.0, + "step": 11028 + }, + { + "epoch": 1.2111794421260706, + "grad_norm": 2.1965057849884033, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7187246084213257, + "num_tokens": 285361596.0, + "step": 11029 + }, + { + "epoch": 1.2112892598286844, + "grad_norm": 2.1600534915924072, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7059025764465332, + "num_tokens": 285389838.0, + "step": 11030 + }, + { + "epoch": 1.211399077531298, + "grad_norm": 2.132197380065918, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6991884708404541, + "num_tokens": 285417739.0, + "step": 11031 + }, + { + "epoch": 1.2115088952339117, + "grad_norm": 2.872887134552002, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7315561175346375, + "num_tokens": 285435299.0, + "step": 11032 + }, + { + "epoch": 1.2116187129365255, + "grad_norm": 2.38598370552063, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7140429019927979, + "num_tokens": 285460233.0, + "step": 11033 + }, + { + "epoch": 1.211728530639139, + "grad_norm": 2.3977365493774414, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7135598659515381, + "num_tokens": 285485153.0, + "step": 11034 + }, + { + "epoch": 1.2118383483417527, + "grad_norm": 2.255359649658203, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7317115068435669, + "num_tokens": 285510008.0, + "step": 11035 + }, + { + "epoch": 1.2119481660443663, + "grad_norm": 2.2177889347076416, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7156740427017212, + "num_tokens": 285535003.0, + "step": 11036 + }, + { + "epoch": 1.21205798374698, + "grad_norm": 2.061387538909912, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7115169763565063, + "num_tokens": 285565024.0, + "step": 11037 + }, + { + "epoch": 1.2121678014495938, + "grad_norm": 1.9115053415298462, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.723615288734436, + "num_tokens": 285598590.0, + "step": 11038 + }, + { + "epoch": 1.2122776191522073, + "grad_norm": 2.476559638977051, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7242873907089233, + "num_tokens": 285619570.0, + "step": 11039 + }, + { + "epoch": 1.212387436854821, + "grad_norm": 2.271653175354004, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6989623308181763, + "num_tokens": 285644388.0, + "step": 11040 + }, + { + "epoch": 1.2124972545574346, + "grad_norm": 2.0502419471740723, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6960369348526001, + "num_tokens": 285676890.0, + "step": 11041 + }, + { + "epoch": 1.2126070722600484, + "grad_norm": 2.658604621887207, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7390573024749756, + "num_tokens": 285694848.0, + "step": 11042 + }, + { + "epoch": 1.212716889962662, + "grad_norm": 2.6191585063934326, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7281224727630615, + "num_tokens": 285713171.0, + "step": 11043 + }, + { + "epoch": 1.2128267076652757, + "grad_norm": 2.2689013481140137, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7524583339691162, + "num_tokens": 285736481.0, + "step": 11044 + }, + { + "epoch": 1.2129365253678892, + "grad_norm": 2.0406246185302734, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6924067735671997, + "num_tokens": 285766170.0, + "step": 11045 + }, + { + "epoch": 1.213046343070503, + "grad_norm": 2.9558420181274414, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7412722110748291, + "num_tokens": 285782962.0, + "step": 11046 + }, + { + "epoch": 1.2131561607731167, + "grad_norm": 2.3929669857025146, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7371940612792969, + "num_tokens": 285805223.0, + "step": 11047 + }, + { + "epoch": 1.2132659784757303, + "grad_norm": 2.2844276428222656, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.722603440284729, + "num_tokens": 285830478.0, + "step": 11048 + }, + { + "epoch": 1.213375796178344, + "grad_norm": 2.779031276702881, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.7578201293945312, + "num_tokens": 285848620.0, + "step": 11049 + }, + { + "epoch": 1.2134856138809575, + "grad_norm": 2.1810150146484375, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7086991667747498, + "num_tokens": 285876089.0, + "step": 11050 + }, + { + "epoch": 1.2135954315835713, + "grad_norm": 2.2991323471069336, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7418810725212097, + "num_tokens": 285900376.0, + "step": 11051 + }, + { + "epoch": 1.2137052492861848, + "grad_norm": 2.27315092086792, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7350678443908691, + "num_tokens": 285923703.0, + "step": 11052 + }, + { + "epoch": 1.2138150669887986, + "grad_norm": 2.2794251441955566, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7009694576263428, + "num_tokens": 285950717.0, + "step": 11053 + }, + { + "epoch": 1.2139248846914121, + "grad_norm": 2.3965578079223633, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7038943767547607, + "num_tokens": 285973567.0, + "step": 11054 + }, + { + "epoch": 1.2140347023940259, + "grad_norm": 2.41038179397583, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7096452713012695, + "num_tokens": 285997017.0, + "step": 11055 + }, + { + "epoch": 1.2141445200966396, + "grad_norm": 2.1291282176971436, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.6837241649627686, + "num_tokens": 286028655.0, + "step": 11056 + }, + { + "epoch": 1.2142543377992532, + "grad_norm": 2.2590749263763428, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7129552960395813, + "num_tokens": 286054292.0, + "step": 11057 + }, + { + "epoch": 1.214364155501867, + "grad_norm": 2.2387993335723877, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7284804582595825, + "num_tokens": 286079944.0, + "step": 11058 + }, + { + "epoch": 1.2144739732044805, + "grad_norm": 2.2065627574920654, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.6994956731796265, + "num_tokens": 286108619.0, + "step": 11059 + }, + { + "epoch": 1.2145837909070942, + "grad_norm": 2.1006593704223633, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7321500778198242, + "num_tokens": 286137641.0, + "step": 11060 + }, + { + "epoch": 1.214693608609708, + "grad_norm": 2.168459415435791, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7209958434104919, + "num_tokens": 286163781.0, + "step": 11061 + }, + { + "epoch": 1.2148034263123215, + "grad_norm": 2.0806615352630615, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7090108394622803, + "num_tokens": 286193706.0, + "step": 11062 + }, + { + "epoch": 1.2149132440149353, + "grad_norm": 2.030853509902954, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7055692672729492, + "num_tokens": 286224522.0, + "step": 11063 + }, + { + "epoch": 1.2150230617175488, + "grad_norm": 2.235081195831299, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7338287830352783, + "num_tokens": 286249405.0, + "step": 11064 + }, + { + "epoch": 1.2151328794201626, + "grad_norm": 2.115889549255371, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7225047945976257, + "num_tokens": 286278531.0, + "step": 11065 + }, + { + "epoch": 1.215242697122776, + "grad_norm": 1.9723588228225708, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7142449617385864, + "num_tokens": 286308331.0, + "step": 11066 + }, + { + "epoch": 1.2153525148253899, + "grad_norm": 2.0318615436553955, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7059017419815063, + "num_tokens": 286339208.0, + "step": 11067 + }, + { + "epoch": 1.2154623325280034, + "grad_norm": 2.3664228916168213, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7343692779541016, + "num_tokens": 286361971.0, + "step": 11068 + }, + { + "epoch": 1.2155721502306172, + "grad_norm": 1.9625558853149414, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7187026143074036, + "num_tokens": 286392775.0, + "step": 11069 + }, + { + "epoch": 1.215681967933231, + "grad_norm": 2.620140552520752, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.729653000831604, + "num_tokens": 286411263.0, + "step": 11070 + }, + { + "epoch": 1.2157917856358444, + "grad_norm": 2.1707780361175537, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7137783169746399, + "num_tokens": 286437434.0, + "step": 11071 + }, + { + "epoch": 1.2159016033384582, + "grad_norm": 2.270200252532959, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.723975419998169, + "num_tokens": 286461643.0, + "step": 11072 + }, + { + "epoch": 1.2160114210410717, + "grad_norm": 2.033717632293701, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7071013450622559, + "num_tokens": 286493286.0, + "step": 11073 + }, + { + "epoch": 1.2161212387436855, + "grad_norm": 2.5306336879730225, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7192059755325317, + "num_tokens": 286514586.0, + "step": 11074 + }, + { + "epoch": 1.2162310564462993, + "grad_norm": 2.6320126056671143, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7379561066627502, + "num_tokens": 286533142.0, + "step": 11075 + }, + { + "epoch": 1.2163408741489128, + "grad_norm": 2.2240865230560303, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7164911031723022, + "num_tokens": 286560108.0, + "step": 11076 + }, + { + "epoch": 1.2164506918515265, + "grad_norm": 1.9687440395355225, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6897091269493103, + "num_tokens": 286592101.0, + "step": 11077 + }, + { + "epoch": 1.21656050955414, + "grad_norm": 2.363886594772339, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7122200727462769, + "num_tokens": 286614678.0, + "step": 11078 + }, + { + "epoch": 1.2166703272567538, + "grad_norm": 2.5466737747192383, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7119402885437012, + "num_tokens": 286636211.0, + "step": 11079 + }, + { + "epoch": 1.2167801449593674, + "grad_norm": 2.2128636837005615, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7218056917190552, + "num_tokens": 286662911.0, + "step": 11080 + }, + { + "epoch": 1.2168899626619811, + "grad_norm": 2.129897356033325, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7193546295166016, + "num_tokens": 286690482.0, + "step": 11081 + }, + { + "epoch": 1.2169997803645947, + "grad_norm": 2.274069309234619, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7030516862869263, + "num_tokens": 286716439.0, + "step": 11082 + }, + { + "epoch": 1.2171095980672084, + "grad_norm": 2.0224685668945312, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7268275618553162, + "num_tokens": 286745981.0, + "step": 11083 + }, + { + "epoch": 1.2172194157698222, + "grad_norm": 2.1274445056915283, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7164373397827148, + "num_tokens": 286774086.0, + "step": 11084 + }, + { + "epoch": 1.2173292334724357, + "grad_norm": 2.3290390968322754, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7223755121231079, + "num_tokens": 286797551.0, + "step": 11085 + }, + { + "epoch": 1.2174390511750495, + "grad_norm": 2.4517080783843994, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7176167964935303, + "num_tokens": 286821481.0, + "step": 11086 + }, + { + "epoch": 1.217548868877663, + "grad_norm": 2.0616977214813232, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7023388147354126, + "num_tokens": 286851130.0, + "step": 11087 + }, + { + "epoch": 1.2176586865802768, + "grad_norm": 2.5911214351654053, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7303099036216736, + "num_tokens": 286870940.0, + "step": 11088 + }, + { + "epoch": 1.2177685042828905, + "grad_norm": 2.3797760009765625, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7306450605392456, + "num_tokens": 286893884.0, + "step": 11089 + }, + { + "epoch": 1.217878321985504, + "grad_norm": 2.410341501235962, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7377797365188599, + "num_tokens": 286916003.0, + "step": 11090 + }, + { + "epoch": 1.2179881396881178, + "grad_norm": 2.5095112323760986, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7074149250984192, + "num_tokens": 286938420.0, + "step": 11091 + }, + { + "epoch": 1.2180979573907313, + "grad_norm": 2.720357894897461, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7369921803474426, + "num_tokens": 286958223.0, + "step": 11092 + }, + { + "epoch": 1.218207775093345, + "grad_norm": 2.2679758071899414, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.720963716506958, + "num_tokens": 286982199.0, + "step": 11093 + }, + { + "epoch": 1.2183175927959586, + "grad_norm": 2.206728219985962, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7171551585197449, + "num_tokens": 287010430.0, + "step": 11094 + }, + { + "epoch": 1.2184274104985724, + "grad_norm": 2.548846483230591, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7258647680282593, + "num_tokens": 287030970.0, + "step": 11095 + }, + { + "epoch": 1.218537228201186, + "grad_norm": 2.118514060974121, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7056564688682556, + "num_tokens": 287057132.0, + "step": 11096 + }, + { + "epoch": 1.2186470459037997, + "grad_norm": 2.298490285873413, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7310442328453064, + "num_tokens": 287082012.0, + "step": 11097 + }, + { + "epoch": 1.2187568636064134, + "grad_norm": 2.2690589427948, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7013335227966309, + "num_tokens": 287109465.0, + "step": 11098 + }, + { + "epoch": 1.218866681309027, + "grad_norm": 2.159473419189453, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7483704090118408, + "num_tokens": 287135634.0, + "step": 11099 + }, + { + "epoch": 1.2189764990116407, + "grad_norm": 2.8356359004974365, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7383516430854797, + "num_tokens": 287151167.0, + "step": 11100 + }, + { + "epoch": 1.2190863167142543, + "grad_norm": 2.292048692703247, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7389109134674072, + "num_tokens": 287176289.0, + "step": 11101 + }, + { + "epoch": 1.219196134416868, + "grad_norm": 2.401414632797241, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7120397090911865, + "num_tokens": 287199749.0, + "step": 11102 + }, + { + "epoch": 1.2193059521194818, + "grad_norm": 2.100581645965576, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7052897214889526, + "num_tokens": 287229177.0, + "step": 11103 + }, + { + "epoch": 1.2194157698220953, + "grad_norm": 2.093604326248169, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.709418535232544, + "num_tokens": 287257362.0, + "step": 11104 + }, + { + "epoch": 1.219525587524709, + "grad_norm": 2.0814340114593506, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7235835790634155, + "num_tokens": 287286351.0, + "step": 11105 + }, + { + "epoch": 1.2196354052273226, + "grad_norm": 2.3615641593933105, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.723344624042511, + "num_tokens": 287312114.0, + "step": 11106 + }, + { + "epoch": 1.2197452229299364, + "grad_norm": 2.1540372371673584, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7267329692840576, + "num_tokens": 287340424.0, + "step": 11107 + }, + { + "epoch": 1.21985504063255, + "grad_norm": 2.301586389541626, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7053341865539551, + "num_tokens": 287365830.0, + "step": 11108 + }, + { + "epoch": 1.2199648583351637, + "grad_norm": 2.1574857234954834, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.732357919216156, + "num_tokens": 287392435.0, + "step": 11109 + }, + { + "epoch": 1.2200746760377772, + "grad_norm": 2.6324257850646973, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7314256429672241, + "num_tokens": 287412571.0, + "step": 11110 + }, + { + "epoch": 1.220184493740391, + "grad_norm": 2.0343148708343506, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7167181372642517, + "num_tokens": 287441777.0, + "step": 11111 + }, + { + "epoch": 1.2202943114430047, + "grad_norm": 2.5952937602996826, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7155665159225464, + "num_tokens": 287462481.0, + "step": 11112 + }, + { + "epoch": 1.2204041291456182, + "grad_norm": 2.201860189437866, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7162816524505615, + "num_tokens": 287491128.0, + "step": 11113 + }, + { + "epoch": 1.220513946848232, + "grad_norm": 1.8800734281539917, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7249523401260376, + "num_tokens": 287524819.0, + "step": 11114 + }, + { + "epoch": 1.2206237645508455, + "grad_norm": 1.8517661094665527, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7160228490829468, + "num_tokens": 287562937.0, + "step": 11115 + }, + { + "epoch": 1.2207335822534593, + "grad_norm": 2.413231372833252, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7402342557907104, + "num_tokens": 287584404.0, + "step": 11116 + }, + { + "epoch": 1.2208433999560728, + "grad_norm": 2.181788444519043, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7455624938011169, + "num_tokens": 287609866.0, + "step": 11117 + }, + { + "epoch": 1.2209532176586866, + "grad_norm": 2.057408094406128, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7081097960472107, + "num_tokens": 287639847.0, + "step": 11118 + }, + { + "epoch": 1.2210630353613001, + "grad_norm": 1.967743992805481, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7187052369117737, + "num_tokens": 287670981.0, + "step": 11119 + }, + { + "epoch": 1.2211728530639139, + "grad_norm": 2.0645503997802734, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7110503315925598, + "num_tokens": 287701178.0, + "step": 11120 + }, + { + "epoch": 1.2212826707665276, + "grad_norm": 2.177757740020752, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7025870084762573, + "num_tokens": 287728371.0, + "step": 11121 + }, + { + "epoch": 1.2213924884691412, + "grad_norm": 2.312882900238037, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7204198837280273, + "num_tokens": 287752359.0, + "step": 11122 + }, + { + "epoch": 1.221502306171755, + "grad_norm": 2.447903871536255, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7324793338775635, + "num_tokens": 287774074.0, + "step": 11123 + }, + { + "epoch": 1.2216121238743685, + "grad_norm": 2.0181796550750732, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6999801397323608, + "num_tokens": 287805557.0, + "step": 11124 + }, + { + "epoch": 1.2217219415769822, + "grad_norm": 2.264193058013916, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7038372755050659, + "num_tokens": 287831463.0, + "step": 11125 + }, + { + "epoch": 1.221831759279596, + "grad_norm": 2.442434072494507, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7429911494255066, + "num_tokens": 287852361.0, + "step": 11126 + }, + { + "epoch": 1.2219415769822095, + "grad_norm": 2.5678491592407227, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7252233028411865, + "num_tokens": 287871387.0, + "step": 11127 + }, + { + "epoch": 1.2220513946848233, + "grad_norm": 2.3638086318969727, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7181076407432556, + "num_tokens": 287897123.0, + "step": 11128 + }, + { + "epoch": 1.2221612123874368, + "grad_norm": 2.51287841796875, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7179946899414062, + "num_tokens": 287920282.0, + "step": 11129 + }, + { + "epoch": 1.2222710300900506, + "grad_norm": 2.3110039234161377, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7144759297370911, + "num_tokens": 287945621.0, + "step": 11130 + }, + { + "epoch": 1.222380847792664, + "grad_norm": 2.1784350872039795, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7328194975852966, + "num_tokens": 287972594.0, + "step": 11131 + }, + { + "epoch": 1.2224906654952779, + "grad_norm": 1.9415791034698486, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7044916152954102, + "num_tokens": 288005889.0, + "step": 11132 + }, + { + "epoch": 1.2226004831978914, + "grad_norm": 2.354147434234619, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7325952649116516, + "num_tokens": 288026620.0, + "step": 11133 + }, + { + "epoch": 1.2227103009005051, + "grad_norm": 2.1838274002075195, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7138682007789612, + "num_tokens": 288055937.0, + "step": 11134 + }, + { + "epoch": 1.222820118603119, + "grad_norm": 2.2441623210906982, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7219207882881165, + "num_tokens": 288082276.0, + "step": 11135 + }, + { + "epoch": 1.2229299363057324, + "grad_norm": 2.4694437980651855, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.715360164642334, + "num_tokens": 288105292.0, + "step": 11136 + }, + { + "epoch": 1.2230397540083462, + "grad_norm": 2.205454111099243, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.725399374961853, + "num_tokens": 288131219.0, + "step": 11137 + }, + { + "epoch": 1.2231495717109597, + "grad_norm": 2.006361722946167, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7122251987457275, + "num_tokens": 288164176.0, + "step": 11138 + }, + { + "epoch": 1.2232593894135735, + "grad_norm": 2.4276740550994873, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7183309197425842, + "num_tokens": 288188719.0, + "step": 11139 + }, + { + "epoch": 1.2233692071161872, + "grad_norm": 2.2018096446990967, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7080208659172058, + "num_tokens": 288216109.0, + "step": 11140 + }, + { + "epoch": 1.2234790248188008, + "grad_norm": 2.4532411098480225, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7150493860244751, + "num_tokens": 288240464.0, + "step": 11141 + }, + { + "epoch": 1.2235888425214145, + "grad_norm": 2.134683847427368, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7275339961051941, + "num_tokens": 288266942.0, + "step": 11142 + }, + { + "epoch": 1.223698660224028, + "grad_norm": 2.225310802459717, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7039141654968262, + "num_tokens": 288295463.0, + "step": 11143 + }, + { + "epoch": 1.2238084779266418, + "grad_norm": 2.235438585281372, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7284929752349854, + "num_tokens": 288319423.0, + "step": 11144 + }, + { + "epoch": 1.2239182956292554, + "grad_norm": 2.774003267288208, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7370110750198364, + "num_tokens": 288338142.0, + "step": 11145 + }, + { + "epoch": 1.2240281133318691, + "grad_norm": 2.0099234580993652, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7326226234436035, + "num_tokens": 288365441.0, + "step": 11146 + }, + { + "epoch": 1.2241379310344827, + "grad_norm": 2.3863942623138428, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7016952037811279, + "num_tokens": 288391090.0, + "step": 11147 + }, + { + "epoch": 1.2242477487370964, + "grad_norm": 1.9783867597579956, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6945102214813232, + "num_tokens": 288423660.0, + "step": 11148 + }, + { + "epoch": 1.2243575664397102, + "grad_norm": 2.127572774887085, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7220942974090576, + "num_tokens": 288452342.0, + "step": 11149 + }, + { + "epoch": 1.2244673841423237, + "grad_norm": 2.144728183746338, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7101213932037354, + "num_tokens": 288480932.0, + "step": 11150 + }, + { + "epoch": 1.2245772018449375, + "grad_norm": 2.7755868434906006, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7464837431907654, + "num_tokens": 288498077.0, + "step": 11151 + }, + { + "epoch": 1.224687019547551, + "grad_norm": 2.298410654067993, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7063295245170593, + "num_tokens": 288524421.0, + "step": 11152 + }, + { + "epoch": 1.2247968372501647, + "grad_norm": 2.181030035018921, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7223570942878723, + "num_tokens": 288551768.0, + "step": 11153 + }, + { + "epoch": 1.2249066549527785, + "grad_norm": 2.354595899581909, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7262462377548218, + "num_tokens": 288575893.0, + "step": 11154 + }, + { + "epoch": 1.225016472655392, + "grad_norm": 2.351170301437378, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7261338233947754, + "num_tokens": 288598787.0, + "step": 11155 + }, + { + "epoch": 1.2251262903580058, + "grad_norm": 2.4875986576080322, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7162732481956482, + "num_tokens": 288620878.0, + "step": 11156 + }, + { + "epoch": 1.2252361080606193, + "grad_norm": 2.1752519607543945, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7219589948654175, + "num_tokens": 288648869.0, + "step": 11157 + }, + { + "epoch": 1.225345925763233, + "grad_norm": 2.5265958309173584, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7474737167358398, + "num_tokens": 288668783.0, + "step": 11158 + }, + { + "epoch": 1.2254557434658466, + "grad_norm": 2.208858013153076, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.720240592956543, + "num_tokens": 288696825.0, + "step": 11159 + }, + { + "epoch": 1.2255655611684604, + "grad_norm": 1.9772547483444214, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7255659103393555, + "num_tokens": 288729299.0, + "step": 11160 + }, + { + "epoch": 1.225675378871074, + "grad_norm": 2.3435726165771484, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7246111631393433, + "num_tokens": 288754789.0, + "step": 11161 + }, + { + "epoch": 1.2257851965736877, + "grad_norm": 2.325425386428833, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7022668719291687, + "num_tokens": 288780435.0, + "step": 11162 + }, + { + "epoch": 1.2258950142763014, + "grad_norm": 2.2900753021240234, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.71285080909729, + "num_tokens": 288803995.0, + "step": 11163 + }, + { + "epoch": 1.226004831978915, + "grad_norm": 2.1092467308044434, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7048260569572449, + "num_tokens": 288832003.0, + "step": 11164 + }, + { + "epoch": 1.2261146496815287, + "grad_norm": 2.1044652462005615, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7064613103866577, + "num_tokens": 288862135.0, + "step": 11165 + }, + { + "epoch": 1.2262244673841423, + "grad_norm": 2.065572738647461, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7025724649429321, + "num_tokens": 288890024.0, + "step": 11166 + }, + { + "epoch": 1.226334285086756, + "grad_norm": 2.3138203620910645, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7205169200897217, + "num_tokens": 288913336.0, + "step": 11167 + }, + { + "epoch": 1.2264441027893696, + "grad_norm": 2.0373873710632324, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7191678285598755, + "num_tokens": 288944772.0, + "step": 11168 + }, + { + "epoch": 1.2265539204919833, + "grad_norm": 2.0871291160583496, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7187900543212891, + "num_tokens": 288972703.0, + "step": 11169 + }, + { + "epoch": 1.226663738194597, + "grad_norm": 2.2108266353607178, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7421743273735046, + "num_tokens": 288997926.0, + "step": 11170 + }, + { + "epoch": 1.2267735558972106, + "grad_norm": 2.3737785816192627, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.703938364982605, + "num_tokens": 289020373.0, + "step": 11171 + }, + { + "epoch": 1.2268833735998244, + "grad_norm": 2.207064151763916, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6979690790176392, + "num_tokens": 289048954.0, + "step": 11172 + }, + { + "epoch": 1.226993191302438, + "grad_norm": 2.3400001525878906, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7187883853912354, + "num_tokens": 289073845.0, + "step": 11173 + }, + { + "epoch": 1.2271030090050516, + "grad_norm": 2.4903836250305176, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7275742292404175, + "num_tokens": 289094320.0, + "step": 11174 + }, + { + "epoch": 1.2272128267076652, + "grad_norm": 2.2763454914093018, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.741558313369751, + "num_tokens": 289123281.0, + "step": 11175 + }, + { + "epoch": 1.227322644410279, + "grad_norm": 2.514151096343994, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7069119215011597, + "num_tokens": 289146395.0, + "step": 11176 + }, + { + "epoch": 1.2274324621128927, + "grad_norm": 2.5369768142700195, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7219944596290588, + "num_tokens": 289166789.0, + "step": 11177 + }, + { + "epoch": 1.2275422798155062, + "grad_norm": 2.297642230987549, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7031533718109131, + "num_tokens": 289190811.0, + "step": 11178 + }, + { + "epoch": 1.22765209751812, + "grad_norm": 2.281078815460205, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7220088243484497, + "num_tokens": 289216576.0, + "step": 11179 + }, + { + "epoch": 1.2277619152207335, + "grad_norm": 2.2767720222473145, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7213459610939026, + "num_tokens": 289241926.0, + "step": 11180 + }, + { + "epoch": 1.2278717329233473, + "grad_norm": 2.5087664127349854, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7453633546829224, + "num_tokens": 289261528.0, + "step": 11181 + }, + { + "epoch": 1.2279815506259608, + "grad_norm": 2.5073440074920654, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6925568580627441, + "num_tokens": 289284892.0, + "step": 11182 + }, + { + "epoch": 1.2280913683285746, + "grad_norm": 2.2171945571899414, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7092524766921997, + "num_tokens": 289311750.0, + "step": 11183 + }, + { + "epoch": 1.228201186031188, + "grad_norm": 2.3082165718078613, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7181733846664429, + "num_tokens": 289336750.0, + "step": 11184 + }, + { + "epoch": 1.2283110037338019, + "grad_norm": 1.8798556327819824, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7069189548492432, + "num_tokens": 289372893.0, + "step": 11185 + }, + { + "epoch": 1.2284208214364156, + "grad_norm": 2.1007437705993652, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7075638771057129, + "num_tokens": 289400788.0, + "step": 11186 + }, + { + "epoch": 1.2285306391390292, + "grad_norm": 2.2314751148223877, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7159351110458374, + "num_tokens": 289428204.0, + "step": 11187 + }, + { + "epoch": 1.228640456841643, + "grad_norm": 2.5368576049804688, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7385385036468506, + "num_tokens": 289448773.0, + "step": 11188 + }, + { + "epoch": 1.2287502745442564, + "grad_norm": 2.383502721786499, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.726352870464325, + "num_tokens": 289472201.0, + "step": 11189 + }, + { + "epoch": 1.2288600922468702, + "grad_norm": 2.313832998275757, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.741859495639801, + "num_tokens": 289493334.0, + "step": 11190 + }, + { + "epoch": 1.228969909949484, + "grad_norm": 2.205399751663208, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7008626461029053, + "num_tokens": 289522313.0, + "step": 11191 + }, + { + "epoch": 1.2290797276520975, + "grad_norm": 1.9423339366912842, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6884516477584839, + "num_tokens": 289555995.0, + "step": 11192 + }, + { + "epoch": 1.2291895453547113, + "grad_norm": 2.049973487854004, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7064817547798157, + "num_tokens": 289588907.0, + "step": 11193 + }, + { + "epoch": 1.2292993630573248, + "grad_norm": 2.2846384048461914, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7162315249443054, + "num_tokens": 289613880.0, + "step": 11194 + }, + { + "epoch": 1.2294091807599385, + "grad_norm": 2.4686858654022217, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.724105715751648, + "num_tokens": 289635850.0, + "step": 11195 + }, + { + "epoch": 1.229518998462552, + "grad_norm": 2.151853084564209, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7200367450714111, + "num_tokens": 289661786.0, + "step": 11196 + }, + { + "epoch": 1.2296288161651658, + "grad_norm": 2.1635515689849854, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7288309335708618, + "num_tokens": 289690312.0, + "step": 11197 + }, + { + "epoch": 1.2297386338677794, + "grad_norm": 2.32025408744812, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7071307897567749, + "num_tokens": 289716247.0, + "step": 11198 + }, + { + "epoch": 1.2298484515703931, + "grad_norm": 2.492717742919922, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7057568430900574, + "num_tokens": 289737054.0, + "step": 11199 + }, + { + "epoch": 1.2299582692730069, + "grad_norm": 2.1775877475738525, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6912092566490173, + "num_tokens": 289765541.0, + "step": 11200 + }, + { + "epoch": 1.2300680869756204, + "grad_norm": 2.1965954303741455, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7204473614692688, + "num_tokens": 289793721.0, + "step": 11201 + }, + { + "epoch": 1.2301779046782342, + "grad_norm": 2.154432773590088, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7264675498008728, + "num_tokens": 289818886.0, + "step": 11202 + }, + { + "epoch": 1.2302877223808477, + "grad_norm": 2.524576425552368, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7240270376205444, + "num_tokens": 289839342.0, + "step": 11203 + }, + { + "epoch": 1.2303975400834615, + "grad_norm": 2.4941561222076416, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7097264528274536, + "num_tokens": 289859828.0, + "step": 11204 + }, + { + "epoch": 1.2305073577860752, + "grad_norm": 2.623785972595215, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7384745478630066, + "num_tokens": 289878027.0, + "step": 11205 + }, + { + "epoch": 1.2306171754886888, + "grad_norm": 2.317837953567505, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7025370597839355, + "num_tokens": 289902423.0, + "step": 11206 + }, + { + "epoch": 1.2307269931913025, + "grad_norm": 2.3581736087799072, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7072045803070068, + "num_tokens": 289926283.0, + "step": 11207 + }, + { + "epoch": 1.230836810893916, + "grad_norm": 2.494006395339966, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.724694013595581, + "num_tokens": 289947044.0, + "step": 11208 + }, + { + "epoch": 1.2309466285965298, + "grad_norm": 2.146423816680908, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7064741849899292, + "num_tokens": 289973967.0, + "step": 11209 + }, + { + "epoch": 1.2310564462991433, + "grad_norm": 2.4143574237823486, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7175394892692566, + "num_tokens": 289997008.0, + "step": 11210 + }, + { + "epoch": 1.231166264001757, + "grad_norm": 2.2010574340820312, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.698849081993103, + "num_tokens": 290025815.0, + "step": 11211 + }, + { + "epoch": 1.2312760817043706, + "grad_norm": 2.3618826866149902, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7097622752189636, + "num_tokens": 290049001.0, + "step": 11212 + }, + { + "epoch": 1.2313858994069844, + "grad_norm": 2.4147989749908447, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7186206579208374, + "num_tokens": 290072534.0, + "step": 11213 + }, + { + "epoch": 1.2314957171095982, + "grad_norm": 2.414302110671997, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.6995837688446045, + "num_tokens": 290098182.0, + "step": 11214 + }, + { + "epoch": 1.2316055348122117, + "grad_norm": 2.4128670692443848, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7288382053375244, + "num_tokens": 290121546.0, + "step": 11215 + }, + { + "epoch": 1.2317153525148254, + "grad_norm": 2.3555490970611572, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7417306900024414, + "num_tokens": 290144929.0, + "step": 11216 + }, + { + "epoch": 1.231825170217439, + "grad_norm": 2.336510419845581, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7223231792449951, + "num_tokens": 290169041.0, + "step": 11217 + }, + { + "epoch": 1.2319349879200527, + "grad_norm": 2.123826742172241, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7297981977462769, + "num_tokens": 290196061.0, + "step": 11218 + }, + { + "epoch": 1.2320448056226665, + "grad_norm": 2.235136032104492, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7101500034332275, + "num_tokens": 290222130.0, + "step": 11219 + }, + { + "epoch": 1.23215462332528, + "grad_norm": 2.458303928375244, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7291285395622253, + "num_tokens": 290245566.0, + "step": 11220 + }, + { + "epoch": 1.2322644410278938, + "grad_norm": 2.4375040531158447, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.708598256111145, + "num_tokens": 290268633.0, + "step": 11221 + }, + { + "epoch": 1.2323742587305073, + "grad_norm": 2.047811269760132, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7385187149047852, + "num_tokens": 290298052.0, + "step": 11222 + }, + { + "epoch": 1.232484076433121, + "grad_norm": 2.311471939086914, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7343482375144958, + "num_tokens": 290323306.0, + "step": 11223 + }, + { + "epoch": 1.2325938941357346, + "grad_norm": 2.1601040363311768, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7224909067153931, + "num_tokens": 290351222.0, + "step": 11224 + }, + { + "epoch": 1.2327037118383484, + "grad_norm": 2.139009475708008, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6911700963973999, + "num_tokens": 290380297.0, + "step": 11225 + }, + { + "epoch": 1.232813529540962, + "grad_norm": 2.0666732788085938, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7319048643112183, + "num_tokens": 290410462.0, + "step": 11226 + }, + { + "epoch": 1.2329233472435757, + "grad_norm": 2.1447880268096924, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.717038631439209, + "num_tokens": 290438048.0, + "step": 11227 + }, + { + "epoch": 1.2330331649461894, + "grad_norm": 1.9827276468276978, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7060885429382324, + "num_tokens": 290469821.0, + "step": 11228 + }, + { + "epoch": 1.233142982648803, + "grad_norm": 2.1664209365844727, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7210716605186462, + "num_tokens": 290495476.0, + "step": 11229 + }, + { + "epoch": 1.2332528003514167, + "grad_norm": 2.327511787414551, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.701725959777832, + "num_tokens": 290519504.0, + "step": 11230 + }, + { + "epoch": 1.2333626180540302, + "grad_norm": 2.080700397491455, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7068231701850891, + "num_tokens": 290547917.0, + "step": 11231 + }, + { + "epoch": 1.233472435756644, + "grad_norm": 2.362198829650879, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7079412937164307, + "num_tokens": 290575889.0, + "step": 11232 + }, + { + "epoch": 1.2335822534592575, + "grad_norm": 2.4311206340789795, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7330100536346436, + "num_tokens": 290597692.0, + "step": 11233 + }, + { + "epoch": 1.2336920711618713, + "grad_norm": 2.2471399307250977, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.700904369354248, + "num_tokens": 290624743.0, + "step": 11234 + }, + { + "epoch": 1.2338018888644848, + "grad_norm": 2.24560809135437, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7079639434814453, + "num_tokens": 290651588.0, + "step": 11235 + }, + { + "epoch": 1.2339117065670986, + "grad_norm": 2.1869070529937744, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7083991169929504, + "num_tokens": 290679796.0, + "step": 11236 + }, + { + "epoch": 1.2340215242697123, + "grad_norm": 2.847996711730957, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7214916944503784, + "num_tokens": 290698190.0, + "step": 11237 + }, + { + "epoch": 1.2341313419723259, + "grad_norm": 2.251377582550049, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7284746766090393, + "num_tokens": 290722165.0, + "step": 11238 + }, + { + "epoch": 1.2342411596749396, + "grad_norm": 1.955626130104065, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6961003541946411, + "num_tokens": 290755251.0, + "step": 11239 + }, + { + "epoch": 1.2343509773775532, + "grad_norm": 2.0677855014801025, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.717197060585022, + "num_tokens": 290782938.0, + "step": 11240 + }, + { + "epoch": 1.234460795080167, + "grad_norm": 1.9454004764556885, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7309328317642212, + "num_tokens": 290814105.0, + "step": 11241 + }, + { + "epoch": 1.2345706127827807, + "grad_norm": 2.0747604370117188, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7348372936248779, + "num_tokens": 290841887.0, + "step": 11242 + }, + { + "epoch": 1.2346804304853942, + "grad_norm": 2.128722906112671, + "learning_rate": 1e-06, + "loss": 1.0666, + "mean_token_accuracy": 0.6731541156768799, + "num_tokens": 290872908.0, + "step": 11243 + }, + { + "epoch": 1.234790248188008, + "grad_norm": 2.1120080947875977, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7135151624679565, + "num_tokens": 290902229.0, + "step": 11244 + }, + { + "epoch": 1.2349000658906215, + "grad_norm": 2.1336679458618164, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6947680711746216, + "num_tokens": 290930274.0, + "step": 11245 + }, + { + "epoch": 1.2350098835932353, + "grad_norm": 2.232252597808838, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6854568719863892, + "num_tokens": 290957237.0, + "step": 11246 + }, + { + "epoch": 1.2351197012958488, + "grad_norm": 2.2365105152130127, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7254691123962402, + "num_tokens": 290982019.0, + "step": 11247 + }, + { + "epoch": 1.2352295189984626, + "grad_norm": 2.440911293029785, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7364836931228638, + "num_tokens": 291004914.0, + "step": 11248 + }, + { + "epoch": 1.235339336701076, + "grad_norm": 2.088935613632202, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7143176794052124, + "num_tokens": 291035370.0, + "step": 11249 + }, + { + "epoch": 1.2354491544036899, + "grad_norm": 2.177096366882324, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6972709894180298, + "num_tokens": 291064938.0, + "step": 11250 + }, + { + "epoch": 1.2355589721063036, + "grad_norm": 2.169299602508545, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7100764513015747, + "num_tokens": 291090672.0, + "step": 11251 + }, + { + "epoch": 1.2356687898089171, + "grad_norm": 2.455603837966919, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.719167172908783, + "num_tokens": 291113233.0, + "step": 11252 + }, + { + "epoch": 1.235778607511531, + "grad_norm": 2.0970358848571777, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7175731658935547, + "num_tokens": 291142475.0, + "step": 11253 + }, + { + "epoch": 1.2358884252141444, + "grad_norm": 2.094365119934082, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.717229425907135, + "num_tokens": 291169504.0, + "step": 11254 + }, + { + "epoch": 1.2359982429167582, + "grad_norm": 2.0368146896362305, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7116894721984863, + "num_tokens": 291202195.0, + "step": 11255 + }, + { + "epoch": 1.236108060619372, + "grad_norm": 2.260387420654297, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6894804239273071, + "num_tokens": 291228469.0, + "step": 11256 + }, + { + "epoch": 1.2362178783219855, + "grad_norm": 2.106593608856201, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6909609436988831, + "num_tokens": 291259022.0, + "step": 11257 + }, + { + "epoch": 1.2363276960245992, + "grad_norm": 2.264216899871826, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7282438278198242, + "num_tokens": 291283767.0, + "step": 11258 + }, + { + "epoch": 1.2364375137272128, + "grad_norm": 2.4301674365997314, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7275177240371704, + "num_tokens": 291306139.0, + "step": 11259 + }, + { + "epoch": 1.2365473314298265, + "grad_norm": 1.9987990856170654, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7156926989555359, + "num_tokens": 291336887.0, + "step": 11260 + }, + { + "epoch": 1.23665714913244, + "grad_norm": 2.2824556827545166, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7008230090141296, + "num_tokens": 291363600.0, + "step": 11261 + }, + { + "epoch": 1.2367669668350538, + "grad_norm": 2.2183773517608643, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7207232713699341, + "num_tokens": 291389067.0, + "step": 11262 + }, + { + "epoch": 1.2368767845376674, + "grad_norm": 2.422074556350708, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.6945108771324158, + "num_tokens": 291413585.0, + "step": 11263 + }, + { + "epoch": 1.2369866022402811, + "grad_norm": 2.3680920600891113, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7363402843475342, + "num_tokens": 291434867.0, + "step": 11264 + }, + { + "epoch": 1.2370964199428949, + "grad_norm": 2.266216516494751, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7204502820968628, + "num_tokens": 291461764.0, + "step": 11265 + }, + { + "epoch": 1.2372062376455084, + "grad_norm": 2.224564790725708, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.709082841873169, + "num_tokens": 291488950.0, + "step": 11266 + }, + { + "epoch": 1.2373160553481222, + "grad_norm": 2.1383683681488037, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6829302906990051, + "num_tokens": 291516738.0, + "step": 11267 + }, + { + "epoch": 1.2374258730507357, + "grad_norm": 2.0029296875, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7130091786384583, + "num_tokens": 291548428.0, + "step": 11268 + }, + { + "epoch": 1.2375356907533495, + "grad_norm": 2.2486915588378906, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7265909314155579, + "num_tokens": 291572552.0, + "step": 11269 + }, + { + "epoch": 1.2376455084559632, + "grad_norm": 2.1933491230010986, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7390316724777222, + "num_tokens": 291597888.0, + "step": 11270 + }, + { + "epoch": 1.2377553261585768, + "grad_norm": 2.174086332321167, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.715442419052124, + "num_tokens": 291624458.0, + "step": 11271 + }, + { + "epoch": 1.2378651438611905, + "grad_norm": 2.2779812812805176, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7128129601478577, + "num_tokens": 291649258.0, + "step": 11272 + }, + { + "epoch": 1.237974961563804, + "grad_norm": 2.2014105319976807, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7366604208946228, + "num_tokens": 291673698.0, + "step": 11273 + }, + { + "epoch": 1.2380847792664178, + "grad_norm": 2.313952684402466, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7248417139053345, + "num_tokens": 291697675.0, + "step": 11274 + }, + { + "epoch": 1.2381945969690313, + "grad_norm": 2.2953860759735107, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7111465930938721, + "num_tokens": 291724573.0, + "step": 11275 + }, + { + "epoch": 1.238304414671645, + "grad_norm": 2.6609928607940674, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7324333786964417, + "num_tokens": 291743913.0, + "step": 11276 + }, + { + "epoch": 1.2384142323742586, + "grad_norm": 2.2206149101257324, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7360566258430481, + "num_tokens": 291768326.0, + "step": 11277 + }, + { + "epoch": 1.2385240500768724, + "grad_norm": 2.2382469177246094, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7070327401161194, + "num_tokens": 291796062.0, + "step": 11278 + }, + { + "epoch": 1.2386338677794861, + "grad_norm": 2.1943938732147217, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7171275019645691, + "num_tokens": 291825094.0, + "step": 11279 + }, + { + "epoch": 1.2387436854820997, + "grad_norm": 2.21701979637146, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7196529507637024, + "num_tokens": 291851180.0, + "step": 11280 + }, + { + "epoch": 1.2388535031847134, + "grad_norm": 2.2584829330444336, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7304834127426147, + "num_tokens": 291875864.0, + "step": 11281 + }, + { + "epoch": 1.238963320887327, + "grad_norm": 2.2308197021484375, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7216348052024841, + "num_tokens": 291901757.0, + "step": 11282 + }, + { + "epoch": 1.2390731385899407, + "grad_norm": 2.321493148803711, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7354795932769775, + "num_tokens": 291924418.0, + "step": 11283 + }, + { + "epoch": 1.2391829562925545, + "grad_norm": 2.131878137588501, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7118285894393921, + "num_tokens": 291953676.0, + "step": 11284 + }, + { + "epoch": 1.239292773995168, + "grad_norm": 2.194218635559082, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6878799200057983, + "num_tokens": 291982067.0, + "step": 11285 + }, + { + "epoch": 1.2394025916977818, + "grad_norm": 2.557305097579956, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7254657745361328, + "num_tokens": 292002505.0, + "step": 11286 + }, + { + "epoch": 1.2395124094003953, + "grad_norm": 2.1061291694641113, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7156904339790344, + "num_tokens": 292030540.0, + "step": 11287 + }, + { + "epoch": 1.239622227103009, + "grad_norm": 2.3356356620788574, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7024518847465515, + "num_tokens": 292055186.0, + "step": 11288 + }, + { + "epoch": 1.2397320448056226, + "grad_norm": 2.449476718902588, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7416695356369019, + "num_tokens": 292075507.0, + "step": 11289 + }, + { + "epoch": 1.2398418625082364, + "grad_norm": 2.194457530975342, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.7007440328598022, + "num_tokens": 292103532.0, + "step": 11290 + }, + { + "epoch": 1.23995168021085, + "grad_norm": 2.1199042797088623, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.733193576335907, + "num_tokens": 292131001.0, + "step": 11291 + }, + { + "epoch": 1.2400614979134637, + "grad_norm": 2.0885472297668457, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7172908782958984, + "num_tokens": 292159966.0, + "step": 11292 + }, + { + "epoch": 1.2401713156160774, + "grad_norm": 2.2522833347320557, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7276686429977417, + "num_tokens": 292185311.0, + "step": 11293 + }, + { + "epoch": 1.240281133318691, + "grad_norm": 2.2000668048858643, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7266415953636169, + "num_tokens": 292210456.0, + "step": 11294 + }, + { + "epoch": 1.2403909510213047, + "grad_norm": 2.144695520401001, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7085331082344055, + "num_tokens": 292238954.0, + "step": 11295 + }, + { + "epoch": 1.2405007687239182, + "grad_norm": 2.0551862716674805, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7515318989753723, + "num_tokens": 292268336.0, + "step": 11296 + }, + { + "epoch": 1.240610586426532, + "grad_norm": 2.3929624557495117, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7337640523910522, + "num_tokens": 292290510.0, + "step": 11297 + }, + { + "epoch": 1.2407204041291455, + "grad_norm": 2.1545896530151367, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7329839468002319, + "num_tokens": 292319469.0, + "step": 11298 + }, + { + "epoch": 1.2408302218317593, + "grad_norm": 2.329488515853882, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.710986852645874, + "num_tokens": 292345632.0, + "step": 11299 + }, + { + "epoch": 1.2409400395343728, + "grad_norm": 2.2514376640319824, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7094832062721252, + "num_tokens": 292372304.0, + "step": 11300 + }, + { + "epoch": 1.2410498572369866, + "grad_norm": 2.2934513092041016, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7522145509719849, + "num_tokens": 292395329.0, + "step": 11301 + }, + { + "epoch": 1.2411596749396003, + "grad_norm": 2.355215311050415, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.721332311630249, + "num_tokens": 292420030.0, + "step": 11302 + }, + { + "epoch": 1.2412694926422139, + "grad_norm": 2.2446587085723877, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7128736972808838, + "num_tokens": 292444679.0, + "step": 11303 + }, + { + "epoch": 1.2413793103448276, + "grad_norm": 2.2040417194366455, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7208936214447021, + "num_tokens": 292471191.0, + "step": 11304 + }, + { + "epoch": 1.2414891280474412, + "grad_norm": 2.5503056049346924, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7216506004333496, + "num_tokens": 292493761.0, + "step": 11305 + }, + { + "epoch": 1.241598945750055, + "grad_norm": 2.4493932723999023, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7134923934936523, + "num_tokens": 292516844.0, + "step": 11306 + }, + { + "epoch": 1.2417087634526687, + "grad_norm": 2.4615719318389893, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.702923059463501, + "num_tokens": 292540030.0, + "step": 11307 + }, + { + "epoch": 1.2418185811552822, + "grad_norm": 2.131943702697754, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.72101891040802, + "num_tokens": 292568982.0, + "step": 11308 + }, + { + "epoch": 1.241928398857896, + "grad_norm": 2.5483193397521973, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7194202542304993, + "num_tokens": 292589148.0, + "step": 11309 + }, + { + "epoch": 1.2420382165605095, + "grad_norm": 2.180880308151245, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7063372731208801, + "num_tokens": 292617921.0, + "step": 11310 + }, + { + "epoch": 1.2421480342631233, + "grad_norm": 2.1603918075561523, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7378666400909424, + "num_tokens": 292644301.0, + "step": 11311 + }, + { + "epoch": 1.2422578519657368, + "grad_norm": 2.0476579666137695, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7214269638061523, + "num_tokens": 292674415.0, + "step": 11312 + }, + { + "epoch": 1.2423676696683505, + "grad_norm": 2.347334623336792, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7222487330436707, + "num_tokens": 292698202.0, + "step": 11313 + }, + { + "epoch": 1.242477487370964, + "grad_norm": 2.0732176303863525, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7161504030227661, + "num_tokens": 292727895.0, + "step": 11314 + }, + { + "epoch": 1.2425873050735778, + "grad_norm": 2.4596176147460938, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.6971883773803711, + "num_tokens": 292751726.0, + "step": 11315 + }, + { + "epoch": 1.2426971227761916, + "grad_norm": 2.24153733253479, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6785096526145935, + "num_tokens": 292782025.0, + "step": 11316 + }, + { + "epoch": 1.2428069404788051, + "grad_norm": 2.176541328430176, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7185460329055786, + "num_tokens": 292809852.0, + "step": 11317 + }, + { + "epoch": 1.242916758181419, + "grad_norm": 2.0145974159240723, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7064180374145508, + "num_tokens": 292840054.0, + "step": 11318 + }, + { + "epoch": 1.2430265758840324, + "grad_norm": 2.651873826980591, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7434792518615723, + "num_tokens": 292858529.0, + "step": 11319 + }, + { + "epoch": 1.2431363935866462, + "grad_norm": 2.0264768600463867, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7059862613677979, + "num_tokens": 292888213.0, + "step": 11320 + }, + { + "epoch": 1.24324621128926, + "grad_norm": 1.9029494524002075, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7171972990036011, + "num_tokens": 292921633.0, + "step": 11321 + }, + { + "epoch": 1.2433560289918735, + "grad_norm": 2.08656907081604, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7156901359558105, + "num_tokens": 292950474.0, + "step": 11322 + }, + { + "epoch": 1.2434658466944872, + "grad_norm": 2.348021984100342, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7004806995391846, + "num_tokens": 292975140.0, + "step": 11323 + }, + { + "epoch": 1.2435756643971008, + "grad_norm": 2.0404226779937744, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.724044144153595, + "num_tokens": 293005180.0, + "step": 11324 + }, + { + "epoch": 1.2436854820997145, + "grad_norm": 2.338224411010742, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7272069454193115, + "num_tokens": 293028606.0, + "step": 11325 + }, + { + "epoch": 1.243795299802328, + "grad_norm": 2.4786291122436523, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7314144372940063, + "num_tokens": 293049332.0, + "step": 11326 + }, + { + "epoch": 1.2439051175049418, + "grad_norm": 1.924608588218689, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7164410352706909, + "num_tokens": 293079125.0, + "step": 11327 + }, + { + "epoch": 1.2440149352075554, + "grad_norm": 2.1601505279541016, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7004343271255493, + "num_tokens": 293106284.0, + "step": 11328 + }, + { + "epoch": 1.244124752910169, + "grad_norm": 1.9101253747940063, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7244948744773865, + "num_tokens": 293137777.0, + "step": 11329 + }, + { + "epoch": 1.2442345706127829, + "grad_norm": 2.6954164505004883, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7402402758598328, + "num_tokens": 293157064.0, + "step": 11330 + }, + { + "epoch": 1.2443443883153964, + "grad_norm": 2.391911029815674, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6968401074409485, + "num_tokens": 293181353.0, + "step": 11331 + }, + { + "epoch": 1.2444542060180102, + "grad_norm": 2.0716969966888428, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7056609988212585, + "num_tokens": 293211082.0, + "step": 11332 + }, + { + "epoch": 1.2445640237206237, + "grad_norm": 2.197720766067505, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7299587726593018, + "num_tokens": 293238078.0, + "step": 11333 + }, + { + "epoch": 1.2446738414232374, + "grad_norm": 2.814317226409912, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7307854890823364, + "num_tokens": 293255608.0, + "step": 11334 + }, + { + "epoch": 1.2447836591258512, + "grad_norm": 2.0564568042755127, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7127312421798706, + "num_tokens": 293285103.0, + "step": 11335 + }, + { + "epoch": 1.2448934768284647, + "grad_norm": 2.266925811767578, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7108609676361084, + "num_tokens": 293309605.0, + "step": 11336 + }, + { + "epoch": 1.2450032945310785, + "grad_norm": 2.204350471496582, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6893952488899231, + "num_tokens": 293336245.0, + "step": 11337 + }, + { + "epoch": 1.245113112233692, + "grad_norm": 2.3264102935791016, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7030669450759888, + "num_tokens": 293361734.0, + "step": 11338 + }, + { + "epoch": 1.2452229299363058, + "grad_norm": 2.12652325630188, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7244664430618286, + "num_tokens": 293389235.0, + "step": 11339 + }, + { + "epoch": 1.2453327476389193, + "grad_norm": 2.4828250408172607, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7178115844726562, + "num_tokens": 293410452.0, + "step": 11340 + }, + { + "epoch": 1.245442565341533, + "grad_norm": 2.251951217651367, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7366944551467896, + "num_tokens": 293435361.0, + "step": 11341 + }, + { + "epoch": 1.2455523830441466, + "grad_norm": 2.4584598541259766, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7364184856414795, + "num_tokens": 293457915.0, + "step": 11342 + }, + { + "epoch": 1.2456622007467604, + "grad_norm": 2.4438633918762207, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7459263205528259, + "num_tokens": 293478810.0, + "step": 11343 + }, + { + "epoch": 1.2457720184493741, + "grad_norm": 2.322676181793213, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7325481176376343, + "num_tokens": 293502615.0, + "step": 11344 + }, + { + "epoch": 1.2458818361519877, + "grad_norm": 2.2321457862854004, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.724259614944458, + "num_tokens": 293527291.0, + "step": 11345 + }, + { + "epoch": 1.2459916538546014, + "grad_norm": 2.151475429534912, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7359939217567444, + "num_tokens": 293554129.0, + "step": 11346 + }, + { + "epoch": 1.246101471557215, + "grad_norm": 2.180788278579712, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7245697379112244, + "num_tokens": 293579967.0, + "step": 11347 + }, + { + "epoch": 1.2462112892598287, + "grad_norm": 2.155660629272461, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7339707612991333, + "num_tokens": 293606901.0, + "step": 11348 + }, + { + "epoch": 1.2463211069624422, + "grad_norm": 2.2222678661346436, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7041425108909607, + "num_tokens": 293634345.0, + "step": 11349 + }, + { + "epoch": 1.246430924665056, + "grad_norm": 2.1880130767822266, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7116488218307495, + "num_tokens": 293660439.0, + "step": 11350 + }, + { + "epoch": 1.2465407423676698, + "grad_norm": 2.2339413166046143, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.736851692199707, + "num_tokens": 293685799.0, + "step": 11351 + }, + { + "epoch": 1.2466505600702833, + "grad_norm": 2.3465824127197266, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7311510443687439, + "num_tokens": 293710985.0, + "step": 11352 + }, + { + "epoch": 1.246760377772897, + "grad_norm": 2.109865665435791, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6897068023681641, + "num_tokens": 293742124.0, + "step": 11353 + }, + { + "epoch": 1.2468701954755106, + "grad_norm": 2.4115798473358154, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.715540885925293, + "num_tokens": 293764646.0, + "step": 11354 + }, + { + "epoch": 1.2469800131781243, + "grad_norm": 2.4591894149780273, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7262659072875977, + "num_tokens": 293788042.0, + "step": 11355 + }, + { + "epoch": 1.2470898308807379, + "grad_norm": 2.376117467880249, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7334540486335754, + "num_tokens": 293811110.0, + "step": 11356 + }, + { + "epoch": 1.2471996485833516, + "grad_norm": 2.1212451457977295, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7013198733329773, + "num_tokens": 293839656.0, + "step": 11357 + }, + { + "epoch": 1.2473094662859654, + "grad_norm": 2.3386356830596924, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.698700487613678, + "num_tokens": 293864921.0, + "step": 11358 + }, + { + "epoch": 1.247419283988579, + "grad_norm": 2.2697505950927734, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7133468985557556, + "num_tokens": 293891997.0, + "step": 11359 + }, + { + "epoch": 1.2475291016911927, + "grad_norm": 1.9287244081497192, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.6980141997337341, + "num_tokens": 293928415.0, + "step": 11360 + }, + { + "epoch": 1.2476389193938062, + "grad_norm": 1.881883144378662, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7103995084762573, + "num_tokens": 293964945.0, + "step": 11361 + }, + { + "epoch": 1.24774873709642, + "grad_norm": 2.174760103225708, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7091994285583496, + "num_tokens": 293991970.0, + "step": 11362 + }, + { + "epoch": 1.2478585547990335, + "grad_norm": 2.240029811859131, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7205088138580322, + "num_tokens": 294018661.0, + "step": 11363 + }, + { + "epoch": 1.2479683725016473, + "grad_norm": 2.1335408687591553, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7488034963607788, + "num_tokens": 294046716.0, + "step": 11364 + }, + { + "epoch": 1.2480781902042608, + "grad_norm": 2.176689863204956, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7504650950431824, + "num_tokens": 294071987.0, + "step": 11365 + }, + { + "epoch": 1.2481880079068746, + "grad_norm": 2.342149257659912, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7416244149208069, + "num_tokens": 294095503.0, + "step": 11366 + }, + { + "epoch": 1.2482978256094883, + "grad_norm": 2.259896993637085, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.696395754814148, + "num_tokens": 294122424.0, + "step": 11367 + }, + { + "epoch": 1.2484076433121019, + "grad_norm": 2.1111738681793213, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7163223624229431, + "num_tokens": 294150376.0, + "step": 11368 + }, + { + "epoch": 1.2485174610147156, + "grad_norm": 2.1890859603881836, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7209349870681763, + "num_tokens": 294177393.0, + "step": 11369 + }, + { + "epoch": 1.2486272787173291, + "grad_norm": 2.225262403488159, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7115257978439331, + "num_tokens": 294203511.0, + "step": 11370 + }, + { + "epoch": 1.248737096419943, + "grad_norm": 2.047027826309204, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7163950204849243, + "num_tokens": 294232230.0, + "step": 11371 + }, + { + "epoch": 1.2488469141225567, + "grad_norm": 2.071186065673828, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.742090106010437, + "num_tokens": 294261107.0, + "step": 11372 + }, + { + "epoch": 1.2489567318251702, + "grad_norm": 2.6011857986450195, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7301239967346191, + "num_tokens": 294283492.0, + "step": 11373 + }, + { + "epoch": 1.249066549527784, + "grad_norm": 2.261993169784546, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7235570549964905, + "num_tokens": 294308887.0, + "step": 11374 + }, + { + "epoch": 1.2491763672303975, + "grad_norm": 2.267171621322632, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7101128101348877, + "num_tokens": 294336422.0, + "step": 11375 + }, + { + "epoch": 1.2492861849330112, + "grad_norm": 2.1963698863983154, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7001336812973022, + "num_tokens": 294363697.0, + "step": 11376 + }, + { + "epoch": 1.2493960026356248, + "grad_norm": 2.2492263317108154, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7323170900344849, + "num_tokens": 294387983.0, + "step": 11377 + }, + { + "epoch": 1.2495058203382385, + "grad_norm": 2.2136900424957275, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7097112536430359, + "num_tokens": 294414661.0, + "step": 11378 + }, + { + "epoch": 1.249615638040852, + "grad_norm": 2.171088457107544, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.729293704032898, + "num_tokens": 294440692.0, + "step": 11379 + }, + { + "epoch": 1.2497254557434658, + "grad_norm": 2.4247210025787354, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7160789966583252, + "num_tokens": 294468741.0, + "step": 11380 + }, + { + "epoch": 1.2498352734460796, + "grad_norm": 2.511108160018921, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7257552146911621, + "num_tokens": 294489488.0, + "step": 11381 + }, + { + "epoch": 1.2499450911486931, + "grad_norm": 2.341322898864746, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7156563997268677, + "num_tokens": 294513430.0, + "step": 11382 + }, + { + "epoch": 1.2500549088513069, + "grad_norm": 2.0475380420684814, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7198969721794128, + "num_tokens": 294542188.0, + "step": 11383 + }, + { + "epoch": 1.2501647265539204, + "grad_norm": 3.7502601146698, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7160021066665649, + "num_tokens": 294566002.0, + "step": 11384 + }, + { + "epoch": 1.2502745442565342, + "grad_norm": 2.1786341667175293, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7144291400909424, + "num_tokens": 294591308.0, + "step": 11385 + }, + { + "epoch": 1.250384361959148, + "grad_norm": 2.269507646560669, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7022576332092285, + "num_tokens": 294615500.0, + "step": 11386 + }, + { + "epoch": 1.2504941796617615, + "grad_norm": 2.237178087234497, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7195878624916077, + "num_tokens": 294640524.0, + "step": 11387 + }, + { + "epoch": 1.2506039973643752, + "grad_norm": 2.0147740840911865, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7098327875137329, + "num_tokens": 294670595.0, + "step": 11388 + }, + { + "epoch": 1.2507138150669888, + "grad_norm": 2.240680694580078, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7044755220413208, + "num_tokens": 294696095.0, + "step": 11389 + }, + { + "epoch": 1.2508236327696025, + "grad_norm": 2.2540674209594727, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.727138876914978, + "num_tokens": 294722676.0, + "step": 11390 + }, + { + "epoch": 1.250933450472216, + "grad_norm": 2.0528295040130615, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7180959582328796, + "num_tokens": 294752360.0, + "step": 11391 + }, + { + "epoch": 1.2510432681748298, + "grad_norm": 2.249631881713867, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7087280750274658, + "num_tokens": 294777603.0, + "step": 11392 + }, + { + "epoch": 1.2511530858774433, + "grad_norm": 2.2395851612091064, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.73260897397995, + "num_tokens": 294802529.0, + "step": 11393 + }, + { + "epoch": 1.251262903580057, + "grad_norm": 2.3076412677764893, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.700962245464325, + "num_tokens": 294827373.0, + "step": 11394 + }, + { + "epoch": 1.2513727212826709, + "grad_norm": 2.12646222114563, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7205512523651123, + "num_tokens": 294855380.0, + "step": 11395 + }, + { + "epoch": 1.2514825389852844, + "grad_norm": 2.3176305294036865, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7480471134185791, + "num_tokens": 294877037.0, + "step": 11396 + }, + { + "epoch": 1.2515923566878981, + "grad_norm": 2.291036367416382, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7133769989013672, + "num_tokens": 294901232.0, + "step": 11397 + }, + { + "epoch": 1.2517021743905117, + "grad_norm": 2.4431958198547363, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7310613393783569, + "num_tokens": 294921628.0, + "step": 11398 + }, + { + "epoch": 1.2518119920931254, + "grad_norm": 2.2585768699645996, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7287901639938354, + "num_tokens": 294947376.0, + "step": 11399 + }, + { + "epoch": 1.2519218097957392, + "grad_norm": 2.1521360874176025, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7316092848777771, + "num_tokens": 294976345.0, + "step": 11400 + }, + { + "epoch": 1.2520316274983527, + "grad_norm": 2.2845475673675537, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7230057120323181, + "num_tokens": 295000527.0, + "step": 11401 + }, + { + "epoch": 1.2521414452009663, + "grad_norm": 2.526270866394043, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7012048959732056, + "num_tokens": 295021857.0, + "step": 11402 + }, + { + "epoch": 1.25225126290358, + "grad_norm": 2.3786797523498535, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7491488456726074, + "num_tokens": 295045127.0, + "step": 11403 + }, + { + "epoch": 1.2523610806061938, + "grad_norm": 2.2333054542541504, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7372204065322876, + "num_tokens": 295070148.0, + "step": 11404 + }, + { + "epoch": 1.2524708983088073, + "grad_norm": 2.5064733028411865, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.71699458360672, + "num_tokens": 295089819.0, + "step": 11405 + }, + { + "epoch": 1.252580716011421, + "grad_norm": 2.7853288650512695, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7219704985618591, + "num_tokens": 295107843.0, + "step": 11406 + }, + { + "epoch": 1.2526905337140346, + "grad_norm": 2.308957099914551, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7312988638877869, + "num_tokens": 295133757.0, + "step": 11407 + }, + { + "epoch": 1.2528003514166484, + "grad_norm": 2.2218685150146484, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7091665863990784, + "num_tokens": 295161640.0, + "step": 11408 + }, + { + "epoch": 1.2529101691192621, + "grad_norm": 2.095850706100464, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7320661544799805, + "num_tokens": 295189968.0, + "step": 11409 + }, + { + "epoch": 1.2530199868218757, + "grad_norm": 2.406766653060913, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7340584397315979, + "num_tokens": 295214500.0, + "step": 11410 + }, + { + "epoch": 1.2531298045244894, + "grad_norm": 2.3954551219940186, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7103073596954346, + "num_tokens": 295237895.0, + "step": 11411 + }, + { + "epoch": 1.253239622227103, + "grad_norm": 2.314516305923462, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.73021399974823, + "num_tokens": 295260666.0, + "step": 11412 + }, + { + "epoch": 1.2533494399297167, + "grad_norm": 2.002472400665283, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7034429311752319, + "num_tokens": 295293423.0, + "step": 11413 + }, + { + "epoch": 1.2534592576323305, + "grad_norm": 2.2355942726135254, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7236213088035583, + "num_tokens": 295318031.0, + "step": 11414 + }, + { + "epoch": 1.253569075334944, + "grad_norm": 2.1689395904541016, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6915941834449768, + "num_tokens": 295345764.0, + "step": 11415 + }, + { + "epoch": 1.2536788930375575, + "grad_norm": 2.3662688732147217, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6907224655151367, + "num_tokens": 295369222.0, + "step": 11416 + }, + { + "epoch": 1.2537887107401713, + "grad_norm": 2.486896276473999, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7426289319992065, + "num_tokens": 295390231.0, + "step": 11417 + }, + { + "epoch": 1.253898528442785, + "grad_norm": 2.5961852073669434, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.725968599319458, + "num_tokens": 295411044.0, + "step": 11418 + }, + { + "epoch": 1.2540083461453986, + "grad_norm": 2.6353492736816406, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7321633100509644, + "num_tokens": 295432250.0, + "step": 11419 + }, + { + "epoch": 1.2541181638480123, + "grad_norm": 2.150921583175659, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7203273773193359, + "num_tokens": 295462432.0, + "step": 11420 + }, + { + "epoch": 1.2542279815506259, + "grad_norm": 2.104067325592041, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7022287249565125, + "num_tokens": 295490547.0, + "step": 11421 + }, + { + "epoch": 1.2543377992532396, + "grad_norm": 2.3296701908111572, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7190742492675781, + "num_tokens": 295513531.0, + "step": 11422 + }, + { + "epoch": 1.2544476169558534, + "grad_norm": 2.399078130722046, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7407183051109314, + "num_tokens": 295535425.0, + "step": 11423 + }, + { + "epoch": 1.254557434658467, + "grad_norm": 2.2899835109710693, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7220833897590637, + "num_tokens": 295561283.0, + "step": 11424 + }, + { + "epoch": 1.2546672523610807, + "grad_norm": 2.163076400756836, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7215971946716309, + "num_tokens": 295589515.0, + "step": 11425 + }, + { + "epoch": 1.2547770700636942, + "grad_norm": 2.3166117668151855, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7096462845802307, + "num_tokens": 295615953.0, + "step": 11426 + }, + { + "epoch": 1.254886887766308, + "grad_norm": 2.2169883251190186, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7175453901290894, + "num_tokens": 295644770.0, + "step": 11427 + }, + { + "epoch": 1.2549967054689217, + "grad_norm": 2.4686710834503174, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7263867259025574, + "num_tokens": 295665904.0, + "step": 11428 + }, + { + "epoch": 1.2551065231715353, + "grad_norm": 1.91775381565094, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.712463915348053, + "num_tokens": 295698262.0, + "step": 11429 + }, + { + "epoch": 1.2552163408741488, + "grad_norm": 2.2886440753936768, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6890997886657715, + "num_tokens": 295723447.0, + "step": 11430 + }, + { + "epoch": 1.2553261585767626, + "grad_norm": 2.4185023307800293, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7329139709472656, + "num_tokens": 295744698.0, + "step": 11431 + }, + { + "epoch": 1.2554359762793763, + "grad_norm": 2.6099507808685303, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7287336587905884, + "num_tokens": 295764393.0, + "step": 11432 + }, + { + "epoch": 1.2555457939819898, + "grad_norm": 2.3015801906585693, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7123328447341919, + "num_tokens": 295790134.0, + "step": 11433 + }, + { + "epoch": 1.2556556116846036, + "grad_norm": 2.4949967861175537, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7226685285568237, + "num_tokens": 295811072.0, + "step": 11434 + }, + { + "epoch": 1.2557654293872171, + "grad_norm": 2.123478412628174, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.746271014213562, + "num_tokens": 295837017.0, + "step": 11435 + }, + { + "epoch": 1.255875247089831, + "grad_norm": 2.448148012161255, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7172582149505615, + "num_tokens": 295859040.0, + "step": 11436 + }, + { + "epoch": 1.2559850647924446, + "grad_norm": 2.783701181411743, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7465434074401855, + "num_tokens": 295875491.0, + "step": 11437 + }, + { + "epoch": 1.2560948824950582, + "grad_norm": 2.2771735191345215, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7333410978317261, + "num_tokens": 295897898.0, + "step": 11438 + }, + { + "epoch": 1.256204700197672, + "grad_norm": 2.039764642715454, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6964871287345886, + "num_tokens": 295928789.0, + "step": 11439 + }, + { + "epoch": 1.2563145179002855, + "grad_norm": 2.2199573516845703, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7229026556015015, + "num_tokens": 295953425.0, + "step": 11440 + }, + { + "epoch": 1.2564243356028992, + "grad_norm": 2.082458257675171, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.6989727020263672, + "num_tokens": 295982609.0, + "step": 11441 + }, + { + "epoch": 1.2565341533055128, + "grad_norm": 2.3753323554992676, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6930969953536987, + "num_tokens": 296008120.0, + "step": 11442 + }, + { + "epoch": 1.2566439710081265, + "grad_norm": 2.714437961578369, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7364513278007507, + "num_tokens": 296025613.0, + "step": 11443 + }, + { + "epoch": 1.25675378871074, + "grad_norm": 2.1233322620391846, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7152140140533447, + "num_tokens": 296053874.0, + "step": 11444 + }, + { + "epoch": 1.2568636064133538, + "grad_norm": 2.6499836444854736, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.732752799987793, + "num_tokens": 296074215.0, + "step": 11445 + }, + { + "epoch": 1.2569734241159676, + "grad_norm": 2.169492483139038, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7063099145889282, + "num_tokens": 296102474.0, + "step": 11446 + }, + { + "epoch": 1.257083241818581, + "grad_norm": 2.094484567642212, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6977945566177368, + "num_tokens": 296131694.0, + "step": 11447 + }, + { + "epoch": 1.2571930595211949, + "grad_norm": 2.421529531478882, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7392115592956543, + "num_tokens": 296153126.0, + "step": 11448 + }, + { + "epoch": 1.2573028772238084, + "grad_norm": 2.3324851989746094, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.701675534248352, + "num_tokens": 296179671.0, + "step": 11449 + }, + { + "epoch": 1.2574126949264222, + "grad_norm": 2.4876973628997803, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7026009559631348, + "num_tokens": 296204581.0, + "step": 11450 + }, + { + "epoch": 1.257522512629036, + "grad_norm": 2.6707310676574707, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7196612358093262, + "num_tokens": 296224412.0, + "step": 11451 + }, + { + "epoch": 1.2576323303316495, + "grad_norm": 2.5030159950256348, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7176409959793091, + "num_tokens": 296246127.0, + "step": 11452 + }, + { + "epoch": 1.257742148034263, + "grad_norm": 2.299778461456299, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7104686498641968, + "num_tokens": 296269949.0, + "step": 11453 + }, + { + "epoch": 1.2578519657368767, + "grad_norm": 2.537950277328491, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6944806575775146, + "num_tokens": 296291955.0, + "step": 11454 + }, + { + "epoch": 1.2579617834394905, + "grad_norm": 2.2387988567352295, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7242962121963501, + "num_tokens": 296319512.0, + "step": 11455 + }, + { + "epoch": 1.258071601142104, + "grad_norm": 2.320396900177002, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7131389379501343, + "num_tokens": 296344515.0, + "step": 11456 + }, + { + "epoch": 1.2581814188447178, + "grad_norm": 2.239645481109619, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7084441184997559, + "num_tokens": 296371510.0, + "step": 11457 + }, + { + "epoch": 1.2582912365473313, + "grad_norm": 2.306849956512451, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7248479127883911, + "num_tokens": 296395037.0, + "step": 11458 + }, + { + "epoch": 1.258401054249945, + "grad_norm": 2.2610907554626465, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.720877468585968, + "num_tokens": 296420556.0, + "step": 11459 + }, + { + "epoch": 1.2585108719525588, + "grad_norm": 2.1285905838012695, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7018242478370667, + "num_tokens": 296449227.0, + "step": 11460 + }, + { + "epoch": 1.2586206896551724, + "grad_norm": 2.1451590061187744, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7220632433891296, + "num_tokens": 296474535.0, + "step": 11461 + }, + { + "epoch": 1.2587305073577861, + "grad_norm": 2.340717315673828, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7049257159233093, + "num_tokens": 296498255.0, + "step": 11462 + }, + { + "epoch": 1.2588403250603997, + "grad_norm": 2.0791356563568115, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7289905548095703, + "num_tokens": 296527472.0, + "step": 11463 + }, + { + "epoch": 1.2589501427630134, + "grad_norm": 2.4008915424346924, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7137477397918701, + "num_tokens": 296551741.0, + "step": 11464 + }, + { + "epoch": 1.2590599604656272, + "grad_norm": 2.1195335388183594, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.745580792427063, + "num_tokens": 296577777.0, + "step": 11465 + }, + { + "epoch": 1.2591697781682407, + "grad_norm": 1.9899002313613892, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7170939445495605, + "num_tokens": 296611448.0, + "step": 11466 + }, + { + "epoch": 1.2592795958708543, + "grad_norm": 2.506382942199707, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7168325781822205, + "num_tokens": 296638612.0, + "step": 11467 + }, + { + "epoch": 1.259389413573468, + "grad_norm": 2.2455925941467285, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7194850444793701, + "num_tokens": 296663290.0, + "step": 11468 + }, + { + "epoch": 1.2594992312760818, + "grad_norm": 2.243177652359009, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.732609212398529, + "num_tokens": 296688612.0, + "step": 11469 + }, + { + "epoch": 1.2596090489786953, + "grad_norm": 2.4192347526550293, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6922092437744141, + "num_tokens": 296718311.0, + "step": 11470 + }, + { + "epoch": 1.259718866681309, + "grad_norm": 2.335294723510742, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7125804424285889, + "num_tokens": 296741384.0, + "step": 11471 + }, + { + "epoch": 1.2598286843839226, + "grad_norm": 2.325627088546753, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7354776263237, + "num_tokens": 296764291.0, + "step": 11472 + }, + { + "epoch": 1.2599385020865363, + "grad_norm": 2.2377045154571533, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7150725722312927, + "num_tokens": 296790654.0, + "step": 11473 + }, + { + "epoch": 1.26004831978915, + "grad_norm": 2.011556625366211, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7294226884841919, + "num_tokens": 296821205.0, + "step": 11474 + }, + { + "epoch": 1.2601581374917636, + "grad_norm": 2.0177133083343506, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7314530611038208, + "num_tokens": 296849946.0, + "step": 11475 + }, + { + "epoch": 1.2602679551943774, + "grad_norm": 2.2378225326538086, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7146157026290894, + "num_tokens": 296879116.0, + "step": 11476 + }, + { + "epoch": 1.260377772896991, + "grad_norm": 2.2228477001190186, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6969805955886841, + "num_tokens": 296905714.0, + "step": 11477 + }, + { + "epoch": 1.2604875905996047, + "grad_norm": 2.2150843143463135, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7136793732643127, + "num_tokens": 296932215.0, + "step": 11478 + }, + { + "epoch": 1.2605974083022184, + "grad_norm": 2.06067156791687, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7623361945152283, + "num_tokens": 296959360.0, + "step": 11479 + }, + { + "epoch": 1.260707226004832, + "grad_norm": 2.419551372528076, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7050590515136719, + "num_tokens": 296981937.0, + "step": 11480 + }, + { + "epoch": 1.2608170437074455, + "grad_norm": 1.9444564580917358, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7119355201721191, + "num_tokens": 297013783.0, + "step": 11481 + }, + { + "epoch": 1.2609268614100593, + "grad_norm": 1.9934500455856323, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7016088962554932, + "num_tokens": 297045639.0, + "step": 11482 + }, + { + "epoch": 1.261036679112673, + "grad_norm": 2.0927774906158447, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7056920528411865, + "num_tokens": 297074810.0, + "step": 11483 + }, + { + "epoch": 1.2611464968152866, + "grad_norm": 2.1824021339416504, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7388503551483154, + "num_tokens": 297100410.0, + "step": 11484 + }, + { + "epoch": 1.2612563145179003, + "grad_norm": 1.9558992385864258, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7262985110282898, + "num_tokens": 297132999.0, + "step": 11485 + }, + { + "epoch": 1.2613661322205139, + "grad_norm": 2.3678195476531982, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7220061421394348, + "num_tokens": 297157221.0, + "step": 11486 + }, + { + "epoch": 1.2614759499231276, + "grad_norm": 2.281726360321045, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7083557844161987, + "num_tokens": 297183062.0, + "step": 11487 + }, + { + "epoch": 1.2615857676257414, + "grad_norm": 2.303402900695801, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7319185733795166, + "num_tokens": 297206154.0, + "step": 11488 + }, + { + "epoch": 1.261695585328355, + "grad_norm": 2.3510968685150146, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.733396053314209, + "num_tokens": 297228651.0, + "step": 11489 + }, + { + "epoch": 1.2618054030309687, + "grad_norm": 2.098771333694458, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7155295610427856, + "num_tokens": 297259028.0, + "step": 11490 + }, + { + "epoch": 1.2619152207335822, + "grad_norm": 2.158909797668457, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7219213247299194, + "num_tokens": 297285857.0, + "step": 11491 + }, + { + "epoch": 1.262025038436196, + "grad_norm": 2.3166298866271973, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.69537752866745, + "num_tokens": 297312496.0, + "step": 11492 + }, + { + "epoch": 1.2621348561388097, + "grad_norm": 2.258512258529663, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7010172605514526, + "num_tokens": 297338471.0, + "step": 11493 + }, + { + "epoch": 1.2622446738414232, + "grad_norm": 2.3947975635528564, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.725293755531311, + "num_tokens": 297363122.0, + "step": 11494 + }, + { + "epoch": 1.2623544915440368, + "grad_norm": 2.093181610107422, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7112356424331665, + "num_tokens": 297393466.0, + "step": 11495 + }, + { + "epoch": 1.2624643092466505, + "grad_norm": 2.2306900024414062, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7198630571365356, + "num_tokens": 297422391.0, + "step": 11496 + }, + { + "epoch": 1.2625741269492643, + "grad_norm": 2.1082019805908203, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6957692503929138, + "num_tokens": 297451099.0, + "step": 11497 + }, + { + "epoch": 1.2626839446518778, + "grad_norm": 2.2384800910949707, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7066934108734131, + "num_tokens": 297476115.0, + "step": 11498 + }, + { + "epoch": 1.2627937623544916, + "grad_norm": 2.4411041736602783, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7025673389434814, + "num_tokens": 297498409.0, + "step": 11499 + }, + { + "epoch": 1.2629035800571051, + "grad_norm": 2.35671067237854, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7316264510154724, + "num_tokens": 297521611.0, + "step": 11500 + }, + { + "epoch": 1.2630133977597189, + "grad_norm": 2.1027631759643555, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.732439398765564, + "num_tokens": 297548205.0, + "step": 11501 + }, + { + "epoch": 1.2631232154623326, + "grad_norm": 2.315014123916626, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.6987344026565552, + "num_tokens": 297573850.0, + "step": 11502 + }, + { + "epoch": 1.2632330331649462, + "grad_norm": 2.703819513320923, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7262523770332336, + "num_tokens": 297593424.0, + "step": 11503 + }, + { + "epoch": 1.26334285086756, + "grad_norm": 2.3186392784118652, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7041031718254089, + "num_tokens": 297618308.0, + "step": 11504 + }, + { + "epoch": 1.2634526685701735, + "grad_norm": 2.3579702377319336, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7198811173439026, + "num_tokens": 297642742.0, + "step": 11505 + }, + { + "epoch": 1.2635624862727872, + "grad_norm": 2.313410520553589, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7275746464729309, + "num_tokens": 297667547.0, + "step": 11506 + }, + { + "epoch": 1.2636723039754008, + "grad_norm": 2.2102205753326416, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7465168237686157, + "num_tokens": 297691845.0, + "step": 11507 + }, + { + "epoch": 1.2637821216780145, + "grad_norm": 2.327777624130249, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7264364957809448, + "num_tokens": 297716930.0, + "step": 11508 + }, + { + "epoch": 1.263891939380628, + "grad_norm": 2.054178476333618, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7070367336273193, + "num_tokens": 297747549.0, + "step": 11509 + }, + { + "epoch": 1.2640017570832418, + "grad_norm": 2.699673891067505, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.735022246837616, + "num_tokens": 297766898.0, + "step": 11510 + }, + { + "epoch": 1.2641115747858556, + "grad_norm": 2.329292058944702, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7031877636909485, + "num_tokens": 297791316.0, + "step": 11511 + }, + { + "epoch": 1.264221392488469, + "grad_norm": 2.5745394229888916, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7175477147102356, + "num_tokens": 297811952.0, + "step": 11512 + }, + { + "epoch": 1.2643312101910829, + "grad_norm": 2.4385576248168945, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7238240242004395, + "num_tokens": 297833174.0, + "step": 11513 + }, + { + "epoch": 1.2644410278936964, + "grad_norm": 2.5988104343414307, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7152758836746216, + "num_tokens": 297853684.0, + "step": 11514 + }, + { + "epoch": 1.2645508455963101, + "grad_norm": 2.208857536315918, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7022812366485596, + "num_tokens": 297880080.0, + "step": 11515 + }, + { + "epoch": 1.264660663298924, + "grad_norm": 2.0832321643829346, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7012323141098022, + "num_tokens": 297909560.0, + "step": 11516 + }, + { + "epoch": 1.2647704810015374, + "grad_norm": 2.331814765930176, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.717613935470581, + "num_tokens": 297932492.0, + "step": 11517 + }, + { + "epoch": 1.264880298704151, + "grad_norm": 2.0641121864318848, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7011605501174927, + "num_tokens": 297965319.0, + "step": 11518 + }, + { + "epoch": 1.2649901164067647, + "grad_norm": 2.243727445602417, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7002801299095154, + "num_tokens": 297993903.0, + "step": 11519 + }, + { + "epoch": 1.2650999341093785, + "grad_norm": 2.211374044418335, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7062664031982422, + "num_tokens": 298022800.0, + "step": 11520 + }, + { + "epoch": 1.265209751811992, + "grad_norm": 2.099745035171509, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7324076294898987, + "num_tokens": 298052026.0, + "step": 11521 + }, + { + "epoch": 1.2653195695146058, + "grad_norm": 2.3017055988311768, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7290357947349548, + "num_tokens": 298076619.0, + "step": 11522 + }, + { + "epoch": 1.2654293872172193, + "grad_norm": 2.3852109909057617, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.724014163017273, + "num_tokens": 298100965.0, + "step": 11523 + }, + { + "epoch": 1.265539204919833, + "grad_norm": 2.376405954360962, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6902362704277039, + "num_tokens": 298124809.0, + "step": 11524 + }, + { + "epoch": 1.2656490226224468, + "grad_norm": 2.0849449634552, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7050453424453735, + "num_tokens": 298155550.0, + "step": 11525 + }, + { + "epoch": 1.2657588403250604, + "grad_norm": 2.1388134956359863, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.710879921913147, + "num_tokens": 298181998.0, + "step": 11526 + }, + { + "epoch": 1.2658686580276741, + "grad_norm": 2.006664276123047, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7279976606369019, + "num_tokens": 298211803.0, + "step": 11527 + }, + { + "epoch": 1.2659784757302877, + "grad_norm": 2.5369932651519775, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7077386379241943, + "num_tokens": 298234110.0, + "step": 11528 + }, + { + "epoch": 1.2660882934329014, + "grad_norm": 1.9878920316696167, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.6959705352783203, + "num_tokens": 298265890.0, + "step": 11529 + }, + { + "epoch": 1.2661981111355152, + "grad_norm": 2.273200273513794, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7246246337890625, + "num_tokens": 298290478.0, + "step": 11530 + }, + { + "epoch": 1.2663079288381287, + "grad_norm": 2.1163458824157715, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7315414547920227, + "num_tokens": 298318961.0, + "step": 11531 + }, + { + "epoch": 1.2664177465407422, + "grad_norm": 2.3550164699554443, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7349838018417358, + "num_tokens": 298341171.0, + "step": 11532 + }, + { + "epoch": 1.266527564243356, + "grad_norm": 2.3156626224517822, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7272320985794067, + "num_tokens": 298365083.0, + "step": 11533 + }, + { + "epoch": 1.2666373819459698, + "grad_norm": 2.031672716140747, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7117847204208374, + "num_tokens": 298396270.0, + "step": 11534 + }, + { + "epoch": 1.2667471996485833, + "grad_norm": 2.447528600692749, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7125188112258911, + "num_tokens": 298420344.0, + "step": 11535 + }, + { + "epoch": 1.266857017351197, + "grad_norm": 1.9530748128890991, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7167351245880127, + "num_tokens": 298451386.0, + "step": 11536 + }, + { + "epoch": 1.2669668350538106, + "grad_norm": 2.4627246856689453, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.708550751209259, + "num_tokens": 298473767.0, + "step": 11537 + }, + { + "epoch": 1.2670766527564243, + "grad_norm": 2.4778034687042236, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7295613288879395, + "num_tokens": 298495330.0, + "step": 11538 + }, + { + "epoch": 1.267186470459038, + "grad_norm": 2.1878654956817627, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7049899101257324, + "num_tokens": 298524502.0, + "step": 11539 + }, + { + "epoch": 1.2672962881616516, + "grad_norm": 2.1760661602020264, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7151148915290833, + "num_tokens": 298552364.0, + "step": 11540 + }, + { + "epoch": 1.2674061058642654, + "grad_norm": 2.2248382568359375, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7177866697311401, + "num_tokens": 298578078.0, + "step": 11541 + }, + { + "epoch": 1.267515923566879, + "grad_norm": 2.117325782775879, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7177315950393677, + "num_tokens": 298605107.0, + "step": 11542 + }, + { + "epoch": 1.2676257412694927, + "grad_norm": 2.468944787979126, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.731921911239624, + "num_tokens": 298625913.0, + "step": 11543 + }, + { + "epoch": 1.2677355589721064, + "grad_norm": 2.0646414756774902, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6931607723236084, + "num_tokens": 298656895.0, + "step": 11544 + }, + { + "epoch": 1.26784537667472, + "grad_norm": 2.4481797218322754, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7178837060928345, + "num_tokens": 298680044.0, + "step": 11545 + }, + { + "epoch": 1.2679551943773335, + "grad_norm": 2.2107834815979004, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7106845378875732, + "num_tokens": 298707118.0, + "step": 11546 + }, + { + "epoch": 1.2680650120799473, + "grad_norm": 2.1612467765808105, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7045573592185974, + "num_tokens": 298731089.0, + "step": 11547 + }, + { + "epoch": 1.268174829782561, + "grad_norm": 2.136673927307129, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7398682832717896, + "num_tokens": 298757668.0, + "step": 11548 + }, + { + "epoch": 1.2682846474851746, + "grad_norm": 2.0929105281829834, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7301754951477051, + "num_tokens": 298785147.0, + "step": 11549 + }, + { + "epoch": 1.2683944651877883, + "grad_norm": 2.092393159866333, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7088915109634399, + "num_tokens": 298812858.0, + "step": 11550 + }, + { + "epoch": 1.2685042828904018, + "grad_norm": 2.2468652725219727, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7280199527740479, + "num_tokens": 298835848.0, + "step": 11551 + }, + { + "epoch": 1.2686141005930156, + "grad_norm": 2.483879327774048, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7151306867599487, + "num_tokens": 298856536.0, + "step": 11552 + }, + { + "epoch": 1.2687239182956294, + "grad_norm": 2.1355140209198, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7307323217391968, + "num_tokens": 298883334.0, + "step": 11553 + }, + { + "epoch": 1.268833735998243, + "grad_norm": 2.6385409832000732, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.7451585531234741, + "num_tokens": 298902690.0, + "step": 11554 + }, + { + "epoch": 1.2689435537008567, + "grad_norm": 1.969785451889038, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7090303897857666, + "num_tokens": 298935723.0, + "step": 11555 + }, + { + "epoch": 1.2690533714034702, + "grad_norm": 2.26576566696167, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7170999050140381, + "num_tokens": 298961129.0, + "step": 11556 + }, + { + "epoch": 1.269163189106084, + "grad_norm": 2.085184097290039, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7278575301170349, + "num_tokens": 298989359.0, + "step": 11557 + }, + { + "epoch": 1.2692730068086975, + "grad_norm": 2.1180098056793213, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7292251586914062, + "num_tokens": 299018271.0, + "step": 11558 + }, + { + "epoch": 1.2693828245113112, + "grad_norm": 2.419024705886841, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6961197853088379, + "num_tokens": 299041732.0, + "step": 11559 + }, + { + "epoch": 1.2694926422139248, + "grad_norm": 2.309058904647827, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.727127194404602, + "num_tokens": 299066729.0, + "step": 11560 + }, + { + "epoch": 1.2696024599165385, + "grad_norm": 2.4332528114318848, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7082107663154602, + "num_tokens": 299088342.0, + "step": 11561 + }, + { + "epoch": 1.2697122776191523, + "grad_norm": 2.3562588691711426, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.6983668208122253, + "num_tokens": 299113587.0, + "step": 11562 + }, + { + "epoch": 1.2698220953217658, + "grad_norm": 2.275222063064575, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7106952667236328, + "num_tokens": 299140214.0, + "step": 11563 + }, + { + "epoch": 1.2699319130243796, + "grad_norm": 1.9282970428466797, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7072243094444275, + "num_tokens": 299174699.0, + "step": 11564 + }, + { + "epoch": 1.2700417307269931, + "grad_norm": 2.352372646331787, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7290188074111938, + "num_tokens": 299206309.0, + "step": 11565 + }, + { + "epoch": 1.2701515484296069, + "grad_norm": 2.5937118530273438, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7252187728881836, + "num_tokens": 299225998.0, + "step": 11566 + }, + { + "epoch": 1.2702613661322206, + "grad_norm": 2.0213253498077393, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.716080904006958, + "num_tokens": 299258633.0, + "step": 11567 + }, + { + "epoch": 1.2703711838348342, + "grad_norm": 2.484180212020874, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.710107684135437, + "num_tokens": 299280740.0, + "step": 11568 + }, + { + "epoch": 1.270481001537448, + "grad_norm": 2.421504020690918, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7173532247543335, + "num_tokens": 299304369.0, + "step": 11569 + }, + { + "epoch": 1.2705908192400615, + "grad_norm": 2.215188503265381, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7012408971786499, + "num_tokens": 299332152.0, + "step": 11570 + }, + { + "epoch": 1.2707006369426752, + "grad_norm": 2.1602463722229004, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6994689106941223, + "num_tokens": 299361670.0, + "step": 11571 + }, + { + "epoch": 1.2708104546452887, + "grad_norm": 2.253718376159668, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6989842057228088, + "num_tokens": 299389803.0, + "step": 11572 + }, + { + "epoch": 1.2709202723479025, + "grad_norm": 2.0010669231414795, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7166895270347595, + "num_tokens": 299420332.0, + "step": 11573 + }, + { + "epoch": 1.271030090050516, + "grad_norm": 2.3451521396636963, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.6988666653633118, + "num_tokens": 299444266.0, + "step": 11574 + }, + { + "epoch": 1.2711399077531298, + "grad_norm": 1.9486860036849976, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7121917009353638, + "num_tokens": 299476084.0, + "step": 11575 + }, + { + "epoch": 1.2712497254557436, + "grad_norm": 2.2810304164886475, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7189611196517944, + "num_tokens": 299501733.0, + "step": 11576 + }, + { + "epoch": 1.271359543158357, + "grad_norm": 2.4027953147888184, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7108198404312134, + "num_tokens": 299524569.0, + "step": 11577 + }, + { + "epoch": 1.2714693608609708, + "grad_norm": 2.2813382148742676, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7456997036933899, + "num_tokens": 299546818.0, + "step": 11578 + }, + { + "epoch": 1.2715791785635844, + "grad_norm": 2.480726718902588, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7183239459991455, + "num_tokens": 299568630.0, + "step": 11579 + }, + { + "epoch": 1.2716889962661981, + "grad_norm": 2.2335517406463623, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7057470083236694, + "num_tokens": 299595091.0, + "step": 11580 + }, + { + "epoch": 1.271798813968812, + "grad_norm": 2.077449321746826, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7200408577919006, + "num_tokens": 299624124.0, + "step": 11581 + }, + { + "epoch": 1.2719086316714254, + "grad_norm": 2.3429477214813232, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.701430082321167, + "num_tokens": 299649454.0, + "step": 11582 + }, + { + "epoch": 1.272018449374039, + "grad_norm": 2.3538143634796143, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.7066446542739868, + "num_tokens": 299675046.0, + "step": 11583 + }, + { + "epoch": 1.2721282670766527, + "grad_norm": 2.1482882499694824, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7008676528930664, + "num_tokens": 299704160.0, + "step": 11584 + }, + { + "epoch": 1.2722380847792665, + "grad_norm": 2.273021697998047, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6875671148300171, + "num_tokens": 299730361.0, + "step": 11585 + }, + { + "epoch": 1.27234790248188, + "grad_norm": 2.166841983795166, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7180556058883667, + "num_tokens": 299757251.0, + "step": 11586 + }, + { + "epoch": 1.2724577201844938, + "grad_norm": 2.6937718391418457, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.732275664806366, + "num_tokens": 299777812.0, + "step": 11587 + }, + { + "epoch": 1.2725675378871073, + "grad_norm": 2.3732995986938477, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7449747920036316, + "num_tokens": 299799763.0, + "step": 11588 + }, + { + "epoch": 1.272677355589721, + "grad_norm": 2.2386834621429443, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7113346457481384, + "num_tokens": 299823944.0, + "step": 11589 + }, + { + "epoch": 1.2727871732923348, + "grad_norm": 2.4204089641571045, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7383260726928711, + "num_tokens": 299846384.0, + "step": 11590 + }, + { + "epoch": 1.2728969909949484, + "grad_norm": 2.0548338890075684, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6792001724243164, + "num_tokens": 299877965.0, + "step": 11591 + }, + { + "epoch": 1.273006808697562, + "grad_norm": 2.0686559677124023, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7298920750617981, + "num_tokens": 299907103.0, + "step": 11592 + }, + { + "epoch": 1.2731166264001756, + "grad_norm": 2.0595123767852783, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7056494355201721, + "num_tokens": 299939433.0, + "step": 11593 + }, + { + "epoch": 1.2732264441027894, + "grad_norm": 2.1886935234069824, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7443195581436157, + "num_tokens": 299965674.0, + "step": 11594 + }, + { + "epoch": 1.2733362618054032, + "grad_norm": 2.4484193325042725, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.709528923034668, + "num_tokens": 299988125.0, + "step": 11595 + }, + { + "epoch": 1.2734460795080167, + "grad_norm": 2.5640640258789062, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7399090528488159, + "num_tokens": 300007609.0, + "step": 11596 + }, + { + "epoch": 1.2735558972106302, + "grad_norm": 1.9762924909591675, + "learning_rate": 1e-06, + "loss": 1.0766, + "mean_token_accuracy": 0.6750285029411316, + "num_tokens": 300042810.0, + "step": 11597 + }, + { + "epoch": 1.273665714913244, + "grad_norm": 2.4299819469451904, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7105302810668945, + "num_tokens": 300065000.0, + "step": 11598 + }, + { + "epoch": 1.2737755326158577, + "grad_norm": 2.125121593475342, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7269961833953857, + "num_tokens": 300093292.0, + "step": 11599 + }, + { + "epoch": 1.2738853503184713, + "grad_norm": 2.2658803462982178, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7326174974441528, + "num_tokens": 300116894.0, + "step": 11600 + }, + { + "epoch": 1.273995168021085, + "grad_norm": 2.549473762512207, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.705957293510437, + "num_tokens": 300138074.0, + "step": 11601 + }, + { + "epoch": 1.2741049857236986, + "grad_norm": 2.0408782958984375, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7274752855300903, + "num_tokens": 300167547.0, + "step": 11602 + }, + { + "epoch": 1.2742148034263123, + "grad_norm": 2.0537893772125244, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7080967426300049, + "num_tokens": 300197864.0, + "step": 11603 + }, + { + "epoch": 1.274324621128926, + "grad_norm": 2.2705655097961426, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7068840265274048, + "num_tokens": 300223913.0, + "step": 11604 + }, + { + "epoch": 1.2744344388315396, + "grad_norm": 2.0870039463043213, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7225626707077026, + "num_tokens": 300253707.0, + "step": 11605 + }, + { + "epoch": 1.2745442565341534, + "grad_norm": 2.277254343032837, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7328127026557922, + "num_tokens": 300277603.0, + "step": 11606 + }, + { + "epoch": 1.274654074236767, + "grad_norm": 2.030346155166626, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.715316653251648, + "num_tokens": 300308542.0, + "step": 11607 + }, + { + "epoch": 1.2747638919393807, + "grad_norm": 2.7360732555389404, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7210275530815125, + "num_tokens": 300326984.0, + "step": 11608 + }, + { + "epoch": 1.2748737096419944, + "grad_norm": 2.1632461547851562, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7188063859939575, + "num_tokens": 300353479.0, + "step": 11609 + }, + { + "epoch": 1.274983527344608, + "grad_norm": 2.3598084449768066, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6985298991203308, + "num_tokens": 300377842.0, + "step": 11610 + }, + { + "epoch": 1.2750933450472215, + "grad_norm": 2.5771923065185547, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7189338207244873, + "num_tokens": 300399800.0, + "step": 11611 + }, + { + "epoch": 1.2752031627498353, + "grad_norm": 2.5961496829986572, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7007066607475281, + "num_tokens": 300420136.0, + "step": 11612 + }, + { + "epoch": 1.275312980452449, + "grad_norm": 2.3491368293762207, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7373597025871277, + "num_tokens": 300443520.0, + "step": 11613 + }, + { + "epoch": 1.2754227981550625, + "grad_norm": 2.2938337326049805, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7338765859603882, + "num_tokens": 300468267.0, + "step": 11614 + }, + { + "epoch": 1.2755326158576763, + "grad_norm": 2.148571729660034, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7111226320266724, + "num_tokens": 300498103.0, + "step": 11615 + }, + { + "epoch": 1.2756424335602898, + "grad_norm": 2.2476460933685303, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.730800986289978, + "num_tokens": 300521745.0, + "step": 11616 + }, + { + "epoch": 1.2757522512629036, + "grad_norm": 2.0890660285949707, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7262246608734131, + "num_tokens": 300550409.0, + "step": 11617 + }, + { + "epoch": 1.2758620689655173, + "grad_norm": 2.3751237392425537, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7292569875717163, + "num_tokens": 300573697.0, + "step": 11618 + }, + { + "epoch": 1.2759718866681309, + "grad_norm": 2.475452184677124, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7222495675086975, + "num_tokens": 300595692.0, + "step": 11619 + }, + { + "epoch": 1.2760817043707446, + "grad_norm": 2.1705925464630127, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7210767269134521, + "num_tokens": 300621855.0, + "step": 11620 + }, + { + "epoch": 1.2761915220733582, + "grad_norm": 2.225184679031372, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7433080077171326, + "num_tokens": 300648529.0, + "step": 11621 + }, + { + "epoch": 1.276301339775972, + "grad_norm": 2.0936472415924072, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7057029008865356, + "num_tokens": 300680264.0, + "step": 11622 + }, + { + "epoch": 1.2764111574785855, + "grad_norm": 2.481536865234375, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7141236662864685, + "num_tokens": 300703359.0, + "step": 11623 + }, + { + "epoch": 1.2765209751811992, + "grad_norm": 3.003314256668091, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.708225429058075, + "num_tokens": 300722624.0, + "step": 11624 + }, + { + "epoch": 1.2766307928838128, + "grad_norm": 2.1571459770202637, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7234882116317749, + "num_tokens": 300749852.0, + "step": 11625 + }, + { + "epoch": 1.2767406105864265, + "grad_norm": 2.4013965129852295, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7423451542854309, + "num_tokens": 300772727.0, + "step": 11626 + }, + { + "epoch": 1.2768504282890403, + "grad_norm": 2.32808518409729, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7149348855018616, + "num_tokens": 300796476.0, + "step": 11627 + }, + { + "epoch": 1.2769602459916538, + "grad_norm": 2.4529385566711426, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7134597897529602, + "num_tokens": 300821612.0, + "step": 11628 + }, + { + "epoch": 1.2770700636942676, + "grad_norm": 2.2362265586853027, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7399114966392517, + "num_tokens": 300847114.0, + "step": 11629 + }, + { + "epoch": 1.277179881396881, + "grad_norm": 2.582887887954712, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7180193662643433, + "num_tokens": 300867258.0, + "step": 11630 + }, + { + "epoch": 1.2772896990994949, + "grad_norm": 2.0951461791992188, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6888203620910645, + "num_tokens": 300895077.0, + "step": 11631 + }, + { + "epoch": 1.2773995168021086, + "grad_norm": 2.0993447303771973, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7056781053543091, + "num_tokens": 300926950.0, + "step": 11632 + }, + { + "epoch": 1.2775093345047221, + "grad_norm": 2.068329334259033, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7223848104476929, + "num_tokens": 300954676.0, + "step": 11633 + }, + { + "epoch": 1.2776191522073357, + "grad_norm": 2.32869553565979, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7227555513381958, + "num_tokens": 300979637.0, + "step": 11634 + }, + { + "epoch": 1.2777289699099494, + "grad_norm": 2.0800249576568604, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.707421064376831, + "num_tokens": 301008969.0, + "step": 11635 + }, + { + "epoch": 1.2778387876125632, + "grad_norm": 2.118521213531494, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.705742597579956, + "num_tokens": 301039515.0, + "step": 11636 + }, + { + "epoch": 1.2779486053151767, + "grad_norm": 2.1544625759124756, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7252463102340698, + "num_tokens": 301066854.0, + "step": 11637 + }, + { + "epoch": 1.2780584230177905, + "grad_norm": 2.0701565742492676, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.704057514667511, + "num_tokens": 301097544.0, + "step": 11638 + }, + { + "epoch": 1.278168240720404, + "grad_norm": 1.954484462738037, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7195016145706177, + "num_tokens": 301128404.0, + "step": 11639 + }, + { + "epoch": 1.2782780584230178, + "grad_norm": 2.2804198265075684, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7154871225357056, + "num_tokens": 301154182.0, + "step": 11640 + }, + { + "epoch": 1.2783878761256315, + "grad_norm": 2.555582046508789, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7277019619941711, + "num_tokens": 301174498.0, + "step": 11641 + }, + { + "epoch": 1.278497693828245, + "grad_norm": 2.355551242828369, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7212129831314087, + "num_tokens": 301198593.0, + "step": 11642 + }, + { + "epoch": 1.2786075115308588, + "grad_norm": 2.3994674682617188, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7343858480453491, + "num_tokens": 301219877.0, + "step": 11643 + }, + { + "epoch": 1.2787173292334724, + "grad_norm": 2.7094171047210693, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7445101737976074, + "num_tokens": 301237174.0, + "step": 11644 + }, + { + "epoch": 1.2788271469360861, + "grad_norm": 2.2008090019226074, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7266255021095276, + "num_tokens": 301264961.0, + "step": 11645 + }, + { + "epoch": 1.2789369646386999, + "grad_norm": 2.226680278778076, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7224732637405396, + "num_tokens": 301288854.0, + "step": 11646 + }, + { + "epoch": 1.2790467823413134, + "grad_norm": 2.1303746700286865, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7217475175857544, + "num_tokens": 301317681.0, + "step": 11647 + }, + { + "epoch": 1.279156600043927, + "grad_norm": 2.175241231918335, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.732545793056488, + "num_tokens": 301342976.0, + "step": 11648 + }, + { + "epoch": 1.2792664177465407, + "grad_norm": 2.076096534729004, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7151318192481995, + "num_tokens": 301371968.0, + "step": 11649 + }, + { + "epoch": 1.2793762354491545, + "grad_norm": 2.0941529273986816, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7123810052871704, + "num_tokens": 301399863.0, + "step": 11650 + }, + { + "epoch": 1.279486053151768, + "grad_norm": 2.5374088287353516, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7157982587814331, + "num_tokens": 301421566.0, + "step": 11651 + }, + { + "epoch": 1.2795958708543818, + "grad_norm": 2.17657470703125, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7312353849411011, + "num_tokens": 301448732.0, + "step": 11652 + }, + { + "epoch": 1.2797056885569953, + "grad_norm": 2.32373309135437, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.6960667371749878, + "num_tokens": 301472964.0, + "step": 11653 + }, + { + "epoch": 1.279815506259609, + "grad_norm": 2.2747292518615723, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.721387505531311, + "num_tokens": 301499186.0, + "step": 11654 + }, + { + "epoch": 1.2799253239622228, + "grad_norm": 2.075430393218994, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7192389369010925, + "num_tokens": 301526014.0, + "step": 11655 + }, + { + "epoch": 1.2800351416648363, + "grad_norm": 2.2460880279541016, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7090030908584595, + "num_tokens": 301551649.0, + "step": 11656 + }, + { + "epoch": 1.28014495936745, + "grad_norm": 2.1755731105804443, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6942715048789978, + "num_tokens": 301580729.0, + "step": 11657 + }, + { + "epoch": 1.2802547770700636, + "grad_norm": 2.058336019515991, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6980192065238953, + "num_tokens": 301610803.0, + "step": 11658 + }, + { + "epoch": 1.2803645947726774, + "grad_norm": 2.328359603881836, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7259098291397095, + "num_tokens": 301634563.0, + "step": 11659 + }, + { + "epoch": 1.2804744124752911, + "grad_norm": 2.445596218109131, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7333411574363708, + "num_tokens": 301655487.0, + "step": 11660 + }, + { + "epoch": 1.2805842301779047, + "grad_norm": 1.852474331855774, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7029149532318115, + "num_tokens": 301692870.0, + "step": 11661 + }, + { + "epoch": 1.2806940478805182, + "grad_norm": 2.3006417751312256, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7033230066299438, + "num_tokens": 301718774.0, + "step": 11662 + }, + { + "epoch": 1.280803865583132, + "grad_norm": 2.014726400375366, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7403991222381592, + "num_tokens": 301745583.0, + "step": 11663 + }, + { + "epoch": 1.2809136832857457, + "grad_norm": 2.6643013954162598, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7190240025520325, + "num_tokens": 301766400.0, + "step": 11664 + }, + { + "epoch": 1.2810235009883593, + "grad_norm": 2.3790276050567627, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7343116998672485, + "num_tokens": 301788337.0, + "step": 11665 + }, + { + "epoch": 1.281133318690973, + "grad_norm": 2.354861259460449, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.711860179901123, + "num_tokens": 301812213.0, + "step": 11666 + }, + { + "epoch": 1.2812431363935866, + "grad_norm": 2.41274094581604, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7411055564880371, + "num_tokens": 301834284.0, + "step": 11667 + }, + { + "epoch": 1.2813529540962003, + "grad_norm": 1.9994335174560547, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7096424102783203, + "num_tokens": 301866714.0, + "step": 11668 + }, + { + "epoch": 1.281462771798814, + "grad_norm": 2.1513471603393555, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7185901403427124, + "num_tokens": 301895602.0, + "step": 11669 + }, + { + "epoch": 1.2815725895014276, + "grad_norm": 2.2830591201782227, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7109087705612183, + "num_tokens": 301922347.0, + "step": 11670 + }, + { + "epoch": 1.2816824072040414, + "grad_norm": 2.189288854598999, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7236951589584351, + "num_tokens": 301948200.0, + "step": 11671 + }, + { + "epoch": 1.281792224906655, + "grad_norm": 2.373124599456787, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7173635959625244, + "num_tokens": 301972825.0, + "step": 11672 + }, + { + "epoch": 1.2819020426092687, + "grad_norm": 2.6369659900665283, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7305185198783875, + "num_tokens": 301992609.0, + "step": 11673 + }, + { + "epoch": 1.2820118603118824, + "grad_norm": 2.12398099899292, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7065073251724243, + "num_tokens": 302019194.0, + "step": 11674 + }, + { + "epoch": 1.282121678014496, + "grad_norm": 2.3046798706054688, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7067304253578186, + "num_tokens": 302045526.0, + "step": 11675 + }, + { + "epoch": 1.2822314957171095, + "grad_norm": 2.3284337520599365, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.707297146320343, + "num_tokens": 302071242.0, + "step": 11676 + }, + { + "epoch": 1.2823413134197232, + "grad_norm": 2.221299648284912, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7064680457115173, + "num_tokens": 302099235.0, + "step": 11677 + }, + { + "epoch": 1.282451131122337, + "grad_norm": 2.2949118614196777, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7167003750801086, + "num_tokens": 302125416.0, + "step": 11678 + }, + { + "epoch": 1.2825609488249505, + "grad_norm": 2.067220687866211, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7151495814323425, + "num_tokens": 302155002.0, + "step": 11679 + }, + { + "epoch": 1.2826707665275643, + "grad_norm": 2.383312463760376, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.717462420463562, + "num_tokens": 302177355.0, + "step": 11680 + }, + { + "epoch": 1.2827805842301778, + "grad_norm": 2.3438169956207275, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7283731698989868, + "num_tokens": 302199821.0, + "step": 11681 + }, + { + "epoch": 1.2828904019327916, + "grad_norm": 2.182806968688965, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7124425768852234, + "num_tokens": 302228048.0, + "step": 11682 + }, + { + "epoch": 1.2830002196354053, + "grad_norm": 2.3174729347229004, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7329351902008057, + "num_tokens": 302250998.0, + "step": 11683 + }, + { + "epoch": 1.2831100373380189, + "grad_norm": 2.3074917793273926, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7352146506309509, + "num_tokens": 302275876.0, + "step": 11684 + }, + { + "epoch": 1.2832198550406326, + "grad_norm": 2.1494157314300537, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.724819540977478, + "num_tokens": 302303707.0, + "step": 11685 + }, + { + "epoch": 1.2833296727432462, + "grad_norm": 2.1278345584869385, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.742379903793335, + "num_tokens": 302329954.0, + "step": 11686 + }, + { + "epoch": 1.28343949044586, + "grad_norm": 2.534414291381836, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7248831987380981, + "num_tokens": 302352435.0, + "step": 11687 + }, + { + "epoch": 1.2835493081484735, + "grad_norm": 1.9839117527008057, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7248508334159851, + "num_tokens": 302383080.0, + "step": 11688 + }, + { + "epoch": 1.2836591258510872, + "grad_norm": 2.274362564086914, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7001778483390808, + "num_tokens": 302407828.0, + "step": 11689 + }, + { + "epoch": 1.2837689435537007, + "grad_norm": 2.2643847465515137, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7170354723930359, + "num_tokens": 302434324.0, + "step": 11690 + }, + { + "epoch": 1.2838787612563145, + "grad_norm": 1.964972734451294, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.717729926109314, + "num_tokens": 302466784.0, + "step": 11691 + }, + { + "epoch": 1.2839885789589283, + "grad_norm": 2.3645436763763428, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7387534379959106, + "num_tokens": 302490422.0, + "step": 11692 + }, + { + "epoch": 1.2840983966615418, + "grad_norm": 2.1332805156707764, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6926300525665283, + "num_tokens": 302521864.0, + "step": 11693 + }, + { + "epoch": 1.2842082143641556, + "grad_norm": 2.1215906143188477, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7175527811050415, + "num_tokens": 302550059.0, + "step": 11694 + }, + { + "epoch": 1.284318032066769, + "grad_norm": 2.336503267288208, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7247451543807983, + "num_tokens": 302572257.0, + "step": 11695 + }, + { + "epoch": 1.2844278497693828, + "grad_norm": 2.2483127117156982, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7191780805587769, + "num_tokens": 302597184.0, + "step": 11696 + }, + { + "epoch": 1.2845376674719966, + "grad_norm": 2.311291217803955, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7244499921798706, + "num_tokens": 302625370.0, + "step": 11697 + }, + { + "epoch": 1.2846474851746101, + "grad_norm": 2.527960777282715, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7140787243843079, + "num_tokens": 302647698.0, + "step": 11698 + }, + { + "epoch": 1.2847573028772237, + "grad_norm": 2.210629940032959, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7227058410644531, + "num_tokens": 302675217.0, + "step": 11699 + }, + { + "epoch": 1.2848671205798374, + "grad_norm": 2.2452707290649414, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7134252786636353, + "num_tokens": 302701118.0, + "step": 11700 + }, + { + "epoch": 1.2849769382824512, + "grad_norm": 2.6638336181640625, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7067029476165771, + "num_tokens": 302721586.0, + "step": 11701 + }, + { + "epoch": 1.2850867559850647, + "grad_norm": 2.158637762069702, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7193591594696045, + "num_tokens": 302749143.0, + "step": 11702 + }, + { + "epoch": 1.2851965736876785, + "grad_norm": 2.3909528255462646, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7218074798583984, + "num_tokens": 302771892.0, + "step": 11703 + }, + { + "epoch": 1.285306391390292, + "grad_norm": 2.264728546142578, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7186267375946045, + "num_tokens": 302795399.0, + "step": 11704 + }, + { + "epoch": 1.2854162090929058, + "grad_norm": 2.2181451320648193, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7290598154067993, + "num_tokens": 302820330.0, + "step": 11705 + }, + { + "epoch": 1.2855260267955195, + "grad_norm": 2.2812321186065674, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7010596990585327, + "num_tokens": 302844470.0, + "step": 11706 + }, + { + "epoch": 1.285635844498133, + "grad_norm": 2.0993456840515137, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6916298270225525, + "num_tokens": 302875201.0, + "step": 11707 + }, + { + "epoch": 1.2857456622007468, + "grad_norm": 2.2520132064819336, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7297514081001282, + "num_tokens": 302900116.0, + "step": 11708 + }, + { + "epoch": 1.2858554799033604, + "grad_norm": 2.226184606552124, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7007806301116943, + "num_tokens": 302926066.0, + "step": 11709 + }, + { + "epoch": 1.2859652976059741, + "grad_norm": 2.32365083694458, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7190749049186707, + "num_tokens": 302951977.0, + "step": 11710 + }, + { + "epoch": 1.2860751153085879, + "grad_norm": 2.3560678958892822, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7157512903213501, + "num_tokens": 302976941.0, + "step": 11711 + }, + { + "epoch": 1.2861849330112014, + "grad_norm": 2.5208070278167725, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7209497094154358, + "num_tokens": 302999808.0, + "step": 11712 + }, + { + "epoch": 1.286294750713815, + "grad_norm": 1.9907485246658325, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7454550266265869, + "num_tokens": 303030124.0, + "step": 11713 + }, + { + "epoch": 1.2864045684164287, + "grad_norm": 2.0077197551727295, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7300318479537964, + "num_tokens": 303062520.0, + "step": 11714 + }, + { + "epoch": 1.2865143861190425, + "grad_norm": 2.340207815170288, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.718326210975647, + "num_tokens": 303087211.0, + "step": 11715 + }, + { + "epoch": 1.286624203821656, + "grad_norm": 2.1202094554901123, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7233046293258667, + "num_tokens": 303114010.0, + "step": 11716 + }, + { + "epoch": 1.2867340215242697, + "grad_norm": 2.821061134338379, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7397952079772949, + "num_tokens": 303132223.0, + "step": 11717 + }, + { + "epoch": 1.2868438392268833, + "grad_norm": 2.3912453651428223, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7261980772018433, + "num_tokens": 303154590.0, + "step": 11718 + }, + { + "epoch": 1.286953656929497, + "grad_norm": 2.1064655780792236, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7091237902641296, + "num_tokens": 303181959.0, + "step": 11719 + }, + { + "epoch": 1.2870634746321108, + "grad_norm": 2.3823482990264893, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7387466430664062, + "num_tokens": 303206582.0, + "step": 11720 + }, + { + "epoch": 1.2871732923347243, + "grad_norm": 2.349696636199951, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7107328772544861, + "num_tokens": 303230746.0, + "step": 11721 + }, + { + "epoch": 1.287283110037338, + "grad_norm": 2.1393539905548096, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.745825469493866, + "num_tokens": 303256952.0, + "step": 11722 + }, + { + "epoch": 1.2873929277399516, + "grad_norm": 2.6414542198181152, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.735232949256897, + "num_tokens": 303276667.0, + "step": 11723 + }, + { + "epoch": 1.2875027454425654, + "grad_norm": 2.0966057777404785, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7361841201782227, + "num_tokens": 303302982.0, + "step": 11724 + }, + { + "epoch": 1.2876125631451791, + "grad_norm": 2.37467622756958, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7086800336837769, + "num_tokens": 303330334.0, + "step": 11725 + }, + { + "epoch": 1.2877223808477927, + "grad_norm": 2.219278573989868, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.69483482837677, + "num_tokens": 303358251.0, + "step": 11726 + }, + { + "epoch": 1.2878321985504062, + "grad_norm": 2.2179014682769775, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7444285750389099, + "num_tokens": 303381982.0, + "step": 11727 + }, + { + "epoch": 1.28794201625302, + "grad_norm": 2.576777696609497, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.734644889831543, + "num_tokens": 303401346.0, + "step": 11728 + }, + { + "epoch": 1.2880518339556337, + "grad_norm": 2.1855180263519287, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7113481760025024, + "num_tokens": 303426493.0, + "step": 11729 + }, + { + "epoch": 1.2881616516582473, + "grad_norm": 2.0739667415618896, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.681878924369812, + "num_tokens": 303458971.0, + "step": 11730 + }, + { + "epoch": 1.288271469360861, + "grad_norm": 2.3647804260253906, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7450917959213257, + "num_tokens": 303482654.0, + "step": 11731 + }, + { + "epoch": 1.2883812870634745, + "grad_norm": 1.9149119853973389, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7235914468765259, + "num_tokens": 303515731.0, + "step": 11732 + }, + { + "epoch": 1.2884911047660883, + "grad_norm": 2.1917173862457275, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7369372844696045, + "num_tokens": 303540606.0, + "step": 11733 + }, + { + "epoch": 1.288600922468702, + "grad_norm": 2.378086805343628, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7245966196060181, + "num_tokens": 303562877.0, + "step": 11734 + }, + { + "epoch": 1.2887107401713156, + "grad_norm": 2.1523635387420654, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7034127116203308, + "num_tokens": 303590329.0, + "step": 11735 + }, + { + "epoch": 1.2888205578739294, + "grad_norm": 1.972669243812561, + "learning_rate": 1e-06, + "loss": 1.0483, + "mean_token_accuracy": 0.6792634725570679, + "num_tokens": 303622275.0, + "step": 11736 + }, + { + "epoch": 1.2889303755765429, + "grad_norm": 2.0256741046905518, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7035009860992432, + "num_tokens": 303652835.0, + "step": 11737 + }, + { + "epoch": 1.2890401932791566, + "grad_norm": 2.285529613494873, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7195966243743896, + "num_tokens": 303677543.0, + "step": 11738 + }, + { + "epoch": 1.2891500109817704, + "grad_norm": 2.1556859016418457, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7158971428871155, + "num_tokens": 303705579.0, + "step": 11739 + }, + { + "epoch": 1.289259828684384, + "grad_norm": 2.327005624771118, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7334331274032593, + "num_tokens": 303727515.0, + "step": 11740 + }, + { + "epoch": 1.2893696463869975, + "grad_norm": 2.0930240154266357, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7167990207672119, + "num_tokens": 303757209.0, + "step": 11741 + }, + { + "epoch": 1.2894794640896112, + "grad_norm": 2.1735103130340576, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7018682956695557, + "num_tokens": 303785195.0, + "step": 11742 + }, + { + "epoch": 1.289589281792225, + "grad_norm": 2.5938847064971924, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7164885997772217, + "num_tokens": 303804918.0, + "step": 11743 + }, + { + "epoch": 1.2896990994948385, + "grad_norm": 2.2630558013916016, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7096426486968994, + "num_tokens": 303829718.0, + "step": 11744 + }, + { + "epoch": 1.2898089171974523, + "grad_norm": 2.1264798641204834, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7113492488861084, + "num_tokens": 303858466.0, + "step": 11745 + }, + { + "epoch": 1.2899187349000658, + "grad_norm": 2.2781364917755127, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7466785311698914, + "num_tokens": 303881926.0, + "step": 11746 + }, + { + "epoch": 1.2900285526026796, + "grad_norm": 2.143078327178955, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7297280430793762, + "num_tokens": 303908711.0, + "step": 11747 + }, + { + "epoch": 1.2901383703052933, + "grad_norm": 2.198061466217041, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7075244188308716, + "num_tokens": 303934977.0, + "step": 11748 + }, + { + "epoch": 1.2902481880079069, + "grad_norm": 2.087568759918213, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7245616316795349, + "num_tokens": 303962166.0, + "step": 11749 + }, + { + "epoch": 1.2903580057105206, + "grad_norm": 2.018205404281616, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7125042676925659, + "num_tokens": 303993221.0, + "step": 11750 + }, + { + "epoch": 1.2904678234131342, + "grad_norm": 2.40640926361084, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6976394653320312, + "num_tokens": 304018005.0, + "step": 11751 + }, + { + "epoch": 1.290577641115748, + "grad_norm": 2.308391809463501, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7252557277679443, + "num_tokens": 304042860.0, + "step": 11752 + }, + { + "epoch": 1.2906874588183614, + "grad_norm": 2.076310157775879, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7097943425178528, + "num_tokens": 304072245.0, + "step": 11753 + }, + { + "epoch": 1.2907972765209752, + "grad_norm": 2.253683567047119, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7208937406539917, + "num_tokens": 304096862.0, + "step": 11754 + }, + { + "epoch": 1.2909070942235887, + "grad_norm": 2.1353819370269775, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7190276980400085, + "num_tokens": 304124158.0, + "step": 11755 + }, + { + "epoch": 1.2910169119262025, + "grad_norm": 2.292593002319336, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6934760808944702, + "num_tokens": 304152468.0, + "step": 11756 + }, + { + "epoch": 1.2911267296288162, + "grad_norm": 2.3805131912231445, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7189081311225891, + "num_tokens": 304175812.0, + "step": 11757 + }, + { + "epoch": 1.2912365473314298, + "grad_norm": 2.309032678604126, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.737913966178894, + "num_tokens": 304200028.0, + "step": 11758 + }, + { + "epoch": 1.2913463650340435, + "grad_norm": 2.3348469734191895, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7321299314498901, + "num_tokens": 304224531.0, + "step": 11759 + }, + { + "epoch": 1.291456182736657, + "grad_norm": 1.9610928297042847, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6947430372238159, + "num_tokens": 304256347.0, + "step": 11760 + }, + { + "epoch": 1.2915660004392708, + "grad_norm": 2.2642300128936768, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7041784524917603, + "num_tokens": 304281763.0, + "step": 11761 + }, + { + "epoch": 1.2916758181418846, + "grad_norm": 2.427455425262451, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7562271356582642, + "num_tokens": 304304235.0, + "step": 11762 + }, + { + "epoch": 1.2917856358444981, + "grad_norm": 2.326573610305786, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7258384227752686, + "num_tokens": 304329706.0, + "step": 11763 + }, + { + "epoch": 1.2918954535471117, + "grad_norm": 2.0850377082824707, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7352735996246338, + "num_tokens": 304360223.0, + "step": 11764 + }, + { + "epoch": 1.2920052712497254, + "grad_norm": 2.3129377365112305, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.729099690914154, + "num_tokens": 304382198.0, + "step": 11765 + }, + { + "epoch": 1.2921150889523392, + "grad_norm": 1.9373220205307007, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7317359447479248, + "num_tokens": 304415819.0, + "step": 11766 + }, + { + "epoch": 1.2922249066549527, + "grad_norm": 2.196866035461426, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7101162672042847, + "num_tokens": 304443269.0, + "step": 11767 + }, + { + "epoch": 1.2923347243575665, + "grad_norm": 2.3832905292510986, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7027847766876221, + "num_tokens": 304467737.0, + "step": 11768 + }, + { + "epoch": 1.29244454206018, + "grad_norm": 2.02012300491333, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7060078382492065, + "num_tokens": 304498891.0, + "step": 11769 + }, + { + "epoch": 1.2925543597627938, + "grad_norm": 2.567105293273926, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7188903093338013, + "num_tokens": 304521902.0, + "step": 11770 + }, + { + "epoch": 1.2926641774654075, + "grad_norm": 2.3332183361053467, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7202642560005188, + "num_tokens": 304546070.0, + "step": 11771 + }, + { + "epoch": 1.292773995168021, + "grad_norm": 2.500385284423828, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7165219187736511, + "num_tokens": 304567669.0, + "step": 11772 + }, + { + "epoch": 1.2928838128706348, + "grad_norm": 2.156160354614258, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7167950868606567, + "num_tokens": 304594245.0, + "step": 11773 + }, + { + "epoch": 1.2929936305732483, + "grad_norm": 2.184736728668213, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7285125255584717, + "num_tokens": 304622361.0, + "step": 11774 + }, + { + "epoch": 1.293103448275862, + "grad_norm": 2.664005756378174, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7604438662528992, + "num_tokens": 304641523.0, + "step": 11775 + }, + { + "epoch": 1.2932132659784759, + "grad_norm": 2.1830554008483887, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7292954325675964, + "num_tokens": 304666763.0, + "step": 11776 + }, + { + "epoch": 1.2933230836810894, + "grad_norm": 2.2479569911956787, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.712220311164856, + "num_tokens": 304691674.0, + "step": 11777 + }, + { + "epoch": 1.293432901383703, + "grad_norm": 2.4447569847106934, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7240394353866577, + "num_tokens": 304713565.0, + "step": 11778 + }, + { + "epoch": 1.2935427190863167, + "grad_norm": 2.2460739612579346, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7119742631912231, + "num_tokens": 304738147.0, + "step": 11779 + }, + { + "epoch": 1.2936525367889304, + "grad_norm": 2.329988956451416, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7045845985412598, + "num_tokens": 304763544.0, + "step": 11780 + }, + { + "epoch": 1.293762354491544, + "grad_norm": 2.132417678833008, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.699888288974762, + "num_tokens": 304793019.0, + "step": 11781 + }, + { + "epoch": 1.2938721721941577, + "grad_norm": 2.0825135707855225, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7082948088645935, + "num_tokens": 304821249.0, + "step": 11782 + }, + { + "epoch": 1.2939819898967713, + "grad_norm": 2.112591505050659, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6995255351066589, + "num_tokens": 304849658.0, + "step": 11783 + }, + { + "epoch": 1.294091807599385, + "grad_norm": 2.373013496398926, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7215972542762756, + "num_tokens": 304872325.0, + "step": 11784 + }, + { + "epoch": 1.2942016253019988, + "grad_norm": 2.209216356277466, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7460534572601318, + "num_tokens": 304896426.0, + "step": 11785 + }, + { + "epoch": 1.2943114430046123, + "grad_norm": 2.3823471069335938, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7025634050369263, + "num_tokens": 304920851.0, + "step": 11786 + }, + { + "epoch": 1.294421260707226, + "grad_norm": 2.345492124557495, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7664496898651123, + "num_tokens": 304943334.0, + "step": 11787 + }, + { + "epoch": 1.2945310784098396, + "grad_norm": 2.2851595878601074, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.6991482973098755, + "num_tokens": 304969500.0, + "step": 11788 + }, + { + "epoch": 1.2946408961124534, + "grad_norm": 2.3990588188171387, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7235544919967651, + "num_tokens": 304991110.0, + "step": 11789 + }, + { + "epoch": 1.2947507138150671, + "grad_norm": 2.439141035079956, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7187424898147583, + "num_tokens": 305015955.0, + "step": 11790 + }, + { + "epoch": 1.2948605315176807, + "grad_norm": 2.189906120300293, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7158435583114624, + "num_tokens": 305042520.0, + "step": 11791 + }, + { + "epoch": 1.2949703492202942, + "grad_norm": 2.224741220474243, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7076709270477295, + "num_tokens": 305071471.0, + "step": 11792 + }, + { + "epoch": 1.295080166922908, + "grad_norm": 2.2836320400238037, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7500221729278564, + "num_tokens": 305095514.0, + "step": 11793 + }, + { + "epoch": 1.2951899846255217, + "grad_norm": 2.3960039615631104, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.706610918045044, + "num_tokens": 305117920.0, + "step": 11794 + }, + { + "epoch": 1.2952998023281352, + "grad_norm": 2.4405128955841064, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7451130151748657, + "num_tokens": 305139672.0, + "step": 11795 + }, + { + "epoch": 1.295409620030749, + "grad_norm": 2.264394998550415, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7297422289848328, + "num_tokens": 305164213.0, + "step": 11796 + }, + { + "epoch": 1.2955194377333625, + "grad_norm": 2.346043109893799, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7127865552902222, + "num_tokens": 305188419.0, + "step": 11797 + }, + { + "epoch": 1.2956292554359763, + "grad_norm": 2.1184935569763184, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7082072496414185, + "num_tokens": 305217696.0, + "step": 11798 + }, + { + "epoch": 1.29573907313859, + "grad_norm": 2.2094063758850098, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7207235097885132, + "num_tokens": 305244888.0, + "step": 11799 + }, + { + "epoch": 1.2958488908412036, + "grad_norm": 2.1890463829040527, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7498654127120972, + "num_tokens": 305271407.0, + "step": 11800 + }, + { + "epoch": 1.2959587085438173, + "grad_norm": 2.44978928565979, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7210198640823364, + "num_tokens": 305293901.0, + "step": 11801 + }, + { + "epoch": 1.2960685262464309, + "grad_norm": 2.251046657562256, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.716191291809082, + "num_tokens": 305319982.0, + "step": 11802 + }, + { + "epoch": 1.2961783439490446, + "grad_norm": 2.407975912094116, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7220169305801392, + "num_tokens": 305341512.0, + "step": 11803 + }, + { + "epoch": 1.2962881616516582, + "grad_norm": 2.203690767288208, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.704765260219574, + "num_tokens": 305369414.0, + "step": 11804 + }, + { + "epoch": 1.296397979354272, + "grad_norm": 2.3935468196868896, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7373051643371582, + "num_tokens": 305392887.0, + "step": 11805 + }, + { + "epoch": 1.2965077970568855, + "grad_norm": 2.020631790161133, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6931136846542358, + "num_tokens": 305423865.0, + "step": 11806 + }, + { + "epoch": 1.2966176147594992, + "grad_norm": 2.024033546447754, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7044464945793152, + "num_tokens": 305454469.0, + "step": 11807 + }, + { + "epoch": 1.296727432462113, + "grad_norm": 2.06088924407959, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7191433906555176, + "num_tokens": 305483821.0, + "step": 11808 + }, + { + "epoch": 1.2968372501647265, + "grad_norm": 2.410259962081909, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7349230051040649, + "num_tokens": 305506685.0, + "step": 11809 + }, + { + "epoch": 1.2969470678673403, + "grad_norm": 2.3309898376464844, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7352807521820068, + "num_tokens": 305530489.0, + "step": 11810 + }, + { + "epoch": 1.2970568855699538, + "grad_norm": 2.362048625946045, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7323713898658752, + "num_tokens": 305552683.0, + "step": 11811 + }, + { + "epoch": 1.2971667032725676, + "grad_norm": 2.3137753009796143, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7656896114349365, + "num_tokens": 305575005.0, + "step": 11812 + }, + { + "epoch": 1.2972765209751813, + "grad_norm": 2.289541244506836, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7343802452087402, + "num_tokens": 305601169.0, + "step": 11813 + }, + { + "epoch": 1.2973863386777948, + "grad_norm": 2.2526562213897705, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7319303154945374, + "num_tokens": 305628213.0, + "step": 11814 + }, + { + "epoch": 1.2974961563804086, + "grad_norm": 2.185175657272339, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.712509036064148, + "num_tokens": 305655998.0, + "step": 11815 + }, + { + "epoch": 1.2976059740830221, + "grad_norm": 2.1859471797943115, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7068606615066528, + "num_tokens": 305682178.0, + "step": 11816 + }, + { + "epoch": 1.297715791785636, + "grad_norm": 2.4065067768096924, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7092581391334534, + "num_tokens": 305706363.0, + "step": 11817 + }, + { + "epoch": 1.2978256094882494, + "grad_norm": 1.9105199575424194, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7257548570632935, + "num_tokens": 305736478.0, + "step": 11818 + }, + { + "epoch": 1.2979354271908632, + "grad_norm": 2.1704063415527344, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7121340036392212, + "num_tokens": 305765312.0, + "step": 11819 + }, + { + "epoch": 1.2980452448934767, + "grad_norm": 2.3364601135253906, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7381073236465454, + "num_tokens": 305788731.0, + "step": 11820 + }, + { + "epoch": 1.2981550625960905, + "grad_norm": 2.2537763118743896, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6946238279342651, + "num_tokens": 305814487.0, + "step": 11821 + }, + { + "epoch": 1.2982648802987042, + "grad_norm": 2.2348902225494385, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7393502593040466, + "num_tokens": 305840110.0, + "step": 11822 + }, + { + "epoch": 1.2983746980013178, + "grad_norm": 2.0296990871429443, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7239616513252258, + "num_tokens": 305870635.0, + "step": 11823 + }, + { + "epoch": 1.2984845157039315, + "grad_norm": 2.127613067626953, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7058017253875732, + "num_tokens": 305900461.0, + "step": 11824 + }, + { + "epoch": 1.298594333406545, + "grad_norm": 2.479013442993164, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7221217751502991, + "num_tokens": 305920842.0, + "step": 11825 + }, + { + "epoch": 1.2987041511091588, + "grad_norm": 2.2582006454467773, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7034489512443542, + "num_tokens": 305947118.0, + "step": 11826 + }, + { + "epoch": 1.2988139688117726, + "grad_norm": 2.0851829051971436, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7241455316543579, + "num_tokens": 305973592.0, + "step": 11827 + }, + { + "epoch": 1.2989237865143861, + "grad_norm": 2.2584142684936523, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.71946120262146, + "num_tokens": 306000339.0, + "step": 11828 + }, + { + "epoch": 1.2990336042169996, + "grad_norm": 3.0362532138824463, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7135099172592163, + "num_tokens": 306018889.0, + "step": 11829 + }, + { + "epoch": 1.2991434219196134, + "grad_norm": 2.473792791366577, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7233981490135193, + "num_tokens": 306041685.0, + "step": 11830 + }, + { + "epoch": 1.2992532396222272, + "grad_norm": 2.1282591819763184, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7103269100189209, + "num_tokens": 306072866.0, + "step": 11831 + }, + { + "epoch": 1.2993630573248407, + "grad_norm": 2.197035551071167, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7172115445137024, + "num_tokens": 306100711.0, + "step": 11832 + }, + { + "epoch": 1.2994728750274545, + "grad_norm": 2.5587403774261475, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7012975215911865, + "num_tokens": 306122537.0, + "step": 11833 + }, + { + "epoch": 1.299582692730068, + "grad_norm": 2.3897719383239746, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7247304916381836, + "num_tokens": 306145470.0, + "step": 11834 + }, + { + "epoch": 1.2996925104326817, + "grad_norm": 2.272592782974243, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.710904598236084, + "num_tokens": 306171805.0, + "step": 11835 + }, + { + "epoch": 1.2998023281352955, + "grad_norm": 2.160231351852417, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7332116961479187, + "num_tokens": 306196865.0, + "step": 11836 + }, + { + "epoch": 1.299912145837909, + "grad_norm": 2.083892345428467, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7207807898521423, + "num_tokens": 306224739.0, + "step": 11837 + }, + { + "epoch": 1.3000219635405228, + "grad_norm": 2.353522300720215, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7034687995910645, + "num_tokens": 306248787.0, + "step": 11838 + }, + { + "epoch": 1.3001317812431363, + "grad_norm": 2.1879544258117676, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7114997506141663, + "num_tokens": 306274373.0, + "step": 11839 + }, + { + "epoch": 1.30024159894575, + "grad_norm": 2.1769416332244873, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7182867527008057, + "num_tokens": 306299504.0, + "step": 11840 + }, + { + "epoch": 1.3003514166483638, + "grad_norm": 2.027920722961426, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7075443267822266, + "num_tokens": 306328299.0, + "step": 11841 + }, + { + "epoch": 1.3004612343509774, + "grad_norm": 2.449864625930786, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7361563444137573, + "num_tokens": 306351122.0, + "step": 11842 + }, + { + "epoch": 1.300571052053591, + "grad_norm": 2.4578306674957275, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7340576648712158, + "num_tokens": 306371039.0, + "step": 11843 + }, + { + "epoch": 1.3006808697562047, + "grad_norm": 2.14644718170166, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7158607244491577, + "num_tokens": 306398171.0, + "step": 11844 + }, + { + "epoch": 1.3007906874588184, + "grad_norm": 2.091979742050171, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7178930044174194, + "num_tokens": 306426726.0, + "step": 11845 + }, + { + "epoch": 1.300900505161432, + "grad_norm": 2.081132650375366, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.6968146562576294, + "num_tokens": 306459056.0, + "step": 11846 + }, + { + "epoch": 1.3010103228640457, + "grad_norm": 2.0469188690185547, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7039303183555603, + "num_tokens": 306488203.0, + "step": 11847 + }, + { + "epoch": 1.3011201405666593, + "grad_norm": 2.243894338607788, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7387106418609619, + "num_tokens": 306514586.0, + "step": 11848 + }, + { + "epoch": 1.301229958269273, + "grad_norm": 2.2810308933258057, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7340118885040283, + "num_tokens": 306541307.0, + "step": 11849 + }, + { + "epoch": 1.3013397759718868, + "grad_norm": 2.188948392868042, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.727733850479126, + "num_tokens": 306568190.0, + "step": 11850 + }, + { + "epoch": 1.3014495936745003, + "grad_norm": 2.038745403289795, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.722108781337738, + "num_tokens": 306598005.0, + "step": 11851 + }, + { + "epoch": 1.301559411377114, + "grad_norm": 2.1338107585906982, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7248656749725342, + "num_tokens": 306626348.0, + "step": 11852 + }, + { + "epoch": 1.3016692290797276, + "grad_norm": 2.5515851974487305, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7164517641067505, + "num_tokens": 306648604.0, + "step": 11853 + }, + { + "epoch": 1.3017790467823414, + "grad_norm": 2.1318233013153076, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7076511979103088, + "num_tokens": 306677033.0, + "step": 11854 + }, + { + "epoch": 1.301888864484955, + "grad_norm": 2.104989528656006, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7304982542991638, + "num_tokens": 306704591.0, + "step": 11855 + }, + { + "epoch": 1.3019986821875686, + "grad_norm": 2.5548839569091797, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7328964471817017, + "num_tokens": 306722762.0, + "step": 11856 + }, + { + "epoch": 1.3021084998901822, + "grad_norm": 1.9660720825195312, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7045137882232666, + "num_tokens": 306755451.0, + "step": 11857 + }, + { + "epoch": 1.302218317592796, + "grad_norm": 2.391648530960083, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7466892004013062, + "num_tokens": 306777213.0, + "step": 11858 + }, + { + "epoch": 1.3023281352954097, + "grad_norm": 1.9333713054656982, + "learning_rate": 1e-06, + "loss": 1.0302, + "mean_token_accuracy": 0.6902449727058411, + "num_tokens": 306811483.0, + "step": 11859 + }, + { + "epoch": 1.3024379529980232, + "grad_norm": 2.271409511566162, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7175321578979492, + "num_tokens": 306835767.0, + "step": 11860 + }, + { + "epoch": 1.302547770700637, + "grad_norm": 2.3408896923065186, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.733971357345581, + "num_tokens": 306858057.0, + "step": 11861 + }, + { + "epoch": 1.3026575884032505, + "grad_norm": 2.4381020069122314, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7381712198257446, + "num_tokens": 306881002.0, + "step": 11862 + }, + { + "epoch": 1.3027674061058643, + "grad_norm": 2.2320680618286133, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7138388752937317, + "num_tokens": 306907122.0, + "step": 11863 + }, + { + "epoch": 1.302877223808478, + "grad_norm": 2.2894105911254883, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7229349613189697, + "num_tokens": 306931095.0, + "step": 11864 + }, + { + "epoch": 1.3029870415110916, + "grad_norm": 2.383585214614868, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7280242443084717, + "num_tokens": 306953333.0, + "step": 11865 + }, + { + "epoch": 1.3030968592137053, + "grad_norm": 2.3436172008514404, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.713072657585144, + "num_tokens": 306979310.0, + "step": 11866 + }, + { + "epoch": 1.3032066769163189, + "grad_norm": 2.267180919647217, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7256193161010742, + "num_tokens": 307005402.0, + "step": 11867 + }, + { + "epoch": 1.3033164946189326, + "grad_norm": 2.2749037742614746, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7172155976295471, + "num_tokens": 307031279.0, + "step": 11868 + }, + { + "epoch": 1.3034263123215462, + "grad_norm": 2.4452426433563232, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.725272536277771, + "num_tokens": 307054553.0, + "step": 11869 + }, + { + "epoch": 1.30353613002416, + "grad_norm": 2.1751532554626465, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7089281678199768, + "num_tokens": 307081607.0, + "step": 11870 + }, + { + "epoch": 1.3036459477267734, + "grad_norm": 2.1434590816497803, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7165602445602417, + "num_tokens": 307108952.0, + "step": 11871 + }, + { + "epoch": 1.3037557654293872, + "grad_norm": 2.5915637016296387, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7112905979156494, + "num_tokens": 307129054.0, + "step": 11872 + }, + { + "epoch": 1.303865583132001, + "grad_norm": 2.063812017440796, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7449811100959778, + "num_tokens": 307155838.0, + "step": 11873 + }, + { + "epoch": 1.3039754008346145, + "grad_norm": 2.1493709087371826, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7032144069671631, + "num_tokens": 307182441.0, + "step": 11874 + }, + { + "epoch": 1.3040852185372283, + "grad_norm": 2.5759661197662354, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7272840142250061, + "num_tokens": 307204790.0, + "step": 11875 + }, + { + "epoch": 1.3041950362398418, + "grad_norm": 2.2431867122650146, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7103269100189209, + "num_tokens": 307230438.0, + "step": 11876 + }, + { + "epoch": 1.3043048539424555, + "grad_norm": 2.072591543197632, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7087430953979492, + "num_tokens": 307262556.0, + "step": 11877 + }, + { + "epoch": 1.3044146716450693, + "grad_norm": 2.5952212810516357, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.702984094619751, + "num_tokens": 307283816.0, + "step": 11878 + }, + { + "epoch": 1.3045244893476828, + "grad_norm": 2.6459484100341797, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.710762619972229, + "num_tokens": 307304357.0, + "step": 11879 + }, + { + "epoch": 1.3046343070502964, + "grad_norm": 2.2517778873443604, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7119265794754028, + "num_tokens": 307329642.0, + "step": 11880 + }, + { + "epoch": 1.3047441247529101, + "grad_norm": 2.0213029384613037, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7251979112625122, + "num_tokens": 307357631.0, + "step": 11881 + }, + { + "epoch": 1.3048539424555239, + "grad_norm": 2.147479295730591, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7170365452766418, + "num_tokens": 307384204.0, + "step": 11882 + }, + { + "epoch": 1.3049637601581374, + "grad_norm": 2.121675491333008, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7014979124069214, + "num_tokens": 307411603.0, + "step": 11883 + }, + { + "epoch": 1.3050735778607512, + "grad_norm": 2.093015670776367, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.686171293258667, + "num_tokens": 307440959.0, + "step": 11884 + }, + { + "epoch": 1.3051833955633647, + "grad_norm": 2.031773567199707, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.6974176168441772, + "num_tokens": 307472194.0, + "step": 11885 + }, + { + "epoch": 1.3052932132659785, + "grad_norm": 2.0398547649383545, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7200515270233154, + "num_tokens": 307502651.0, + "step": 11886 + }, + { + "epoch": 1.3054030309685922, + "grad_norm": 2.4371204376220703, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7293912768363953, + "num_tokens": 307525708.0, + "step": 11887 + }, + { + "epoch": 1.3055128486712058, + "grad_norm": 2.311115026473999, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7051007151603699, + "num_tokens": 307551774.0, + "step": 11888 + }, + { + "epoch": 1.3056226663738195, + "grad_norm": 2.141757011413574, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7283728122711182, + "num_tokens": 307577389.0, + "step": 11889 + }, + { + "epoch": 1.305732484076433, + "grad_norm": 2.1079487800598145, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7133837938308716, + "num_tokens": 307605984.0, + "step": 11890 + }, + { + "epoch": 1.3058423017790468, + "grad_norm": 2.5262253284454346, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7230520844459534, + "num_tokens": 307627950.0, + "step": 11891 + }, + { + "epoch": 1.3059521194816606, + "grad_norm": 2.23402738571167, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7159068584442139, + "num_tokens": 307653276.0, + "step": 11892 + }, + { + "epoch": 1.306061937184274, + "grad_norm": 2.1156599521636963, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7058950066566467, + "num_tokens": 307681643.0, + "step": 11893 + }, + { + "epoch": 1.3061717548868876, + "grad_norm": 2.1373708248138428, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7323976755142212, + "num_tokens": 307707993.0, + "step": 11894 + }, + { + "epoch": 1.3062815725895014, + "grad_norm": 2.443305492401123, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7117675542831421, + "num_tokens": 307730284.0, + "step": 11895 + }, + { + "epoch": 1.3063913902921152, + "grad_norm": 2.58953595161438, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7384085655212402, + "num_tokens": 307750876.0, + "step": 11896 + }, + { + "epoch": 1.3065012079947287, + "grad_norm": 2.4594573974609375, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7356042265892029, + "num_tokens": 307771352.0, + "step": 11897 + }, + { + "epoch": 1.3066110256973424, + "grad_norm": 2.2314937114715576, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.6970332860946655, + "num_tokens": 307799490.0, + "step": 11898 + }, + { + "epoch": 1.306720843399956, + "grad_norm": 2.4934473037719727, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7421442270278931, + "num_tokens": 307822092.0, + "step": 11899 + }, + { + "epoch": 1.3068306611025697, + "grad_norm": 2.2359066009521484, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7203463315963745, + "num_tokens": 307847578.0, + "step": 11900 + }, + { + "epoch": 1.3069404788051835, + "grad_norm": 2.0239250659942627, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7093231678009033, + "num_tokens": 307879184.0, + "step": 11901 + }, + { + "epoch": 1.307050296507797, + "grad_norm": 2.3465726375579834, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7052913904190063, + "num_tokens": 307902946.0, + "step": 11902 + }, + { + "epoch": 1.3071601142104108, + "grad_norm": 2.412914514541626, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7015880346298218, + "num_tokens": 307925745.0, + "step": 11903 + }, + { + "epoch": 1.3072699319130243, + "grad_norm": 2.370554208755493, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7048553824424744, + "num_tokens": 307950738.0, + "step": 11904 + }, + { + "epoch": 1.307379749615638, + "grad_norm": 2.055968761444092, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7091385722160339, + "num_tokens": 307980167.0, + "step": 11905 + }, + { + "epoch": 1.3074895673182518, + "grad_norm": 2.559324026107788, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7205770611763, + "num_tokens": 308003106.0, + "step": 11906 + }, + { + "epoch": 1.3075993850208654, + "grad_norm": 2.143436908721924, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7258988618850708, + "num_tokens": 308029924.0, + "step": 11907 + }, + { + "epoch": 1.307709202723479, + "grad_norm": 2.33341121673584, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7044617533683777, + "num_tokens": 308053551.0, + "step": 11908 + }, + { + "epoch": 1.3078190204260927, + "grad_norm": 2.435324192047119, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7251704335212708, + "num_tokens": 308075839.0, + "step": 11909 + }, + { + "epoch": 1.3079288381287064, + "grad_norm": 2.367774724960327, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7430615425109863, + "num_tokens": 308098003.0, + "step": 11910 + }, + { + "epoch": 1.30803865583132, + "grad_norm": 1.9584269523620605, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7062729001045227, + "num_tokens": 308131484.0, + "step": 11911 + }, + { + "epoch": 1.3081484735339337, + "grad_norm": 2.4517698287963867, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7232470512390137, + "num_tokens": 308152707.0, + "step": 11912 + }, + { + "epoch": 1.3082582912365472, + "grad_norm": 2.166311502456665, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7163382768630981, + "num_tokens": 308183132.0, + "step": 11913 + }, + { + "epoch": 1.308368108939161, + "grad_norm": 2.682133197784424, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7480568885803223, + "num_tokens": 308201717.0, + "step": 11914 + }, + { + "epoch": 1.3084779266417748, + "grad_norm": 2.581463098526001, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7234866619110107, + "num_tokens": 308221777.0, + "step": 11915 + }, + { + "epoch": 1.3085877443443883, + "grad_norm": 2.0538177490234375, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7123429775238037, + "num_tokens": 308252221.0, + "step": 11916 + }, + { + "epoch": 1.308697562047002, + "grad_norm": 2.3821117877960205, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7288386821746826, + "num_tokens": 308274420.0, + "step": 11917 + }, + { + "epoch": 1.3088073797496156, + "grad_norm": 1.920046091079712, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6916427612304688, + "num_tokens": 308309469.0, + "step": 11918 + }, + { + "epoch": 1.3089171974522293, + "grad_norm": 2.1559109687805176, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.737135112285614, + "num_tokens": 308334493.0, + "step": 11919 + }, + { + "epoch": 1.309027015154843, + "grad_norm": 2.2916572093963623, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7128309011459351, + "num_tokens": 308362169.0, + "step": 11920 + }, + { + "epoch": 1.3091368328574566, + "grad_norm": 2.435513973236084, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7333475947380066, + "num_tokens": 308383792.0, + "step": 11921 + }, + { + "epoch": 1.3092466505600702, + "grad_norm": 2.2647488117218018, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7208157181739807, + "num_tokens": 308408941.0, + "step": 11922 + }, + { + "epoch": 1.309356468262684, + "grad_norm": 2.222109317779541, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7025835514068604, + "num_tokens": 308437407.0, + "step": 11923 + }, + { + "epoch": 1.3094662859652977, + "grad_norm": 2.1154539585113525, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7091282606124878, + "num_tokens": 308464422.0, + "step": 11924 + }, + { + "epoch": 1.3095761036679112, + "grad_norm": 2.1342122554779053, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.696617841720581, + "num_tokens": 308493901.0, + "step": 11925 + }, + { + "epoch": 1.309685921370525, + "grad_norm": 1.9521151781082153, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7222543358802795, + "num_tokens": 308527056.0, + "step": 11926 + }, + { + "epoch": 1.3097957390731385, + "grad_norm": 2.0026211738586426, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7147751450538635, + "num_tokens": 308559454.0, + "step": 11927 + }, + { + "epoch": 1.3099055567757523, + "grad_norm": 2.4312727451324463, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7241520881652832, + "num_tokens": 308582588.0, + "step": 11928 + }, + { + "epoch": 1.310015374478366, + "grad_norm": 2.364130973815918, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7136830687522888, + "num_tokens": 308607198.0, + "step": 11929 + }, + { + "epoch": 1.3101251921809796, + "grad_norm": 2.269887685775757, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7137793302536011, + "num_tokens": 308633119.0, + "step": 11930 + }, + { + "epoch": 1.3102350098835933, + "grad_norm": 2.30826735496521, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7143967151641846, + "num_tokens": 308658039.0, + "step": 11931 + }, + { + "epoch": 1.3103448275862069, + "grad_norm": 2.4357104301452637, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7409155964851379, + "num_tokens": 308678716.0, + "step": 11932 + }, + { + "epoch": 1.3104546452888206, + "grad_norm": 2.06907057762146, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7111195921897888, + "num_tokens": 308708883.0, + "step": 11933 + }, + { + "epoch": 1.3105644629914341, + "grad_norm": 2.2135074138641357, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7270092964172363, + "num_tokens": 308733671.0, + "step": 11934 + }, + { + "epoch": 1.310674280694048, + "grad_norm": 2.261897563934326, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7121328115463257, + "num_tokens": 308760280.0, + "step": 11935 + }, + { + "epoch": 1.3107840983966614, + "grad_norm": 2.511263608932495, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7357156276702881, + "num_tokens": 308780609.0, + "step": 11936 + }, + { + "epoch": 1.3108939160992752, + "grad_norm": 2.37274169921875, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7024049162864685, + "num_tokens": 308804795.0, + "step": 11937 + }, + { + "epoch": 1.311003733801889, + "grad_norm": 2.5493805408477783, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7272182106971741, + "num_tokens": 308827372.0, + "step": 11938 + }, + { + "epoch": 1.3111135515045025, + "grad_norm": 2.216688871383667, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7290477752685547, + "num_tokens": 308851157.0, + "step": 11939 + }, + { + "epoch": 1.3112233692071162, + "grad_norm": 2.3287477493286133, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7162687182426453, + "num_tokens": 308876515.0, + "step": 11940 + }, + { + "epoch": 1.3113331869097298, + "grad_norm": 2.4538002014160156, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7032202482223511, + "num_tokens": 308898561.0, + "step": 11941 + }, + { + "epoch": 1.3114430046123435, + "grad_norm": 2.0254218578338623, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7157868146896362, + "num_tokens": 308928792.0, + "step": 11942 + }, + { + "epoch": 1.3115528223149573, + "grad_norm": 2.111168146133423, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7361499071121216, + "num_tokens": 308955868.0, + "step": 11943 + }, + { + "epoch": 1.3116626400175708, + "grad_norm": 2.305501699447632, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7184067964553833, + "num_tokens": 308979800.0, + "step": 11944 + }, + { + "epoch": 1.3117724577201844, + "grad_norm": 2.1162354946136475, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7111533880233765, + "num_tokens": 309009514.0, + "step": 11945 + }, + { + "epoch": 1.3118822754227981, + "grad_norm": 2.581609010696411, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7208723425865173, + "num_tokens": 309030237.0, + "step": 11946 + }, + { + "epoch": 1.3119920931254119, + "grad_norm": 2.200688362121582, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7281303405761719, + "num_tokens": 309058352.0, + "step": 11947 + }, + { + "epoch": 1.3121019108280254, + "grad_norm": 2.5272700786590576, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7202427387237549, + "num_tokens": 309078411.0, + "step": 11948 + }, + { + "epoch": 1.3122117285306392, + "grad_norm": 2.413065195083618, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7396655678749084, + "num_tokens": 309099362.0, + "step": 11949 + }, + { + "epoch": 1.3123215462332527, + "grad_norm": 2.5122263431549072, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7393133640289307, + "num_tokens": 309120392.0, + "step": 11950 + }, + { + "epoch": 1.3124313639358665, + "grad_norm": 2.0743520259857178, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7258789539337158, + "num_tokens": 309149883.0, + "step": 11951 + }, + { + "epoch": 1.3125411816384802, + "grad_norm": 2.223055601119995, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7121275663375854, + "num_tokens": 309176404.0, + "step": 11952 + }, + { + "epoch": 1.3126509993410937, + "grad_norm": 1.9893914461135864, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7030397057533264, + "num_tokens": 309204974.0, + "step": 11953 + }, + { + "epoch": 1.3127608170437075, + "grad_norm": 2.064124345779419, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7211316227912903, + "num_tokens": 309233538.0, + "step": 11954 + }, + { + "epoch": 1.312870634746321, + "grad_norm": 1.9688477516174316, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7083992958068848, + "num_tokens": 309265874.0, + "step": 11955 + }, + { + "epoch": 1.3129804524489348, + "grad_norm": 2.328268051147461, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6949959993362427, + "num_tokens": 309292134.0, + "step": 11956 + }, + { + "epoch": 1.3130902701515486, + "grad_norm": 2.4641590118408203, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7403162121772766, + "num_tokens": 309314202.0, + "step": 11957 + }, + { + "epoch": 1.313200087854162, + "grad_norm": 1.929742455482483, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7214992642402649, + "num_tokens": 309345545.0, + "step": 11958 + }, + { + "epoch": 1.3133099055567756, + "grad_norm": 2.428800582885742, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7057709693908691, + "num_tokens": 309368933.0, + "step": 11959 + }, + { + "epoch": 1.3134197232593894, + "grad_norm": 2.1104836463928223, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7158631086349487, + "num_tokens": 309395945.0, + "step": 11960 + }, + { + "epoch": 1.3135295409620031, + "grad_norm": 2.2854812145233154, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7045348882675171, + "num_tokens": 309421344.0, + "step": 11961 + }, + { + "epoch": 1.3136393586646167, + "grad_norm": 2.457563638687134, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7138756513595581, + "num_tokens": 309444483.0, + "step": 11962 + }, + { + "epoch": 1.3137491763672304, + "grad_norm": 2.3633460998535156, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7229601740837097, + "num_tokens": 309468167.0, + "step": 11963 + }, + { + "epoch": 1.313858994069844, + "grad_norm": 1.921743631362915, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7070543766021729, + "num_tokens": 309503149.0, + "step": 11964 + }, + { + "epoch": 1.3139688117724577, + "grad_norm": 2.2329630851745605, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7354240417480469, + "num_tokens": 309528160.0, + "step": 11965 + }, + { + "epoch": 1.3140786294750715, + "grad_norm": 2.0854406356811523, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.726142942905426, + "num_tokens": 309556572.0, + "step": 11966 + }, + { + "epoch": 1.314188447177685, + "grad_norm": 2.233908176422119, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7248469591140747, + "num_tokens": 309582611.0, + "step": 11967 + }, + { + "epoch": 1.3142982648802988, + "grad_norm": 2.2908947467803955, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7403013110160828, + "num_tokens": 309609761.0, + "step": 11968 + }, + { + "epoch": 1.3144080825829123, + "grad_norm": 2.1510982513427734, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.716711699962616, + "num_tokens": 309635992.0, + "step": 11969 + }, + { + "epoch": 1.314517900285526, + "grad_norm": 2.4896557331085205, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7166922688484192, + "num_tokens": 309659002.0, + "step": 11970 + }, + { + "epoch": 1.3146277179881398, + "grad_norm": 2.130035877227783, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7086650133132935, + "num_tokens": 309689015.0, + "step": 11971 + }, + { + "epoch": 1.3147375356907534, + "grad_norm": 2.26326584815979, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6961679458618164, + "num_tokens": 309715038.0, + "step": 11972 + }, + { + "epoch": 1.314847353393367, + "grad_norm": 2.4085404872894287, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7123569250106812, + "num_tokens": 309739071.0, + "step": 11973 + }, + { + "epoch": 1.3149571710959806, + "grad_norm": 2.193525791168213, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6937034130096436, + "num_tokens": 309769186.0, + "step": 11974 + }, + { + "epoch": 1.3150669887985944, + "grad_norm": 2.092154026031494, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7200034260749817, + "num_tokens": 309800157.0, + "step": 11975 + }, + { + "epoch": 1.315176806501208, + "grad_norm": 2.209455728530884, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6966390013694763, + "num_tokens": 309827522.0, + "step": 11976 + }, + { + "epoch": 1.3152866242038217, + "grad_norm": 2.126274824142456, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6939359307289124, + "num_tokens": 309855405.0, + "step": 11977 + }, + { + "epoch": 1.3153964419064352, + "grad_norm": 2.0168447494506836, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.715786337852478, + "num_tokens": 309885220.0, + "step": 11978 + }, + { + "epoch": 1.315506259609049, + "grad_norm": 2.2214081287384033, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7288481593132019, + "num_tokens": 309912524.0, + "step": 11979 + }, + { + "epoch": 1.3156160773116627, + "grad_norm": 2.01165509223938, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6894067525863647, + "num_tokens": 309944880.0, + "step": 11980 + }, + { + "epoch": 1.3157258950142763, + "grad_norm": 2.3396944999694824, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7061301469802856, + "num_tokens": 309971327.0, + "step": 11981 + }, + { + "epoch": 1.31583571271689, + "grad_norm": 2.4604876041412354, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7040098905563354, + "num_tokens": 309996491.0, + "step": 11982 + }, + { + "epoch": 1.3159455304195036, + "grad_norm": 2.6316065788269043, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7385060787200928, + "num_tokens": 310016335.0, + "step": 11983 + }, + { + "epoch": 1.3160553481221173, + "grad_norm": 1.8999035358428955, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.70367032289505, + "num_tokens": 310051258.0, + "step": 11984 + }, + { + "epoch": 1.3161651658247309, + "grad_norm": 1.9574792385101318, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7121302485466003, + "num_tokens": 310084879.0, + "step": 11985 + }, + { + "epoch": 1.3162749835273446, + "grad_norm": 2.184323310852051, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7404662370681763, + "num_tokens": 310110181.0, + "step": 11986 + }, + { + "epoch": 1.3163848012299582, + "grad_norm": 2.2470803260803223, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7516031861305237, + "num_tokens": 310132279.0, + "step": 11987 + }, + { + "epoch": 1.316494618932572, + "grad_norm": 2.313023090362549, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7339924573898315, + "num_tokens": 310154905.0, + "step": 11988 + }, + { + "epoch": 1.3166044366351857, + "grad_norm": 2.2234115600585938, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.720562219619751, + "num_tokens": 310182554.0, + "step": 11989 + }, + { + "epoch": 1.3167142543377992, + "grad_norm": 2.0189361572265625, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7200992703437805, + "num_tokens": 310216138.0, + "step": 11990 + }, + { + "epoch": 1.316824072040413, + "grad_norm": 2.3162851333618164, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7202705144882202, + "num_tokens": 310243173.0, + "step": 11991 + }, + { + "epoch": 1.3169338897430265, + "grad_norm": 2.3770737648010254, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7200753688812256, + "num_tokens": 310266769.0, + "step": 11992 + }, + { + "epoch": 1.3170437074456403, + "grad_norm": 2.039857864379883, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6955589056015015, + "num_tokens": 310299363.0, + "step": 11993 + }, + { + "epoch": 1.317153525148254, + "grad_norm": 2.0747733116149902, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6949399709701538, + "num_tokens": 310327603.0, + "step": 11994 + }, + { + "epoch": 1.3172633428508675, + "grad_norm": 2.302844285964966, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7221630811691284, + "num_tokens": 310352373.0, + "step": 11995 + }, + { + "epoch": 1.3173731605534813, + "grad_norm": 2.1928157806396484, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7309232354164124, + "num_tokens": 310378023.0, + "step": 11996 + }, + { + "epoch": 1.3174829782560948, + "grad_norm": 1.9574631452560425, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7019544243812561, + "num_tokens": 310408772.0, + "step": 11997 + }, + { + "epoch": 1.3175927959587086, + "grad_norm": 2.167410373687744, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7131337523460388, + "num_tokens": 310437185.0, + "step": 11998 + }, + { + "epoch": 1.3177026136613221, + "grad_norm": 1.8600237369537354, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7221029996871948, + "num_tokens": 310472318.0, + "step": 11999 + }, + { + "epoch": 1.3178124313639359, + "grad_norm": 2.10033917427063, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7351632714271545, + "num_tokens": 310501902.0, + "step": 12000 + }, + { + "epoch": 1.3179222490665494, + "grad_norm": 2.4285714626312256, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7520551681518555, + "num_tokens": 310523070.0, + "step": 12001 + }, + { + "epoch": 1.3180320667691632, + "grad_norm": 2.1853506565093994, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7186720371246338, + "num_tokens": 310548975.0, + "step": 12002 + }, + { + "epoch": 1.318141884471777, + "grad_norm": 1.9511512517929077, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.753838837146759, + "num_tokens": 310578440.0, + "step": 12003 + }, + { + "epoch": 1.3182517021743905, + "grad_norm": 2.1372153759002686, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7022684812545776, + "num_tokens": 310608050.0, + "step": 12004 + }, + { + "epoch": 1.3183615198770042, + "grad_norm": 2.2161967754364014, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7179032564163208, + "num_tokens": 310634990.0, + "step": 12005 + }, + { + "epoch": 1.3184713375796178, + "grad_norm": 2.185202121734619, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.718698263168335, + "num_tokens": 310663671.0, + "step": 12006 + }, + { + "epoch": 1.3185811552822315, + "grad_norm": 2.379713296890259, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7435307502746582, + "num_tokens": 310686032.0, + "step": 12007 + }, + { + "epoch": 1.3186909729848453, + "grad_norm": 2.0904085636138916, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7134203910827637, + "num_tokens": 310715174.0, + "step": 12008 + }, + { + "epoch": 1.3188007906874588, + "grad_norm": 2.2836742401123047, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.697624921798706, + "num_tokens": 310741269.0, + "step": 12009 + }, + { + "epoch": 1.3189106083900723, + "grad_norm": 2.58935809135437, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7358226776123047, + "num_tokens": 310761641.0, + "step": 12010 + }, + { + "epoch": 1.319020426092686, + "grad_norm": 2.2989211082458496, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7304293513298035, + "num_tokens": 310784991.0, + "step": 12011 + }, + { + "epoch": 1.3191302437952999, + "grad_norm": 2.52412486076355, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7004765272140503, + "num_tokens": 310807560.0, + "step": 12012 + }, + { + "epoch": 1.3192400614979134, + "grad_norm": 2.19063138961792, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7242162823677063, + "num_tokens": 310833251.0, + "step": 12013 + }, + { + "epoch": 1.3193498792005272, + "grad_norm": 1.9644414186477661, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7151780128479004, + "num_tokens": 310865625.0, + "step": 12014 + }, + { + "epoch": 1.3194596969031407, + "grad_norm": 2.362342596054077, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7210675477981567, + "num_tokens": 310889811.0, + "step": 12015 + }, + { + "epoch": 1.3195695146057544, + "grad_norm": 2.7004854679107666, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7236121296882629, + "num_tokens": 310908529.0, + "step": 12016 + }, + { + "epoch": 1.3196793323083682, + "grad_norm": 1.9990582466125488, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.682121217250824, + "num_tokens": 310940143.0, + "step": 12017 + }, + { + "epoch": 1.3197891500109817, + "grad_norm": 2.5173864364624023, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7233887314796448, + "num_tokens": 310960261.0, + "step": 12018 + }, + { + "epoch": 1.3198989677135955, + "grad_norm": 2.4687113761901855, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7163079977035522, + "num_tokens": 310983025.0, + "step": 12019 + }, + { + "epoch": 1.320008785416209, + "grad_norm": 2.3023345470428467, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7102088928222656, + "num_tokens": 311008978.0, + "step": 12020 + }, + { + "epoch": 1.3201186031188228, + "grad_norm": 2.1296911239624023, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7298740148544312, + "num_tokens": 311037576.0, + "step": 12021 + }, + { + "epoch": 1.3202284208214365, + "grad_norm": 2.376051187515259, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.6986221075057983, + "num_tokens": 311061893.0, + "step": 12022 + }, + { + "epoch": 1.32033823852405, + "grad_norm": 2.5004630088806152, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7380650043487549, + "num_tokens": 311084726.0, + "step": 12023 + }, + { + "epoch": 1.3204480562266636, + "grad_norm": 2.18105149269104, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7248207330703735, + "num_tokens": 311110099.0, + "step": 12024 + }, + { + "epoch": 1.3205578739292774, + "grad_norm": 2.583055257797241, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7374887466430664, + "num_tokens": 311130690.0, + "step": 12025 + }, + { + "epoch": 1.3206676916318911, + "grad_norm": 2.275073528289795, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.6952043175697327, + "num_tokens": 311157366.0, + "step": 12026 + }, + { + "epoch": 1.3207775093345047, + "grad_norm": 2.4741291999816895, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.711419403553009, + "num_tokens": 311179686.0, + "step": 12027 + }, + { + "epoch": 1.3208873270371184, + "grad_norm": 2.343258857727051, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.706684947013855, + "num_tokens": 311203656.0, + "step": 12028 + }, + { + "epoch": 1.320997144739732, + "grad_norm": 2.179158926010132, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7155937552452087, + "num_tokens": 311230369.0, + "step": 12029 + }, + { + "epoch": 1.3211069624423457, + "grad_norm": 1.9942243099212646, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7363802194595337, + "num_tokens": 311260256.0, + "step": 12030 + }, + { + "epoch": 1.3212167801449595, + "grad_norm": 2.484744071960449, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.702646017074585, + "num_tokens": 311282552.0, + "step": 12031 + }, + { + "epoch": 1.321326597847573, + "grad_norm": 2.3418314456939697, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7431135773658752, + "num_tokens": 311304267.0, + "step": 12032 + }, + { + "epoch": 1.3214364155501868, + "grad_norm": 2.2146074771881104, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7024499773979187, + "num_tokens": 311330001.0, + "step": 12033 + }, + { + "epoch": 1.3215462332528003, + "grad_norm": 2.532454490661621, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7194521427154541, + "num_tokens": 311355827.0, + "step": 12034 + }, + { + "epoch": 1.321656050955414, + "grad_norm": 2.3325564861297607, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7229169011116028, + "num_tokens": 311379536.0, + "step": 12035 + }, + { + "epoch": 1.3217658686580278, + "grad_norm": 2.2601842880249023, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7259379625320435, + "num_tokens": 311404203.0, + "step": 12036 + }, + { + "epoch": 1.3218756863606413, + "grad_norm": 2.23512864112854, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7275354862213135, + "num_tokens": 311429686.0, + "step": 12037 + }, + { + "epoch": 1.3219855040632549, + "grad_norm": 2.0663235187530518, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7109841108322144, + "num_tokens": 311459298.0, + "step": 12038 + }, + { + "epoch": 1.3220953217658686, + "grad_norm": 2.463902711868286, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7473584413528442, + "num_tokens": 311479998.0, + "step": 12039 + }, + { + "epoch": 1.3222051394684824, + "grad_norm": 2.143491744995117, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7013171911239624, + "num_tokens": 311508414.0, + "step": 12040 + }, + { + "epoch": 1.322314957171096, + "grad_norm": 2.334521532058716, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7339178323745728, + "num_tokens": 311531813.0, + "step": 12041 + }, + { + "epoch": 1.3224247748737097, + "grad_norm": 1.989354133605957, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7197359800338745, + "num_tokens": 311565499.0, + "step": 12042 + }, + { + "epoch": 1.3225345925763232, + "grad_norm": 2.6011643409729004, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7197809219360352, + "num_tokens": 311585634.0, + "step": 12043 + }, + { + "epoch": 1.322644410278937, + "grad_norm": 2.1171748638153076, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7192666530609131, + "num_tokens": 311612145.0, + "step": 12044 + }, + { + "epoch": 1.3227542279815507, + "grad_norm": 2.189462423324585, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7014429569244385, + "num_tokens": 311637952.0, + "step": 12045 + }, + { + "epoch": 1.3228640456841643, + "grad_norm": 1.9668254852294922, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7303789854049683, + "num_tokens": 311667932.0, + "step": 12046 + }, + { + "epoch": 1.322973863386778, + "grad_norm": 2.26798415184021, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.731769323348999, + "num_tokens": 311691471.0, + "step": 12047 + }, + { + "epoch": 1.3230836810893916, + "grad_norm": 2.620652437210083, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7278156280517578, + "num_tokens": 311710910.0, + "step": 12048 + }, + { + "epoch": 1.3231934987920053, + "grad_norm": 2.377500057220459, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7398512363433838, + "num_tokens": 311733474.0, + "step": 12049 + }, + { + "epoch": 1.3233033164946189, + "grad_norm": 2.586690664291382, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7154830098152161, + "num_tokens": 311754050.0, + "step": 12050 + }, + { + "epoch": 1.3234131341972326, + "grad_norm": 2.276480197906494, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7034631967544556, + "num_tokens": 311782576.0, + "step": 12051 + }, + { + "epoch": 1.3235229518998461, + "grad_norm": 2.110997438430786, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.730730414390564, + "num_tokens": 311810613.0, + "step": 12052 + }, + { + "epoch": 1.32363276960246, + "grad_norm": 1.9461255073547363, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6943948268890381, + "num_tokens": 311845004.0, + "step": 12053 + }, + { + "epoch": 1.3237425873050737, + "grad_norm": 2.472400188446045, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7434086799621582, + "num_tokens": 311865498.0, + "step": 12054 + }, + { + "epoch": 1.3238524050076872, + "grad_norm": 2.018550157546997, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7109593152999878, + "num_tokens": 311895993.0, + "step": 12055 + }, + { + "epoch": 1.323962222710301, + "grad_norm": 2.189047336578369, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7203161716461182, + "num_tokens": 311922674.0, + "step": 12056 + }, + { + "epoch": 1.3240720404129145, + "grad_norm": 2.4605064392089844, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7582027912139893, + "num_tokens": 311944228.0, + "step": 12057 + }, + { + "epoch": 1.3241818581155282, + "grad_norm": 2.6137499809265137, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7145577073097229, + "num_tokens": 311967283.0, + "step": 12058 + }, + { + "epoch": 1.324291675818142, + "grad_norm": 2.6565921306610107, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7270260453224182, + "num_tokens": 311986322.0, + "step": 12059 + }, + { + "epoch": 1.3244014935207555, + "grad_norm": 2.2882025241851807, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7174863815307617, + "num_tokens": 312011426.0, + "step": 12060 + }, + { + "epoch": 1.324511311223369, + "grad_norm": 2.056346893310547, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.6972655653953552, + "num_tokens": 312043022.0, + "step": 12061 + }, + { + "epoch": 1.3246211289259828, + "grad_norm": 2.145270586013794, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6959871053695679, + "num_tokens": 312073029.0, + "step": 12062 + }, + { + "epoch": 1.3247309466285966, + "grad_norm": 2.3415920734405518, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7301679849624634, + "num_tokens": 312098938.0, + "step": 12063 + }, + { + "epoch": 1.3248407643312101, + "grad_norm": 2.4839181900024414, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7181398868560791, + "num_tokens": 312122336.0, + "step": 12064 + }, + { + "epoch": 1.3249505820338239, + "grad_norm": 2.3839855194091797, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7090538740158081, + "num_tokens": 312145917.0, + "step": 12065 + }, + { + "epoch": 1.3250603997364374, + "grad_norm": 2.5300300121307373, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7035560607910156, + "num_tokens": 312167339.0, + "step": 12066 + }, + { + "epoch": 1.3251702174390512, + "grad_norm": 2.3517205715179443, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7263627052307129, + "num_tokens": 312189994.0, + "step": 12067 + }, + { + "epoch": 1.325280035141665, + "grad_norm": 2.190898895263672, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7163586616516113, + "num_tokens": 312220602.0, + "step": 12068 + }, + { + "epoch": 1.3253898528442785, + "grad_norm": 2.3175783157348633, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7135840654373169, + "num_tokens": 312244048.0, + "step": 12069 + }, + { + "epoch": 1.3254996705468922, + "grad_norm": 2.097851037979126, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7133428454399109, + "num_tokens": 312272462.0, + "step": 12070 + }, + { + "epoch": 1.3256094882495058, + "grad_norm": 2.0178451538085938, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6930859684944153, + "num_tokens": 312304011.0, + "step": 12071 + }, + { + "epoch": 1.3257193059521195, + "grad_norm": 1.911210536956787, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7069684267044067, + "num_tokens": 312339018.0, + "step": 12072 + }, + { + "epoch": 1.3258291236547333, + "grad_norm": 2.229466676712036, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.710442304611206, + "num_tokens": 312365181.0, + "step": 12073 + }, + { + "epoch": 1.3259389413573468, + "grad_norm": 2.9280691146850586, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7435972690582275, + "num_tokens": 312380926.0, + "step": 12074 + }, + { + "epoch": 1.3260487590599603, + "grad_norm": 2.103912115097046, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7246139645576477, + "num_tokens": 312408282.0, + "step": 12075 + }, + { + "epoch": 1.326158576762574, + "grad_norm": 2.279937267303467, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7300265431404114, + "num_tokens": 312432842.0, + "step": 12076 + }, + { + "epoch": 1.3262683944651878, + "grad_norm": 2.1764323711395264, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7329120635986328, + "num_tokens": 312457932.0, + "step": 12077 + }, + { + "epoch": 1.3263782121678014, + "grad_norm": 2.2330594062805176, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7494115233421326, + "num_tokens": 312482510.0, + "step": 12078 + }, + { + "epoch": 1.3264880298704151, + "grad_norm": 1.8427419662475586, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.694930374622345, + "num_tokens": 312522526.0, + "step": 12079 + }, + { + "epoch": 1.3265978475730287, + "grad_norm": 2.2647860050201416, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7104254961013794, + "num_tokens": 312547872.0, + "step": 12080 + }, + { + "epoch": 1.3267076652756424, + "grad_norm": 2.5395400524139404, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7493642568588257, + "num_tokens": 312567048.0, + "step": 12081 + }, + { + "epoch": 1.3268174829782562, + "grad_norm": 2.3362808227539062, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7213972210884094, + "num_tokens": 312590450.0, + "step": 12082 + }, + { + "epoch": 1.3269273006808697, + "grad_norm": 2.0631256103515625, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7025426030158997, + "num_tokens": 312621839.0, + "step": 12083 + }, + { + "epoch": 1.3270371183834835, + "grad_norm": 2.183285713195801, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7004297971725464, + "num_tokens": 312649493.0, + "step": 12084 + }, + { + "epoch": 1.327146936086097, + "grad_norm": 2.1703898906707764, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7065279483795166, + "num_tokens": 312678801.0, + "step": 12085 + }, + { + "epoch": 1.3272567537887108, + "grad_norm": 2.249933958053589, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7316281795501709, + "num_tokens": 312702894.0, + "step": 12086 + }, + { + "epoch": 1.3273665714913245, + "grad_norm": 2.252739429473877, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7174549102783203, + "num_tokens": 312726396.0, + "step": 12087 + }, + { + "epoch": 1.327476389193938, + "grad_norm": 2.0628559589385986, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7053529620170593, + "num_tokens": 312755864.0, + "step": 12088 + }, + { + "epoch": 1.3275862068965516, + "grad_norm": 2.342862844467163, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.6978586912155151, + "num_tokens": 312780319.0, + "step": 12089 + }, + { + "epoch": 1.3276960245991654, + "grad_norm": 2.126774787902832, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7175912857055664, + "num_tokens": 312808674.0, + "step": 12090 + }, + { + "epoch": 1.3278058423017791, + "grad_norm": 2.252610921859741, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7040393352508545, + "num_tokens": 312834242.0, + "step": 12091 + }, + { + "epoch": 1.3279156600043927, + "grad_norm": 2.3898351192474365, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7320917844772339, + "num_tokens": 312856847.0, + "step": 12092 + }, + { + "epoch": 1.3280254777070064, + "grad_norm": 2.1116371154785156, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7026292681694031, + "num_tokens": 312885152.0, + "step": 12093 + }, + { + "epoch": 1.32813529540962, + "grad_norm": 2.4958736896514893, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7132844924926758, + "num_tokens": 312907134.0, + "step": 12094 + }, + { + "epoch": 1.3282451131122337, + "grad_norm": 2.5742709636688232, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7408022880554199, + "num_tokens": 312929375.0, + "step": 12095 + }, + { + "epoch": 1.3283549308148475, + "grad_norm": 2.126880168914795, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6891885995864868, + "num_tokens": 312959592.0, + "step": 12096 + }, + { + "epoch": 1.328464748517461, + "grad_norm": 2.1947810649871826, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7285464406013489, + "num_tokens": 312984791.0, + "step": 12097 + }, + { + "epoch": 1.3285745662200747, + "grad_norm": 2.3719310760498047, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7367374897003174, + "num_tokens": 313008554.0, + "step": 12098 + }, + { + "epoch": 1.3286843839226883, + "grad_norm": 2.139045476913452, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7116613388061523, + "num_tokens": 313036100.0, + "step": 12099 + }, + { + "epoch": 1.328794201625302, + "grad_norm": 2.3917343616485596, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7205451130867004, + "num_tokens": 313060742.0, + "step": 12100 + }, + { + "epoch": 1.3289040193279158, + "grad_norm": 2.2649295330047607, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7274937629699707, + "num_tokens": 313085469.0, + "step": 12101 + }, + { + "epoch": 1.3290138370305293, + "grad_norm": 2.1223456859588623, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7084378004074097, + "num_tokens": 313112793.0, + "step": 12102 + }, + { + "epoch": 1.3291236547331429, + "grad_norm": 2.5582494735717773, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7367404699325562, + "num_tokens": 313134516.0, + "step": 12103 + }, + { + "epoch": 1.3292334724357566, + "grad_norm": 2.15242075920105, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7160079479217529, + "num_tokens": 313163023.0, + "step": 12104 + }, + { + "epoch": 1.3293432901383704, + "grad_norm": 1.9709826707839966, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7213916778564453, + "num_tokens": 313196724.0, + "step": 12105 + }, + { + "epoch": 1.329453107840984, + "grad_norm": 1.9884274005889893, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7150290012359619, + "num_tokens": 313227346.0, + "step": 12106 + }, + { + "epoch": 1.3295629255435977, + "grad_norm": 2.1462364196777344, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7107942700386047, + "num_tokens": 313255908.0, + "step": 12107 + }, + { + "epoch": 1.3296727432462112, + "grad_norm": 2.294255495071411, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7163034081459045, + "num_tokens": 313282227.0, + "step": 12108 + }, + { + "epoch": 1.329782560948825, + "grad_norm": 2.4101788997650146, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7321482300758362, + "num_tokens": 313303207.0, + "step": 12109 + }, + { + "epoch": 1.3298923786514387, + "grad_norm": 2.438997983932495, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7062777280807495, + "num_tokens": 313325047.0, + "step": 12110 + }, + { + "epoch": 1.3300021963540523, + "grad_norm": 2.050171375274658, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6889455914497375, + "num_tokens": 313356491.0, + "step": 12111 + }, + { + "epoch": 1.330112014056666, + "grad_norm": 1.9787834882736206, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7051362991333008, + "num_tokens": 313389447.0, + "step": 12112 + }, + { + "epoch": 1.3302218317592795, + "grad_norm": 2.296008348464966, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7157336473464966, + "num_tokens": 313415681.0, + "step": 12113 + }, + { + "epoch": 1.3303316494618933, + "grad_norm": 2.298504114151001, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.725914478302002, + "num_tokens": 313440534.0, + "step": 12114 + }, + { + "epoch": 1.3304414671645068, + "grad_norm": 2.00797700881958, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7160942554473877, + "num_tokens": 313472238.0, + "step": 12115 + }, + { + "epoch": 1.3305512848671206, + "grad_norm": 2.3231632709503174, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7109097242355347, + "num_tokens": 313496901.0, + "step": 12116 + }, + { + "epoch": 1.3306611025697341, + "grad_norm": 2.288942813873291, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7006961703300476, + "num_tokens": 313521690.0, + "step": 12117 + }, + { + "epoch": 1.330770920272348, + "grad_norm": 2.2361671924591064, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7220754027366638, + "num_tokens": 313546813.0, + "step": 12118 + }, + { + "epoch": 1.3308807379749616, + "grad_norm": 2.124540090560913, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.726125180721283, + "num_tokens": 313574997.0, + "step": 12119 + }, + { + "epoch": 1.3309905556775752, + "grad_norm": 1.958698034286499, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7053472995758057, + "num_tokens": 313607098.0, + "step": 12120 + }, + { + "epoch": 1.331100373380189, + "grad_norm": 2.3029308319091797, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7479217052459717, + "num_tokens": 313630093.0, + "step": 12121 + }, + { + "epoch": 1.3312101910828025, + "grad_norm": 2.137225389480591, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7199956774711609, + "num_tokens": 313657207.0, + "step": 12122 + }, + { + "epoch": 1.3313200087854162, + "grad_norm": 2.0839195251464844, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.711556077003479, + "num_tokens": 313686686.0, + "step": 12123 + }, + { + "epoch": 1.33142982648803, + "grad_norm": 1.9358123540878296, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7101590037345886, + "num_tokens": 313718020.0, + "step": 12124 + }, + { + "epoch": 1.3315396441906435, + "grad_norm": 2.1523101329803467, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7275631427764893, + "num_tokens": 313744954.0, + "step": 12125 + }, + { + "epoch": 1.331649461893257, + "grad_norm": 2.3534421920776367, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7042493224143982, + "num_tokens": 313767651.0, + "step": 12126 + }, + { + "epoch": 1.3317592795958708, + "grad_norm": 2.214062452316284, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7205427289009094, + "num_tokens": 313793030.0, + "step": 12127 + }, + { + "epoch": 1.3318690972984846, + "grad_norm": 2.183213949203491, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7161830067634583, + "num_tokens": 313819612.0, + "step": 12128 + }, + { + "epoch": 1.331978915001098, + "grad_norm": 2.091411590576172, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7287355661392212, + "num_tokens": 313847069.0, + "step": 12129 + }, + { + "epoch": 1.3320887327037119, + "grad_norm": 2.1316723823547363, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7295054197311401, + "num_tokens": 313875458.0, + "step": 12130 + }, + { + "epoch": 1.3321985504063254, + "grad_norm": 2.4965546131134033, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7313103675842285, + "num_tokens": 313896527.0, + "step": 12131 + }, + { + "epoch": 1.3323083681089392, + "grad_norm": 2.4113855361938477, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7181534767150879, + "num_tokens": 313918697.0, + "step": 12132 + }, + { + "epoch": 1.332418185811553, + "grad_norm": 2.1385879516601562, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7197672724723816, + "num_tokens": 313946077.0, + "step": 12133 + }, + { + "epoch": 1.3325280035141664, + "grad_norm": 2.184666395187378, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7090038657188416, + "num_tokens": 313972542.0, + "step": 12134 + }, + { + "epoch": 1.3326378212167802, + "grad_norm": 1.9212616682052612, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7265857458114624, + "num_tokens": 314003809.0, + "step": 12135 + }, + { + "epoch": 1.3327476389193937, + "grad_norm": 2.2439961433410645, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.693164050579071, + "num_tokens": 314030491.0, + "step": 12136 + }, + { + "epoch": 1.3328574566220075, + "grad_norm": 1.9314903020858765, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7237504720687866, + "num_tokens": 314062339.0, + "step": 12137 + }, + { + "epoch": 1.3329672743246213, + "grad_norm": 2.11625599861145, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7311379313468933, + "num_tokens": 314089700.0, + "step": 12138 + }, + { + "epoch": 1.3330770920272348, + "grad_norm": 2.1184985637664795, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7076781392097473, + "num_tokens": 314120020.0, + "step": 12139 + }, + { + "epoch": 1.3331869097298483, + "grad_norm": 2.2901129722595215, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7368446588516235, + "num_tokens": 314143549.0, + "step": 12140 + }, + { + "epoch": 1.333296727432462, + "grad_norm": 2.0418050289154053, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7214058637619019, + "num_tokens": 314173193.0, + "step": 12141 + }, + { + "epoch": 1.3334065451350758, + "grad_norm": 2.1154041290283203, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7128165364265442, + "num_tokens": 314200701.0, + "step": 12142 + }, + { + "epoch": 1.3335163628376894, + "grad_norm": 2.2252326011657715, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7292044162750244, + "num_tokens": 314224618.0, + "step": 12143 + }, + { + "epoch": 1.3336261805403031, + "grad_norm": 2.082402229309082, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7031757831573486, + "num_tokens": 314254988.0, + "step": 12144 + }, + { + "epoch": 1.3337359982429167, + "grad_norm": 2.1501708030700684, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7059364318847656, + "num_tokens": 314282456.0, + "step": 12145 + }, + { + "epoch": 1.3338458159455304, + "grad_norm": 2.414094924926758, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.6994415521621704, + "num_tokens": 314305152.0, + "step": 12146 + }, + { + "epoch": 1.3339556336481442, + "grad_norm": 2.2044711112976074, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.713812530040741, + "num_tokens": 314333641.0, + "step": 12147 + }, + { + "epoch": 1.3340654513507577, + "grad_norm": 1.9554613828659058, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7032222747802734, + "num_tokens": 314368978.0, + "step": 12148 + }, + { + "epoch": 1.3341752690533715, + "grad_norm": 2.1313588619232178, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7315045595169067, + "num_tokens": 314394530.0, + "step": 12149 + }, + { + "epoch": 1.334285086755985, + "grad_norm": 2.404371500015259, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7175086736679077, + "num_tokens": 314417140.0, + "step": 12150 + }, + { + "epoch": 1.3343949044585988, + "grad_norm": 2.3367972373962402, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7133262753486633, + "num_tokens": 314440596.0, + "step": 12151 + }, + { + "epoch": 1.3345047221612125, + "grad_norm": 2.28232741355896, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7281033992767334, + "num_tokens": 314462690.0, + "step": 12152 + }, + { + "epoch": 1.334614539863826, + "grad_norm": 2.464040756225586, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7077872157096863, + "num_tokens": 314485297.0, + "step": 12153 + }, + { + "epoch": 1.3347243575664396, + "grad_norm": 2.3359012603759766, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7163246870040894, + "num_tokens": 314511086.0, + "step": 12154 + }, + { + "epoch": 1.3348341752690533, + "grad_norm": 2.333897352218628, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7436714768409729, + "num_tokens": 314534220.0, + "step": 12155 + }, + { + "epoch": 1.334943992971667, + "grad_norm": 2.2587194442749023, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7037439346313477, + "num_tokens": 314559951.0, + "step": 12156 + }, + { + "epoch": 1.3350538106742806, + "grad_norm": 2.289501190185547, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7218344807624817, + "num_tokens": 314584385.0, + "step": 12157 + }, + { + "epoch": 1.3351636283768944, + "grad_norm": 2.3502602577209473, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7314351797103882, + "num_tokens": 314608107.0, + "step": 12158 + }, + { + "epoch": 1.335273446079508, + "grad_norm": 2.2112841606140137, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6893928647041321, + "num_tokens": 314636606.0, + "step": 12159 + }, + { + "epoch": 1.3353832637821217, + "grad_norm": 2.2899999618530273, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7283031940460205, + "num_tokens": 314660635.0, + "step": 12160 + }, + { + "epoch": 1.3354930814847354, + "grad_norm": 2.172978401184082, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7252126932144165, + "num_tokens": 314687780.0, + "step": 12161 + }, + { + "epoch": 1.335602899187349, + "grad_norm": 2.233887195587158, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7340248823165894, + "num_tokens": 314713069.0, + "step": 12162 + }, + { + "epoch": 1.3357127168899627, + "grad_norm": 2.2581658363342285, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7154719829559326, + "num_tokens": 314737852.0, + "step": 12163 + }, + { + "epoch": 1.3358225345925763, + "grad_norm": 2.177900552749634, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.688487708568573, + "num_tokens": 314764199.0, + "step": 12164 + }, + { + "epoch": 1.33593235229519, + "grad_norm": 2.276888132095337, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7275768518447876, + "num_tokens": 314788466.0, + "step": 12165 + }, + { + "epoch": 1.3360421699978036, + "grad_norm": 2.2586004734039307, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7057779431343079, + "num_tokens": 314814228.0, + "step": 12166 + }, + { + "epoch": 1.3361519877004173, + "grad_norm": 1.9037092924118042, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.6874492168426514, + "num_tokens": 314847990.0, + "step": 12167 + }, + { + "epoch": 1.3362618054030309, + "grad_norm": 2.3650062084198, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7376158833503723, + "num_tokens": 314871778.0, + "step": 12168 + }, + { + "epoch": 1.3363716231056446, + "grad_norm": 2.507750988006592, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7181243300437927, + "num_tokens": 314894790.0, + "step": 12169 + }, + { + "epoch": 1.3364814408082584, + "grad_norm": 2.375624895095825, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7151716947555542, + "num_tokens": 314918890.0, + "step": 12170 + }, + { + "epoch": 1.336591258510872, + "grad_norm": 2.2384307384490967, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7289177179336548, + "num_tokens": 314944008.0, + "step": 12171 + }, + { + "epoch": 1.3367010762134857, + "grad_norm": 1.9208735227584839, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7027195692062378, + "num_tokens": 314979013.0, + "step": 12172 + }, + { + "epoch": 1.3368108939160992, + "grad_norm": 2.234025239944458, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7436856031417847, + "num_tokens": 315003209.0, + "step": 12173 + }, + { + "epoch": 1.336920711618713, + "grad_norm": 2.163142204284668, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7352210283279419, + "num_tokens": 315029803.0, + "step": 12174 + }, + { + "epoch": 1.3370305293213267, + "grad_norm": 2.2827467918395996, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7132997512817383, + "num_tokens": 315058104.0, + "step": 12175 + }, + { + "epoch": 1.3371403470239402, + "grad_norm": 2.3907313346862793, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7325514554977417, + "num_tokens": 315080546.0, + "step": 12176 + }, + { + "epoch": 1.337250164726554, + "grad_norm": 2.292772054672241, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7244073152542114, + "num_tokens": 315105139.0, + "step": 12177 + }, + { + "epoch": 1.3373599824291675, + "grad_norm": 2.3690786361694336, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7100588083267212, + "num_tokens": 315128254.0, + "step": 12178 + }, + { + "epoch": 1.3374698001317813, + "grad_norm": 2.2200775146484375, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7155215740203857, + "num_tokens": 315151826.0, + "step": 12179 + }, + { + "epoch": 1.3375796178343948, + "grad_norm": 2.0473034381866455, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7120314836502075, + "num_tokens": 315180659.0, + "step": 12180 + }, + { + "epoch": 1.3376894355370086, + "grad_norm": 2.2075788974761963, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7195833325386047, + "num_tokens": 315208419.0, + "step": 12181 + }, + { + "epoch": 1.3377992532396221, + "grad_norm": 2.1998825073242188, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.724732518196106, + "num_tokens": 315233736.0, + "step": 12182 + }, + { + "epoch": 1.3379090709422359, + "grad_norm": 2.257505178451538, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7483026385307312, + "num_tokens": 315258558.0, + "step": 12183 + }, + { + "epoch": 1.3380188886448496, + "grad_norm": 2.2647767066955566, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7157851457595825, + "num_tokens": 315284503.0, + "step": 12184 + }, + { + "epoch": 1.3381287063474632, + "grad_norm": 2.1260576248168945, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.713120698928833, + "num_tokens": 315314227.0, + "step": 12185 + }, + { + "epoch": 1.338238524050077, + "grad_norm": 2.0270862579345703, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7453780174255371, + "num_tokens": 315341861.0, + "step": 12186 + }, + { + "epoch": 1.3383483417526905, + "grad_norm": 2.686323642730713, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7374072074890137, + "num_tokens": 315362198.0, + "step": 12187 + }, + { + "epoch": 1.3384581594553042, + "grad_norm": 2.40075945854187, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7142733931541443, + "num_tokens": 315384800.0, + "step": 12188 + }, + { + "epoch": 1.338567977157918, + "grad_norm": 2.1618075370788574, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.720125675201416, + "num_tokens": 315412593.0, + "step": 12189 + }, + { + "epoch": 1.3386777948605315, + "grad_norm": 2.2528936862945557, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7211897373199463, + "num_tokens": 315437974.0, + "step": 12190 + }, + { + "epoch": 1.338787612563145, + "grad_norm": 2.0930752754211426, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6936110854148865, + "num_tokens": 315466659.0, + "step": 12191 + }, + { + "epoch": 1.3388974302657588, + "grad_norm": 1.940476655960083, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7297102212905884, + "num_tokens": 315498402.0, + "step": 12192 + }, + { + "epoch": 1.3390072479683726, + "grad_norm": 2.4122660160064697, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7280949354171753, + "num_tokens": 315521133.0, + "step": 12193 + }, + { + "epoch": 1.339117065670986, + "grad_norm": 2.2513506412506104, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7408558130264282, + "num_tokens": 315545156.0, + "step": 12194 + }, + { + "epoch": 1.3392268833735999, + "grad_norm": 2.1435317993164062, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7165452241897583, + "num_tokens": 315571491.0, + "step": 12195 + }, + { + "epoch": 1.3393367010762134, + "grad_norm": 2.303699254989624, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7268068194389343, + "num_tokens": 315597388.0, + "step": 12196 + }, + { + "epoch": 1.3394465187788271, + "grad_norm": 2.6103508472442627, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7260706424713135, + "num_tokens": 315618018.0, + "step": 12197 + }, + { + "epoch": 1.339556336481441, + "grad_norm": 2.195948600769043, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7281780242919922, + "num_tokens": 315644214.0, + "step": 12198 + }, + { + "epoch": 1.3396661541840544, + "grad_norm": 2.216094493865967, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7268987894058228, + "num_tokens": 315670775.0, + "step": 12199 + }, + { + "epoch": 1.3397759718866682, + "grad_norm": 2.4899749755859375, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7246108055114746, + "num_tokens": 315692463.0, + "step": 12200 + }, + { + "epoch": 1.3398857895892817, + "grad_norm": 2.287926435470581, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6934123039245605, + "num_tokens": 315719338.0, + "step": 12201 + }, + { + "epoch": 1.3399956072918955, + "grad_norm": 2.168985605239868, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.6939548850059509, + "num_tokens": 315748717.0, + "step": 12202 + }, + { + "epoch": 1.3401054249945092, + "grad_norm": 2.2981479167938232, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7486758232116699, + "num_tokens": 315771304.0, + "step": 12203 + }, + { + "epoch": 1.3402152426971228, + "grad_norm": 2.1667118072509766, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7101526856422424, + "num_tokens": 315801733.0, + "step": 12204 + }, + { + "epoch": 1.3403250603997363, + "grad_norm": 2.279181718826294, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7152290940284729, + "num_tokens": 315825516.0, + "step": 12205 + }, + { + "epoch": 1.34043487810235, + "grad_norm": 2.122446298599243, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7075620889663696, + "num_tokens": 315854523.0, + "step": 12206 + }, + { + "epoch": 1.3405446958049638, + "grad_norm": 2.2139973640441895, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7226345539093018, + "num_tokens": 315881088.0, + "step": 12207 + }, + { + "epoch": 1.3406545135075774, + "grad_norm": 2.1239562034606934, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6826637983322144, + "num_tokens": 315913033.0, + "step": 12208 + }, + { + "epoch": 1.3407643312101911, + "grad_norm": 2.206700325012207, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7093059420585632, + "num_tokens": 315939853.0, + "step": 12209 + }, + { + "epoch": 1.3408741489128047, + "grad_norm": 2.208794116973877, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7108456492424011, + "num_tokens": 315964538.0, + "step": 12210 + }, + { + "epoch": 1.3409839666154184, + "grad_norm": 1.8624264001846313, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7089231610298157, + "num_tokens": 316001696.0, + "step": 12211 + }, + { + "epoch": 1.3410937843180322, + "grad_norm": 2.096914529800415, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7167491912841797, + "num_tokens": 316027487.0, + "step": 12212 + }, + { + "epoch": 1.3412036020206457, + "grad_norm": 2.2125794887542725, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7275266051292419, + "num_tokens": 316052502.0, + "step": 12213 + }, + { + "epoch": 1.3413134197232595, + "grad_norm": 2.3599624633789062, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7416286468505859, + "num_tokens": 316074290.0, + "step": 12214 + }, + { + "epoch": 1.341423237425873, + "grad_norm": 2.560718059539795, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7134485244750977, + "num_tokens": 316095713.0, + "step": 12215 + }, + { + "epoch": 1.3415330551284868, + "grad_norm": 2.501213550567627, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7128154635429382, + "num_tokens": 316119090.0, + "step": 12216 + }, + { + "epoch": 1.3416428728311005, + "grad_norm": 2.1690101623535156, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6797206401824951, + "num_tokens": 316151908.0, + "step": 12217 + }, + { + "epoch": 1.341752690533714, + "grad_norm": 2.781309127807617, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7404189109802246, + "num_tokens": 316173112.0, + "step": 12218 + }, + { + "epoch": 1.3418625082363276, + "grad_norm": 2.455758810043335, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7374148964881897, + "num_tokens": 316195721.0, + "step": 12219 + }, + { + "epoch": 1.3419723259389413, + "grad_norm": 2.050997257232666, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7126112580299377, + "num_tokens": 316224989.0, + "step": 12220 + }, + { + "epoch": 1.342082143641555, + "grad_norm": 1.9138987064361572, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7085530757904053, + "num_tokens": 316259046.0, + "step": 12221 + }, + { + "epoch": 1.3421919613441686, + "grad_norm": 2.3150863647460938, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7333993911743164, + "num_tokens": 316281973.0, + "step": 12222 + }, + { + "epoch": 1.3423017790467824, + "grad_norm": 2.3260977268218994, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7128572463989258, + "num_tokens": 316307353.0, + "step": 12223 + }, + { + "epoch": 1.342411596749396, + "grad_norm": 2.3366527557373047, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7167181968688965, + "num_tokens": 316332042.0, + "step": 12224 + }, + { + "epoch": 1.3425214144520097, + "grad_norm": 2.0480897426605225, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7256385684013367, + "num_tokens": 316361949.0, + "step": 12225 + }, + { + "epoch": 1.3426312321546234, + "grad_norm": 2.101245164871216, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7419489622116089, + "num_tokens": 316389388.0, + "step": 12226 + }, + { + "epoch": 1.342741049857237, + "grad_norm": 2.5088043212890625, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7288379669189453, + "num_tokens": 316411616.0, + "step": 12227 + }, + { + "epoch": 1.3428508675598507, + "grad_norm": 2.298293113708496, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7234135866165161, + "num_tokens": 316436210.0, + "step": 12228 + }, + { + "epoch": 1.3429606852624643, + "grad_norm": 2.3302435874938965, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7027477622032166, + "num_tokens": 316460338.0, + "step": 12229 + }, + { + "epoch": 1.343070502965078, + "grad_norm": 1.9359396696090698, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7239867448806763, + "num_tokens": 316492222.0, + "step": 12230 + }, + { + "epoch": 1.3431803206676916, + "grad_norm": 2.1115851402282715, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7303433418273926, + "num_tokens": 316520060.0, + "step": 12231 + }, + { + "epoch": 1.3432901383703053, + "grad_norm": 1.9570518732070923, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6917368173599243, + "num_tokens": 316557495.0, + "step": 12232 + }, + { + "epoch": 1.3433999560729188, + "grad_norm": 2.3231139183044434, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7110629081726074, + "num_tokens": 316580479.0, + "step": 12233 + }, + { + "epoch": 1.3435097737755326, + "grad_norm": 2.277892589569092, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.725196123123169, + "num_tokens": 316604763.0, + "step": 12234 + }, + { + "epoch": 1.3436195914781464, + "grad_norm": 2.266031503677368, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7424657940864563, + "num_tokens": 316628819.0, + "step": 12235 + }, + { + "epoch": 1.34372940918076, + "grad_norm": 2.5090525150299072, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7282782793045044, + "num_tokens": 316649717.0, + "step": 12236 + }, + { + "epoch": 1.3438392268833736, + "grad_norm": 2.298231840133667, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7173179388046265, + "num_tokens": 316673641.0, + "step": 12237 + }, + { + "epoch": 1.3439490445859872, + "grad_norm": 2.1243717670440674, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7118873000144958, + "num_tokens": 316703302.0, + "step": 12238 + }, + { + "epoch": 1.344058862288601, + "grad_norm": 2.099534511566162, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7207841277122498, + "num_tokens": 316731974.0, + "step": 12239 + }, + { + "epoch": 1.3441686799912147, + "grad_norm": 2.3989553451538086, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7280466556549072, + "num_tokens": 316753735.0, + "step": 12240 + }, + { + "epoch": 1.3442784976938282, + "grad_norm": 2.362638235092163, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7342877984046936, + "num_tokens": 316777376.0, + "step": 12241 + }, + { + "epoch": 1.3443883153964418, + "grad_norm": 2.2740530967712402, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7364886999130249, + "num_tokens": 316803200.0, + "step": 12242 + }, + { + "epoch": 1.3444981330990555, + "grad_norm": 2.374756336212158, + "learning_rate": 1e-06, + "loss": 1.0591, + "mean_token_accuracy": 0.6721189022064209, + "num_tokens": 316830502.0, + "step": 12243 + }, + { + "epoch": 1.3446079508016693, + "grad_norm": 2.1858456134796143, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.729388952255249, + "num_tokens": 316858731.0, + "step": 12244 + }, + { + "epoch": 1.3447177685042828, + "grad_norm": 2.0470597743988037, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7026958465576172, + "num_tokens": 316890304.0, + "step": 12245 + }, + { + "epoch": 1.3448275862068966, + "grad_norm": 2.225339889526367, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7225769758224487, + "num_tokens": 316914171.0, + "step": 12246 + }, + { + "epoch": 1.34493740390951, + "grad_norm": 2.0288000106811523, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7269011735916138, + "num_tokens": 316945920.0, + "step": 12247 + }, + { + "epoch": 1.3450472216121239, + "grad_norm": 2.542914628982544, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7297372817993164, + "num_tokens": 316967233.0, + "step": 12248 + }, + { + "epoch": 1.3451570393147376, + "grad_norm": 1.9988203048706055, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7073507308959961, + "num_tokens": 316996454.0, + "step": 12249 + }, + { + "epoch": 1.3452668570173512, + "grad_norm": 2.3114593029022217, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7159897089004517, + "num_tokens": 317021126.0, + "step": 12250 + }, + { + "epoch": 1.345376674719965, + "grad_norm": 2.108449697494507, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.703225314617157, + "num_tokens": 317049288.0, + "step": 12251 + }, + { + "epoch": 1.3454864924225785, + "grad_norm": 2.2726526260375977, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7068937420845032, + "num_tokens": 317075763.0, + "step": 12252 + }, + { + "epoch": 1.3455963101251922, + "grad_norm": 2.1045048236846924, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7242289781570435, + "num_tokens": 317105490.0, + "step": 12253 + }, + { + "epoch": 1.345706127827806, + "grad_norm": 2.3403539657592773, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.723986029624939, + "num_tokens": 317128532.0, + "step": 12254 + }, + { + "epoch": 1.3458159455304195, + "grad_norm": 2.463106393814087, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7250423431396484, + "num_tokens": 317149149.0, + "step": 12255 + }, + { + "epoch": 1.345925763233033, + "grad_norm": 2.073331356048584, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7297539710998535, + "num_tokens": 317179995.0, + "step": 12256 + }, + { + "epoch": 1.3460355809356468, + "grad_norm": 2.1495258808135986, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7207980155944824, + "num_tokens": 317208166.0, + "step": 12257 + }, + { + "epoch": 1.3461453986382605, + "grad_norm": 2.365513324737549, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6914409399032593, + "num_tokens": 317235731.0, + "step": 12258 + }, + { + "epoch": 1.346255216340874, + "grad_norm": 2.301654100418091, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7340855598449707, + "num_tokens": 317259161.0, + "step": 12259 + }, + { + "epoch": 1.3463650340434878, + "grad_norm": 2.4413280487060547, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.716195821762085, + "num_tokens": 317280911.0, + "step": 12260 + }, + { + "epoch": 1.3464748517461014, + "grad_norm": 2.22682523727417, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7184489369392395, + "num_tokens": 317307150.0, + "step": 12261 + }, + { + "epoch": 1.3465846694487151, + "grad_norm": 2.101308822631836, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7185875177383423, + "num_tokens": 317336466.0, + "step": 12262 + }, + { + "epoch": 1.346694487151329, + "grad_norm": 2.503164768218994, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7434148788452148, + "num_tokens": 317358547.0, + "step": 12263 + }, + { + "epoch": 1.3468043048539424, + "grad_norm": 1.851330041885376, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7286016941070557, + "num_tokens": 317395401.0, + "step": 12264 + }, + { + "epoch": 1.3469141225565562, + "grad_norm": 2.0894737243652344, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7177319526672363, + "num_tokens": 317423197.0, + "step": 12265 + }, + { + "epoch": 1.3470239402591697, + "grad_norm": 2.273167848587036, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7084245681762695, + "num_tokens": 317446824.0, + "step": 12266 + }, + { + "epoch": 1.3471337579617835, + "grad_norm": 2.297494649887085, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7194132208824158, + "num_tokens": 317471013.0, + "step": 12267 + }, + { + "epoch": 1.3472435756643972, + "grad_norm": 2.3538944721221924, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7342185974121094, + "num_tokens": 317494213.0, + "step": 12268 + }, + { + "epoch": 1.3473533933670108, + "grad_norm": 2.1313881874084473, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7256275415420532, + "num_tokens": 317521586.0, + "step": 12269 + }, + { + "epoch": 1.3474632110696243, + "grad_norm": 2.036198377609253, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7226496934890747, + "num_tokens": 317550203.0, + "step": 12270 + }, + { + "epoch": 1.347573028772238, + "grad_norm": 2.047574520111084, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7260557413101196, + "num_tokens": 317577786.0, + "step": 12271 + }, + { + "epoch": 1.3476828464748518, + "grad_norm": 2.133249044418335, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.705571174621582, + "num_tokens": 317605820.0, + "step": 12272 + }, + { + "epoch": 1.3477926641774653, + "grad_norm": 2.1738691329956055, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6953439712524414, + "num_tokens": 317632019.0, + "step": 12273 + }, + { + "epoch": 1.347902481880079, + "grad_norm": 2.1744184494018555, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7148940563201904, + "num_tokens": 317660099.0, + "step": 12274 + }, + { + "epoch": 1.3480122995826926, + "grad_norm": 2.299285411834717, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7095645666122437, + "num_tokens": 317686365.0, + "step": 12275 + }, + { + "epoch": 1.3481221172853064, + "grad_norm": 2.4937000274658203, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7215437889099121, + "num_tokens": 317707757.0, + "step": 12276 + }, + { + "epoch": 1.3482319349879202, + "grad_norm": 2.3450465202331543, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7386702299118042, + "num_tokens": 317731141.0, + "step": 12277 + }, + { + "epoch": 1.3483417526905337, + "grad_norm": 2.2433481216430664, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7371100783348083, + "num_tokens": 317758618.0, + "step": 12278 + }, + { + "epoch": 1.3484515703931474, + "grad_norm": 2.3575637340545654, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7161529660224915, + "num_tokens": 317780188.0, + "step": 12279 + }, + { + "epoch": 1.348561388095761, + "grad_norm": 2.3061113357543945, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7443547248840332, + "num_tokens": 317803852.0, + "step": 12280 + }, + { + "epoch": 1.3486712057983747, + "grad_norm": 1.990071177482605, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7134294509887695, + "num_tokens": 317834475.0, + "step": 12281 + }, + { + "epoch": 1.3487810235009885, + "grad_norm": 2.1403377056121826, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.744393527507782, + "num_tokens": 317860962.0, + "step": 12282 + }, + { + "epoch": 1.348890841203602, + "grad_norm": 2.1831250190734863, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7237551212310791, + "num_tokens": 317887167.0, + "step": 12283 + }, + { + "epoch": 1.3490006589062156, + "grad_norm": 2.672384738922119, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7384830713272095, + "num_tokens": 317907342.0, + "step": 12284 + }, + { + "epoch": 1.3491104766088293, + "grad_norm": 2.634023666381836, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7180944085121155, + "num_tokens": 317929542.0, + "step": 12285 + }, + { + "epoch": 1.349220294311443, + "grad_norm": 2.5984129905700684, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7150148749351501, + "num_tokens": 317950297.0, + "step": 12286 + }, + { + "epoch": 1.3493301120140566, + "grad_norm": 2.1232166290283203, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7043776512145996, + "num_tokens": 317978347.0, + "step": 12287 + }, + { + "epoch": 1.3494399297166704, + "grad_norm": 2.2306344509124756, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7364705801010132, + "num_tokens": 318003315.0, + "step": 12288 + }, + { + "epoch": 1.349549747419284, + "grad_norm": 2.3149290084838867, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7288378477096558, + "num_tokens": 318027089.0, + "step": 12289 + }, + { + "epoch": 1.3496595651218977, + "grad_norm": 2.4324755668640137, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7237098217010498, + "num_tokens": 318049983.0, + "step": 12290 + }, + { + "epoch": 1.3497693828245114, + "grad_norm": 2.276510238647461, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7106680870056152, + "num_tokens": 318076446.0, + "step": 12291 + }, + { + "epoch": 1.349879200527125, + "grad_norm": 2.3927595615386963, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7272767424583435, + "num_tokens": 318099282.0, + "step": 12292 + }, + { + "epoch": 1.3499890182297387, + "grad_norm": 2.276486396789551, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7138998508453369, + "num_tokens": 318125169.0, + "step": 12293 + }, + { + "epoch": 1.3500988359323522, + "grad_norm": 1.9322410821914673, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.740745484828949, + "num_tokens": 318157779.0, + "step": 12294 + }, + { + "epoch": 1.350208653634966, + "grad_norm": 1.9438519477844238, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7469073534011841, + "num_tokens": 318187301.0, + "step": 12295 + }, + { + "epoch": 1.3503184713375795, + "grad_norm": 2.196526288986206, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7150954008102417, + "num_tokens": 318212475.0, + "step": 12296 + }, + { + "epoch": 1.3504282890401933, + "grad_norm": 2.020076036453247, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7078850269317627, + "num_tokens": 318245545.0, + "step": 12297 + }, + { + "epoch": 1.3505381067428068, + "grad_norm": 2.355522394180298, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7332205772399902, + "num_tokens": 318268347.0, + "step": 12298 + }, + { + "epoch": 1.3506479244454206, + "grad_norm": 2.016960620880127, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7259119749069214, + "num_tokens": 318296588.0, + "step": 12299 + }, + { + "epoch": 1.3507577421480343, + "grad_norm": 2.4184353351593018, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7246121764183044, + "num_tokens": 318318387.0, + "step": 12300 + }, + { + "epoch": 1.3508675598506479, + "grad_norm": 2.2411491870880127, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7555785179138184, + "num_tokens": 318340938.0, + "step": 12301 + }, + { + "epoch": 1.3509773775532616, + "grad_norm": 2.361440896987915, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7248269319534302, + "num_tokens": 318362842.0, + "step": 12302 + }, + { + "epoch": 1.3510871952558752, + "grad_norm": 2.2037527561187744, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.734898567199707, + "num_tokens": 318388176.0, + "step": 12303 + }, + { + "epoch": 1.351197012958489, + "grad_norm": 2.362889051437378, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7348032593727112, + "num_tokens": 318411027.0, + "step": 12304 + }, + { + "epoch": 1.3513068306611027, + "grad_norm": 2.4580328464508057, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7311869859695435, + "num_tokens": 318431454.0, + "step": 12305 + }, + { + "epoch": 1.3514166483637162, + "grad_norm": 2.348133087158203, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7093136310577393, + "num_tokens": 318456081.0, + "step": 12306 + }, + { + "epoch": 1.3515264660663298, + "grad_norm": 2.1511952877044678, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7159748077392578, + "num_tokens": 318484595.0, + "step": 12307 + }, + { + "epoch": 1.3516362837689435, + "grad_norm": 2.511349678039551, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7069332599639893, + "num_tokens": 318507267.0, + "step": 12308 + }, + { + "epoch": 1.3517461014715573, + "grad_norm": 2.3121984004974365, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7345554828643799, + "num_tokens": 318530460.0, + "step": 12309 + }, + { + "epoch": 1.3518559191741708, + "grad_norm": 2.1587936878204346, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7146115899085999, + "num_tokens": 318557169.0, + "step": 12310 + }, + { + "epoch": 1.3519657368767846, + "grad_norm": 2.1142585277557373, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7054694890975952, + "num_tokens": 318585845.0, + "step": 12311 + }, + { + "epoch": 1.352075554579398, + "grad_norm": 2.526329278945923, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7049511671066284, + "num_tokens": 318607694.0, + "step": 12312 + }, + { + "epoch": 1.3521853722820119, + "grad_norm": 2.4091036319732666, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.713695764541626, + "num_tokens": 318630886.0, + "step": 12313 + }, + { + "epoch": 1.3522951899846256, + "grad_norm": 2.530264139175415, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7358220815658569, + "num_tokens": 318651318.0, + "step": 12314 + }, + { + "epoch": 1.3524050076872391, + "grad_norm": 2.06691575050354, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.709306001663208, + "num_tokens": 318681369.0, + "step": 12315 + }, + { + "epoch": 1.352514825389853, + "grad_norm": 2.262789726257324, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.70677250623703, + "num_tokens": 318708443.0, + "step": 12316 + }, + { + "epoch": 1.3526246430924664, + "grad_norm": 2.479353666305542, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7225058078765869, + "num_tokens": 318728753.0, + "step": 12317 + }, + { + "epoch": 1.3527344607950802, + "grad_norm": 2.243882656097412, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7108443975448608, + "num_tokens": 318752398.0, + "step": 12318 + }, + { + "epoch": 1.352844278497694, + "grad_norm": 2.2805354595184326, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.699524998664856, + "num_tokens": 318777523.0, + "step": 12319 + }, + { + "epoch": 1.3529540962003075, + "grad_norm": 2.1710193157196045, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.716549813747406, + "num_tokens": 318804092.0, + "step": 12320 + }, + { + "epoch": 1.353063913902921, + "grad_norm": 2.2474613189697266, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.727487325668335, + "num_tokens": 318830073.0, + "step": 12321 + }, + { + "epoch": 1.3531737316055348, + "grad_norm": 2.2493984699249268, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7200972437858582, + "num_tokens": 318855429.0, + "step": 12322 + }, + { + "epoch": 1.3532835493081485, + "grad_norm": 2.299278736114502, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7325960397720337, + "num_tokens": 318878567.0, + "step": 12323 + }, + { + "epoch": 1.353393367010762, + "grad_norm": 2.1161108016967773, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7377310395240784, + "num_tokens": 318905612.0, + "step": 12324 + }, + { + "epoch": 1.3535031847133758, + "grad_norm": 2.2117605209350586, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.715396523475647, + "num_tokens": 318932808.0, + "step": 12325 + }, + { + "epoch": 1.3536130024159894, + "grad_norm": 2.2377092838287354, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7136448621749878, + "num_tokens": 318958431.0, + "step": 12326 + }, + { + "epoch": 1.3537228201186031, + "grad_norm": 2.2674944400787354, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7336289882659912, + "num_tokens": 318981995.0, + "step": 12327 + }, + { + "epoch": 1.3538326378212169, + "grad_norm": 2.1922826766967773, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6982031464576721, + "num_tokens": 319011387.0, + "step": 12328 + }, + { + "epoch": 1.3539424555238304, + "grad_norm": 2.0330636501312256, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.691906213760376, + "num_tokens": 319044796.0, + "step": 12329 + }, + { + "epoch": 1.3540522732264442, + "grad_norm": 1.994065284729004, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7145536541938782, + "num_tokens": 319074717.0, + "step": 12330 + }, + { + "epoch": 1.3541620909290577, + "grad_norm": 2.377204179763794, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7002272009849548, + "num_tokens": 319099557.0, + "step": 12331 + }, + { + "epoch": 1.3542719086316715, + "grad_norm": 2.4477663040161133, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.729327917098999, + "num_tokens": 319123101.0, + "step": 12332 + }, + { + "epoch": 1.3543817263342852, + "grad_norm": 2.4752821922302246, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7215074300765991, + "num_tokens": 319145466.0, + "step": 12333 + }, + { + "epoch": 1.3544915440368988, + "grad_norm": 1.7669636011123657, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6886134743690491, + "num_tokens": 319184665.0, + "step": 12334 + }, + { + "epoch": 1.3546013617395123, + "grad_norm": 2.5037355422973633, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7063888311386108, + "num_tokens": 319205907.0, + "step": 12335 + }, + { + "epoch": 1.354711179442126, + "grad_norm": 2.413769245147705, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7243388891220093, + "num_tokens": 319227066.0, + "step": 12336 + }, + { + "epoch": 1.3548209971447398, + "grad_norm": 2.301154613494873, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7040467262268066, + "num_tokens": 319252101.0, + "step": 12337 + }, + { + "epoch": 1.3549308148473533, + "grad_norm": 2.3571135997772217, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7211397886276245, + "num_tokens": 319275052.0, + "step": 12338 + }, + { + "epoch": 1.355040632549967, + "grad_norm": 2.5791702270507812, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7515286803245544, + "num_tokens": 319295313.0, + "step": 12339 + }, + { + "epoch": 1.3551504502525806, + "grad_norm": 2.2459912300109863, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7113133668899536, + "num_tokens": 319321658.0, + "step": 12340 + }, + { + "epoch": 1.3552602679551944, + "grad_norm": 2.453321933746338, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7214090824127197, + "num_tokens": 319344828.0, + "step": 12341 + }, + { + "epoch": 1.3553700856578081, + "grad_norm": 2.188533306121826, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7202422618865967, + "num_tokens": 319372964.0, + "step": 12342 + }, + { + "epoch": 1.3554799033604217, + "grad_norm": 2.391308546066284, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7230046987533569, + "num_tokens": 319397019.0, + "step": 12343 + }, + { + "epoch": 1.3555897210630354, + "grad_norm": 2.2503578662872314, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.695929765701294, + "num_tokens": 319423150.0, + "step": 12344 + }, + { + "epoch": 1.355699538765649, + "grad_norm": 2.4831838607788086, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7394041419029236, + "num_tokens": 319444028.0, + "step": 12345 + }, + { + "epoch": 1.3558093564682627, + "grad_norm": 2.2427167892456055, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7074928283691406, + "num_tokens": 319471726.0, + "step": 12346 + }, + { + "epoch": 1.3559191741708763, + "grad_norm": 2.2473807334899902, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.711927056312561, + "num_tokens": 319496920.0, + "step": 12347 + }, + { + "epoch": 1.35602899187349, + "grad_norm": 2.146024703979492, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7440376281738281, + "num_tokens": 319522394.0, + "step": 12348 + }, + { + "epoch": 1.3561388095761036, + "grad_norm": 2.0407910346984863, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7367944121360779, + "num_tokens": 319549366.0, + "step": 12349 + }, + { + "epoch": 1.3562486272787173, + "grad_norm": 2.256983757019043, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7388672828674316, + "num_tokens": 319573807.0, + "step": 12350 + }, + { + "epoch": 1.356358444981331, + "grad_norm": 2.400444507598877, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.717597484588623, + "num_tokens": 319597163.0, + "step": 12351 + }, + { + "epoch": 1.3564682626839446, + "grad_norm": 2.092228651046753, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7331132888793945, + "num_tokens": 319624851.0, + "step": 12352 + }, + { + "epoch": 1.3565780803865584, + "grad_norm": 2.3678338527679443, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7052597999572754, + "num_tokens": 319648659.0, + "step": 12353 + }, + { + "epoch": 1.356687898089172, + "grad_norm": 2.651488780975342, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7120533585548401, + "num_tokens": 319668633.0, + "step": 12354 + }, + { + "epoch": 1.3567977157917857, + "grad_norm": 2.031177282333374, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.716942310333252, + "num_tokens": 319697321.0, + "step": 12355 + }, + { + "epoch": 1.3569075334943994, + "grad_norm": 2.05344295501709, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6946905851364136, + "num_tokens": 319727386.0, + "step": 12356 + }, + { + "epoch": 1.357017351197013, + "grad_norm": 2.283883810043335, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7205273509025574, + "num_tokens": 319751705.0, + "step": 12357 + }, + { + "epoch": 1.3571271688996267, + "grad_norm": 2.3119726181030273, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7293915152549744, + "num_tokens": 319777094.0, + "step": 12358 + }, + { + "epoch": 1.3572369866022402, + "grad_norm": 2.016937494277954, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.715782105922699, + "num_tokens": 319806021.0, + "step": 12359 + }, + { + "epoch": 1.357346804304854, + "grad_norm": 2.1630656719207764, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7121269702911377, + "num_tokens": 319834341.0, + "step": 12360 + }, + { + "epoch": 1.3574566220074675, + "grad_norm": 2.1094367504119873, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6862776279449463, + "num_tokens": 319863895.0, + "step": 12361 + }, + { + "epoch": 1.3575664397100813, + "grad_norm": 2.0820868015289307, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7048318386077881, + "num_tokens": 319893533.0, + "step": 12362 + }, + { + "epoch": 1.3576762574126948, + "grad_norm": 2.0510575771331787, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.6981328129768372, + "num_tokens": 319925237.0, + "step": 12363 + }, + { + "epoch": 1.3577860751153086, + "grad_norm": 2.2411539554595947, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.712165355682373, + "num_tokens": 319953740.0, + "step": 12364 + }, + { + "epoch": 1.3578958928179223, + "grad_norm": 2.4604504108428955, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.740851879119873, + "num_tokens": 319975003.0, + "step": 12365 + }, + { + "epoch": 1.3580057105205359, + "grad_norm": 2.0277161598205566, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7162518501281738, + "num_tokens": 320005624.0, + "step": 12366 + }, + { + "epoch": 1.3581155282231496, + "grad_norm": 2.4382176399230957, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7188678979873657, + "num_tokens": 320030292.0, + "step": 12367 + }, + { + "epoch": 1.3582253459257632, + "grad_norm": 2.296912431716919, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7088733911514282, + "num_tokens": 320054812.0, + "step": 12368 + }, + { + "epoch": 1.358335163628377, + "grad_norm": 2.1459567546844482, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7188584208488464, + "num_tokens": 320083315.0, + "step": 12369 + }, + { + "epoch": 1.3584449813309907, + "grad_norm": 2.333289384841919, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7111608982086182, + "num_tokens": 320107574.0, + "step": 12370 + }, + { + "epoch": 1.3585547990336042, + "grad_norm": 2.0465681552886963, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7020658850669861, + "num_tokens": 320137930.0, + "step": 12371 + }, + { + "epoch": 1.3586646167362177, + "grad_norm": 2.529939889907837, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7149102091789246, + "num_tokens": 320160452.0, + "step": 12372 + }, + { + "epoch": 1.3587744344388315, + "grad_norm": 2.3752663135528564, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7077172994613647, + "num_tokens": 320186376.0, + "step": 12373 + }, + { + "epoch": 1.3588842521414453, + "grad_norm": 2.3387742042541504, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.7063248753547668, + "num_tokens": 320212668.0, + "step": 12374 + }, + { + "epoch": 1.3589940698440588, + "grad_norm": 2.028930187225342, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7224889993667603, + "num_tokens": 320242462.0, + "step": 12375 + }, + { + "epoch": 1.3591038875466726, + "grad_norm": 2.0824718475341797, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7309830188751221, + "num_tokens": 320270761.0, + "step": 12376 + }, + { + "epoch": 1.359213705249286, + "grad_norm": 2.149299383163452, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7307790517807007, + "num_tokens": 320299388.0, + "step": 12377 + }, + { + "epoch": 1.3593235229518998, + "grad_norm": 1.9836540222167969, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7249656915664673, + "num_tokens": 320329246.0, + "step": 12378 + }, + { + "epoch": 1.3594333406545136, + "grad_norm": 2.1655313968658447, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7177602648735046, + "num_tokens": 320357784.0, + "step": 12379 + }, + { + "epoch": 1.3595431583571271, + "grad_norm": 2.0534350872039795, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7201339602470398, + "num_tokens": 320388559.0, + "step": 12380 + }, + { + "epoch": 1.359652976059741, + "grad_norm": 2.177297592163086, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7079054117202759, + "num_tokens": 320413902.0, + "step": 12381 + }, + { + "epoch": 1.3597627937623544, + "grad_norm": 2.6016080379486084, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7200015783309937, + "num_tokens": 320433528.0, + "step": 12382 + }, + { + "epoch": 1.3598726114649682, + "grad_norm": 2.2156190872192383, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7007067799568176, + "num_tokens": 320461910.0, + "step": 12383 + }, + { + "epoch": 1.359982429167582, + "grad_norm": 2.360792398452759, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7151412963867188, + "num_tokens": 320488243.0, + "step": 12384 + }, + { + "epoch": 1.3600922468701955, + "grad_norm": 2.3397083282470703, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7000986337661743, + "num_tokens": 320513492.0, + "step": 12385 + }, + { + "epoch": 1.360202064572809, + "grad_norm": 2.1903154850006104, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7038216590881348, + "num_tokens": 320541935.0, + "step": 12386 + }, + { + "epoch": 1.3603118822754228, + "grad_norm": 2.7069835662841797, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7495608925819397, + "num_tokens": 320559125.0, + "step": 12387 + }, + { + "epoch": 1.3604216999780365, + "grad_norm": 2.27433180809021, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7181446552276611, + "num_tokens": 320583098.0, + "step": 12388 + }, + { + "epoch": 1.36053151768065, + "grad_norm": 2.4521853923797607, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7178816199302673, + "num_tokens": 320603576.0, + "step": 12389 + }, + { + "epoch": 1.3606413353832638, + "grad_norm": 1.7868850231170654, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.6970844864845276, + "num_tokens": 320644237.0, + "step": 12390 + }, + { + "epoch": 1.3607511530858774, + "grad_norm": 2.1572625637054443, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7011646628379822, + "num_tokens": 320673042.0, + "step": 12391 + }, + { + "epoch": 1.360860970788491, + "grad_norm": 2.957064151763916, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7205085158348083, + "num_tokens": 320696520.0, + "step": 12392 + }, + { + "epoch": 1.3609707884911049, + "grad_norm": 2.3438475131988525, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7425402402877808, + "num_tokens": 320719902.0, + "step": 12393 + }, + { + "epoch": 1.3610806061937184, + "grad_norm": 1.9356460571289062, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7192064523696899, + "num_tokens": 320753425.0, + "step": 12394 + }, + { + "epoch": 1.3611904238963322, + "grad_norm": 2.1529455184936523, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7175952792167664, + "num_tokens": 320780438.0, + "step": 12395 + }, + { + "epoch": 1.3613002415989457, + "grad_norm": 2.420520067214966, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7350672483444214, + "num_tokens": 320801567.0, + "step": 12396 + }, + { + "epoch": 1.3614100593015594, + "grad_norm": 2.385309934616089, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7249847650527954, + "num_tokens": 320825226.0, + "step": 12397 + }, + { + "epoch": 1.3615198770041732, + "grad_norm": 2.468698263168335, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7376481294631958, + "num_tokens": 320845317.0, + "step": 12398 + }, + { + "epoch": 1.3616296947067867, + "grad_norm": 2.769294500350952, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7277200818061829, + "num_tokens": 320863226.0, + "step": 12399 + }, + { + "epoch": 1.3617395124094003, + "grad_norm": 2.584095001220703, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7329574227333069, + "num_tokens": 320884355.0, + "step": 12400 + }, + { + "epoch": 1.361849330112014, + "grad_norm": 2.366696357727051, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7168118357658386, + "num_tokens": 320909532.0, + "step": 12401 + }, + { + "epoch": 1.3619591478146278, + "grad_norm": 2.285203695297241, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7111645936965942, + "num_tokens": 320937048.0, + "step": 12402 + }, + { + "epoch": 1.3620689655172413, + "grad_norm": 2.269672155380249, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7208367586135864, + "num_tokens": 320960287.0, + "step": 12403 + }, + { + "epoch": 1.362178783219855, + "grad_norm": 2.6983745098114014, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7078854441642761, + "num_tokens": 320979821.0, + "step": 12404 + }, + { + "epoch": 1.3622886009224686, + "grad_norm": 2.438014030456543, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7033568620681763, + "num_tokens": 321002289.0, + "step": 12405 + }, + { + "epoch": 1.3623984186250824, + "grad_norm": 2.1239519119262695, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.6987086534500122, + "num_tokens": 321031301.0, + "step": 12406 + }, + { + "epoch": 1.3625082363276961, + "grad_norm": 2.416644334793091, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7080515623092651, + "num_tokens": 321054918.0, + "step": 12407 + }, + { + "epoch": 1.3626180540303097, + "grad_norm": 2.1497802734375, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7299349308013916, + "num_tokens": 321081261.0, + "step": 12408 + }, + { + "epoch": 1.3627278717329234, + "grad_norm": 2.1674234867095947, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7328873872756958, + "num_tokens": 321109810.0, + "step": 12409 + }, + { + "epoch": 1.362837689435537, + "grad_norm": 2.4336297512054443, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7223904132843018, + "num_tokens": 321132982.0, + "step": 12410 + }, + { + "epoch": 1.3629475071381507, + "grad_norm": 1.9917402267456055, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7247834801673889, + "num_tokens": 321162779.0, + "step": 12411 + }, + { + "epoch": 1.3630573248407643, + "grad_norm": 2.5613348484039307, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.755446195602417, + "num_tokens": 321181944.0, + "step": 12412 + }, + { + "epoch": 1.363167142543378, + "grad_norm": 2.0680174827575684, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7079266309738159, + "num_tokens": 321210380.0, + "step": 12413 + }, + { + "epoch": 1.3632769602459915, + "grad_norm": 2.227818727493286, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7182202339172363, + "num_tokens": 321235617.0, + "step": 12414 + }, + { + "epoch": 1.3633867779486053, + "grad_norm": 2.306255578994751, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.722719669342041, + "num_tokens": 321260493.0, + "step": 12415 + }, + { + "epoch": 1.363496595651219, + "grad_norm": 2.4467599391937256, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7391161918640137, + "num_tokens": 321282094.0, + "step": 12416 + }, + { + "epoch": 1.3636064133538326, + "grad_norm": 1.9898909330368042, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7112380266189575, + "num_tokens": 321313841.0, + "step": 12417 + }, + { + "epoch": 1.3637162310564463, + "grad_norm": 2.422028064727783, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7222652435302734, + "num_tokens": 321336344.0, + "step": 12418 + }, + { + "epoch": 1.3638260487590599, + "grad_norm": 2.012420177459717, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7029949426651001, + "num_tokens": 321369387.0, + "step": 12419 + }, + { + "epoch": 1.3639358664616736, + "grad_norm": 2.274778366088867, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7282487154006958, + "num_tokens": 321395461.0, + "step": 12420 + }, + { + "epoch": 1.3640456841642874, + "grad_norm": 2.198209762573242, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7325259447097778, + "num_tokens": 321420665.0, + "step": 12421 + }, + { + "epoch": 1.364155501866901, + "grad_norm": 2.20141339302063, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7282352447509766, + "num_tokens": 321446602.0, + "step": 12422 + }, + { + "epoch": 1.3642653195695145, + "grad_norm": 2.243823289871216, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7097471952438354, + "num_tokens": 321472037.0, + "step": 12423 + }, + { + "epoch": 1.3643751372721282, + "grad_norm": 2.1445677280426025, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7205876111984253, + "num_tokens": 321498650.0, + "step": 12424 + }, + { + "epoch": 1.364484954974742, + "grad_norm": 2.2377002239227295, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.7013758420944214, + "num_tokens": 321524056.0, + "step": 12425 + }, + { + "epoch": 1.3645947726773555, + "grad_norm": 2.3622353076934814, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7007531523704529, + "num_tokens": 321548443.0, + "step": 12426 + }, + { + "epoch": 1.3647045903799693, + "grad_norm": 2.1246581077575684, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.704944908618927, + "num_tokens": 321578386.0, + "step": 12427 + }, + { + "epoch": 1.3648144080825828, + "grad_norm": 2.0357723236083984, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7202070355415344, + "num_tokens": 321607798.0, + "step": 12428 + }, + { + "epoch": 1.3649242257851966, + "grad_norm": 2.0620906352996826, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6905096769332886, + "num_tokens": 321637720.0, + "step": 12429 + }, + { + "epoch": 1.3650340434878103, + "grad_norm": 2.2785415649414062, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.6997594833374023, + "num_tokens": 321665072.0, + "step": 12430 + }, + { + "epoch": 1.3651438611904239, + "grad_norm": 2.2032177448272705, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.730857789516449, + "num_tokens": 321690054.0, + "step": 12431 + }, + { + "epoch": 1.3652536788930376, + "grad_norm": 1.9497581720352173, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7296497225761414, + "num_tokens": 321722282.0, + "step": 12432 + }, + { + "epoch": 1.3653634965956511, + "grad_norm": 2.033360719680786, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6974444389343262, + "num_tokens": 321754184.0, + "step": 12433 + }, + { + "epoch": 1.365473314298265, + "grad_norm": 2.546895742416382, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7622629404067993, + "num_tokens": 321774303.0, + "step": 12434 + }, + { + "epoch": 1.3655831320008787, + "grad_norm": 2.2008821964263916, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7197800874710083, + "num_tokens": 321799656.0, + "step": 12435 + }, + { + "epoch": 1.3656929497034922, + "grad_norm": 2.7719552516937256, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.718427836894989, + "num_tokens": 321818801.0, + "step": 12436 + }, + { + "epoch": 1.3658027674061057, + "grad_norm": 2.256575107574463, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7069832682609558, + "num_tokens": 321844992.0, + "step": 12437 + }, + { + "epoch": 1.3659125851087195, + "grad_norm": 2.504014253616333, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7213349342346191, + "num_tokens": 321866970.0, + "step": 12438 + }, + { + "epoch": 1.3660224028113332, + "grad_norm": 2.2387990951538086, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7536547183990479, + "num_tokens": 321891675.0, + "step": 12439 + }, + { + "epoch": 1.3661322205139468, + "grad_norm": 1.8497464656829834, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7231771945953369, + "num_tokens": 321926398.0, + "step": 12440 + }, + { + "epoch": 1.3662420382165605, + "grad_norm": 2.295231580734253, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7334665656089783, + "num_tokens": 321950293.0, + "step": 12441 + }, + { + "epoch": 1.366351855919174, + "grad_norm": 2.1308557987213135, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.724145770072937, + "num_tokens": 321979849.0, + "step": 12442 + }, + { + "epoch": 1.3664616736217878, + "grad_norm": 2.5239126682281494, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7220624685287476, + "num_tokens": 322001928.0, + "step": 12443 + }, + { + "epoch": 1.3665714913244016, + "grad_norm": 2.2089524269104004, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7256446480751038, + "num_tokens": 322029905.0, + "step": 12444 + }, + { + "epoch": 1.3666813090270151, + "grad_norm": 2.280766248703003, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7115400433540344, + "num_tokens": 322055755.0, + "step": 12445 + }, + { + "epoch": 1.3667911267296289, + "grad_norm": 2.3099019527435303, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6986587643623352, + "num_tokens": 322081185.0, + "step": 12446 + }, + { + "epoch": 1.3669009444322424, + "grad_norm": 2.362696647644043, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.712868332862854, + "num_tokens": 322107051.0, + "step": 12447 + }, + { + "epoch": 1.3670107621348562, + "grad_norm": 2.3759727478027344, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7354551553726196, + "num_tokens": 322130687.0, + "step": 12448 + }, + { + "epoch": 1.36712057983747, + "grad_norm": 1.9979273080825806, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6780793070793152, + "num_tokens": 322164725.0, + "step": 12449 + }, + { + "epoch": 1.3672303975400835, + "grad_norm": 1.9769959449768066, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7206152677536011, + "num_tokens": 322193166.0, + "step": 12450 + }, + { + "epoch": 1.367340215242697, + "grad_norm": 2.43436336517334, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7477531433105469, + "num_tokens": 322212005.0, + "step": 12451 + }, + { + "epoch": 1.3674500329453108, + "grad_norm": 2.248054265975952, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7009493708610535, + "num_tokens": 322239553.0, + "step": 12452 + }, + { + "epoch": 1.3675598506479245, + "grad_norm": 2.089665174484253, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.720477819442749, + "num_tokens": 322269202.0, + "step": 12453 + }, + { + "epoch": 1.367669668350538, + "grad_norm": 2.2099223136901855, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7035201787948608, + "num_tokens": 322297129.0, + "step": 12454 + }, + { + "epoch": 1.3677794860531518, + "grad_norm": 1.9610800743103027, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7036267518997192, + "num_tokens": 322331302.0, + "step": 12455 + }, + { + "epoch": 1.3678893037557653, + "grad_norm": 2.2943217754364014, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7156053185462952, + "num_tokens": 322356109.0, + "step": 12456 + }, + { + "epoch": 1.367999121458379, + "grad_norm": 2.2332372665405273, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6997095942497253, + "num_tokens": 322382372.0, + "step": 12457 + }, + { + "epoch": 1.3681089391609929, + "grad_norm": 2.092759370803833, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7385431528091431, + "num_tokens": 322408429.0, + "step": 12458 + }, + { + "epoch": 1.3682187568636064, + "grad_norm": 2.255854368209839, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7248396277427673, + "num_tokens": 322433048.0, + "step": 12459 + }, + { + "epoch": 1.3683285745662201, + "grad_norm": 2.2002813816070557, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.729441225528717, + "num_tokens": 322458545.0, + "step": 12460 + }, + { + "epoch": 1.3684383922688337, + "grad_norm": 2.774575710296631, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7405377626419067, + "num_tokens": 322478183.0, + "step": 12461 + }, + { + "epoch": 1.3685482099714474, + "grad_norm": 2.47959566116333, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7159926295280457, + "num_tokens": 322499421.0, + "step": 12462 + }, + { + "epoch": 1.3686580276740612, + "grad_norm": 2.0961601734161377, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.723029613494873, + "num_tokens": 322528571.0, + "step": 12463 + }, + { + "epoch": 1.3687678453766747, + "grad_norm": 2.399676561355591, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7066434621810913, + "num_tokens": 322552969.0, + "step": 12464 + }, + { + "epoch": 1.3688776630792883, + "grad_norm": 2.4436638355255127, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7188382148742676, + "num_tokens": 322576081.0, + "step": 12465 + }, + { + "epoch": 1.368987480781902, + "grad_norm": 2.1931159496307373, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7206668853759766, + "num_tokens": 322605259.0, + "step": 12466 + }, + { + "epoch": 1.3690972984845158, + "grad_norm": 2.179623603820801, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.729025661945343, + "num_tokens": 322632366.0, + "step": 12467 + }, + { + "epoch": 1.3692071161871293, + "grad_norm": 2.3140063285827637, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6859481334686279, + "num_tokens": 322658014.0, + "step": 12468 + }, + { + "epoch": 1.369316933889743, + "grad_norm": 2.2317757606506348, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.718472421169281, + "num_tokens": 322685064.0, + "step": 12469 + }, + { + "epoch": 1.3694267515923566, + "grad_norm": 2.217028856277466, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7302504777908325, + "num_tokens": 322711501.0, + "step": 12470 + }, + { + "epoch": 1.3695365692949704, + "grad_norm": 2.2019832134246826, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7349805235862732, + "num_tokens": 322738897.0, + "step": 12471 + }, + { + "epoch": 1.3696463869975841, + "grad_norm": 2.340341329574585, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7250226736068726, + "num_tokens": 322762890.0, + "step": 12472 + }, + { + "epoch": 1.3697562047001977, + "grad_norm": 2.2615954875946045, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.706203281879425, + "num_tokens": 322788541.0, + "step": 12473 + }, + { + "epoch": 1.3698660224028114, + "grad_norm": 2.301485061645508, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7148581743240356, + "num_tokens": 322812278.0, + "step": 12474 + }, + { + "epoch": 1.369975840105425, + "grad_norm": 2.0793988704681396, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6876161098480225, + "num_tokens": 322844266.0, + "step": 12475 + }, + { + "epoch": 1.3700856578080387, + "grad_norm": 2.361701011657715, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6990017890930176, + "num_tokens": 322867518.0, + "step": 12476 + }, + { + "epoch": 1.3701954755106522, + "grad_norm": 2.4081859588623047, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7236432433128357, + "num_tokens": 322891856.0, + "step": 12477 + }, + { + "epoch": 1.370305293213266, + "grad_norm": 2.34063720703125, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6978373527526855, + "num_tokens": 322916582.0, + "step": 12478 + }, + { + "epoch": 1.3704151109158795, + "grad_norm": 2.6011340618133545, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7177122831344604, + "num_tokens": 322938281.0, + "step": 12479 + }, + { + "epoch": 1.3705249286184933, + "grad_norm": 2.3417444229125977, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7197816967964172, + "num_tokens": 322961516.0, + "step": 12480 + }, + { + "epoch": 1.370634746321107, + "grad_norm": 2.738264560699463, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7505494952201843, + "num_tokens": 322979819.0, + "step": 12481 + }, + { + "epoch": 1.3707445640237206, + "grad_norm": 1.8959345817565918, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7103936672210693, + "num_tokens": 323013915.0, + "step": 12482 + }, + { + "epoch": 1.3708543817263343, + "grad_norm": 2.384111166000366, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7396727800369263, + "num_tokens": 323036826.0, + "step": 12483 + }, + { + "epoch": 1.3709641994289479, + "grad_norm": 2.0345618724823, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7094597816467285, + "num_tokens": 323067041.0, + "step": 12484 + }, + { + "epoch": 1.3710740171315616, + "grad_norm": 2.3664543628692627, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.718474268913269, + "num_tokens": 323090266.0, + "step": 12485 + }, + { + "epoch": 1.3711838348341754, + "grad_norm": 2.355145215988159, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7031699419021606, + "num_tokens": 323116084.0, + "step": 12486 + }, + { + "epoch": 1.371293652536789, + "grad_norm": 2.0777175426483154, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7212753295898438, + "num_tokens": 323145061.0, + "step": 12487 + }, + { + "epoch": 1.3714034702394025, + "grad_norm": 2.4211294651031494, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7232770919799805, + "num_tokens": 323167555.0, + "step": 12488 + }, + { + "epoch": 1.3715132879420162, + "grad_norm": 2.4589972496032715, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7306731343269348, + "num_tokens": 323189204.0, + "step": 12489 + }, + { + "epoch": 1.37162310564463, + "grad_norm": 2.3440825939178467, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7287071943283081, + "num_tokens": 323212801.0, + "step": 12490 + }, + { + "epoch": 1.3717329233472435, + "grad_norm": 2.011786460876465, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7312748432159424, + "num_tokens": 323239307.0, + "step": 12491 + }, + { + "epoch": 1.3718427410498573, + "grad_norm": 2.3107526302337646, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7106849551200867, + "num_tokens": 323264180.0, + "step": 12492 + }, + { + "epoch": 1.3719525587524708, + "grad_norm": 2.14878511428833, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.705015242099762, + "num_tokens": 323292103.0, + "step": 12493 + }, + { + "epoch": 1.3720623764550846, + "grad_norm": 2.315641403198242, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7164021730422974, + "num_tokens": 323315127.0, + "step": 12494 + }, + { + "epoch": 1.3721721941576983, + "grad_norm": 2.308809280395508, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6975146532058716, + "num_tokens": 323340495.0, + "step": 12495 + }, + { + "epoch": 1.3722820118603118, + "grad_norm": 2.3549697399139404, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.695382833480835, + "num_tokens": 323366720.0, + "step": 12496 + }, + { + "epoch": 1.3723918295629256, + "grad_norm": 2.356091260910034, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7286098599433899, + "num_tokens": 323390323.0, + "step": 12497 + }, + { + "epoch": 1.3725016472655391, + "grad_norm": 2.356435537338257, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7294799089431763, + "num_tokens": 323412899.0, + "step": 12498 + }, + { + "epoch": 1.372611464968153, + "grad_norm": 2.8911056518554688, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.733225405216217, + "num_tokens": 323430627.0, + "step": 12499 + }, + { + "epoch": 1.3727212826707667, + "grad_norm": 2.5899648666381836, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7347029447555542, + "num_tokens": 323451115.0, + "step": 12500 + }, + { + "epoch": 1.3728311003733802, + "grad_norm": 2.289740800857544, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7131076455116272, + "num_tokens": 323477932.0, + "step": 12501 + }, + { + "epoch": 1.3729409180759937, + "grad_norm": 2.039592981338501, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7115333676338196, + "num_tokens": 323507554.0, + "step": 12502 + }, + { + "epoch": 1.3730507357786075, + "grad_norm": 2.1827807426452637, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7054228186607361, + "num_tokens": 323536096.0, + "step": 12503 + }, + { + "epoch": 1.3731605534812212, + "grad_norm": 2.062486171722412, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6861238479614258, + "num_tokens": 323568211.0, + "step": 12504 + }, + { + "epoch": 1.3732703711838348, + "grad_norm": 2.1220946311950684, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7036200761795044, + "num_tokens": 323597033.0, + "step": 12505 + }, + { + "epoch": 1.3733801888864485, + "grad_norm": 2.1851556301116943, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7353500127792358, + "num_tokens": 323623987.0, + "step": 12506 + }, + { + "epoch": 1.373490006589062, + "grad_norm": 2.1005759239196777, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7206900119781494, + "num_tokens": 323651896.0, + "step": 12507 + }, + { + "epoch": 1.3735998242916758, + "grad_norm": 2.1788058280944824, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7028648853302002, + "num_tokens": 323679655.0, + "step": 12508 + }, + { + "epoch": 1.3737096419942896, + "grad_norm": 2.2766592502593994, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7122960686683655, + "num_tokens": 323704418.0, + "step": 12509 + }, + { + "epoch": 1.3738194596969031, + "grad_norm": 2.5534980297088623, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7123727202415466, + "num_tokens": 323727290.0, + "step": 12510 + }, + { + "epoch": 1.3739292773995169, + "grad_norm": 2.1390469074249268, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7005859017372131, + "num_tokens": 323756261.0, + "step": 12511 + }, + { + "epoch": 1.3740390951021304, + "grad_norm": 1.9977469444274902, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7023845911026001, + "num_tokens": 323785724.0, + "step": 12512 + }, + { + "epoch": 1.3741489128047442, + "grad_norm": 2.170656442642212, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6996471881866455, + "num_tokens": 323812965.0, + "step": 12513 + }, + { + "epoch": 1.374258730507358, + "grad_norm": 2.193890333175659, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7392042875289917, + "num_tokens": 323838244.0, + "step": 12514 + }, + { + "epoch": 1.3743685482099715, + "grad_norm": 2.470778226852417, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7468478083610535, + "num_tokens": 323858223.0, + "step": 12515 + }, + { + "epoch": 1.374478365912585, + "grad_norm": 1.97995924949646, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7078111171722412, + "num_tokens": 323890650.0, + "step": 12516 + }, + { + "epoch": 1.3745881836151987, + "grad_norm": 2.187351942062378, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7185915112495422, + "num_tokens": 323915138.0, + "step": 12517 + }, + { + "epoch": 1.3746980013178125, + "grad_norm": 2.403900146484375, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7011716365814209, + "num_tokens": 323939118.0, + "step": 12518 + }, + { + "epoch": 1.374807819020426, + "grad_norm": 1.997517704963684, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7161177396774292, + "num_tokens": 323967959.0, + "step": 12519 + }, + { + "epoch": 1.3749176367230398, + "grad_norm": 2.4579949378967285, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7215335369110107, + "num_tokens": 323989323.0, + "step": 12520 + }, + { + "epoch": 1.3750274544256533, + "grad_norm": 2.4786012172698975, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7285149693489075, + "num_tokens": 324010929.0, + "step": 12521 + }, + { + "epoch": 1.375137272128267, + "grad_norm": 2.650254249572754, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7482723593711853, + "num_tokens": 324029764.0, + "step": 12522 + }, + { + "epoch": 1.3752470898308808, + "grad_norm": 2.4525070190429688, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7377490401268005, + "num_tokens": 324051254.0, + "step": 12523 + }, + { + "epoch": 1.3753569075334944, + "grad_norm": 2.4631967544555664, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.724547803401947, + "num_tokens": 324073161.0, + "step": 12524 + }, + { + "epoch": 1.3754667252361081, + "grad_norm": 2.2628543376922607, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.723182201385498, + "num_tokens": 324096249.0, + "step": 12525 + }, + { + "epoch": 1.3755765429387217, + "grad_norm": 2.3708784580230713, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7158941626548767, + "num_tokens": 324119618.0, + "step": 12526 + }, + { + "epoch": 1.3756863606413354, + "grad_norm": 2.130723476409912, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7168000936508179, + "num_tokens": 324149285.0, + "step": 12527 + }, + { + "epoch": 1.3757961783439492, + "grad_norm": 2.5374863147735596, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7182115316390991, + "num_tokens": 324170406.0, + "step": 12528 + }, + { + "epoch": 1.3759059960465627, + "grad_norm": 2.2332606315612793, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7013906240463257, + "num_tokens": 324198163.0, + "step": 12529 + }, + { + "epoch": 1.3760158137491763, + "grad_norm": 2.1732187271118164, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7153229117393494, + "num_tokens": 324226223.0, + "step": 12530 + }, + { + "epoch": 1.37612563145179, + "grad_norm": 2.2085299491882324, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7050585746765137, + "num_tokens": 324252547.0, + "step": 12531 + }, + { + "epoch": 1.3762354491544038, + "grad_norm": 2.8695106506347656, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7296687960624695, + "num_tokens": 324269247.0, + "step": 12532 + }, + { + "epoch": 1.3763452668570173, + "grad_norm": 2.163778066635132, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7082551717758179, + "num_tokens": 324294598.0, + "step": 12533 + }, + { + "epoch": 1.376455084559631, + "grad_norm": 2.162240505218506, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7081674337387085, + "num_tokens": 324322960.0, + "step": 12534 + }, + { + "epoch": 1.3765649022622446, + "grad_norm": 2.3929502964019775, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7467685341835022, + "num_tokens": 324345051.0, + "step": 12535 + }, + { + "epoch": 1.3766747199648584, + "grad_norm": 2.273902416229248, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7053048014640808, + "num_tokens": 324370134.0, + "step": 12536 + }, + { + "epoch": 1.376784537667472, + "grad_norm": 2.1630146503448486, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6938204765319824, + "num_tokens": 324396827.0, + "step": 12537 + }, + { + "epoch": 1.3768943553700856, + "grad_norm": 2.3990516662597656, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7125571966171265, + "num_tokens": 324419711.0, + "step": 12538 + }, + { + "epoch": 1.3770041730726994, + "grad_norm": 2.3570220470428467, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7171318531036377, + "num_tokens": 324442608.0, + "step": 12539 + }, + { + "epoch": 1.377113990775313, + "grad_norm": 2.422056198120117, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7204127907752991, + "num_tokens": 324463862.0, + "step": 12540 + }, + { + "epoch": 1.3772238084779267, + "grad_norm": 2.446462631225586, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7246754169464111, + "num_tokens": 324485779.0, + "step": 12541 + }, + { + "epoch": 1.3773336261805402, + "grad_norm": 2.2110936641693115, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6999916434288025, + "num_tokens": 324513413.0, + "step": 12542 + }, + { + "epoch": 1.377443443883154, + "grad_norm": 1.9637311697006226, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7271209359169006, + "num_tokens": 324543901.0, + "step": 12543 + }, + { + "epoch": 1.3775532615857675, + "grad_norm": 2.4558615684509277, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7071851491928101, + "num_tokens": 324566134.0, + "step": 12544 + }, + { + "epoch": 1.3776630792883813, + "grad_norm": 2.1454858779907227, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6935621500015259, + "num_tokens": 324594770.0, + "step": 12545 + }, + { + "epoch": 1.377772896990995, + "grad_norm": 2.2112278938293457, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7096945643424988, + "num_tokens": 324619443.0, + "step": 12546 + }, + { + "epoch": 1.3778827146936086, + "grad_norm": 2.558415174484253, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7259758710861206, + "num_tokens": 324639903.0, + "step": 12547 + }, + { + "epoch": 1.3779925323962223, + "grad_norm": 2.3264312744140625, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7307440042495728, + "num_tokens": 324661797.0, + "step": 12548 + }, + { + "epoch": 1.3781023500988359, + "grad_norm": 2.16473650932312, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7373559474945068, + "num_tokens": 324686076.0, + "step": 12549 + }, + { + "epoch": 1.3782121678014496, + "grad_norm": 2.3065600395202637, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7379882335662842, + "num_tokens": 324709315.0, + "step": 12550 + }, + { + "epoch": 1.3783219855040634, + "grad_norm": 2.234945297241211, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7318472862243652, + "num_tokens": 324733142.0, + "step": 12551 + }, + { + "epoch": 1.378431803206677, + "grad_norm": 2.3103442192077637, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7394490242004395, + "num_tokens": 324756808.0, + "step": 12552 + }, + { + "epoch": 1.3785416209092904, + "grad_norm": 2.621798515319824, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7165539264678955, + "num_tokens": 324777993.0, + "step": 12553 + }, + { + "epoch": 1.3786514386119042, + "grad_norm": 2.0540106296539307, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7436447739601135, + "num_tokens": 324806322.0, + "step": 12554 + }, + { + "epoch": 1.378761256314518, + "grad_norm": 1.995446801185608, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7022942304611206, + "num_tokens": 324837557.0, + "step": 12555 + }, + { + "epoch": 1.3788710740171315, + "grad_norm": 2.61069917678833, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7034047842025757, + "num_tokens": 324857980.0, + "step": 12556 + }, + { + "epoch": 1.3789808917197452, + "grad_norm": 2.2418830394744873, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7084617018699646, + "num_tokens": 324887487.0, + "step": 12557 + }, + { + "epoch": 1.3790907094223588, + "grad_norm": 2.448190689086914, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.744933009147644, + "num_tokens": 324908476.0, + "step": 12558 + }, + { + "epoch": 1.3792005271249725, + "grad_norm": 2.3599681854248047, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7253377437591553, + "num_tokens": 324934794.0, + "step": 12559 + }, + { + "epoch": 1.3793103448275863, + "grad_norm": 2.0259361267089844, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7362572550773621, + "num_tokens": 324965560.0, + "step": 12560 + }, + { + "epoch": 1.3794201625301998, + "grad_norm": 2.2011380195617676, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7182588577270508, + "num_tokens": 324992511.0, + "step": 12561 + }, + { + "epoch": 1.3795299802328136, + "grad_norm": 2.212496280670166, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7266980409622192, + "num_tokens": 325019487.0, + "step": 12562 + }, + { + "epoch": 1.3796397979354271, + "grad_norm": 2.34382700920105, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7202188968658447, + "num_tokens": 325043901.0, + "step": 12563 + }, + { + "epoch": 1.3797496156380409, + "grad_norm": 2.2151644229888916, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7293732762336731, + "num_tokens": 325068013.0, + "step": 12564 + }, + { + "epoch": 1.3798594333406546, + "grad_norm": 2.2953319549560547, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7082908153533936, + "num_tokens": 325094329.0, + "step": 12565 + }, + { + "epoch": 1.3799692510432682, + "grad_norm": 2.094421148300171, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7226617336273193, + "num_tokens": 325121449.0, + "step": 12566 + }, + { + "epoch": 1.3800790687458817, + "grad_norm": 2.156275987625122, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7240590453147888, + "num_tokens": 325146118.0, + "step": 12567 + }, + { + "epoch": 1.3801888864484955, + "grad_norm": 2.0357284545898438, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7104772329330444, + "num_tokens": 325177571.0, + "step": 12568 + }, + { + "epoch": 1.3802987041511092, + "grad_norm": 2.077744245529175, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.695111095905304, + "num_tokens": 325207045.0, + "step": 12569 + }, + { + "epoch": 1.3804085218537228, + "grad_norm": 2.204822301864624, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7152653932571411, + "num_tokens": 325233060.0, + "step": 12570 + }, + { + "epoch": 1.3805183395563365, + "grad_norm": 2.353079080581665, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7329174876213074, + "num_tokens": 325255279.0, + "step": 12571 + }, + { + "epoch": 1.38062815725895, + "grad_norm": 2.348616600036621, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7162554860115051, + "num_tokens": 325279494.0, + "step": 12572 + }, + { + "epoch": 1.3807379749615638, + "grad_norm": 2.143779993057251, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7506973743438721, + "num_tokens": 325303432.0, + "step": 12573 + }, + { + "epoch": 1.3808477926641776, + "grad_norm": 2.2583487033843994, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7019862532615662, + "num_tokens": 325329242.0, + "step": 12574 + }, + { + "epoch": 1.380957610366791, + "grad_norm": 2.190286159515381, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7130876183509827, + "num_tokens": 325355446.0, + "step": 12575 + }, + { + "epoch": 1.3810674280694049, + "grad_norm": 2.110705852508545, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7124265432357788, + "num_tokens": 325382954.0, + "step": 12576 + }, + { + "epoch": 1.3811772457720184, + "grad_norm": 2.2554547786712646, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7052994966506958, + "num_tokens": 325407603.0, + "step": 12577 + }, + { + "epoch": 1.3812870634746321, + "grad_norm": 2.1373445987701416, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7015776634216309, + "num_tokens": 325434185.0, + "step": 12578 + }, + { + "epoch": 1.381396881177246, + "grad_norm": 2.1124107837677, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7209721207618713, + "num_tokens": 325461343.0, + "step": 12579 + }, + { + "epoch": 1.3815066988798594, + "grad_norm": 2.43206524848938, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7107706665992737, + "num_tokens": 325483556.0, + "step": 12580 + }, + { + "epoch": 1.381616516582473, + "grad_norm": 2.2484450340270996, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7107683420181274, + "num_tokens": 325510162.0, + "step": 12581 + }, + { + "epoch": 1.3817263342850867, + "grad_norm": 2.3418149948120117, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7504165172576904, + "num_tokens": 325533463.0, + "step": 12582 + }, + { + "epoch": 1.3818361519877005, + "grad_norm": 2.22820782661438, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7462579011917114, + "num_tokens": 325558235.0, + "step": 12583 + }, + { + "epoch": 1.381945969690314, + "grad_norm": 2.443734645843506, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7246440649032593, + "num_tokens": 325579757.0, + "step": 12584 + }, + { + "epoch": 1.3820557873929278, + "grad_norm": 2.2659313678741455, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6993182897567749, + "num_tokens": 325605573.0, + "step": 12585 + }, + { + "epoch": 1.3821656050955413, + "grad_norm": 2.3298587799072266, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.717540979385376, + "num_tokens": 325629148.0, + "step": 12586 + }, + { + "epoch": 1.382275422798155, + "grad_norm": 2.1904091835021973, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7167043685913086, + "num_tokens": 325655848.0, + "step": 12587 + }, + { + "epoch": 1.3823852405007688, + "grad_norm": 2.3732733726501465, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7220605611801147, + "num_tokens": 325679168.0, + "step": 12588 + }, + { + "epoch": 1.3824950582033824, + "grad_norm": 2.0712220668792725, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6998545527458191, + "num_tokens": 325711489.0, + "step": 12589 + }, + { + "epoch": 1.3826048759059961, + "grad_norm": 2.211418867111206, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6898159980773926, + "num_tokens": 325740922.0, + "step": 12590 + }, + { + "epoch": 1.3827146936086097, + "grad_norm": 2.074784755706787, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7055188417434692, + "num_tokens": 325770207.0, + "step": 12591 + }, + { + "epoch": 1.3828245113112234, + "grad_norm": 2.041522264480591, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7431348562240601, + "num_tokens": 325798524.0, + "step": 12592 + }, + { + "epoch": 1.382934329013837, + "grad_norm": 2.2666070461273193, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6998622417449951, + "num_tokens": 325825687.0, + "step": 12593 + }, + { + "epoch": 1.3830441467164507, + "grad_norm": 2.136136770248413, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7173268795013428, + "num_tokens": 325854967.0, + "step": 12594 + }, + { + "epoch": 1.3831539644190642, + "grad_norm": 1.9783294200897217, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.694709062576294, + "num_tokens": 325887490.0, + "step": 12595 + }, + { + "epoch": 1.383263782121678, + "grad_norm": 2.591012954711914, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7053040862083435, + "num_tokens": 325908570.0, + "step": 12596 + }, + { + "epoch": 1.3833735998242918, + "grad_norm": 2.571112871170044, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7228930592536926, + "num_tokens": 325928528.0, + "step": 12597 + }, + { + "epoch": 1.3834834175269053, + "grad_norm": 2.182588577270508, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7070091962814331, + "num_tokens": 325956713.0, + "step": 12598 + }, + { + "epoch": 1.383593235229519, + "grad_norm": 2.1318795680999756, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7151392698287964, + "num_tokens": 325984839.0, + "step": 12599 + }, + { + "epoch": 1.3837030529321326, + "grad_norm": 2.1004533767700195, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7292885780334473, + "num_tokens": 326011963.0, + "step": 12600 + }, + { + "epoch": 1.3838128706347463, + "grad_norm": 2.2649996280670166, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.725888729095459, + "num_tokens": 326037627.0, + "step": 12601 + }, + { + "epoch": 1.38392268833736, + "grad_norm": 1.930673599243164, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7166520357131958, + "num_tokens": 326068689.0, + "step": 12602 + }, + { + "epoch": 1.3840325060399736, + "grad_norm": 2.130354166030884, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.711498498916626, + "num_tokens": 326099103.0, + "step": 12603 + }, + { + "epoch": 1.3841423237425874, + "grad_norm": 2.260530710220337, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6887024641036987, + "num_tokens": 326125592.0, + "step": 12604 + }, + { + "epoch": 1.384252141445201, + "grad_norm": 2.087655544281006, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7185755372047424, + "num_tokens": 326154108.0, + "step": 12605 + }, + { + "epoch": 1.3843619591478147, + "grad_norm": 2.331138849258423, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7256138920783997, + "num_tokens": 326178614.0, + "step": 12606 + }, + { + "epoch": 1.3844717768504282, + "grad_norm": 2.1361300945281982, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7030231952667236, + "num_tokens": 326205864.0, + "step": 12607 + }, + { + "epoch": 1.384581594553042, + "grad_norm": 2.390363931655884, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7102969884872437, + "num_tokens": 326229546.0, + "step": 12608 + }, + { + "epoch": 1.3846914122556555, + "grad_norm": 2.0866897106170654, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7045354843139648, + "num_tokens": 326257447.0, + "step": 12609 + }, + { + "epoch": 1.3848012299582693, + "grad_norm": 2.3447206020355225, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7251850962638855, + "num_tokens": 326279636.0, + "step": 12610 + }, + { + "epoch": 1.384911047660883, + "grad_norm": 2.1366703510284424, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7047972679138184, + "num_tokens": 326307847.0, + "step": 12611 + }, + { + "epoch": 1.3850208653634966, + "grad_norm": 2.330293893814087, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7110209465026855, + "num_tokens": 326334739.0, + "step": 12612 + }, + { + "epoch": 1.3851306830661103, + "grad_norm": 2.634904146194458, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7429947853088379, + "num_tokens": 326354685.0, + "step": 12613 + }, + { + "epoch": 1.3852405007687238, + "grad_norm": 2.119842529296875, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7002082467079163, + "num_tokens": 326384519.0, + "step": 12614 + }, + { + "epoch": 1.3853503184713376, + "grad_norm": 2.4462332725524902, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7394335269927979, + "num_tokens": 326407163.0, + "step": 12615 + }, + { + "epoch": 1.3854601361739514, + "grad_norm": 2.2222185134887695, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7125210762023926, + "num_tokens": 326432569.0, + "step": 12616 + }, + { + "epoch": 1.385569953876565, + "grad_norm": 2.1259491443634033, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7007931470870972, + "num_tokens": 326462060.0, + "step": 12617 + }, + { + "epoch": 1.3856797715791784, + "grad_norm": 2.259859085083008, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7324459552764893, + "num_tokens": 326487727.0, + "step": 12618 + }, + { + "epoch": 1.3857895892817922, + "grad_norm": 2.43076753616333, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7332452535629272, + "num_tokens": 326510858.0, + "step": 12619 + }, + { + "epoch": 1.385899406984406, + "grad_norm": 2.398076057434082, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7293089628219604, + "num_tokens": 326532694.0, + "step": 12620 + }, + { + "epoch": 1.3860092246870195, + "grad_norm": 2.1753628253936768, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7286819219589233, + "num_tokens": 326560597.0, + "step": 12621 + }, + { + "epoch": 1.3861190423896332, + "grad_norm": 2.515075445175171, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6910641193389893, + "num_tokens": 326582964.0, + "step": 12622 + }, + { + "epoch": 1.3862288600922468, + "grad_norm": 2.188469886779785, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.7021292448043823, + "num_tokens": 326609534.0, + "step": 12623 + }, + { + "epoch": 1.3863386777948605, + "grad_norm": 2.2716000080108643, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7351421117782593, + "num_tokens": 326634252.0, + "step": 12624 + }, + { + "epoch": 1.3864484954974743, + "grad_norm": 2.3093082904815674, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7335895895957947, + "num_tokens": 326658361.0, + "step": 12625 + }, + { + "epoch": 1.3865583132000878, + "grad_norm": 2.5375964641571045, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7359854578971863, + "num_tokens": 326678597.0, + "step": 12626 + }, + { + "epoch": 1.3866681309027016, + "grad_norm": 2.1485061645507812, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7070110440254211, + "num_tokens": 326707463.0, + "step": 12627 + }, + { + "epoch": 1.3867779486053151, + "grad_norm": 2.0203146934509277, + "learning_rate": 1e-06, + "loss": 1.0328, + "mean_token_accuracy": 0.6858212947845459, + "num_tokens": 326740808.0, + "step": 12628 + }, + { + "epoch": 1.3868877663079289, + "grad_norm": 2.0249321460723877, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7112560868263245, + "num_tokens": 326770351.0, + "step": 12629 + }, + { + "epoch": 1.3869975840105426, + "grad_norm": 2.191795825958252, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7277331948280334, + "num_tokens": 326795628.0, + "step": 12630 + }, + { + "epoch": 1.3871074017131562, + "grad_norm": 2.006326198577881, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7107236981391907, + "num_tokens": 326828612.0, + "step": 12631 + }, + { + "epoch": 1.3872172194157697, + "grad_norm": 2.322965383529663, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7286035418510437, + "num_tokens": 326852073.0, + "step": 12632 + }, + { + "epoch": 1.3873270371183835, + "grad_norm": 2.195405960083008, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7435027360916138, + "num_tokens": 326877643.0, + "step": 12633 + }, + { + "epoch": 1.3874368548209972, + "grad_norm": 2.2882516384124756, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7091653943061829, + "num_tokens": 326902042.0, + "step": 12634 + }, + { + "epoch": 1.3875466725236107, + "grad_norm": 2.2458159923553467, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7175359129905701, + "num_tokens": 326927492.0, + "step": 12635 + }, + { + "epoch": 1.3876564902262245, + "grad_norm": 2.1659531593322754, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7177647948265076, + "num_tokens": 326955640.0, + "step": 12636 + }, + { + "epoch": 1.387766307928838, + "grad_norm": 2.1444904804229736, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7139033675193787, + "num_tokens": 326982934.0, + "step": 12637 + }, + { + "epoch": 1.3878761256314518, + "grad_norm": 2.6781082153320312, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7109137177467346, + "num_tokens": 327003284.0, + "step": 12638 + }, + { + "epoch": 1.3879859433340656, + "grad_norm": 1.9123958349227905, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6850093603134155, + "num_tokens": 327039297.0, + "step": 12639 + }, + { + "epoch": 1.388095761036679, + "grad_norm": 2.1200530529022217, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6931201219558716, + "num_tokens": 327068485.0, + "step": 12640 + }, + { + "epoch": 1.3882055787392928, + "grad_norm": 2.2198433876037598, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7158231735229492, + "num_tokens": 327094322.0, + "step": 12641 + }, + { + "epoch": 1.3883153964419064, + "grad_norm": 2.327364444732666, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7492669224739075, + "num_tokens": 327118567.0, + "step": 12642 + }, + { + "epoch": 1.3884252141445201, + "grad_norm": 2.173619031906128, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.720518946647644, + "num_tokens": 327144865.0, + "step": 12643 + }, + { + "epoch": 1.388535031847134, + "grad_norm": 2.0001635551452637, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7097887992858887, + "num_tokens": 327174998.0, + "step": 12644 + }, + { + "epoch": 1.3886448495497474, + "grad_norm": 2.227738618850708, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.702392041683197, + "num_tokens": 327200509.0, + "step": 12645 + }, + { + "epoch": 1.388754667252361, + "grad_norm": 2.1654255390167236, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7294813394546509, + "num_tokens": 327225745.0, + "step": 12646 + }, + { + "epoch": 1.3888644849549747, + "grad_norm": 2.098893642425537, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6969360113143921, + "num_tokens": 327254984.0, + "step": 12647 + }, + { + "epoch": 1.3889743026575885, + "grad_norm": 2.1774470806121826, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7268867492675781, + "num_tokens": 327279223.0, + "step": 12648 + }, + { + "epoch": 1.389084120360202, + "grad_norm": 2.481353282928467, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7200028896331787, + "num_tokens": 327303319.0, + "step": 12649 + }, + { + "epoch": 1.3891939380628158, + "grad_norm": 2.080674648284912, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7116398811340332, + "num_tokens": 327331675.0, + "step": 12650 + }, + { + "epoch": 1.3893037557654293, + "grad_norm": 2.2955775260925293, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7193799614906311, + "num_tokens": 327355990.0, + "step": 12651 + }, + { + "epoch": 1.389413573468043, + "grad_norm": 2.3904061317443848, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7298976182937622, + "num_tokens": 327379324.0, + "step": 12652 + }, + { + "epoch": 1.3895233911706568, + "grad_norm": 2.397613763809204, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7092125415802002, + "num_tokens": 327402498.0, + "step": 12653 + }, + { + "epoch": 1.3896332088732704, + "grad_norm": 2.3534536361694336, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7103583216667175, + "num_tokens": 327426181.0, + "step": 12654 + }, + { + "epoch": 1.389743026575884, + "grad_norm": 2.022554874420166, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7127668261528015, + "num_tokens": 327455846.0, + "step": 12655 + }, + { + "epoch": 1.3898528442784976, + "grad_norm": 2.177565574645996, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6871685981750488, + "num_tokens": 327486203.0, + "step": 12656 + }, + { + "epoch": 1.3899626619811114, + "grad_norm": 2.2389161586761475, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7411340475082397, + "num_tokens": 327510687.0, + "step": 12657 + }, + { + "epoch": 1.390072479683725, + "grad_norm": 2.21850323677063, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7198221683502197, + "num_tokens": 327536849.0, + "step": 12658 + }, + { + "epoch": 1.3901822973863387, + "grad_norm": 2.3398914337158203, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7135988473892212, + "num_tokens": 327559433.0, + "step": 12659 + }, + { + "epoch": 1.3902921150889522, + "grad_norm": 2.6479709148406982, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7642470598220825, + "num_tokens": 327580154.0, + "step": 12660 + }, + { + "epoch": 1.390401932791566, + "grad_norm": 2.399942398071289, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7363519072532654, + "num_tokens": 327602813.0, + "step": 12661 + }, + { + "epoch": 1.3905117504941797, + "grad_norm": 2.2845356464385986, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7332144379615784, + "num_tokens": 327629603.0, + "step": 12662 + }, + { + "epoch": 1.3906215681967933, + "grad_norm": 2.400099277496338, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7107769250869751, + "num_tokens": 327654648.0, + "step": 12663 + }, + { + "epoch": 1.390731385899407, + "grad_norm": 2.0944771766662598, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7028464078903198, + "num_tokens": 327684120.0, + "step": 12664 + }, + { + "epoch": 1.3908412036020206, + "grad_norm": 2.449032783508301, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.712548017501831, + "num_tokens": 327707176.0, + "step": 12665 + }, + { + "epoch": 1.3909510213046343, + "grad_norm": 2.17218279838562, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7012465596199036, + "num_tokens": 327734720.0, + "step": 12666 + }, + { + "epoch": 1.391060839007248, + "grad_norm": 2.2282235622406006, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7199511528015137, + "num_tokens": 327760150.0, + "step": 12667 + }, + { + "epoch": 1.3911706567098616, + "grad_norm": 2.226997137069702, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7206491231918335, + "num_tokens": 327784149.0, + "step": 12668 + }, + { + "epoch": 1.3912804744124752, + "grad_norm": 2.32261323928833, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7313485145568848, + "num_tokens": 327808125.0, + "step": 12669 + }, + { + "epoch": 1.391390292115089, + "grad_norm": 2.3814780712127686, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7358688116073608, + "num_tokens": 327830077.0, + "step": 12670 + }, + { + "epoch": 1.3915001098177027, + "grad_norm": 2.1982529163360596, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7170255184173584, + "num_tokens": 327857461.0, + "step": 12671 + }, + { + "epoch": 1.3916099275203162, + "grad_norm": 2.5902326107025146, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7258874773979187, + "num_tokens": 327877176.0, + "step": 12672 + }, + { + "epoch": 1.39171974522293, + "grad_norm": 2.674504280090332, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7452141046524048, + "num_tokens": 327895395.0, + "step": 12673 + }, + { + "epoch": 1.3918295629255435, + "grad_norm": 1.959262490272522, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.726062536239624, + "num_tokens": 327927128.0, + "step": 12674 + }, + { + "epoch": 1.3919393806281573, + "grad_norm": 2.2801849842071533, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7433642148971558, + "num_tokens": 327951863.0, + "step": 12675 + }, + { + "epoch": 1.392049198330771, + "grad_norm": 2.1833558082580566, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7343030571937561, + "num_tokens": 327975457.0, + "step": 12676 + }, + { + "epoch": 1.3921590160333845, + "grad_norm": 2.2342348098754883, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7103633284568787, + "num_tokens": 328001139.0, + "step": 12677 + }, + { + "epoch": 1.3922688337359983, + "grad_norm": 2.2186052799224854, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7076781988143921, + "num_tokens": 328027431.0, + "step": 12678 + }, + { + "epoch": 1.3923786514386118, + "grad_norm": 2.1975831985473633, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7015392780303955, + "num_tokens": 328054469.0, + "step": 12679 + }, + { + "epoch": 1.3924884691412256, + "grad_norm": 2.012871742248535, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7100143432617188, + "num_tokens": 328086474.0, + "step": 12680 + }, + { + "epoch": 1.3925982868438394, + "grad_norm": 2.1644833087921143, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7065830826759338, + "num_tokens": 328114272.0, + "step": 12681 + }, + { + "epoch": 1.3927081045464529, + "grad_norm": 2.33478045463562, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7157238721847534, + "num_tokens": 328139879.0, + "step": 12682 + }, + { + "epoch": 1.3928179222490664, + "grad_norm": 2.4394898414611816, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7076328992843628, + "num_tokens": 328163458.0, + "step": 12683 + }, + { + "epoch": 1.3929277399516802, + "grad_norm": 2.209946393966675, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7186859846115112, + "num_tokens": 328189557.0, + "step": 12684 + }, + { + "epoch": 1.393037557654294, + "grad_norm": 2.1635284423828125, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7110333442687988, + "num_tokens": 328216781.0, + "step": 12685 + }, + { + "epoch": 1.3931473753569075, + "grad_norm": 2.05783748626709, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7280049324035645, + "num_tokens": 328248094.0, + "step": 12686 + }, + { + "epoch": 1.3932571930595212, + "grad_norm": 1.9748404026031494, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.698390007019043, + "num_tokens": 328282358.0, + "step": 12687 + }, + { + "epoch": 1.3933670107621348, + "grad_norm": 1.9802663326263428, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.698915421962738, + "num_tokens": 328318452.0, + "step": 12688 + }, + { + "epoch": 1.3934768284647485, + "grad_norm": 2.2277731895446777, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7034699320793152, + "num_tokens": 328344819.0, + "step": 12689 + }, + { + "epoch": 1.3935866461673623, + "grad_norm": 2.6010630130767822, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7004478573799133, + "num_tokens": 328366894.0, + "step": 12690 + }, + { + "epoch": 1.3936964638699758, + "grad_norm": 2.326378345489502, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7118484973907471, + "num_tokens": 328393601.0, + "step": 12691 + }, + { + "epoch": 1.3938062815725896, + "grad_norm": 2.1768016815185547, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7254544496536255, + "num_tokens": 328419230.0, + "step": 12692 + }, + { + "epoch": 1.393916099275203, + "grad_norm": 2.4432528018951416, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7105263471603394, + "num_tokens": 328443372.0, + "step": 12693 + }, + { + "epoch": 1.3940259169778169, + "grad_norm": 2.347266674041748, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7292811274528503, + "num_tokens": 328465375.0, + "step": 12694 + }, + { + "epoch": 1.3941357346804306, + "grad_norm": 2.0213170051574707, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.715407133102417, + "num_tokens": 328497099.0, + "step": 12695 + }, + { + "epoch": 1.3942455523830442, + "grad_norm": 2.3905742168426514, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7435019016265869, + "num_tokens": 328517455.0, + "step": 12696 + }, + { + "epoch": 1.3943553700856577, + "grad_norm": 2.235610008239746, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7044864892959595, + "num_tokens": 328542932.0, + "step": 12697 + }, + { + "epoch": 1.3944651877882714, + "grad_norm": 2.404930830001831, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7296127676963806, + "num_tokens": 328566209.0, + "step": 12698 + }, + { + "epoch": 1.3945750054908852, + "grad_norm": 2.1278135776519775, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7196955680847168, + "num_tokens": 328591339.0, + "step": 12699 + }, + { + "epoch": 1.3946848231934987, + "grad_norm": 1.9490286111831665, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.692572295665741, + "num_tokens": 328624959.0, + "step": 12700 + }, + { + "epoch": 1.3947946408961125, + "grad_norm": 2.2578608989715576, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6920010447502136, + "num_tokens": 328653922.0, + "step": 12701 + }, + { + "epoch": 1.394904458598726, + "grad_norm": 2.486966371536255, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7297572493553162, + "num_tokens": 328674331.0, + "step": 12702 + }, + { + "epoch": 1.3950142763013398, + "grad_norm": 2.2314341068267822, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7331587672233582, + "num_tokens": 328698819.0, + "step": 12703 + }, + { + "epoch": 1.3951240940039535, + "grad_norm": 2.0884082317352295, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7168492078781128, + "num_tokens": 328727751.0, + "step": 12704 + }, + { + "epoch": 1.395233911706567, + "grad_norm": 2.451913356781006, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7358996868133545, + "num_tokens": 328749952.0, + "step": 12705 + }, + { + "epoch": 1.3953437294091808, + "grad_norm": 2.203277111053467, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7420998811721802, + "num_tokens": 328773941.0, + "step": 12706 + }, + { + "epoch": 1.3954535471117944, + "grad_norm": 2.3280446529388428, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.6953880190849304, + "num_tokens": 328799129.0, + "step": 12707 + }, + { + "epoch": 1.3955633648144081, + "grad_norm": 2.118218421936035, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7172377109527588, + "num_tokens": 328826607.0, + "step": 12708 + }, + { + "epoch": 1.3956731825170219, + "grad_norm": 2.27492618560791, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7419134378433228, + "num_tokens": 328851609.0, + "step": 12709 + }, + { + "epoch": 1.3957830002196354, + "grad_norm": 2.1397454738616943, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7167975902557373, + "num_tokens": 328878492.0, + "step": 12710 + }, + { + "epoch": 1.395892817922249, + "grad_norm": 2.3223676681518555, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7228745818138123, + "num_tokens": 328903482.0, + "step": 12711 + }, + { + "epoch": 1.3960026356248627, + "grad_norm": 2.3784549236297607, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7266672253608704, + "num_tokens": 328925219.0, + "step": 12712 + }, + { + "epoch": 1.3961124533274765, + "grad_norm": 2.240267753601074, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7146041393280029, + "num_tokens": 328949992.0, + "step": 12713 + }, + { + "epoch": 1.39622227103009, + "grad_norm": 2.5351929664611816, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7313408255577087, + "num_tokens": 328972712.0, + "step": 12714 + }, + { + "epoch": 1.3963320887327038, + "grad_norm": 2.252948760986328, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7002227902412415, + "num_tokens": 329000792.0, + "step": 12715 + }, + { + "epoch": 1.3964419064353173, + "grad_norm": 2.1230435371398926, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.726676344871521, + "num_tokens": 329028808.0, + "step": 12716 + }, + { + "epoch": 1.396551724137931, + "grad_norm": 2.2477757930755615, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7237467765808105, + "num_tokens": 329056023.0, + "step": 12717 + }, + { + "epoch": 1.3966615418405448, + "grad_norm": 2.0184550285339355, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7023251056671143, + "num_tokens": 329087466.0, + "step": 12718 + }, + { + "epoch": 1.3967713595431583, + "grad_norm": 2.2779793739318848, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7186126708984375, + "num_tokens": 329113202.0, + "step": 12719 + }, + { + "epoch": 1.396881177245772, + "grad_norm": 2.284620761871338, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.705244779586792, + "num_tokens": 329138724.0, + "step": 12720 + }, + { + "epoch": 1.3969909949483856, + "grad_norm": 2.2462751865386963, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.702540934085846, + "num_tokens": 329165168.0, + "step": 12721 + }, + { + "epoch": 1.3971008126509994, + "grad_norm": 2.3475522994995117, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.708305835723877, + "num_tokens": 329189652.0, + "step": 12722 + }, + { + "epoch": 1.397210630353613, + "grad_norm": 2.420804023742676, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7526952028274536, + "num_tokens": 329211107.0, + "step": 12723 + }, + { + "epoch": 1.3973204480562267, + "grad_norm": 2.4848594665527344, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7110867500305176, + "num_tokens": 329233763.0, + "step": 12724 + }, + { + "epoch": 1.3974302657588402, + "grad_norm": 2.447237491607666, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7165957689285278, + "num_tokens": 329256232.0, + "step": 12725 + }, + { + "epoch": 1.397540083461454, + "grad_norm": 2.4536631107330322, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7201440334320068, + "num_tokens": 329278118.0, + "step": 12726 + }, + { + "epoch": 1.3976499011640677, + "grad_norm": 2.167696714401245, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7431442737579346, + "num_tokens": 329306441.0, + "step": 12727 + }, + { + "epoch": 1.3977597188666813, + "grad_norm": 2.0908420085906982, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7064061164855957, + "num_tokens": 329338047.0, + "step": 12728 + }, + { + "epoch": 1.397869536569295, + "grad_norm": 2.4465701580047607, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7330244779586792, + "num_tokens": 329361361.0, + "step": 12729 + }, + { + "epoch": 1.3979793542719086, + "grad_norm": 2.5462591648101807, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7297300696372986, + "num_tokens": 329382056.0, + "step": 12730 + }, + { + "epoch": 1.3980891719745223, + "grad_norm": 2.4993114471435547, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7277818918228149, + "num_tokens": 329404022.0, + "step": 12731 + }, + { + "epoch": 1.398198989677136, + "grad_norm": 2.194931983947754, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7111366987228394, + "num_tokens": 329430905.0, + "step": 12732 + }, + { + "epoch": 1.3983088073797496, + "grad_norm": 2.3776705265045166, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7243891358375549, + "num_tokens": 329453502.0, + "step": 12733 + }, + { + "epoch": 1.3984186250823631, + "grad_norm": 2.2749862670898438, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7059475183486938, + "num_tokens": 329482453.0, + "step": 12734 + }, + { + "epoch": 1.398528442784977, + "grad_norm": 2.0506534576416016, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6923529505729675, + "num_tokens": 329514440.0, + "step": 12735 + }, + { + "epoch": 1.3986382604875907, + "grad_norm": 2.5629515647888184, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7619565725326538, + "num_tokens": 329534295.0, + "step": 12736 + }, + { + "epoch": 1.3987480781902042, + "grad_norm": 2.535137176513672, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7099803686141968, + "num_tokens": 329555527.0, + "step": 12737 + }, + { + "epoch": 1.398857895892818, + "grad_norm": 2.1273856163024902, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6830289363861084, + "num_tokens": 329586322.0, + "step": 12738 + }, + { + "epoch": 1.3989677135954315, + "grad_norm": 2.2862627506256104, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7174428701400757, + "num_tokens": 329611497.0, + "step": 12739 + }, + { + "epoch": 1.3990775312980452, + "grad_norm": 2.285806894302368, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7226937413215637, + "num_tokens": 329636836.0, + "step": 12740 + }, + { + "epoch": 1.399187349000659, + "grad_norm": 2.526470422744751, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7208424806594849, + "num_tokens": 329661010.0, + "step": 12741 + }, + { + "epoch": 1.3992971667032725, + "grad_norm": 2.5200657844543457, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7218195796012878, + "num_tokens": 329682644.0, + "step": 12742 + }, + { + "epoch": 1.3994069844058863, + "grad_norm": 2.2944371700286865, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.726371705532074, + "num_tokens": 329707774.0, + "step": 12743 + }, + { + "epoch": 1.3995168021084998, + "grad_norm": 2.083078384399414, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7207140922546387, + "num_tokens": 329735999.0, + "step": 12744 + }, + { + "epoch": 1.3996266198111136, + "grad_norm": 2.182199001312256, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6977362632751465, + "num_tokens": 329765151.0, + "step": 12745 + }, + { + "epoch": 1.3997364375137273, + "grad_norm": 2.187429189682007, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7405964732170105, + "num_tokens": 329791205.0, + "step": 12746 + }, + { + "epoch": 1.3998462552163409, + "grad_norm": 2.3241982460021973, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.694976806640625, + "num_tokens": 329815303.0, + "step": 12747 + }, + { + "epoch": 1.3999560729189544, + "grad_norm": 2.3290979862213135, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7334553003311157, + "num_tokens": 329838161.0, + "step": 12748 + }, + { + "epoch": 1.4000658906215682, + "grad_norm": 2.1062521934509277, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6966321468353271, + "num_tokens": 329869874.0, + "step": 12749 + }, + { + "epoch": 1.400175708324182, + "grad_norm": 2.5171468257904053, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.742805004119873, + "num_tokens": 329890223.0, + "step": 12750 + }, + { + "epoch": 1.4002855260267955, + "grad_norm": 2.1107938289642334, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7169765830039978, + "num_tokens": 329918639.0, + "step": 12751 + }, + { + "epoch": 1.4003953437294092, + "grad_norm": 2.4915707111358643, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7113018035888672, + "num_tokens": 329943018.0, + "step": 12752 + }, + { + "epoch": 1.4005051614320227, + "grad_norm": 2.2727572917938232, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7233006954193115, + "num_tokens": 329968118.0, + "step": 12753 + }, + { + "epoch": 1.4006149791346365, + "grad_norm": 2.1096222400665283, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7285284996032715, + "num_tokens": 329996407.0, + "step": 12754 + }, + { + "epoch": 1.4007247968372503, + "grad_norm": 2.244734525680542, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7036147117614746, + "num_tokens": 330020797.0, + "step": 12755 + }, + { + "epoch": 1.4008346145398638, + "grad_norm": 2.300529718399048, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7113844156265259, + "num_tokens": 330046151.0, + "step": 12756 + }, + { + "epoch": 1.4009444322424776, + "grad_norm": 2.49180006980896, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7392716407775879, + "num_tokens": 330067459.0, + "step": 12757 + }, + { + "epoch": 1.401054249945091, + "grad_norm": 2.3451035022735596, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7346811294555664, + "num_tokens": 330090836.0, + "step": 12758 + }, + { + "epoch": 1.4011640676477048, + "grad_norm": 2.577727794647217, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.6940797567367554, + "num_tokens": 330112581.0, + "step": 12759 + }, + { + "epoch": 1.4012738853503186, + "grad_norm": 2.393002986907959, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7172729969024658, + "num_tokens": 330134770.0, + "step": 12760 + }, + { + "epoch": 1.4013837030529321, + "grad_norm": 2.23748779296875, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.6890414953231812, + "num_tokens": 330162305.0, + "step": 12761 + }, + { + "epoch": 1.4014935207555457, + "grad_norm": 1.9272955656051636, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.695331871509552, + "num_tokens": 330195513.0, + "step": 12762 + }, + { + "epoch": 1.4016033384581594, + "grad_norm": 2.163795232772827, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.70066899061203, + "num_tokens": 330224612.0, + "step": 12763 + }, + { + "epoch": 1.4017131561607732, + "grad_norm": 2.474058151245117, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7035099267959595, + "num_tokens": 330245872.0, + "step": 12764 + }, + { + "epoch": 1.4018229738633867, + "grad_norm": 2.8670685291290283, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7297925353050232, + "num_tokens": 330263478.0, + "step": 12765 + }, + { + "epoch": 1.4019327915660005, + "grad_norm": 2.161167621612549, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7208991050720215, + "num_tokens": 330292579.0, + "step": 12766 + }, + { + "epoch": 1.402042609268614, + "grad_norm": 2.4380674362182617, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7159007787704468, + "num_tokens": 330315212.0, + "step": 12767 + }, + { + "epoch": 1.4021524269712278, + "grad_norm": 2.1202030181884766, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7135716676712036, + "num_tokens": 330343511.0, + "step": 12768 + }, + { + "epoch": 1.4022622446738415, + "grad_norm": 2.1855366230010986, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7281410098075867, + "num_tokens": 330371013.0, + "step": 12769 + }, + { + "epoch": 1.402372062376455, + "grad_norm": 2.4797165393829346, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7499489784240723, + "num_tokens": 330390565.0, + "step": 12770 + }, + { + "epoch": 1.4024818800790688, + "grad_norm": 2.3439981937408447, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7423012852668762, + "num_tokens": 330414047.0, + "step": 12771 + }, + { + "epoch": 1.4025916977816824, + "grad_norm": 2.1739509105682373, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7224960923194885, + "num_tokens": 330440831.0, + "step": 12772 + }, + { + "epoch": 1.4027015154842961, + "grad_norm": 2.5359299182891846, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7465438842773438, + "num_tokens": 330460607.0, + "step": 12773 + }, + { + "epoch": 1.4028113331869096, + "grad_norm": 1.9746768474578857, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6969996690750122, + "num_tokens": 330493053.0, + "step": 12774 + }, + { + "epoch": 1.4029211508895234, + "grad_norm": 1.8930968046188354, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7192162275314331, + "num_tokens": 330525499.0, + "step": 12775 + }, + { + "epoch": 1.403030968592137, + "grad_norm": 2.2721617221832275, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6875905394554138, + "num_tokens": 330552529.0, + "step": 12776 + }, + { + "epoch": 1.4031407862947507, + "grad_norm": 2.0460801124572754, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.6970129013061523, + "num_tokens": 330584224.0, + "step": 12777 + }, + { + "epoch": 1.4032506039973645, + "grad_norm": 2.312549591064453, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.719801664352417, + "num_tokens": 330608743.0, + "step": 12778 + }, + { + "epoch": 1.403360421699978, + "grad_norm": 2.117086410522461, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7158123850822449, + "num_tokens": 330637403.0, + "step": 12779 + }, + { + "epoch": 1.4034702394025917, + "grad_norm": 2.1065242290496826, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7159990072250366, + "num_tokens": 330669501.0, + "step": 12780 + }, + { + "epoch": 1.4035800571052053, + "grad_norm": 2.0423548221588135, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7081077694892883, + "num_tokens": 330701549.0, + "step": 12781 + }, + { + "epoch": 1.403689874807819, + "grad_norm": 2.1537506580352783, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7041400671005249, + "num_tokens": 330729181.0, + "step": 12782 + }, + { + "epoch": 1.4037996925104328, + "grad_norm": 2.1594860553741455, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7012907862663269, + "num_tokens": 330757862.0, + "step": 12783 + }, + { + "epoch": 1.4039095102130463, + "grad_norm": 2.263878583908081, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7144830226898193, + "num_tokens": 330783695.0, + "step": 12784 + }, + { + "epoch": 1.40401932791566, + "grad_norm": 1.8630554676055908, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7079967856407166, + "num_tokens": 330817814.0, + "step": 12785 + }, + { + "epoch": 1.4041291456182736, + "grad_norm": 2.233800172805786, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.726797342300415, + "num_tokens": 330844176.0, + "step": 12786 + }, + { + "epoch": 1.4042389633208874, + "grad_norm": 2.247821569442749, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7066049575805664, + "num_tokens": 330870796.0, + "step": 12787 + }, + { + "epoch": 1.404348781023501, + "grad_norm": 2.3405072689056396, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7189984321594238, + "num_tokens": 330894438.0, + "step": 12788 + }, + { + "epoch": 1.4044585987261147, + "grad_norm": 2.3290436267852783, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7063071131706238, + "num_tokens": 330921480.0, + "step": 12789 + }, + { + "epoch": 1.4045684164287282, + "grad_norm": 2.3132247924804688, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7083945870399475, + "num_tokens": 330946772.0, + "step": 12790 + }, + { + "epoch": 1.404678234131342, + "grad_norm": 2.0248055458068848, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7176015377044678, + "num_tokens": 330979370.0, + "step": 12791 + }, + { + "epoch": 1.4047880518339557, + "grad_norm": 2.2982773780822754, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7010596990585327, + "num_tokens": 331002954.0, + "step": 12792 + }, + { + "epoch": 1.4048978695365693, + "grad_norm": 1.945290446281433, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.694506049156189, + "num_tokens": 331035129.0, + "step": 12793 + }, + { + "epoch": 1.405007687239183, + "grad_norm": 2.348761558532715, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7111597061157227, + "num_tokens": 331059434.0, + "step": 12794 + }, + { + "epoch": 1.4051175049417965, + "grad_norm": 2.565523862838745, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7055487036705017, + "num_tokens": 331080140.0, + "step": 12795 + }, + { + "epoch": 1.4052273226444103, + "grad_norm": 2.191178560256958, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.72730553150177, + "num_tokens": 331106893.0, + "step": 12796 + }, + { + "epoch": 1.405337140347024, + "grad_norm": 2.574270248413086, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7480432987213135, + "num_tokens": 331126529.0, + "step": 12797 + }, + { + "epoch": 1.4054469580496376, + "grad_norm": 2.039609909057617, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7050638198852539, + "num_tokens": 331158566.0, + "step": 12798 + }, + { + "epoch": 1.4055567757522511, + "grad_norm": 2.0956132411956787, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7157533168792725, + "num_tokens": 331186311.0, + "step": 12799 + }, + { + "epoch": 1.4056665934548649, + "grad_norm": 2.014277696609497, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7405881285667419, + "num_tokens": 331215102.0, + "step": 12800 + }, + { + "epoch": 1.4057764111574786, + "grad_norm": 2.5028164386749268, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7107367515563965, + "num_tokens": 331236980.0, + "step": 12801 + }, + { + "epoch": 1.4058862288600922, + "grad_norm": 2.391065835952759, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6985396146774292, + "num_tokens": 331261948.0, + "step": 12802 + }, + { + "epoch": 1.405996046562706, + "grad_norm": 2.2570314407348633, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7113507986068726, + "num_tokens": 331288192.0, + "step": 12803 + }, + { + "epoch": 1.4061058642653195, + "grad_norm": 2.048953056335449, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7192221879959106, + "num_tokens": 331317084.0, + "step": 12804 + }, + { + "epoch": 1.4062156819679332, + "grad_norm": 2.0061581134796143, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7366328239440918, + "num_tokens": 331345302.0, + "step": 12805 + }, + { + "epoch": 1.406325499670547, + "grad_norm": 2.1780412197113037, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7206878662109375, + "num_tokens": 331372262.0, + "step": 12806 + }, + { + "epoch": 1.4064353173731605, + "grad_norm": 2.1864006519317627, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7281107306480408, + "num_tokens": 331398490.0, + "step": 12807 + }, + { + "epoch": 1.4065451350757743, + "grad_norm": 2.2242319583892822, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6962548494338989, + "num_tokens": 331424470.0, + "step": 12808 + }, + { + "epoch": 1.4066549527783878, + "grad_norm": 2.351816415786743, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7090443968772888, + "num_tokens": 331447838.0, + "step": 12809 + }, + { + "epoch": 1.4067647704810016, + "grad_norm": 2.521489143371582, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.713908314704895, + "num_tokens": 331470010.0, + "step": 12810 + }, + { + "epoch": 1.4068745881836153, + "grad_norm": 2.1529271602630615, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7022171020507812, + "num_tokens": 331496230.0, + "step": 12811 + }, + { + "epoch": 1.4069844058862289, + "grad_norm": 2.1470723152160645, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6924526691436768, + "num_tokens": 331525604.0, + "step": 12812 + }, + { + "epoch": 1.4070942235888424, + "grad_norm": 2.111816644668579, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7159291505813599, + "num_tokens": 331553471.0, + "step": 12813 + }, + { + "epoch": 1.4072040412914562, + "grad_norm": 2.4026923179626465, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7379939556121826, + "num_tokens": 331575373.0, + "step": 12814 + }, + { + "epoch": 1.40731385899407, + "grad_norm": 2.061715602874756, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7144964933395386, + "num_tokens": 331604945.0, + "step": 12815 + }, + { + "epoch": 1.4074236766966834, + "grad_norm": 2.151798963546753, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7003170847892761, + "num_tokens": 331632112.0, + "step": 12816 + }, + { + "epoch": 1.4075334943992972, + "grad_norm": 2.0223193168640137, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7222321629524231, + "num_tokens": 331659631.0, + "step": 12817 + }, + { + "epoch": 1.4076433121019107, + "grad_norm": 2.4077858924865723, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7402452230453491, + "num_tokens": 331681471.0, + "step": 12818 + }, + { + "epoch": 1.4077531298045245, + "grad_norm": 2.342067241668701, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7378570437431335, + "num_tokens": 331704106.0, + "step": 12819 + }, + { + "epoch": 1.4078629475071383, + "grad_norm": 2.026191234588623, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.696526825428009, + "num_tokens": 331736036.0, + "step": 12820 + }, + { + "epoch": 1.4079727652097518, + "grad_norm": 2.3152542114257812, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7169557809829712, + "num_tokens": 331761772.0, + "step": 12821 + }, + { + "epoch": 1.4080825829123655, + "grad_norm": 2.335019111633301, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7229427099227905, + "num_tokens": 331785723.0, + "step": 12822 + }, + { + "epoch": 1.408192400614979, + "grad_norm": 2.38169264793396, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7232472896575928, + "num_tokens": 331809006.0, + "step": 12823 + }, + { + "epoch": 1.4083022183175928, + "grad_norm": 2.5224742889404297, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7389802932739258, + "num_tokens": 331830095.0, + "step": 12824 + }, + { + "epoch": 1.4084120360202066, + "grad_norm": 2.419628858566284, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7450998425483704, + "num_tokens": 331852970.0, + "step": 12825 + }, + { + "epoch": 1.4085218537228201, + "grad_norm": 2.0526793003082275, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7327585220336914, + "num_tokens": 331880021.0, + "step": 12826 + }, + { + "epoch": 1.4086316714254337, + "grad_norm": 2.0961196422576904, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6926758289337158, + "num_tokens": 331909539.0, + "step": 12827 + }, + { + "epoch": 1.4087414891280474, + "grad_norm": 2.0825412273406982, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.691002607345581, + "num_tokens": 331938496.0, + "step": 12828 + }, + { + "epoch": 1.4088513068306612, + "grad_norm": 1.9978948831558228, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7318756580352783, + "num_tokens": 331970350.0, + "step": 12829 + }, + { + "epoch": 1.4089611245332747, + "grad_norm": 2.468015432357788, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7182921171188354, + "num_tokens": 331991869.0, + "step": 12830 + }, + { + "epoch": 1.4090709422358885, + "grad_norm": 2.506312847137451, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.722710907459259, + "num_tokens": 332013233.0, + "step": 12831 + }, + { + "epoch": 1.409180759938502, + "grad_norm": 2.1905369758605957, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7116976976394653, + "num_tokens": 332039007.0, + "step": 12832 + }, + { + "epoch": 1.4092905776411158, + "grad_norm": 2.215799331665039, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7130707502365112, + "num_tokens": 332066044.0, + "step": 12833 + }, + { + "epoch": 1.4094003953437295, + "grad_norm": 2.1496024131774902, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7449226379394531, + "num_tokens": 332089319.0, + "step": 12834 + }, + { + "epoch": 1.409510213046343, + "grad_norm": 2.5920827388763428, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7213956117630005, + "num_tokens": 332109374.0, + "step": 12835 + }, + { + "epoch": 1.4096200307489568, + "grad_norm": 2.125214099884033, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7110929489135742, + "num_tokens": 332137601.0, + "step": 12836 + }, + { + "epoch": 1.4097298484515703, + "grad_norm": 2.3451924324035645, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7141857743263245, + "num_tokens": 332162374.0, + "step": 12837 + }, + { + "epoch": 1.409839666154184, + "grad_norm": 3.002700090408325, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7121718525886536, + "num_tokens": 332191028.0, + "step": 12838 + }, + { + "epoch": 1.4099494838567976, + "grad_norm": 2.4505562782287598, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7246832847595215, + "num_tokens": 332212579.0, + "step": 12839 + }, + { + "epoch": 1.4100593015594114, + "grad_norm": 2.1923680305480957, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7249372601509094, + "num_tokens": 332240660.0, + "step": 12840 + }, + { + "epoch": 1.410169119262025, + "grad_norm": 2.071153402328491, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7087576389312744, + "num_tokens": 332271878.0, + "step": 12841 + }, + { + "epoch": 1.4102789369646387, + "grad_norm": 2.069199323654175, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7236105799674988, + "num_tokens": 332300385.0, + "step": 12842 + }, + { + "epoch": 1.4103887546672524, + "grad_norm": 2.297724723815918, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7230206727981567, + "num_tokens": 332324927.0, + "step": 12843 + }, + { + "epoch": 1.410498572369866, + "grad_norm": 2.5176401138305664, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7208933234214783, + "num_tokens": 332347441.0, + "step": 12844 + }, + { + "epoch": 1.4106083900724797, + "grad_norm": 2.644279718399048, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7226955890655518, + "num_tokens": 332367849.0, + "step": 12845 + }, + { + "epoch": 1.4107182077750933, + "grad_norm": 2.260236978530884, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7102128863334656, + "num_tokens": 332394262.0, + "step": 12846 + }, + { + "epoch": 1.410828025477707, + "grad_norm": 2.4129369258880615, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7000107169151306, + "num_tokens": 332417902.0, + "step": 12847 + }, + { + "epoch": 1.4109378431803208, + "grad_norm": 1.9527767896652222, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7070713043212891, + "num_tokens": 332450973.0, + "step": 12848 + }, + { + "epoch": 1.4110476608829343, + "grad_norm": 2.3186135292053223, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7196295261383057, + "num_tokens": 332474891.0, + "step": 12849 + }, + { + "epoch": 1.4111574785855479, + "grad_norm": 2.173330068588257, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7160398960113525, + "num_tokens": 332501341.0, + "step": 12850 + }, + { + "epoch": 1.4112672962881616, + "grad_norm": 2.1974644660949707, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7099276781082153, + "num_tokens": 332529082.0, + "step": 12851 + }, + { + "epoch": 1.4113771139907754, + "grad_norm": 2.2352468967437744, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6904633045196533, + "num_tokens": 332556574.0, + "step": 12852 + }, + { + "epoch": 1.411486931693389, + "grad_norm": 2.3493781089782715, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7143319845199585, + "num_tokens": 332579569.0, + "step": 12853 + }, + { + "epoch": 1.4115967493960027, + "grad_norm": 2.0507030487060547, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7043726444244385, + "num_tokens": 332611878.0, + "step": 12854 + }, + { + "epoch": 1.4117065670986162, + "grad_norm": 2.284839153289795, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7054556012153625, + "num_tokens": 332638058.0, + "step": 12855 + }, + { + "epoch": 1.41181638480123, + "grad_norm": 2.1501588821411133, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.6983395218849182, + "num_tokens": 332666000.0, + "step": 12856 + }, + { + "epoch": 1.4119262025038437, + "grad_norm": 2.020447254180908, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7255594730377197, + "num_tokens": 332694149.0, + "step": 12857 + }, + { + "epoch": 1.4120360202064572, + "grad_norm": 2.2010271549224854, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6863430738449097, + "num_tokens": 332724188.0, + "step": 12858 + }, + { + "epoch": 1.412145837909071, + "grad_norm": 2.079577922821045, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.6993361711502075, + "num_tokens": 332752306.0, + "step": 12859 + }, + { + "epoch": 1.4122556556116845, + "grad_norm": 2.2702839374542236, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7069356441497803, + "num_tokens": 332780115.0, + "step": 12860 + }, + { + "epoch": 1.4123654733142983, + "grad_norm": 2.5430171489715576, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7153797149658203, + "num_tokens": 332800506.0, + "step": 12861 + }, + { + "epoch": 1.412475291016912, + "grad_norm": 2.165024757385254, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7123934030532837, + "num_tokens": 332830867.0, + "step": 12862 + }, + { + "epoch": 1.4125851087195256, + "grad_norm": 2.1614904403686523, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.684904158115387, + "num_tokens": 332860462.0, + "step": 12863 + }, + { + "epoch": 1.4126949264221391, + "grad_norm": 2.453601598739624, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7295196056365967, + "num_tokens": 332880760.0, + "step": 12864 + }, + { + "epoch": 1.4128047441247529, + "grad_norm": 2.0573954582214355, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7033743858337402, + "num_tokens": 332910473.0, + "step": 12865 + }, + { + "epoch": 1.4129145618273666, + "grad_norm": 2.439142942428589, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6980443000793457, + "num_tokens": 332936802.0, + "step": 12866 + }, + { + "epoch": 1.4130243795299802, + "grad_norm": 2.4057388305664062, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7365204095840454, + "num_tokens": 332959432.0, + "step": 12867 + }, + { + "epoch": 1.413134197232594, + "grad_norm": 2.1018004417419434, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.69610196352005, + "num_tokens": 332987546.0, + "step": 12868 + }, + { + "epoch": 1.4132440149352075, + "grad_norm": 2.245327949523926, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.6993274092674255, + "num_tokens": 333014776.0, + "step": 12869 + }, + { + "epoch": 1.4133538326378212, + "grad_norm": 2.278952121734619, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7155402898788452, + "num_tokens": 333040984.0, + "step": 12870 + }, + { + "epoch": 1.413463650340435, + "grad_norm": 2.3241050243377686, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7276616096496582, + "num_tokens": 333064260.0, + "step": 12871 + }, + { + "epoch": 1.4135734680430485, + "grad_norm": 2.533241033554077, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7284924983978271, + "num_tokens": 333085230.0, + "step": 12872 + }, + { + "epoch": 1.4136832857456623, + "grad_norm": 2.199754476547241, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7051467299461365, + "num_tokens": 333111202.0, + "step": 12873 + }, + { + "epoch": 1.4137931034482758, + "grad_norm": 2.213665246963501, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7110296487808228, + "num_tokens": 333135562.0, + "step": 12874 + }, + { + "epoch": 1.4139029211508896, + "grad_norm": 2.2125911712646484, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7158733606338501, + "num_tokens": 333162233.0, + "step": 12875 + }, + { + "epoch": 1.4140127388535033, + "grad_norm": 2.290739059448242, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7195653915405273, + "num_tokens": 333187389.0, + "step": 12876 + }, + { + "epoch": 1.4141225565561168, + "grad_norm": 2.196604013442993, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7194477915763855, + "num_tokens": 333216193.0, + "step": 12877 + }, + { + "epoch": 1.4142323742587304, + "grad_norm": 2.504290819168091, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7280660271644592, + "num_tokens": 333237036.0, + "step": 12878 + }, + { + "epoch": 1.4143421919613441, + "grad_norm": 2.2054619789123535, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7278921604156494, + "num_tokens": 333261687.0, + "step": 12879 + }, + { + "epoch": 1.414452009663958, + "grad_norm": 2.454728603363037, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7144982218742371, + "num_tokens": 333285475.0, + "step": 12880 + }, + { + "epoch": 1.4145618273665714, + "grad_norm": 2.4013161659240723, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.7082319259643555, + "num_tokens": 333308813.0, + "step": 12881 + }, + { + "epoch": 1.4146716450691852, + "grad_norm": 2.1733644008636475, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7172183990478516, + "num_tokens": 333335279.0, + "step": 12882 + }, + { + "epoch": 1.4147814627717987, + "grad_norm": 2.0937201976776123, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7093682289123535, + "num_tokens": 333364811.0, + "step": 12883 + }, + { + "epoch": 1.4148912804744125, + "grad_norm": 2.6115286350250244, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7268271446228027, + "num_tokens": 333385365.0, + "step": 12884 + }, + { + "epoch": 1.4150010981770262, + "grad_norm": 2.550767421722412, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7281448841094971, + "num_tokens": 333405477.0, + "step": 12885 + }, + { + "epoch": 1.4151109158796398, + "grad_norm": 2.3134560585021973, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7169733643531799, + "num_tokens": 333429113.0, + "step": 12886 + }, + { + "epoch": 1.4152207335822535, + "grad_norm": 2.3771073818206787, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7315841913223267, + "num_tokens": 333451628.0, + "step": 12887 + }, + { + "epoch": 1.415330551284867, + "grad_norm": 2.651508092880249, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7438125610351562, + "num_tokens": 333469094.0, + "step": 12888 + }, + { + "epoch": 1.4154403689874808, + "grad_norm": 2.272888422012329, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.746566891670227, + "num_tokens": 333492280.0, + "step": 12889 + }, + { + "epoch": 1.4155501866900946, + "grad_norm": 2.2179718017578125, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7021733522415161, + "num_tokens": 333518127.0, + "step": 12890 + }, + { + "epoch": 1.4156600043927081, + "grad_norm": 2.1497271060943604, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6816834807395935, + "num_tokens": 333546384.0, + "step": 12891 + }, + { + "epoch": 1.4157698220953217, + "grad_norm": 2.3306705951690674, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7152017951011658, + "num_tokens": 333571273.0, + "step": 12892 + }, + { + "epoch": 1.4158796397979354, + "grad_norm": 2.3491501808166504, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7263955473899841, + "num_tokens": 333594588.0, + "step": 12893 + }, + { + "epoch": 1.4159894575005492, + "grad_norm": 2.036123037338257, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7386110424995422, + "num_tokens": 333622666.0, + "step": 12894 + }, + { + "epoch": 1.4160992752031627, + "grad_norm": 2.3675289154052734, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.732243537902832, + "num_tokens": 333644976.0, + "step": 12895 + }, + { + "epoch": 1.4162090929057765, + "grad_norm": 2.176633834838867, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7138502597808838, + "num_tokens": 333673260.0, + "step": 12896 + }, + { + "epoch": 1.41631891060839, + "grad_norm": 2.2527971267700195, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6900984048843384, + "num_tokens": 333701124.0, + "step": 12897 + }, + { + "epoch": 1.4164287283110037, + "grad_norm": 2.1658241748809814, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7161225080490112, + "num_tokens": 333728031.0, + "step": 12898 + }, + { + "epoch": 1.4165385460136175, + "grad_norm": 2.2773234844207764, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7337374687194824, + "num_tokens": 333753218.0, + "step": 12899 + }, + { + "epoch": 1.416648363716231, + "grad_norm": 2.2557194232940674, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7192561626434326, + "num_tokens": 333780133.0, + "step": 12900 + }, + { + "epoch": 1.4167581814188448, + "grad_norm": 2.2396249771118164, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7085189819335938, + "num_tokens": 333805585.0, + "step": 12901 + }, + { + "epoch": 1.4168679991214583, + "grad_norm": 2.1845109462738037, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7082478404045105, + "num_tokens": 333832788.0, + "step": 12902 + }, + { + "epoch": 1.416977816824072, + "grad_norm": 2.102609872817993, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7247621417045593, + "num_tokens": 333861074.0, + "step": 12903 + }, + { + "epoch": 1.4170876345266856, + "grad_norm": 2.651554822921753, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.714358925819397, + "num_tokens": 333883421.0, + "step": 12904 + }, + { + "epoch": 1.4171974522292994, + "grad_norm": 2.5470223426818848, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7570149898529053, + "num_tokens": 333902876.0, + "step": 12905 + }, + { + "epoch": 1.417307269931913, + "grad_norm": 2.0736255645751953, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7148212194442749, + "num_tokens": 333932878.0, + "step": 12906 + }, + { + "epoch": 1.4174170876345267, + "grad_norm": 2.0221660137176514, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7134799957275391, + "num_tokens": 333964610.0, + "step": 12907 + }, + { + "epoch": 1.4175269053371404, + "grad_norm": 2.329197406768799, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7047949433326721, + "num_tokens": 333990179.0, + "step": 12908 + }, + { + "epoch": 1.417636723039754, + "grad_norm": 2.404998540878296, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.729808509349823, + "num_tokens": 334014098.0, + "step": 12909 + }, + { + "epoch": 1.4177465407423677, + "grad_norm": 2.4961373805999756, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7213298082351685, + "num_tokens": 334035745.0, + "step": 12910 + }, + { + "epoch": 1.4178563584449813, + "grad_norm": 2.0899434089660645, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7200261354446411, + "num_tokens": 334063259.0, + "step": 12911 + }, + { + "epoch": 1.417966176147595, + "grad_norm": 2.029968023300171, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7161073684692383, + "num_tokens": 334092798.0, + "step": 12912 + }, + { + "epoch": 1.4180759938502088, + "grad_norm": 2.066328763961792, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7260609269142151, + "num_tokens": 334121983.0, + "step": 12913 + }, + { + "epoch": 1.4181858115528223, + "grad_norm": 2.1648614406585693, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6978122591972351, + "num_tokens": 334150366.0, + "step": 12914 + }, + { + "epoch": 1.4182956292554358, + "grad_norm": 2.327029228210449, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7293444275856018, + "num_tokens": 334173217.0, + "step": 12915 + }, + { + "epoch": 1.4184054469580496, + "grad_norm": 2.061950206756592, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6940791606903076, + "num_tokens": 334203261.0, + "step": 12916 + }, + { + "epoch": 1.4185152646606634, + "grad_norm": 2.361238718032837, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7019193172454834, + "num_tokens": 334229166.0, + "step": 12917 + }, + { + "epoch": 1.418625082363277, + "grad_norm": 2.1091017723083496, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6934028267860413, + "num_tokens": 334258769.0, + "step": 12918 + }, + { + "epoch": 1.4187349000658906, + "grad_norm": 2.0861687660217285, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7121221423149109, + "num_tokens": 334287469.0, + "step": 12919 + }, + { + "epoch": 1.4188447177685042, + "grad_norm": 2.1716556549072266, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7168847322463989, + "num_tokens": 334314416.0, + "step": 12920 + }, + { + "epoch": 1.418954535471118, + "grad_norm": 2.1291208267211914, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7279317378997803, + "num_tokens": 334339914.0, + "step": 12921 + }, + { + "epoch": 1.4190643531737317, + "grad_norm": 2.360950231552124, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.719371497631073, + "num_tokens": 334362130.0, + "step": 12922 + }, + { + "epoch": 1.4191741708763452, + "grad_norm": 1.8972280025482178, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7052035331726074, + "num_tokens": 334394745.0, + "step": 12923 + }, + { + "epoch": 1.419283988578959, + "grad_norm": 2.1316120624542236, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7232087254524231, + "num_tokens": 334422706.0, + "step": 12924 + }, + { + "epoch": 1.4193938062815725, + "grad_norm": 2.2667551040649414, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7096033096313477, + "num_tokens": 334448475.0, + "step": 12925 + }, + { + "epoch": 1.4195036239841863, + "grad_norm": 2.257642984390259, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7306814193725586, + "num_tokens": 334473179.0, + "step": 12926 + }, + { + "epoch": 1.4196134416868, + "grad_norm": 2.161210060119629, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7187120914459229, + "num_tokens": 334499427.0, + "step": 12927 + }, + { + "epoch": 1.4197232593894136, + "grad_norm": 2.05210542678833, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7044054269790649, + "num_tokens": 334531301.0, + "step": 12928 + }, + { + "epoch": 1.419833077092027, + "grad_norm": 2.2703678607940674, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7188296914100647, + "num_tokens": 334555507.0, + "step": 12929 + }, + { + "epoch": 1.4199428947946409, + "grad_norm": 1.9183905124664307, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.713452935218811, + "num_tokens": 334587652.0, + "step": 12930 + }, + { + "epoch": 1.4200527124972546, + "grad_norm": 2.473386287689209, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7206757664680481, + "num_tokens": 334610022.0, + "step": 12931 + }, + { + "epoch": 1.4201625301998682, + "grad_norm": 2.0361976623535156, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.702792227268219, + "num_tokens": 334639176.0, + "step": 12932 + }, + { + "epoch": 1.420272347902482, + "grad_norm": 2.563683032989502, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7381508350372314, + "num_tokens": 334658933.0, + "step": 12933 + }, + { + "epoch": 1.4203821656050954, + "grad_norm": 2.2323055267333984, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7028581500053406, + "num_tokens": 334687502.0, + "step": 12934 + }, + { + "epoch": 1.4204919833077092, + "grad_norm": 1.9755862951278687, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.704983115196228, + "num_tokens": 334719409.0, + "step": 12935 + }, + { + "epoch": 1.420601801010323, + "grad_norm": 1.9588946104049683, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7058171033859253, + "num_tokens": 334749363.0, + "step": 12936 + }, + { + "epoch": 1.4207116187129365, + "grad_norm": 2.0544393062591553, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6921193599700928, + "num_tokens": 334780552.0, + "step": 12937 + }, + { + "epoch": 1.4208214364155503, + "grad_norm": 2.183215379714966, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7010128498077393, + "num_tokens": 334808206.0, + "step": 12938 + }, + { + "epoch": 1.4209312541181638, + "grad_norm": 2.353030204772949, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7179284691810608, + "num_tokens": 334831561.0, + "step": 12939 + }, + { + "epoch": 1.4210410718207775, + "grad_norm": 2.1691391468048096, + "learning_rate": 1e-06, + "loss": 1.04, + "mean_token_accuracy": 0.6848700046539307, + "num_tokens": 334859188.0, + "step": 12940 + }, + { + "epoch": 1.4211508895233913, + "grad_norm": 2.2805705070495605, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7248789072036743, + "num_tokens": 334884809.0, + "step": 12941 + }, + { + "epoch": 1.4212607072260048, + "grad_norm": 2.2068986892700195, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7133961319923401, + "num_tokens": 334910764.0, + "step": 12942 + }, + { + "epoch": 1.4213705249286184, + "grad_norm": 2.450827121734619, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7216864228248596, + "num_tokens": 334935069.0, + "step": 12943 + }, + { + "epoch": 1.4214803426312321, + "grad_norm": 2.190389394760132, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7144942879676819, + "num_tokens": 334961762.0, + "step": 12944 + }, + { + "epoch": 1.4215901603338459, + "grad_norm": 2.2878975868225098, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7244316339492798, + "num_tokens": 334988095.0, + "step": 12945 + }, + { + "epoch": 1.4216999780364594, + "grad_norm": 2.118215560913086, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.730844259262085, + "num_tokens": 335015639.0, + "step": 12946 + }, + { + "epoch": 1.4218097957390732, + "grad_norm": 2.461127758026123, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7281242609024048, + "num_tokens": 335038331.0, + "step": 12947 + }, + { + "epoch": 1.4219196134416867, + "grad_norm": 2.3199241161346436, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6993224024772644, + "num_tokens": 335064612.0, + "step": 12948 + }, + { + "epoch": 1.4220294311443005, + "grad_norm": 2.3403797149658203, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7245319485664368, + "num_tokens": 335088936.0, + "step": 12949 + }, + { + "epoch": 1.4221392488469142, + "grad_norm": 2.3204498291015625, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7380372285842896, + "num_tokens": 335111531.0, + "step": 12950 + }, + { + "epoch": 1.4222490665495278, + "grad_norm": 2.182681083679199, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7157280445098877, + "num_tokens": 335138758.0, + "step": 12951 + }, + { + "epoch": 1.4223588842521415, + "grad_norm": 2.223797559738159, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.722794771194458, + "num_tokens": 335165080.0, + "step": 12952 + }, + { + "epoch": 1.422468701954755, + "grad_norm": 2.420970916748047, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7047783732414246, + "num_tokens": 335189641.0, + "step": 12953 + }, + { + "epoch": 1.4225785196573688, + "grad_norm": 2.0334861278533936, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7074030637741089, + "num_tokens": 335219361.0, + "step": 12954 + }, + { + "epoch": 1.4226883373599823, + "grad_norm": 2.3539741039276123, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7169876098632812, + "num_tokens": 335242980.0, + "step": 12955 + }, + { + "epoch": 1.422798155062596, + "grad_norm": 2.492668390274048, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7037912607192993, + "num_tokens": 335266402.0, + "step": 12956 + }, + { + "epoch": 1.4229079727652096, + "grad_norm": 2.4471030235290527, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7286632657051086, + "num_tokens": 335287588.0, + "step": 12957 + }, + { + "epoch": 1.4230177904678234, + "grad_norm": 2.353773593902588, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.728213369846344, + "num_tokens": 335312230.0, + "step": 12958 + }, + { + "epoch": 1.4231276081704372, + "grad_norm": 2.022505521774292, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7499973773956299, + "num_tokens": 335339674.0, + "step": 12959 + }, + { + "epoch": 1.4232374258730507, + "grad_norm": 2.2835938930511475, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7141363620758057, + "num_tokens": 335366323.0, + "step": 12960 + }, + { + "epoch": 1.4233472435756644, + "grad_norm": 2.401460647583008, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7368450164794922, + "num_tokens": 335388086.0, + "step": 12961 + }, + { + "epoch": 1.423457061278278, + "grad_norm": 2.0879271030426025, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7327438592910767, + "num_tokens": 335416717.0, + "step": 12962 + }, + { + "epoch": 1.4235668789808917, + "grad_norm": 2.0748000144958496, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7113265991210938, + "num_tokens": 335448787.0, + "step": 12963 + }, + { + "epoch": 1.4236766966835055, + "grad_norm": 2.2508232593536377, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7599236965179443, + "num_tokens": 335471561.0, + "step": 12964 + }, + { + "epoch": 1.423786514386119, + "grad_norm": 2.216757297515869, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7199826836585999, + "num_tokens": 335498861.0, + "step": 12965 + }, + { + "epoch": 1.4238963320887328, + "grad_norm": 2.712786912918091, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.727759599685669, + "num_tokens": 335518377.0, + "step": 12966 + }, + { + "epoch": 1.4240061497913463, + "grad_norm": 2.4619219303131104, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7341516017913818, + "num_tokens": 335539438.0, + "step": 12967 + }, + { + "epoch": 1.42411596749396, + "grad_norm": 2.183830499649048, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7583919763565063, + "num_tokens": 335563992.0, + "step": 12968 + }, + { + "epoch": 1.4242257851965736, + "grad_norm": 2.1085338592529297, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6971871256828308, + "num_tokens": 335595342.0, + "step": 12969 + }, + { + "epoch": 1.4243356028991874, + "grad_norm": 2.093080997467041, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7008379697799683, + "num_tokens": 335623167.0, + "step": 12970 + }, + { + "epoch": 1.424445420601801, + "grad_norm": 2.4651546478271484, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.735337495803833, + "num_tokens": 335645301.0, + "step": 12971 + }, + { + "epoch": 1.4245552383044147, + "grad_norm": 2.4646410942077637, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7307446002960205, + "num_tokens": 335668265.0, + "step": 12972 + }, + { + "epoch": 1.4246650560070284, + "grad_norm": 1.9666337966918945, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6835084557533264, + "num_tokens": 335702380.0, + "step": 12973 + }, + { + "epoch": 1.424774873709642, + "grad_norm": 2.2687015533447266, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7098164558410645, + "num_tokens": 335728746.0, + "step": 12974 + }, + { + "epoch": 1.4248846914122557, + "grad_norm": 2.045273780822754, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.713900625705719, + "num_tokens": 335760911.0, + "step": 12975 + }, + { + "epoch": 1.4249945091148692, + "grad_norm": 2.1982483863830566, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7340183258056641, + "num_tokens": 335787964.0, + "step": 12976 + }, + { + "epoch": 1.425104326817483, + "grad_norm": 2.4714791774749756, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7204611301422119, + "num_tokens": 335808830.0, + "step": 12977 + }, + { + "epoch": 1.4252141445200968, + "grad_norm": 2.0029923915863037, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7064200639724731, + "num_tokens": 335838235.0, + "step": 12978 + }, + { + "epoch": 1.4253239622227103, + "grad_norm": 2.3201193809509277, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7192516326904297, + "num_tokens": 335861638.0, + "step": 12979 + }, + { + "epoch": 1.4254337799253238, + "grad_norm": 2.2627158164978027, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7108503580093384, + "num_tokens": 335887560.0, + "step": 12980 + }, + { + "epoch": 1.4255435976279376, + "grad_norm": 2.040330171585083, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7116508483886719, + "num_tokens": 335916147.0, + "step": 12981 + }, + { + "epoch": 1.4256534153305513, + "grad_norm": 2.094244956970215, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.6955975890159607, + "num_tokens": 335945961.0, + "step": 12982 + }, + { + "epoch": 1.4257632330331649, + "grad_norm": 2.390261650085449, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7270704507827759, + "num_tokens": 335968366.0, + "step": 12983 + }, + { + "epoch": 1.4258730507357786, + "grad_norm": 2.4208197593688965, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7395358681678772, + "num_tokens": 335989642.0, + "step": 12984 + }, + { + "epoch": 1.4259828684383922, + "grad_norm": 2.1375417709350586, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7117030620574951, + "num_tokens": 336017457.0, + "step": 12985 + }, + { + "epoch": 1.426092686141006, + "grad_norm": 2.230902910232544, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7172249555587769, + "num_tokens": 336043602.0, + "step": 12986 + }, + { + "epoch": 1.4262025038436197, + "grad_norm": 2.178720474243164, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7260811924934387, + "num_tokens": 336070210.0, + "step": 12987 + }, + { + "epoch": 1.4263123215462332, + "grad_norm": 2.2001640796661377, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7227567434310913, + "num_tokens": 336097055.0, + "step": 12988 + }, + { + "epoch": 1.426422139248847, + "grad_norm": 2.1931376457214355, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7097935080528259, + "num_tokens": 336124255.0, + "step": 12989 + }, + { + "epoch": 1.4265319569514605, + "grad_norm": 2.237471103668213, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7150996923446655, + "num_tokens": 336151343.0, + "step": 12990 + }, + { + "epoch": 1.4266417746540743, + "grad_norm": 1.9876734018325806, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7156432867050171, + "num_tokens": 336183647.0, + "step": 12991 + }, + { + "epoch": 1.426751592356688, + "grad_norm": 2.5098118782043457, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7293749451637268, + "num_tokens": 336204677.0, + "step": 12992 + }, + { + "epoch": 1.4268614100593016, + "grad_norm": 2.0890302658081055, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7096037268638611, + "num_tokens": 336232008.0, + "step": 12993 + }, + { + "epoch": 1.426971227761915, + "grad_norm": 1.9722408056259155, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6986924409866333, + "num_tokens": 336265650.0, + "step": 12994 + }, + { + "epoch": 1.4270810454645289, + "grad_norm": 2.193321943283081, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.6997898817062378, + "num_tokens": 336293078.0, + "step": 12995 + }, + { + "epoch": 1.4271908631671426, + "grad_norm": 2.002073049545288, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.7004305124282837, + "num_tokens": 336325403.0, + "step": 12996 + }, + { + "epoch": 1.4273006808697561, + "grad_norm": 2.3196582794189453, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7134599089622498, + "num_tokens": 336350773.0, + "step": 12997 + }, + { + "epoch": 1.42741049857237, + "grad_norm": 2.24031925201416, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7229399681091309, + "num_tokens": 336375432.0, + "step": 12998 + }, + { + "epoch": 1.4275203162749834, + "grad_norm": 2.1697473526000977, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7242969870567322, + "num_tokens": 336402467.0, + "step": 12999 + }, + { + "epoch": 1.4276301339775972, + "grad_norm": 2.2393345832824707, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.723456084728241, + "num_tokens": 336426745.0, + "step": 13000 + }, + { + "epoch": 1.427739951680211, + "grad_norm": 1.9429808855056763, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7167177200317383, + "num_tokens": 336456691.0, + "step": 13001 + }, + { + "epoch": 1.4278497693828245, + "grad_norm": 2.1282317638397217, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.736790657043457, + "num_tokens": 336481532.0, + "step": 13002 + }, + { + "epoch": 1.4279595870854382, + "grad_norm": 2.351304769515991, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7473317384719849, + "num_tokens": 336502282.0, + "step": 13003 + }, + { + "epoch": 1.4280694047880518, + "grad_norm": 1.92348051071167, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7042616009712219, + "num_tokens": 336536836.0, + "step": 13004 + }, + { + "epoch": 1.4281792224906655, + "grad_norm": 2.3577816486358643, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7310006618499756, + "num_tokens": 336559622.0, + "step": 13005 + }, + { + "epoch": 1.4282890401932793, + "grad_norm": 2.1711275577545166, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7299439907073975, + "num_tokens": 336586090.0, + "step": 13006 + }, + { + "epoch": 1.4283988578958928, + "grad_norm": 2.2930119037628174, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7263966202735901, + "num_tokens": 336610788.0, + "step": 13007 + }, + { + "epoch": 1.4285086755985064, + "grad_norm": 2.1056833267211914, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7248396277427673, + "num_tokens": 336637114.0, + "step": 13008 + }, + { + "epoch": 1.4286184933011201, + "grad_norm": 2.2504806518554688, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7177208065986633, + "num_tokens": 336662254.0, + "step": 13009 + }, + { + "epoch": 1.4287283110037339, + "grad_norm": 2.010704755783081, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7041090726852417, + "num_tokens": 336692324.0, + "step": 13010 + }, + { + "epoch": 1.4288381287063474, + "grad_norm": 2.304163932800293, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7069598436355591, + "num_tokens": 336719362.0, + "step": 13011 + }, + { + "epoch": 1.4289479464089612, + "grad_norm": 2.578447103500366, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7215856313705444, + "num_tokens": 336739122.0, + "step": 13012 + }, + { + "epoch": 1.4290577641115747, + "grad_norm": 2.0350236892700195, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.69794100522995, + "num_tokens": 336772121.0, + "step": 13013 + }, + { + "epoch": 1.4291675818141885, + "grad_norm": 2.251560926437378, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7064805030822754, + "num_tokens": 336796925.0, + "step": 13014 + }, + { + "epoch": 1.4292773995168022, + "grad_norm": 2.344867706298828, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7123720645904541, + "num_tokens": 336821568.0, + "step": 13015 + }, + { + "epoch": 1.4293872172194158, + "grad_norm": 2.2913308143615723, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7158235311508179, + "num_tokens": 336847377.0, + "step": 13016 + }, + { + "epoch": 1.4294970349220295, + "grad_norm": 2.177659273147583, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6914589405059814, + "num_tokens": 336872968.0, + "step": 13017 + }, + { + "epoch": 1.429606852624643, + "grad_norm": 2.3257997035980225, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7046164274215698, + "num_tokens": 336897484.0, + "step": 13018 + }, + { + "epoch": 1.4297166703272568, + "grad_norm": 2.045416831970215, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7382967472076416, + "num_tokens": 336925818.0, + "step": 13019 + }, + { + "epoch": 1.4298264880298703, + "grad_norm": 2.3196141719818115, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7337860465049744, + "num_tokens": 336949216.0, + "step": 13020 + }, + { + "epoch": 1.429936305732484, + "grad_norm": 2.5242741107940674, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7100785970687866, + "num_tokens": 336970982.0, + "step": 13021 + }, + { + "epoch": 1.4300461234350976, + "grad_norm": 1.9485785961151123, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.7043401598930359, + "num_tokens": 337003995.0, + "step": 13022 + }, + { + "epoch": 1.4301559411377114, + "grad_norm": 2.836388349533081, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7447388172149658, + "num_tokens": 337022070.0, + "step": 13023 + }, + { + "epoch": 1.4302657588403251, + "grad_norm": 2.275547981262207, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7233592867851257, + "num_tokens": 337044754.0, + "step": 13024 + }, + { + "epoch": 1.4303755765429387, + "grad_norm": 2.194600820541382, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7024356126785278, + "num_tokens": 337071397.0, + "step": 13025 + }, + { + "epoch": 1.4304853942455524, + "grad_norm": 2.207102060317993, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7008785009384155, + "num_tokens": 337097440.0, + "step": 13026 + }, + { + "epoch": 1.430595211948166, + "grad_norm": 2.3738090991973877, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7174397110939026, + "num_tokens": 337122037.0, + "step": 13027 + }, + { + "epoch": 1.4307050296507797, + "grad_norm": 2.2314205169677734, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.719165027141571, + "num_tokens": 337147695.0, + "step": 13028 + }, + { + "epoch": 1.4308148473533935, + "grad_norm": 2.268026828765869, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7249705791473389, + "num_tokens": 337174018.0, + "step": 13029 + }, + { + "epoch": 1.430924665056007, + "grad_norm": 2.208787441253662, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7122981548309326, + "num_tokens": 337201036.0, + "step": 13030 + }, + { + "epoch": 1.4310344827586206, + "grad_norm": 2.19923734664917, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7238692045211792, + "num_tokens": 337227358.0, + "step": 13031 + }, + { + "epoch": 1.4311443004612343, + "grad_norm": 2.3058929443359375, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7147345542907715, + "num_tokens": 337251693.0, + "step": 13032 + }, + { + "epoch": 1.431254118163848, + "grad_norm": 2.159046173095703, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7262121438980103, + "num_tokens": 337278719.0, + "step": 13033 + }, + { + "epoch": 1.4313639358664616, + "grad_norm": 2.107473850250244, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6952897310256958, + "num_tokens": 337310203.0, + "step": 13034 + }, + { + "epoch": 1.4314737535690754, + "grad_norm": 2.446495771408081, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7183483839035034, + "num_tokens": 337332944.0, + "step": 13035 + }, + { + "epoch": 1.431583571271689, + "grad_norm": 2.546452045440674, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7170261144638062, + "num_tokens": 337353080.0, + "step": 13036 + }, + { + "epoch": 1.4316933889743026, + "grad_norm": 1.9048887491226196, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6858847141265869, + "num_tokens": 337388005.0, + "step": 13037 + }, + { + "epoch": 1.4318032066769164, + "grad_norm": 2.3272030353546143, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7166866660118103, + "num_tokens": 337412486.0, + "step": 13038 + }, + { + "epoch": 1.43191302437953, + "grad_norm": 2.053961753845215, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7057088613510132, + "num_tokens": 337442562.0, + "step": 13039 + }, + { + "epoch": 1.4320228420821437, + "grad_norm": 2.6149721145629883, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7388977408409119, + "num_tokens": 337462022.0, + "step": 13040 + }, + { + "epoch": 1.4321326597847572, + "grad_norm": 2.1368961334228516, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6863409876823425, + "num_tokens": 337489573.0, + "step": 13041 + }, + { + "epoch": 1.432242477487371, + "grad_norm": 2.2206945419311523, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7066394090652466, + "num_tokens": 337517170.0, + "step": 13042 + }, + { + "epoch": 1.4323522951899847, + "grad_norm": 1.913120985031128, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7121745347976685, + "num_tokens": 337550404.0, + "step": 13043 + }, + { + "epoch": 1.4324621128925983, + "grad_norm": 2.27054500579834, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7255427241325378, + "num_tokens": 337573747.0, + "step": 13044 + }, + { + "epoch": 1.4325719305952118, + "grad_norm": 2.3940680027008057, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7302336692810059, + "num_tokens": 337596783.0, + "step": 13045 + }, + { + "epoch": 1.4326817482978256, + "grad_norm": 2.2854087352752686, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7228125333786011, + "num_tokens": 337619950.0, + "step": 13046 + }, + { + "epoch": 1.4327915660004393, + "grad_norm": 1.9649511575698853, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7362947463989258, + "num_tokens": 337650293.0, + "step": 13047 + }, + { + "epoch": 1.4329013837030529, + "grad_norm": 2.1754660606384277, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.717788577079773, + "num_tokens": 337680476.0, + "step": 13048 + }, + { + "epoch": 1.4330112014056666, + "grad_norm": 2.4180843830108643, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7204718589782715, + "num_tokens": 337702464.0, + "step": 13049 + }, + { + "epoch": 1.4331210191082802, + "grad_norm": 2.155003070831299, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.717864990234375, + "num_tokens": 337729715.0, + "step": 13050 + }, + { + "epoch": 1.433230836810894, + "grad_norm": 2.1146812438964844, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7020958065986633, + "num_tokens": 337757375.0, + "step": 13051 + }, + { + "epoch": 1.4333406545135077, + "grad_norm": 2.235653877258301, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7291722297668457, + "num_tokens": 337784525.0, + "step": 13052 + }, + { + "epoch": 1.4334504722161212, + "grad_norm": 2.5673630237579346, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.731764018535614, + "num_tokens": 337807150.0, + "step": 13053 + }, + { + "epoch": 1.433560289918735, + "grad_norm": 2.4482076168060303, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7089579701423645, + "num_tokens": 337830400.0, + "step": 13054 + }, + { + "epoch": 1.4336701076213485, + "grad_norm": 2.1968753337860107, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7271100282669067, + "num_tokens": 337854411.0, + "step": 13055 + }, + { + "epoch": 1.4337799253239623, + "grad_norm": 2.133085250854492, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7314720153808594, + "num_tokens": 337880853.0, + "step": 13056 + }, + { + "epoch": 1.433889743026576, + "grad_norm": 2.2105600833892822, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7232444882392883, + "num_tokens": 337906239.0, + "step": 13057 + }, + { + "epoch": 1.4339995607291895, + "grad_norm": 2.349560499191284, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7483851313591003, + "num_tokens": 337928118.0, + "step": 13058 + }, + { + "epoch": 1.434109378431803, + "grad_norm": 2.153230905532837, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.709384560585022, + "num_tokens": 337957483.0, + "step": 13059 + }, + { + "epoch": 1.4342191961344168, + "grad_norm": 2.2765378952026367, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7229965925216675, + "num_tokens": 337986256.0, + "step": 13060 + }, + { + "epoch": 1.4343290138370306, + "grad_norm": 2.044701099395752, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7289195656776428, + "num_tokens": 338016447.0, + "step": 13061 + }, + { + "epoch": 1.4344388315396441, + "grad_norm": 2.3976259231567383, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.6948755383491516, + "num_tokens": 338041049.0, + "step": 13062 + }, + { + "epoch": 1.434548649242258, + "grad_norm": 2.0502805709838867, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7404704689979553, + "num_tokens": 338069754.0, + "step": 13063 + }, + { + "epoch": 1.4346584669448714, + "grad_norm": 2.4235565662384033, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7303674221038818, + "num_tokens": 338090336.0, + "step": 13064 + }, + { + "epoch": 1.4347682846474852, + "grad_norm": 2.0948469638824463, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7296652793884277, + "num_tokens": 338118398.0, + "step": 13065 + }, + { + "epoch": 1.434878102350099, + "grad_norm": 2.1456732749938965, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7321954965591431, + "num_tokens": 338144635.0, + "step": 13066 + }, + { + "epoch": 1.4349879200527125, + "grad_norm": 2.140411376953125, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7053176164627075, + "num_tokens": 338172175.0, + "step": 13067 + }, + { + "epoch": 1.4350977377553262, + "grad_norm": 2.139056444168091, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7147995829582214, + "num_tokens": 338200321.0, + "step": 13068 + }, + { + "epoch": 1.4352075554579398, + "grad_norm": 2.394556760787964, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7245824337005615, + "num_tokens": 338224147.0, + "step": 13069 + }, + { + "epoch": 1.4353173731605535, + "grad_norm": 1.9920780658721924, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7271178960800171, + "num_tokens": 338253884.0, + "step": 13070 + }, + { + "epoch": 1.4354271908631673, + "grad_norm": 2.145526170730591, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7127243280410767, + "num_tokens": 338281900.0, + "step": 13071 + }, + { + "epoch": 1.4355370085657808, + "grad_norm": 2.4248290061950684, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7212117910385132, + "num_tokens": 338305899.0, + "step": 13072 + }, + { + "epoch": 1.4356468262683943, + "grad_norm": 1.7692902088165283, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.707693338394165, + "num_tokens": 338344227.0, + "step": 13073 + }, + { + "epoch": 1.435756643971008, + "grad_norm": 2.3566713333129883, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7141804099082947, + "num_tokens": 338369871.0, + "step": 13074 + }, + { + "epoch": 1.4358664616736219, + "grad_norm": 2.3883907794952393, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7139654755592346, + "num_tokens": 338396378.0, + "step": 13075 + }, + { + "epoch": 1.4359762793762354, + "grad_norm": 2.357830047607422, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7166127562522888, + "num_tokens": 338418917.0, + "step": 13076 + }, + { + "epoch": 1.4360860970788492, + "grad_norm": 2.3339526653289795, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7220752239227295, + "num_tokens": 338443366.0, + "step": 13077 + }, + { + "epoch": 1.4361959147814627, + "grad_norm": 2.4323437213897705, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7437783479690552, + "num_tokens": 338464714.0, + "step": 13078 + }, + { + "epoch": 1.4363057324840764, + "grad_norm": 2.1747703552246094, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7072888612747192, + "num_tokens": 338496171.0, + "step": 13079 + }, + { + "epoch": 1.4364155501866902, + "grad_norm": 2.165466785430908, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7389681339263916, + "num_tokens": 338521962.0, + "step": 13080 + }, + { + "epoch": 1.4365253678893037, + "grad_norm": 2.3407013416290283, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7176297307014465, + "num_tokens": 338547278.0, + "step": 13081 + }, + { + "epoch": 1.4366351855919175, + "grad_norm": 2.253458261489868, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7184836864471436, + "num_tokens": 338574124.0, + "step": 13082 + }, + { + "epoch": 1.436745003294531, + "grad_norm": 2.3901078701019287, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7274091243743896, + "num_tokens": 338598037.0, + "step": 13083 + }, + { + "epoch": 1.4368548209971448, + "grad_norm": 2.1241252422332764, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.70302414894104, + "num_tokens": 338629611.0, + "step": 13084 + }, + { + "epoch": 1.4369646386997583, + "grad_norm": 2.20287823677063, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7581261396408081, + "num_tokens": 338652735.0, + "step": 13085 + }, + { + "epoch": 1.437074456402372, + "grad_norm": 2.481381893157959, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.727850079536438, + "num_tokens": 338673571.0, + "step": 13086 + }, + { + "epoch": 1.4371842741049856, + "grad_norm": 2.191195249557495, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.708149790763855, + "num_tokens": 338700594.0, + "step": 13087 + }, + { + "epoch": 1.4372940918075994, + "grad_norm": 2.5527427196502686, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7292245626449585, + "num_tokens": 338721880.0, + "step": 13088 + }, + { + "epoch": 1.4374039095102131, + "grad_norm": 2.3631277084350586, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7347354888916016, + "num_tokens": 338745508.0, + "step": 13089 + }, + { + "epoch": 1.4375137272128267, + "grad_norm": 2.3799455165863037, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7204295992851257, + "num_tokens": 338769394.0, + "step": 13090 + }, + { + "epoch": 1.4376235449154404, + "grad_norm": 2.097351312637329, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7134297490119934, + "num_tokens": 338799700.0, + "step": 13091 + }, + { + "epoch": 1.437733362618054, + "grad_norm": 2.425687789916992, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7169395089149475, + "num_tokens": 338823255.0, + "step": 13092 + }, + { + "epoch": 1.4378431803206677, + "grad_norm": 1.952224612236023, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7106190323829651, + "num_tokens": 338855495.0, + "step": 13093 + }, + { + "epoch": 1.4379529980232815, + "grad_norm": 2.586493492126465, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7671390771865845, + "num_tokens": 338873672.0, + "step": 13094 + }, + { + "epoch": 1.438062815725895, + "grad_norm": 2.136516809463501, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6991400122642517, + "num_tokens": 338900914.0, + "step": 13095 + }, + { + "epoch": 1.4381726334285085, + "grad_norm": 2.326570749282837, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.713821530342102, + "num_tokens": 338926093.0, + "step": 13096 + }, + { + "epoch": 1.4382824511311223, + "grad_norm": 1.8524011373519897, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7044104933738708, + "num_tokens": 338959557.0, + "step": 13097 + }, + { + "epoch": 1.438392268833736, + "grad_norm": 1.9392930269241333, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6942425966262817, + "num_tokens": 338995283.0, + "step": 13098 + }, + { + "epoch": 1.4385020865363496, + "grad_norm": 2.381582260131836, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7028771638870239, + "num_tokens": 339017549.0, + "step": 13099 + }, + { + "epoch": 1.4386119042389633, + "grad_norm": 2.0445587635040283, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7136926054954529, + "num_tokens": 339045339.0, + "step": 13100 + }, + { + "epoch": 1.4387217219415769, + "grad_norm": 2.055445432662964, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7116128206253052, + "num_tokens": 339076105.0, + "step": 13101 + }, + { + "epoch": 1.4388315396441906, + "grad_norm": 2.2183430194854736, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7017834186553955, + "num_tokens": 339101917.0, + "step": 13102 + }, + { + "epoch": 1.4389413573468044, + "grad_norm": 2.361619472503662, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7244277000427246, + "num_tokens": 339125277.0, + "step": 13103 + }, + { + "epoch": 1.439051175049418, + "grad_norm": 2.6202595233917236, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.738941490650177, + "num_tokens": 339142988.0, + "step": 13104 + }, + { + "epoch": 1.4391609927520317, + "grad_norm": 2.51023268699646, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7113940715789795, + "num_tokens": 339164220.0, + "step": 13105 + }, + { + "epoch": 1.4392708104546452, + "grad_norm": 2.09706449508667, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7083929777145386, + "num_tokens": 339194424.0, + "step": 13106 + }, + { + "epoch": 1.439380628157259, + "grad_norm": 2.286496639251709, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7057896852493286, + "num_tokens": 339219591.0, + "step": 13107 + }, + { + "epoch": 1.4394904458598727, + "grad_norm": 2.034052610397339, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7084394693374634, + "num_tokens": 339250645.0, + "step": 13108 + }, + { + "epoch": 1.4396002635624863, + "grad_norm": 2.2027690410614014, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7076160907745361, + "num_tokens": 339276743.0, + "step": 13109 + }, + { + "epoch": 1.4397100812650998, + "grad_norm": 2.427156686782837, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7343473434448242, + "num_tokens": 339297691.0, + "step": 13110 + }, + { + "epoch": 1.4398198989677136, + "grad_norm": 2.38159441947937, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7225968837738037, + "num_tokens": 339320684.0, + "step": 13111 + }, + { + "epoch": 1.4399297166703273, + "grad_norm": 2.322354316711426, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7231203317642212, + "num_tokens": 339345478.0, + "step": 13112 + }, + { + "epoch": 1.4400395343729409, + "grad_norm": 2.486936569213867, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7221240401268005, + "num_tokens": 339368782.0, + "step": 13113 + }, + { + "epoch": 1.4401493520755546, + "grad_norm": 2.2713022232055664, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7045041918754578, + "num_tokens": 339396266.0, + "step": 13114 + }, + { + "epoch": 1.4402591697781681, + "grad_norm": 2.1069841384887695, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.729153037071228, + "num_tokens": 339425471.0, + "step": 13115 + }, + { + "epoch": 1.440368987480782, + "grad_norm": 2.303577423095703, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7403897047042847, + "num_tokens": 339448977.0, + "step": 13116 + }, + { + "epoch": 1.4404788051833957, + "grad_norm": 2.364238739013672, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.729581356048584, + "num_tokens": 339471164.0, + "step": 13117 + }, + { + "epoch": 1.4405886228860092, + "grad_norm": 2.119335651397705, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7199541330337524, + "num_tokens": 339497571.0, + "step": 13118 + }, + { + "epoch": 1.440698440588623, + "grad_norm": 2.2264297008514404, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7255088090896606, + "num_tokens": 339522326.0, + "step": 13119 + }, + { + "epoch": 1.4408082582912365, + "grad_norm": 2.311767816543579, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7122604250907898, + "num_tokens": 339547243.0, + "step": 13120 + }, + { + "epoch": 1.4409180759938502, + "grad_norm": 2.3499255180358887, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7159404754638672, + "num_tokens": 339570558.0, + "step": 13121 + }, + { + "epoch": 1.441027893696464, + "grad_norm": 2.400965929031372, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7396941184997559, + "num_tokens": 339592445.0, + "step": 13122 + }, + { + "epoch": 1.4411377113990775, + "grad_norm": 2.300604820251465, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7096668481826782, + "num_tokens": 339617239.0, + "step": 13123 + }, + { + "epoch": 1.441247529101691, + "grad_norm": 2.630511522293091, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7162148356437683, + "num_tokens": 339639633.0, + "step": 13124 + }, + { + "epoch": 1.4413573468043048, + "grad_norm": 2.325268268585205, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7250637412071228, + "num_tokens": 339663745.0, + "step": 13125 + }, + { + "epoch": 1.4414671645069186, + "grad_norm": 2.510854482650757, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7267009019851685, + "num_tokens": 339683723.0, + "step": 13126 + }, + { + "epoch": 1.4415769822095321, + "grad_norm": 2.677940845489502, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7447558641433716, + "num_tokens": 339702235.0, + "step": 13127 + }, + { + "epoch": 1.4416867999121459, + "grad_norm": 2.176201820373535, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.743175745010376, + "num_tokens": 339727218.0, + "step": 13128 + }, + { + "epoch": 1.4417966176147594, + "grad_norm": 2.143437623977661, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7253523468971252, + "num_tokens": 339755696.0, + "step": 13129 + }, + { + "epoch": 1.4419064353173732, + "grad_norm": 2.14481782913208, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7131339311599731, + "num_tokens": 339781899.0, + "step": 13130 + }, + { + "epoch": 1.442016253019987, + "grad_norm": 2.344674587249756, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.741345226764679, + "num_tokens": 339805352.0, + "step": 13131 + }, + { + "epoch": 1.4421260707226005, + "grad_norm": 2.259606122970581, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7370873689651489, + "num_tokens": 339829741.0, + "step": 13132 + }, + { + "epoch": 1.4422358884252142, + "grad_norm": 2.4232285022735596, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7286229133605957, + "num_tokens": 339852460.0, + "step": 13133 + }, + { + "epoch": 1.4423457061278278, + "grad_norm": 2.2933835983276367, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7084887027740479, + "num_tokens": 339877706.0, + "step": 13134 + }, + { + "epoch": 1.4424555238304415, + "grad_norm": 2.275479793548584, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7368395328521729, + "num_tokens": 339901952.0, + "step": 13135 + }, + { + "epoch": 1.442565341533055, + "grad_norm": 2.4608070850372314, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.724493682384491, + "num_tokens": 339924641.0, + "step": 13136 + }, + { + "epoch": 1.4426751592356688, + "grad_norm": 1.9272325038909912, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7212890386581421, + "num_tokens": 339955650.0, + "step": 13137 + }, + { + "epoch": 1.4427849769382823, + "grad_norm": 2.184138059616089, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7156080007553101, + "num_tokens": 339982083.0, + "step": 13138 + }, + { + "epoch": 1.442894794640896, + "grad_norm": 2.3909900188446045, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6924937963485718, + "num_tokens": 340003890.0, + "step": 13139 + }, + { + "epoch": 1.4430046123435099, + "grad_norm": 2.299161434173584, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7145917415618896, + "num_tokens": 340028979.0, + "step": 13140 + }, + { + "epoch": 1.4431144300461234, + "grad_norm": 2.4023280143737793, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.721505880355835, + "num_tokens": 340051821.0, + "step": 13141 + }, + { + "epoch": 1.4432242477487371, + "grad_norm": 2.2113959789276123, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7297248840332031, + "num_tokens": 340077118.0, + "step": 13142 + }, + { + "epoch": 1.4433340654513507, + "grad_norm": 2.1855247020721436, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7083824872970581, + "num_tokens": 340103599.0, + "step": 13143 + }, + { + "epoch": 1.4434438831539644, + "grad_norm": 2.4184954166412354, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.739189863204956, + "num_tokens": 340125652.0, + "step": 13144 + }, + { + "epoch": 1.4435537008565782, + "grad_norm": 2.284480333328247, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7275635004043579, + "num_tokens": 340150073.0, + "step": 13145 + }, + { + "epoch": 1.4436635185591917, + "grad_norm": 2.4023494720458984, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7367081642150879, + "num_tokens": 340172018.0, + "step": 13146 + }, + { + "epoch": 1.4437733362618055, + "grad_norm": 2.2320573329925537, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7111454010009766, + "num_tokens": 340197922.0, + "step": 13147 + }, + { + "epoch": 1.443883153964419, + "grad_norm": 1.9316390752792358, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6883702278137207, + "num_tokens": 340232043.0, + "step": 13148 + }, + { + "epoch": 1.4439929716670328, + "grad_norm": 1.9507293701171875, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7078474164009094, + "num_tokens": 340263950.0, + "step": 13149 + }, + { + "epoch": 1.4441027893696463, + "grad_norm": 2.2481496334075928, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7205009460449219, + "num_tokens": 340290039.0, + "step": 13150 + }, + { + "epoch": 1.44421260707226, + "grad_norm": 2.4460275173187256, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7070329189300537, + "num_tokens": 340312673.0, + "step": 13151 + }, + { + "epoch": 1.4443224247748736, + "grad_norm": 2.2104437351226807, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7352056503295898, + "num_tokens": 340339078.0, + "step": 13152 + }, + { + "epoch": 1.4444322424774874, + "grad_norm": 2.2546744346618652, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7150660753250122, + "num_tokens": 340364148.0, + "step": 13153 + }, + { + "epoch": 1.4445420601801011, + "grad_norm": 2.154618263244629, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7283446788787842, + "num_tokens": 340390357.0, + "step": 13154 + }, + { + "epoch": 1.4446518778827147, + "grad_norm": 2.3232836723327637, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7289866209030151, + "num_tokens": 340414598.0, + "step": 13155 + }, + { + "epoch": 1.4447616955853284, + "grad_norm": 2.3835809230804443, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7154984474182129, + "num_tokens": 340436416.0, + "step": 13156 + }, + { + "epoch": 1.444871513287942, + "grad_norm": 2.001460552215576, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7133458256721497, + "num_tokens": 340468002.0, + "step": 13157 + }, + { + "epoch": 1.4449813309905557, + "grad_norm": 2.607959508895874, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7123867869377136, + "num_tokens": 340490276.0, + "step": 13158 + }, + { + "epoch": 1.4450911486931695, + "grad_norm": 2.530581474304199, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7304019927978516, + "num_tokens": 340511926.0, + "step": 13159 + }, + { + "epoch": 1.445200966395783, + "grad_norm": 2.173210620880127, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7275005578994751, + "num_tokens": 340537993.0, + "step": 13160 + }, + { + "epoch": 1.4453107840983965, + "grad_norm": 2.24814510345459, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7118611931800842, + "num_tokens": 340562051.0, + "step": 13161 + }, + { + "epoch": 1.4454206018010103, + "grad_norm": 2.3773651123046875, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7098034620285034, + "num_tokens": 340584541.0, + "step": 13162 + }, + { + "epoch": 1.445530419503624, + "grad_norm": 2.448882818222046, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7305424809455872, + "num_tokens": 340607199.0, + "step": 13163 + }, + { + "epoch": 1.4456402372062376, + "grad_norm": 2.352450132369995, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7154397964477539, + "num_tokens": 340631608.0, + "step": 13164 + }, + { + "epoch": 1.4457500549088513, + "grad_norm": 2.111628293991089, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7005000114440918, + "num_tokens": 340659819.0, + "step": 13165 + }, + { + "epoch": 1.4458598726114649, + "grad_norm": 2.6407864093780518, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7378185987472534, + "num_tokens": 340678889.0, + "step": 13166 + }, + { + "epoch": 1.4459696903140786, + "grad_norm": 2.3661816120147705, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7190310955047607, + "num_tokens": 340704348.0, + "step": 13167 + }, + { + "epoch": 1.4460795080166924, + "grad_norm": 2.2128348350524902, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7281487584114075, + "num_tokens": 340728379.0, + "step": 13168 + }, + { + "epoch": 1.446189325719306, + "grad_norm": 2.2847702503204346, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7053353190422058, + "num_tokens": 340753423.0, + "step": 13169 + }, + { + "epoch": 1.4462991434219197, + "grad_norm": 2.1641485691070557, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7095869183540344, + "num_tokens": 340779801.0, + "step": 13170 + }, + { + "epoch": 1.4464089611245332, + "grad_norm": 2.387025833129883, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.706264317035675, + "num_tokens": 340805614.0, + "step": 13171 + }, + { + "epoch": 1.446518778827147, + "grad_norm": 1.876460075378418, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7130748629570007, + "num_tokens": 340840393.0, + "step": 13172 + }, + { + "epoch": 1.4466285965297607, + "grad_norm": 2.0211267471313477, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7050833106040955, + "num_tokens": 340871054.0, + "step": 13173 + }, + { + "epoch": 1.4467384142323743, + "grad_norm": 2.19743275642395, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6913391351699829, + "num_tokens": 340898566.0, + "step": 13174 + }, + { + "epoch": 1.4468482319349878, + "grad_norm": 2.309300184249878, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7431459426879883, + "num_tokens": 340921348.0, + "step": 13175 + }, + { + "epoch": 1.4469580496376016, + "grad_norm": 2.2047505378723145, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7180150747299194, + "num_tokens": 340948757.0, + "step": 13176 + }, + { + "epoch": 1.4470678673402153, + "grad_norm": 2.343592405319214, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7115938067436218, + "num_tokens": 340975774.0, + "step": 13177 + }, + { + "epoch": 1.4471776850428288, + "grad_norm": 2.505460500717163, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7401260137557983, + "num_tokens": 340996031.0, + "step": 13178 + }, + { + "epoch": 1.4472875027454426, + "grad_norm": 2.4156763553619385, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7125441431999207, + "num_tokens": 341020197.0, + "step": 13179 + }, + { + "epoch": 1.4473973204480561, + "grad_norm": 2.171259641647339, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7228009104728699, + "num_tokens": 341047284.0, + "step": 13180 + }, + { + "epoch": 1.44750713815067, + "grad_norm": 2.316209316253662, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.706904947757721, + "num_tokens": 341074120.0, + "step": 13181 + }, + { + "epoch": 1.4476169558532836, + "grad_norm": 2.038621664047241, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6987422704696655, + "num_tokens": 341106844.0, + "step": 13182 + }, + { + "epoch": 1.4477267735558972, + "grad_norm": 2.13252329826355, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6915972828865051, + "num_tokens": 341133552.0, + "step": 13183 + }, + { + "epoch": 1.447836591258511, + "grad_norm": 2.182687282562256, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7422983646392822, + "num_tokens": 341159404.0, + "step": 13184 + }, + { + "epoch": 1.4479464089611245, + "grad_norm": 2.085383653640747, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7135674953460693, + "num_tokens": 341188828.0, + "step": 13185 + }, + { + "epoch": 1.4480562266637382, + "grad_norm": 2.260096788406372, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7308428883552551, + "num_tokens": 341212905.0, + "step": 13186 + }, + { + "epoch": 1.448166044366352, + "grad_norm": 2.196007013320923, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.746208667755127, + "num_tokens": 341237002.0, + "step": 13187 + }, + { + "epoch": 1.4482758620689655, + "grad_norm": 2.051588296890259, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7124723196029663, + "num_tokens": 341268766.0, + "step": 13188 + }, + { + "epoch": 1.448385679771579, + "grad_norm": 2.542576789855957, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7267622351646423, + "num_tokens": 341290040.0, + "step": 13189 + }, + { + "epoch": 1.4484954974741928, + "grad_norm": 1.9664241075515747, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6913957595825195, + "num_tokens": 341322524.0, + "step": 13190 + }, + { + "epoch": 1.4486053151768066, + "grad_norm": 2.381704330444336, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7075936198234558, + "num_tokens": 341345369.0, + "step": 13191 + }, + { + "epoch": 1.44871513287942, + "grad_norm": 2.267669916152954, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7223255634307861, + "num_tokens": 341370538.0, + "step": 13192 + }, + { + "epoch": 1.4488249505820339, + "grad_norm": 1.9588569402694702, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7080267667770386, + "num_tokens": 341402489.0, + "step": 13193 + }, + { + "epoch": 1.4489347682846474, + "grad_norm": 2.1688201427459717, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7197178602218628, + "num_tokens": 341428296.0, + "step": 13194 + }, + { + "epoch": 1.4490445859872612, + "grad_norm": 2.192375659942627, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7065190672874451, + "num_tokens": 341455598.0, + "step": 13195 + }, + { + "epoch": 1.449154403689875, + "grad_norm": 2.1601951122283936, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7151634693145752, + "num_tokens": 341485051.0, + "step": 13196 + }, + { + "epoch": 1.4492642213924884, + "grad_norm": 2.2398016452789307, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7169761657714844, + "num_tokens": 341508687.0, + "step": 13197 + }, + { + "epoch": 1.4493740390951022, + "grad_norm": 2.1170578002929688, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7477484941482544, + "num_tokens": 341536130.0, + "step": 13198 + }, + { + "epoch": 1.4494838567977157, + "grad_norm": 2.2789573669433594, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7215858697891235, + "num_tokens": 341560952.0, + "step": 13199 + }, + { + "epoch": 1.4495936745003295, + "grad_norm": 2.0537829399108887, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7096320986747742, + "num_tokens": 341591777.0, + "step": 13200 + }, + { + "epoch": 1.449703492202943, + "grad_norm": 2.207426071166992, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7129281759262085, + "num_tokens": 341621448.0, + "step": 13201 + }, + { + "epoch": 1.4498133099055568, + "grad_norm": 2.350586414337158, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7002718448638916, + "num_tokens": 341647764.0, + "step": 13202 + }, + { + "epoch": 1.4499231276081703, + "grad_norm": 2.1745903491973877, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6853163242340088, + "num_tokens": 341677722.0, + "step": 13203 + }, + { + "epoch": 1.450032945310784, + "grad_norm": 2.4360575675964355, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7085568904876709, + "num_tokens": 341699976.0, + "step": 13204 + }, + { + "epoch": 1.4501427630133978, + "grad_norm": 2.2804970741271973, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7330713868141174, + "num_tokens": 341727137.0, + "step": 13205 + }, + { + "epoch": 1.4502525807160114, + "grad_norm": 2.025099992752075, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7178364396095276, + "num_tokens": 341756721.0, + "step": 13206 + }, + { + "epoch": 1.4503623984186251, + "grad_norm": 2.3651907444000244, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7029211521148682, + "num_tokens": 341781790.0, + "step": 13207 + }, + { + "epoch": 1.4504722161212387, + "grad_norm": 2.3648476600646973, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7262166738510132, + "num_tokens": 341804509.0, + "step": 13208 + }, + { + "epoch": 1.4505820338238524, + "grad_norm": 2.5219874382019043, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7029989957809448, + "num_tokens": 341828413.0, + "step": 13209 + }, + { + "epoch": 1.4506918515264662, + "grad_norm": 2.2532989978790283, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7319326996803284, + "num_tokens": 341851612.0, + "step": 13210 + }, + { + "epoch": 1.4508016692290797, + "grad_norm": 2.4349639415740967, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7555216550827026, + "num_tokens": 341873403.0, + "step": 13211 + }, + { + "epoch": 1.4509114869316933, + "grad_norm": 2.0324487686157227, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7205281853675842, + "num_tokens": 341903244.0, + "step": 13212 + }, + { + "epoch": 1.451021304634307, + "grad_norm": 2.797550916671753, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7224386930465698, + "num_tokens": 341921413.0, + "step": 13213 + }, + { + "epoch": 1.4511311223369208, + "grad_norm": 2.264763355255127, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7262325882911682, + "num_tokens": 341946752.0, + "step": 13214 + }, + { + "epoch": 1.4512409400395343, + "grad_norm": 2.170487880706787, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7238224744796753, + "num_tokens": 341974994.0, + "step": 13215 + }, + { + "epoch": 1.451350757742148, + "grad_norm": 2.2402749061584473, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7109511494636536, + "num_tokens": 342001400.0, + "step": 13216 + }, + { + "epoch": 1.4514605754447616, + "grad_norm": 2.4120523929595947, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.70917809009552, + "num_tokens": 342025161.0, + "step": 13217 + }, + { + "epoch": 1.4515703931473753, + "grad_norm": 2.1963346004486084, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7274117469787598, + "num_tokens": 342052036.0, + "step": 13218 + }, + { + "epoch": 1.451680210849989, + "grad_norm": 2.0252532958984375, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7101461291313171, + "num_tokens": 342081345.0, + "step": 13219 + }, + { + "epoch": 1.4517900285526026, + "grad_norm": 2.4280409812927246, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7177733778953552, + "num_tokens": 342103168.0, + "step": 13220 + }, + { + "epoch": 1.4518998462552164, + "grad_norm": 2.259878635406494, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7170565128326416, + "num_tokens": 342128722.0, + "step": 13221 + }, + { + "epoch": 1.45200966395783, + "grad_norm": 2.2359445095062256, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.712056040763855, + "num_tokens": 342153962.0, + "step": 13222 + }, + { + "epoch": 1.4521194816604437, + "grad_norm": 2.437391996383667, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7014079093933105, + "num_tokens": 342177007.0, + "step": 13223 + }, + { + "epoch": 1.4522292993630574, + "grad_norm": 2.251771926879883, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7195296287536621, + "num_tokens": 342204681.0, + "step": 13224 + }, + { + "epoch": 1.452339117065671, + "grad_norm": 2.5356647968292236, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7447173595428467, + "num_tokens": 342223924.0, + "step": 13225 + }, + { + "epoch": 1.4524489347682845, + "grad_norm": 2.1884732246398926, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7325706481933594, + "num_tokens": 342250807.0, + "step": 13226 + }, + { + "epoch": 1.4525587524708983, + "grad_norm": 2.1448493003845215, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7262649536132812, + "num_tokens": 342275965.0, + "step": 13227 + }, + { + "epoch": 1.452668570173512, + "grad_norm": 2.0256941318511963, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6865711808204651, + "num_tokens": 342307194.0, + "step": 13228 + }, + { + "epoch": 1.4527783878761256, + "grad_norm": 2.1577682495117188, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7150474786758423, + "num_tokens": 342335970.0, + "step": 13229 + }, + { + "epoch": 1.4528882055787393, + "grad_norm": 2.3294296264648438, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.719889223575592, + "num_tokens": 342359907.0, + "step": 13230 + }, + { + "epoch": 1.4529980232813529, + "grad_norm": 2.2714648246765137, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7260974645614624, + "num_tokens": 342384495.0, + "step": 13231 + }, + { + "epoch": 1.4531078409839666, + "grad_norm": 1.9518132209777832, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7003341913223267, + "num_tokens": 342419582.0, + "step": 13232 + }, + { + "epoch": 1.4532176586865804, + "grad_norm": 2.236677408218384, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6963605880737305, + "num_tokens": 342446851.0, + "step": 13233 + }, + { + "epoch": 1.453327476389194, + "grad_norm": 2.343900680541992, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7359062433242798, + "num_tokens": 342470904.0, + "step": 13234 + }, + { + "epoch": 1.4534372940918077, + "grad_norm": 2.482064962387085, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.729052722454071, + "num_tokens": 342491848.0, + "step": 13235 + }, + { + "epoch": 1.4535471117944212, + "grad_norm": 2.0052475929260254, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7137278318405151, + "num_tokens": 342521773.0, + "step": 13236 + }, + { + "epoch": 1.453656929497035, + "grad_norm": 2.2420742511749268, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6947738528251648, + "num_tokens": 342548782.0, + "step": 13237 + }, + { + "epoch": 1.4537667471996487, + "grad_norm": 2.070833444595337, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.688541829586029, + "num_tokens": 342581713.0, + "step": 13238 + }, + { + "epoch": 1.4538765649022622, + "grad_norm": 2.1178436279296875, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7301495671272278, + "num_tokens": 342608817.0, + "step": 13239 + }, + { + "epoch": 1.4539863826048758, + "grad_norm": 2.0207455158233643, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7239490151405334, + "num_tokens": 342635871.0, + "step": 13240 + }, + { + "epoch": 1.4540962003074895, + "grad_norm": 2.2519824504852295, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7190614342689514, + "num_tokens": 342660502.0, + "step": 13241 + }, + { + "epoch": 1.4542060180101033, + "grad_norm": 2.4509048461914062, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7305681109428406, + "num_tokens": 342681603.0, + "step": 13242 + }, + { + "epoch": 1.4543158357127168, + "grad_norm": 2.165100574493408, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7308893799781799, + "num_tokens": 342707872.0, + "step": 13243 + }, + { + "epoch": 1.4544256534153306, + "grad_norm": 2.218064546585083, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6943817138671875, + "num_tokens": 342738552.0, + "step": 13244 + }, + { + "epoch": 1.4545354711179441, + "grad_norm": 2.302919864654541, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7243342399597168, + "num_tokens": 342762795.0, + "step": 13245 + }, + { + "epoch": 1.4546452888205579, + "grad_norm": 2.3501949310302734, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7275198698043823, + "num_tokens": 342787005.0, + "step": 13246 + }, + { + "epoch": 1.4547551065231716, + "grad_norm": 2.474027395248413, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7115696668624878, + "num_tokens": 342807908.0, + "step": 13247 + }, + { + "epoch": 1.4548649242257852, + "grad_norm": 2.06917405128479, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.700860857963562, + "num_tokens": 342836125.0, + "step": 13248 + }, + { + "epoch": 1.454974741928399, + "grad_norm": 1.9187419414520264, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.733960747718811, + "num_tokens": 342868122.0, + "step": 13249 + }, + { + "epoch": 1.4550845596310125, + "grad_norm": 2.114513874053955, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7276476621627808, + "num_tokens": 342895977.0, + "step": 13250 + }, + { + "epoch": 1.4551943773336262, + "grad_norm": 2.227661609649658, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7071581482887268, + "num_tokens": 342921265.0, + "step": 13251 + }, + { + "epoch": 1.45530419503624, + "grad_norm": 2.2139387130737305, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7221567630767822, + "num_tokens": 342948013.0, + "step": 13252 + }, + { + "epoch": 1.4554140127388535, + "grad_norm": 1.9699422121047974, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7192922234535217, + "num_tokens": 342979030.0, + "step": 13253 + }, + { + "epoch": 1.455523830441467, + "grad_norm": 2.3966896533966064, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7267356514930725, + "num_tokens": 343001334.0, + "step": 13254 + }, + { + "epoch": 1.4556336481440808, + "grad_norm": 2.141514301300049, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7095916867256165, + "num_tokens": 343030971.0, + "step": 13255 + }, + { + "epoch": 1.4557434658466946, + "grad_norm": 2.290158987045288, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7039112448692322, + "num_tokens": 343057286.0, + "step": 13256 + }, + { + "epoch": 1.455853283549308, + "grad_norm": 2.090250015258789, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7078552842140198, + "num_tokens": 343087943.0, + "step": 13257 + }, + { + "epoch": 1.4559631012519219, + "grad_norm": 2.410160541534424, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7313817143440247, + "num_tokens": 343110953.0, + "step": 13258 + }, + { + "epoch": 1.4560729189545354, + "grad_norm": 2.445817708969116, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.71905517578125, + "num_tokens": 343133106.0, + "step": 13259 + }, + { + "epoch": 1.4561827366571491, + "grad_norm": 2.3145384788513184, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7262855172157288, + "num_tokens": 343157036.0, + "step": 13260 + }, + { + "epoch": 1.456292554359763, + "grad_norm": 2.1371688842773438, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6917104125022888, + "num_tokens": 343186413.0, + "step": 13261 + }, + { + "epoch": 1.4564023720623764, + "grad_norm": 1.9963228702545166, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7013154625892639, + "num_tokens": 343218981.0, + "step": 13262 + }, + { + "epoch": 1.4565121897649902, + "grad_norm": 2.134742259979248, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7070897221565247, + "num_tokens": 343245759.0, + "step": 13263 + }, + { + "epoch": 1.4566220074676037, + "grad_norm": 2.4750661849975586, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7641835808753967, + "num_tokens": 343265382.0, + "step": 13264 + }, + { + "epoch": 1.4567318251702175, + "grad_norm": 2.51662015914917, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7043856382369995, + "num_tokens": 343287786.0, + "step": 13265 + }, + { + "epoch": 1.456841642872831, + "grad_norm": 2.256293296813965, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.725593090057373, + "num_tokens": 343312571.0, + "step": 13266 + }, + { + "epoch": 1.4569514605754448, + "grad_norm": 2.1826283931732178, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.6998796463012695, + "num_tokens": 343338845.0, + "step": 13267 + }, + { + "epoch": 1.4570612782780583, + "grad_norm": 2.0913236141204834, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7101504802703857, + "num_tokens": 343366259.0, + "step": 13268 + }, + { + "epoch": 1.457171095980672, + "grad_norm": 2.332426071166992, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7079533338546753, + "num_tokens": 343389709.0, + "step": 13269 + }, + { + "epoch": 1.4572809136832858, + "grad_norm": 2.5001795291900635, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7534598708152771, + "num_tokens": 343409779.0, + "step": 13270 + }, + { + "epoch": 1.4573907313858994, + "grad_norm": 2.562770366668701, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7258436679840088, + "num_tokens": 343433056.0, + "step": 13271 + }, + { + "epoch": 1.4575005490885131, + "grad_norm": 2.4219727516174316, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7191634774208069, + "num_tokens": 343455858.0, + "step": 13272 + }, + { + "epoch": 1.4576103667911267, + "grad_norm": 2.7189745903015137, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7256897687911987, + "num_tokens": 343474974.0, + "step": 13273 + }, + { + "epoch": 1.4577201844937404, + "grad_norm": 2.538632392883301, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7187986373901367, + "num_tokens": 343494711.0, + "step": 13274 + }, + { + "epoch": 1.4578300021963542, + "grad_norm": 2.216679096221924, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7126128077507019, + "num_tokens": 343523387.0, + "step": 13275 + }, + { + "epoch": 1.4579398198989677, + "grad_norm": 2.5069704055786133, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7365524768829346, + "num_tokens": 343544079.0, + "step": 13276 + }, + { + "epoch": 1.4580496376015812, + "grad_norm": 2.0544047355651855, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7219337224960327, + "num_tokens": 343571845.0, + "step": 13277 + }, + { + "epoch": 1.458159455304195, + "grad_norm": 2.1870763301849365, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.725753903388977, + "num_tokens": 343596591.0, + "step": 13278 + }, + { + "epoch": 1.4582692730068088, + "grad_norm": 2.1415815353393555, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7308440208435059, + "num_tokens": 343622319.0, + "step": 13279 + }, + { + "epoch": 1.4583790907094223, + "grad_norm": 2.2691571712493896, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7201635837554932, + "num_tokens": 343645477.0, + "step": 13280 + }, + { + "epoch": 1.458488908412036, + "grad_norm": 2.583404541015625, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7163487672805786, + "num_tokens": 343664693.0, + "step": 13281 + }, + { + "epoch": 1.4585987261146496, + "grad_norm": 2.576310634613037, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7469390630722046, + "num_tokens": 343683758.0, + "step": 13282 + }, + { + "epoch": 1.4587085438172633, + "grad_norm": 2.2725751399993896, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7092388868331909, + "num_tokens": 343708334.0, + "step": 13283 + }, + { + "epoch": 1.458818361519877, + "grad_norm": 2.0438809394836426, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7130929231643677, + "num_tokens": 343740386.0, + "step": 13284 + }, + { + "epoch": 1.4589281792224906, + "grad_norm": 2.2658915519714355, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7318048477172852, + "num_tokens": 343765478.0, + "step": 13285 + }, + { + "epoch": 1.4590379969251044, + "grad_norm": 2.389298915863037, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7316200137138367, + "num_tokens": 343790678.0, + "step": 13286 + }, + { + "epoch": 1.459147814627718, + "grad_norm": 2.0823731422424316, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7150384187698364, + "num_tokens": 343818608.0, + "step": 13287 + }, + { + "epoch": 1.4592576323303317, + "grad_norm": 2.114305019378662, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.724681556224823, + "num_tokens": 343844894.0, + "step": 13288 + }, + { + "epoch": 1.4593674500329454, + "grad_norm": 2.390336036682129, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7200336456298828, + "num_tokens": 343866536.0, + "step": 13289 + }, + { + "epoch": 1.459477267735559, + "grad_norm": 2.002747058868408, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7160770297050476, + "num_tokens": 343894798.0, + "step": 13290 + }, + { + "epoch": 1.4595870854381725, + "grad_norm": 2.2072856426239014, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7007802128791809, + "num_tokens": 343921430.0, + "step": 13291 + }, + { + "epoch": 1.4596969031407863, + "grad_norm": 2.080627202987671, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7340254783630371, + "num_tokens": 343949715.0, + "step": 13292 + }, + { + "epoch": 1.4598067208434, + "grad_norm": 2.3038506507873535, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7346502542495728, + "num_tokens": 343974416.0, + "step": 13293 + }, + { + "epoch": 1.4599165385460136, + "grad_norm": 2.1339175701141357, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6971575021743774, + "num_tokens": 344004519.0, + "step": 13294 + }, + { + "epoch": 1.4600263562486273, + "grad_norm": 2.217170476913452, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7639729380607605, + "num_tokens": 344028304.0, + "step": 13295 + }, + { + "epoch": 1.4601361739512408, + "grad_norm": 2.2486541271209717, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7348503470420837, + "num_tokens": 344053256.0, + "step": 13296 + }, + { + "epoch": 1.4602459916538546, + "grad_norm": 2.7083842754364014, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7238097190856934, + "num_tokens": 344073480.0, + "step": 13297 + }, + { + "epoch": 1.4603558093564684, + "grad_norm": 2.039292335510254, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.70174241065979, + "num_tokens": 344103404.0, + "step": 13298 + }, + { + "epoch": 1.460465627059082, + "grad_norm": 2.153834819793701, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7241003513336182, + "num_tokens": 344129978.0, + "step": 13299 + }, + { + "epoch": 1.4605754447616957, + "grad_norm": 2.247481346130371, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.742434024810791, + "num_tokens": 344153913.0, + "step": 13300 + }, + { + "epoch": 1.4606852624643092, + "grad_norm": 2.363933563232422, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7206674814224243, + "num_tokens": 344179762.0, + "step": 13301 + }, + { + "epoch": 1.460795080166923, + "grad_norm": 2.385338068008423, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7247693538665771, + "num_tokens": 344203479.0, + "step": 13302 + }, + { + "epoch": 1.4609048978695367, + "grad_norm": 2.3740272521972656, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7140821218490601, + "num_tokens": 344227702.0, + "step": 13303 + }, + { + "epoch": 1.4610147155721502, + "grad_norm": 2.5111286640167236, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7236418128013611, + "num_tokens": 344250615.0, + "step": 13304 + }, + { + "epoch": 1.4611245332747638, + "grad_norm": 2.084968328475952, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6993175745010376, + "num_tokens": 344279535.0, + "step": 13305 + }, + { + "epoch": 1.4612343509773775, + "grad_norm": 1.8677870035171509, + "learning_rate": 1e-06, + "loss": 1.1072, + "mean_token_accuracy": 0.6679448485374451, + "num_tokens": 344315144.0, + "step": 13306 + }, + { + "epoch": 1.4613441686799913, + "grad_norm": 2.2986271381378174, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7147839665412903, + "num_tokens": 344338533.0, + "step": 13307 + }, + { + "epoch": 1.4614539863826048, + "grad_norm": 2.231013774871826, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7102521657943726, + "num_tokens": 344364248.0, + "step": 13308 + }, + { + "epoch": 1.4615638040852186, + "grad_norm": 2.505545139312744, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7299549579620361, + "num_tokens": 344386105.0, + "step": 13309 + }, + { + "epoch": 1.4616736217878321, + "grad_norm": 2.465510129928589, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.723305881023407, + "num_tokens": 344407088.0, + "step": 13310 + }, + { + "epoch": 1.4617834394904459, + "grad_norm": 2.443948745727539, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7149701118469238, + "num_tokens": 344428868.0, + "step": 13311 + }, + { + "epoch": 1.4618932571930596, + "grad_norm": 2.287207841873169, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7065122127532959, + "num_tokens": 344454809.0, + "step": 13312 + }, + { + "epoch": 1.4620030748956732, + "grad_norm": 2.162445545196533, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6923233270645142, + "num_tokens": 344487596.0, + "step": 13313 + }, + { + "epoch": 1.462112892598287, + "grad_norm": 2.2660470008850098, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.6977312564849854, + "num_tokens": 344515855.0, + "step": 13314 + }, + { + "epoch": 1.4622227103009005, + "grad_norm": 2.264047861099243, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7284207344055176, + "num_tokens": 344540736.0, + "step": 13315 + }, + { + "epoch": 1.4623325280035142, + "grad_norm": 2.20954966545105, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.700613260269165, + "num_tokens": 344569990.0, + "step": 13316 + }, + { + "epoch": 1.462442345706128, + "grad_norm": 2.0720033645629883, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7301102876663208, + "num_tokens": 344598249.0, + "step": 13317 + }, + { + "epoch": 1.4625521634087415, + "grad_norm": 2.3004000186920166, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.6977630853652954, + "num_tokens": 344623371.0, + "step": 13318 + }, + { + "epoch": 1.462661981111355, + "grad_norm": 2.5867109298706055, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7301170229911804, + "num_tokens": 344645429.0, + "step": 13319 + }, + { + "epoch": 1.4627717988139688, + "grad_norm": 2.2627367973327637, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7293453216552734, + "num_tokens": 344668941.0, + "step": 13320 + }, + { + "epoch": 1.4628816165165826, + "grad_norm": 2.6386616230010986, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.7503441572189331, + "num_tokens": 344687947.0, + "step": 13321 + }, + { + "epoch": 1.462991434219196, + "grad_norm": 2.159137725830078, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7464166879653931, + "num_tokens": 344714531.0, + "step": 13322 + }, + { + "epoch": 1.4631012519218098, + "grad_norm": 1.9693069458007812, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7209212779998779, + "num_tokens": 344745952.0, + "step": 13323 + }, + { + "epoch": 1.4632110696244234, + "grad_norm": 2.3114521503448486, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7222559452056885, + "num_tokens": 344769489.0, + "step": 13324 + }, + { + "epoch": 1.4633208873270371, + "grad_norm": 2.192453384399414, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6888396739959717, + "num_tokens": 344798325.0, + "step": 13325 + }, + { + "epoch": 1.463430705029651, + "grad_norm": 2.125493049621582, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7150262594223022, + "num_tokens": 344826816.0, + "step": 13326 + }, + { + "epoch": 1.4635405227322644, + "grad_norm": 1.9845116138458252, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6998521685600281, + "num_tokens": 344857874.0, + "step": 13327 + }, + { + "epoch": 1.4636503404348782, + "grad_norm": 2.245163917541504, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7063335180282593, + "num_tokens": 344883472.0, + "step": 13328 + }, + { + "epoch": 1.4637601581374917, + "grad_norm": 2.2106151580810547, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7168053388595581, + "num_tokens": 344911717.0, + "step": 13329 + }, + { + "epoch": 1.4638699758401055, + "grad_norm": 2.346039295196533, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7079712748527527, + "num_tokens": 344935917.0, + "step": 13330 + }, + { + "epoch": 1.463979793542719, + "grad_norm": 2.35779070854187, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7253667116165161, + "num_tokens": 344960203.0, + "step": 13331 + }, + { + "epoch": 1.4640896112453328, + "grad_norm": 2.253204584121704, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7298718690872192, + "num_tokens": 344983619.0, + "step": 13332 + }, + { + "epoch": 1.4641994289479463, + "grad_norm": 1.9874361753463745, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7142989635467529, + "num_tokens": 345015122.0, + "step": 13333 + }, + { + "epoch": 1.46430924665056, + "grad_norm": 2.0060322284698486, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7148697376251221, + "num_tokens": 345045631.0, + "step": 13334 + }, + { + "epoch": 1.4644190643531738, + "grad_norm": 2.058936357498169, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7186384201049805, + "num_tokens": 345076630.0, + "step": 13335 + }, + { + "epoch": 1.4645288820557874, + "grad_norm": 2.119014024734497, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7398998141288757, + "num_tokens": 345101823.0, + "step": 13336 + }, + { + "epoch": 1.464638699758401, + "grad_norm": 2.3027544021606445, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7210391759872437, + "num_tokens": 345128795.0, + "step": 13337 + }, + { + "epoch": 1.4647485174610146, + "grad_norm": 2.184420585632324, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7163927555084229, + "num_tokens": 345156852.0, + "step": 13338 + }, + { + "epoch": 1.4648583351636284, + "grad_norm": 2.3200418949127197, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7292934656143188, + "num_tokens": 345182349.0, + "step": 13339 + }, + { + "epoch": 1.4649681528662422, + "grad_norm": 2.1102590560913086, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7325431108474731, + "num_tokens": 345210182.0, + "step": 13340 + }, + { + "epoch": 1.4650779705688557, + "grad_norm": 2.5033318996429443, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.753675103187561, + "num_tokens": 345230677.0, + "step": 13341 + }, + { + "epoch": 1.4651877882714692, + "grad_norm": 2.1915314197540283, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7240860462188721, + "num_tokens": 345256864.0, + "step": 13342 + }, + { + "epoch": 1.465297605974083, + "grad_norm": 2.1740548610687256, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.739727795124054, + "num_tokens": 345283478.0, + "step": 13343 + }, + { + "epoch": 1.4654074236766967, + "grad_norm": 2.2699670791625977, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.709777295589447, + "num_tokens": 345307144.0, + "step": 13344 + }, + { + "epoch": 1.4655172413793103, + "grad_norm": 2.0028700828552246, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.7016085982322693, + "num_tokens": 345336359.0, + "step": 13345 + }, + { + "epoch": 1.465627059081924, + "grad_norm": 2.2320616245269775, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7328732013702393, + "num_tokens": 345359927.0, + "step": 13346 + }, + { + "epoch": 1.4657368767845376, + "grad_norm": 2.5974111557006836, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7342942953109741, + "num_tokens": 345380985.0, + "step": 13347 + }, + { + "epoch": 1.4658466944871513, + "grad_norm": 2.157451629638672, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.6921893954277039, + "num_tokens": 345409928.0, + "step": 13348 + }, + { + "epoch": 1.465956512189765, + "grad_norm": 2.162858486175537, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7206052541732788, + "num_tokens": 345435622.0, + "step": 13349 + }, + { + "epoch": 1.4660663298923786, + "grad_norm": 2.432619571685791, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7174919843673706, + "num_tokens": 345457381.0, + "step": 13350 + }, + { + "epoch": 1.4661761475949924, + "grad_norm": 2.283090114593506, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7302967309951782, + "num_tokens": 345480374.0, + "step": 13351 + }, + { + "epoch": 1.466285965297606, + "grad_norm": 2.26884126663208, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.752266526222229, + "num_tokens": 345505238.0, + "step": 13352 + }, + { + "epoch": 1.4663957830002197, + "grad_norm": 2.2598562240600586, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6999322772026062, + "num_tokens": 345531950.0, + "step": 13353 + }, + { + "epoch": 1.4665056007028334, + "grad_norm": 2.364896774291992, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6962804198265076, + "num_tokens": 345556257.0, + "step": 13354 + }, + { + "epoch": 1.466615418405447, + "grad_norm": 2.2839245796203613, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6928764581680298, + "num_tokens": 345586451.0, + "step": 13355 + }, + { + "epoch": 1.4667252361080605, + "grad_norm": 2.498908042907715, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7182509899139404, + "num_tokens": 345610752.0, + "step": 13356 + }, + { + "epoch": 1.4668350538106742, + "grad_norm": 2.656827926635742, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7323462963104248, + "num_tokens": 345631872.0, + "step": 13357 + }, + { + "epoch": 1.466944871513288, + "grad_norm": 2.1105430126190186, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7013201713562012, + "num_tokens": 345661342.0, + "step": 13358 + }, + { + "epoch": 1.4670546892159015, + "grad_norm": 2.4053854942321777, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7556748390197754, + "num_tokens": 345683908.0, + "step": 13359 + }, + { + "epoch": 1.4671645069185153, + "grad_norm": 2.2354393005371094, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7087440490722656, + "num_tokens": 345709061.0, + "step": 13360 + }, + { + "epoch": 1.4672743246211288, + "grad_norm": 2.29062819480896, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7204002141952515, + "num_tokens": 345734093.0, + "step": 13361 + }, + { + "epoch": 1.4673841423237426, + "grad_norm": 2.184818983078003, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7053719758987427, + "num_tokens": 345762128.0, + "step": 13362 + }, + { + "epoch": 1.4674939600263563, + "grad_norm": 2.2562832832336426, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7216070890426636, + "num_tokens": 345785034.0, + "step": 13363 + }, + { + "epoch": 1.4676037777289699, + "grad_norm": 2.2850685119628906, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7206876277923584, + "num_tokens": 345808752.0, + "step": 13364 + }, + { + "epoch": 1.4677135954315836, + "grad_norm": 2.1563973426818848, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7110649943351746, + "num_tokens": 345838267.0, + "step": 13365 + }, + { + "epoch": 1.4678234131341972, + "grad_norm": 2.2917721271514893, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.736836850643158, + "num_tokens": 345861573.0, + "step": 13366 + }, + { + "epoch": 1.467933230836811, + "grad_norm": 2.1306912899017334, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7181071639060974, + "num_tokens": 345888270.0, + "step": 13367 + }, + { + "epoch": 1.4680430485394247, + "grad_norm": 2.5258307456970215, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7338643074035645, + "num_tokens": 345909122.0, + "step": 13368 + }, + { + "epoch": 1.4681528662420382, + "grad_norm": 1.9327629804611206, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7063125371932983, + "num_tokens": 345941407.0, + "step": 13369 + }, + { + "epoch": 1.4682626839446518, + "grad_norm": 2.1436383724212646, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.71112060546875, + "num_tokens": 345967749.0, + "step": 13370 + }, + { + "epoch": 1.4683725016472655, + "grad_norm": 2.6140661239624023, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7436563968658447, + "num_tokens": 345986427.0, + "step": 13371 + }, + { + "epoch": 1.4684823193498793, + "grad_norm": 2.263126850128174, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7198692560195923, + "num_tokens": 346014159.0, + "step": 13372 + }, + { + "epoch": 1.4685921370524928, + "grad_norm": 2.09635329246521, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7250288724899292, + "num_tokens": 346043607.0, + "step": 13373 + }, + { + "epoch": 1.4687019547551066, + "grad_norm": 2.3458969593048096, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7254449129104614, + "num_tokens": 346068148.0, + "step": 13374 + }, + { + "epoch": 1.46881177245772, + "grad_norm": 2.4907662868499756, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.717508852481842, + "num_tokens": 346089553.0, + "step": 13375 + }, + { + "epoch": 1.4689215901603339, + "grad_norm": 2.1665117740631104, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7094462513923645, + "num_tokens": 346119179.0, + "step": 13376 + }, + { + "epoch": 1.4690314078629476, + "grad_norm": 2.4946255683898926, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7500346899032593, + "num_tokens": 346137783.0, + "step": 13377 + }, + { + "epoch": 1.4691412255655611, + "grad_norm": 2.313323736190796, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7131729125976562, + "num_tokens": 346163726.0, + "step": 13378 + }, + { + "epoch": 1.469251043268175, + "grad_norm": 2.044644355773926, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7106497883796692, + "num_tokens": 346192312.0, + "step": 13379 + }, + { + "epoch": 1.4693608609707884, + "grad_norm": 2.127211332321167, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7055721282958984, + "num_tokens": 346221468.0, + "step": 13380 + }, + { + "epoch": 1.4694706786734022, + "grad_norm": 2.1404316425323486, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7169607281684875, + "num_tokens": 346247660.0, + "step": 13381 + }, + { + "epoch": 1.4695804963760157, + "grad_norm": 2.169978380203247, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7150341272354126, + "num_tokens": 346275419.0, + "step": 13382 + }, + { + "epoch": 1.4696903140786295, + "grad_norm": 2.3157801628112793, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7308589220046997, + "num_tokens": 346299583.0, + "step": 13383 + }, + { + "epoch": 1.469800131781243, + "grad_norm": 2.66209077835083, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7350180745124817, + "num_tokens": 346319314.0, + "step": 13384 + }, + { + "epoch": 1.4699099494838568, + "grad_norm": 2.2267680168151855, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.69974285364151, + "num_tokens": 346344487.0, + "step": 13385 + }, + { + "epoch": 1.4700197671864705, + "grad_norm": 2.1303958892822266, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.728520393371582, + "num_tokens": 346371288.0, + "step": 13386 + }, + { + "epoch": 1.470129584889084, + "grad_norm": 2.2595043182373047, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7300140857696533, + "num_tokens": 346397949.0, + "step": 13387 + }, + { + "epoch": 1.4702394025916978, + "grad_norm": 2.5729377269744873, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.725605845451355, + "num_tokens": 346418085.0, + "step": 13388 + }, + { + "epoch": 1.4703492202943114, + "grad_norm": 2.1060404777526855, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7126609683036804, + "num_tokens": 346445774.0, + "step": 13389 + }, + { + "epoch": 1.4704590379969251, + "grad_norm": 1.918112874031067, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7111167907714844, + "num_tokens": 346479092.0, + "step": 13390 + }, + { + "epoch": 1.4705688556995389, + "grad_norm": 2.6347873210906982, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7092360854148865, + "num_tokens": 346500975.0, + "step": 13391 + }, + { + "epoch": 1.4706786734021524, + "grad_norm": 2.3089253902435303, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.720995306968689, + "num_tokens": 346524780.0, + "step": 13392 + }, + { + "epoch": 1.4707884911047662, + "grad_norm": 2.1820642948150635, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7022082805633545, + "num_tokens": 346554229.0, + "step": 13393 + }, + { + "epoch": 1.4708983088073797, + "grad_norm": 2.103834867477417, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7347314953804016, + "num_tokens": 346581471.0, + "step": 13394 + }, + { + "epoch": 1.4710081265099935, + "grad_norm": 2.2562971115112305, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7278324365615845, + "num_tokens": 346606506.0, + "step": 13395 + }, + { + "epoch": 1.471117944212607, + "grad_norm": 2.2514021396636963, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7060011625289917, + "num_tokens": 346632543.0, + "step": 13396 + }, + { + "epoch": 1.4712277619152208, + "grad_norm": 2.498560905456543, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7258560657501221, + "num_tokens": 346654268.0, + "step": 13397 + }, + { + "epoch": 1.4713375796178343, + "grad_norm": 2.0442044734954834, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7561169266700745, + "num_tokens": 346681670.0, + "step": 13398 + }, + { + "epoch": 1.471447397320448, + "grad_norm": 2.2472424507141113, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7428228855133057, + "num_tokens": 346706343.0, + "step": 13399 + }, + { + "epoch": 1.4715572150230618, + "grad_norm": 2.1946351528167725, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7260186672210693, + "num_tokens": 346731063.0, + "step": 13400 + }, + { + "epoch": 1.4716670327256753, + "grad_norm": 2.309364080429077, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7233759164810181, + "num_tokens": 346755843.0, + "step": 13401 + }, + { + "epoch": 1.471776850428289, + "grad_norm": 2.1905112266540527, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7101601958274841, + "num_tokens": 346782314.0, + "step": 13402 + }, + { + "epoch": 1.4718866681309026, + "grad_norm": 2.4518542289733887, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7364588975906372, + "num_tokens": 346804881.0, + "step": 13403 + }, + { + "epoch": 1.4719964858335164, + "grad_norm": 2.2109813690185547, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.6934217214584351, + "num_tokens": 346833038.0, + "step": 13404 + }, + { + "epoch": 1.4721063035361301, + "grad_norm": 2.4330012798309326, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7383674383163452, + "num_tokens": 346854806.0, + "step": 13405 + }, + { + "epoch": 1.4722161212387437, + "grad_norm": 2.0637476444244385, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7004799842834473, + "num_tokens": 346886827.0, + "step": 13406 + }, + { + "epoch": 1.4723259389413572, + "grad_norm": 2.277265787124634, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7192535400390625, + "num_tokens": 346909719.0, + "step": 13407 + }, + { + "epoch": 1.472435756643971, + "grad_norm": 2.102360248565674, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7212427854537964, + "num_tokens": 346936087.0, + "step": 13408 + }, + { + "epoch": 1.4725455743465847, + "grad_norm": 2.3230814933776855, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7022262811660767, + "num_tokens": 346962253.0, + "step": 13409 + }, + { + "epoch": 1.4726553920491983, + "grad_norm": 2.241004228591919, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7168270349502563, + "num_tokens": 346987083.0, + "step": 13410 + }, + { + "epoch": 1.472765209751812, + "grad_norm": 2.466552495956421, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7360273599624634, + "num_tokens": 347008954.0, + "step": 13411 + }, + { + "epoch": 1.4728750274544256, + "grad_norm": 2.212056875228882, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7083539962768555, + "num_tokens": 347037526.0, + "step": 13412 + }, + { + "epoch": 1.4729848451570393, + "grad_norm": 2.26652455329895, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7310170531272888, + "num_tokens": 347060196.0, + "step": 13413 + }, + { + "epoch": 1.473094662859653, + "grad_norm": 2.666121482849121, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7364549040794373, + "num_tokens": 347079279.0, + "step": 13414 + }, + { + "epoch": 1.4732044805622666, + "grad_norm": 2.3789737224578857, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7218152284622192, + "num_tokens": 347101296.0, + "step": 13415 + }, + { + "epoch": 1.4733142982648804, + "grad_norm": 2.2732019424438477, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7280102968215942, + "num_tokens": 347124779.0, + "step": 13416 + }, + { + "epoch": 1.473424115967494, + "grad_norm": 2.357473611831665, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7283432483673096, + "num_tokens": 347148769.0, + "step": 13417 + }, + { + "epoch": 1.4735339336701077, + "grad_norm": 2.0791068077087402, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7321522235870361, + "num_tokens": 347176103.0, + "step": 13418 + }, + { + "epoch": 1.4736437513727214, + "grad_norm": 2.249298095703125, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7197302579879761, + "num_tokens": 347201415.0, + "step": 13419 + }, + { + "epoch": 1.473753569075335, + "grad_norm": 2.1746294498443604, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7376630306243896, + "num_tokens": 347228122.0, + "step": 13420 + }, + { + "epoch": 1.4738633867779485, + "grad_norm": 2.1117749214172363, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.719694972038269, + "num_tokens": 347254149.0, + "step": 13421 + }, + { + "epoch": 1.4739732044805622, + "grad_norm": 2.3532748222351074, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7298995852470398, + "num_tokens": 347278195.0, + "step": 13422 + }, + { + "epoch": 1.474083022183176, + "grad_norm": 2.2872469425201416, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7479997873306274, + "num_tokens": 347301112.0, + "step": 13423 + }, + { + "epoch": 1.4741928398857895, + "grad_norm": 2.2008354663848877, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7175441980361938, + "num_tokens": 347326955.0, + "step": 13424 + }, + { + "epoch": 1.4743026575884033, + "grad_norm": 2.2702646255493164, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7176387906074524, + "num_tokens": 347352680.0, + "step": 13425 + }, + { + "epoch": 1.4744124752910168, + "grad_norm": 2.2408599853515625, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7161429524421692, + "num_tokens": 347379559.0, + "step": 13426 + }, + { + "epoch": 1.4745222929936306, + "grad_norm": 2.2510597705841064, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7352072596549988, + "num_tokens": 347404330.0, + "step": 13427 + }, + { + "epoch": 1.4746321106962443, + "grad_norm": 2.1105051040649414, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7166889905929565, + "num_tokens": 347432882.0, + "step": 13428 + }, + { + "epoch": 1.4747419283988579, + "grad_norm": 2.1430275440216064, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7242288589477539, + "num_tokens": 347460572.0, + "step": 13429 + }, + { + "epoch": 1.4748517461014716, + "grad_norm": 2.3254570960998535, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7102688550949097, + "num_tokens": 347486650.0, + "step": 13430 + }, + { + "epoch": 1.4749615638040852, + "grad_norm": 2.178983688354492, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7287701368331909, + "num_tokens": 347514758.0, + "step": 13431 + }, + { + "epoch": 1.475071381506699, + "grad_norm": 2.2589094638824463, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7181705832481384, + "num_tokens": 347539986.0, + "step": 13432 + }, + { + "epoch": 1.4751811992093127, + "grad_norm": 2.2348787784576416, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7144057750701904, + "num_tokens": 347563513.0, + "step": 13433 + }, + { + "epoch": 1.4752910169119262, + "grad_norm": 2.0942773818969727, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7291745543479919, + "num_tokens": 347591159.0, + "step": 13434 + }, + { + "epoch": 1.4754008346145397, + "grad_norm": 2.179381847381592, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7235426306724548, + "num_tokens": 347618018.0, + "step": 13435 + }, + { + "epoch": 1.4755106523171535, + "grad_norm": 2.1579172611236572, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7022376656532288, + "num_tokens": 347645281.0, + "step": 13436 + }, + { + "epoch": 1.4756204700197673, + "grad_norm": 2.3268611431121826, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7198827862739563, + "num_tokens": 347669321.0, + "step": 13437 + }, + { + "epoch": 1.4757302877223808, + "grad_norm": 2.186605453491211, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7090470194816589, + "num_tokens": 347696628.0, + "step": 13438 + }, + { + "epoch": 1.4758401054249946, + "grad_norm": 2.3257412910461426, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7100127935409546, + "num_tokens": 347720666.0, + "step": 13439 + }, + { + "epoch": 1.475949923127608, + "grad_norm": 2.2785050868988037, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6896215677261353, + "num_tokens": 347748398.0, + "step": 13440 + }, + { + "epoch": 1.4760597408302218, + "grad_norm": 2.0641300678253174, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6990203857421875, + "num_tokens": 347779219.0, + "step": 13441 + }, + { + "epoch": 1.4761695585328356, + "grad_norm": 2.3900461196899414, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7195733785629272, + "num_tokens": 347800401.0, + "step": 13442 + }, + { + "epoch": 1.4762793762354491, + "grad_norm": 2.291203022003174, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7364693284034729, + "num_tokens": 347824024.0, + "step": 13443 + }, + { + "epoch": 1.476389193938063, + "grad_norm": 2.2267391681671143, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7346233129501343, + "num_tokens": 347850139.0, + "step": 13444 + }, + { + "epoch": 1.4764990116406764, + "grad_norm": 2.15632963180542, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.718708336353302, + "num_tokens": 347875516.0, + "step": 13445 + }, + { + "epoch": 1.4766088293432902, + "grad_norm": 1.9928821325302124, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.6986221075057983, + "num_tokens": 347906133.0, + "step": 13446 + }, + { + "epoch": 1.4767186470459037, + "grad_norm": 2.07460618019104, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6951369047164917, + "num_tokens": 347938177.0, + "step": 13447 + }, + { + "epoch": 1.4768284647485175, + "grad_norm": 2.184842824935913, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7055349349975586, + "num_tokens": 347964338.0, + "step": 13448 + }, + { + "epoch": 1.476938282451131, + "grad_norm": 2.6269922256469727, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7234355211257935, + "num_tokens": 347983881.0, + "step": 13449 + }, + { + "epoch": 1.4770481001537448, + "grad_norm": 2.0948965549468994, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7126590013504028, + "num_tokens": 348011842.0, + "step": 13450 + }, + { + "epoch": 1.4771579178563585, + "grad_norm": 2.329228639602661, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7205134630203247, + "num_tokens": 348037002.0, + "step": 13451 + }, + { + "epoch": 1.477267735558972, + "grad_norm": 2.242856740951538, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7219616770744324, + "num_tokens": 348063768.0, + "step": 13452 + }, + { + "epoch": 1.4773775532615858, + "grad_norm": 2.039642333984375, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7436694502830505, + "num_tokens": 348092857.0, + "step": 13453 + }, + { + "epoch": 1.4774873709641994, + "grad_norm": 2.2925822734832764, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7170109748840332, + "num_tokens": 348116796.0, + "step": 13454 + }, + { + "epoch": 1.477597188666813, + "grad_norm": 2.221944808959961, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7285822629928589, + "num_tokens": 348142158.0, + "step": 13455 + }, + { + "epoch": 1.4777070063694269, + "grad_norm": 2.128854274749756, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7118263840675354, + "num_tokens": 348170191.0, + "step": 13456 + }, + { + "epoch": 1.4778168240720404, + "grad_norm": 2.179715394973755, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7366948127746582, + "num_tokens": 348194238.0, + "step": 13457 + }, + { + "epoch": 1.477926641774654, + "grad_norm": 1.929533839225769, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7426997423171997, + "num_tokens": 348226443.0, + "step": 13458 + }, + { + "epoch": 1.4780364594772677, + "grad_norm": 2.211886167526245, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7166519165039062, + "num_tokens": 348253733.0, + "step": 13459 + }, + { + "epoch": 1.4781462771798815, + "grad_norm": 2.2898917198181152, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7163547873497009, + "num_tokens": 348278012.0, + "step": 13460 + }, + { + "epoch": 1.478256094882495, + "grad_norm": 2.5601940155029297, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7268552184104919, + "num_tokens": 348298231.0, + "step": 13461 + }, + { + "epoch": 1.4783659125851087, + "grad_norm": 1.9118069410324097, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7271610498428345, + "num_tokens": 348330234.0, + "step": 13462 + }, + { + "epoch": 1.4784757302877223, + "grad_norm": 2.1371493339538574, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7059109807014465, + "num_tokens": 348359195.0, + "step": 13463 + }, + { + "epoch": 1.478585547990336, + "grad_norm": 2.108920097351074, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7194769382476807, + "num_tokens": 348385778.0, + "step": 13464 + }, + { + "epoch": 1.4786953656929498, + "grad_norm": 2.4045803546905518, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6913962960243225, + "num_tokens": 348409555.0, + "step": 13465 + }, + { + "epoch": 1.4788051833955633, + "grad_norm": 2.133396863937378, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7261349558830261, + "num_tokens": 348436878.0, + "step": 13466 + }, + { + "epoch": 1.478915001098177, + "grad_norm": 2.2230381965637207, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7236672639846802, + "num_tokens": 348460782.0, + "step": 13467 + }, + { + "epoch": 1.4790248188007906, + "grad_norm": 1.9079458713531494, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6893551349639893, + "num_tokens": 348493364.0, + "step": 13468 + }, + { + "epoch": 1.4791346365034044, + "grad_norm": 1.9961795806884766, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7277700901031494, + "num_tokens": 348522052.0, + "step": 13469 + }, + { + "epoch": 1.4792444542060181, + "grad_norm": 2.162458896636963, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.711458683013916, + "num_tokens": 348550787.0, + "step": 13470 + }, + { + "epoch": 1.4793542719086317, + "grad_norm": 2.5226330757141113, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7465208768844604, + "num_tokens": 348569893.0, + "step": 13471 + }, + { + "epoch": 1.4794640896112452, + "grad_norm": 2.5462660789489746, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7404152154922485, + "num_tokens": 348589231.0, + "step": 13472 + }, + { + "epoch": 1.479573907313859, + "grad_norm": 2.3701374530792236, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7444069981575012, + "num_tokens": 348612204.0, + "step": 13473 + }, + { + "epoch": 1.4796837250164727, + "grad_norm": 2.3373632431030273, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.708238959312439, + "num_tokens": 348635785.0, + "step": 13474 + }, + { + "epoch": 1.4797935427190863, + "grad_norm": 2.055891275405884, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7006966471672058, + "num_tokens": 348667896.0, + "step": 13475 + }, + { + "epoch": 1.4799033604217, + "grad_norm": 2.315613031387329, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7272769808769226, + "num_tokens": 348693407.0, + "step": 13476 + }, + { + "epoch": 1.4800131781243135, + "grad_norm": 2.473137378692627, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7378780841827393, + "num_tokens": 348716369.0, + "step": 13477 + }, + { + "epoch": 1.4801229958269273, + "grad_norm": 2.126310348510742, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7070190906524658, + "num_tokens": 348746705.0, + "step": 13478 + }, + { + "epoch": 1.480232813529541, + "grad_norm": 2.5371408462524414, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7499347925186157, + "num_tokens": 348766027.0, + "step": 13479 + }, + { + "epoch": 1.4803426312321546, + "grad_norm": 2.063241720199585, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7207281589508057, + "num_tokens": 348793646.0, + "step": 13480 + }, + { + "epoch": 1.4804524489347684, + "grad_norm": 2.2125773429870605, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7232188582420349, + "num_tokens": 348820525.0, + "step": 13481 + }, + { + "epoch": 1.4805622666373819, + "grad_norm": 2.2425129413604736, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7112441062927246, + "num_tokens": 348847130.0, + "step": 13482 + }, + { + "epoch": 1.4806720843399956, + "grad_norm": 2.3615992069244385, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7265874743461609, + "num_tokens": 348870238.0, + "step": 13483 + }, + { + "epoch": 1.4807819020426094, + "grad_norm": 2.2124440670013428, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6972448229789734, + "num_tokens": 348898170.0, + "step": 13484 + }, + { + "epoch": 1.480891719745223, + "grad_norm": 2.2033166885375977, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7011677026748657, + "num_tokens": 348928003.0, + "step": 13485 + }, + { + "epoch": 1.4810015374478365, + "grad_norm": 2.344918966293335, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7235039472579956, + "num_tokens": 348953028.0, + "step": 13486 + }, + { + "epoch": 1.4811113551504502, + "grad_norm": 2.124089002609253, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7392493486404419, + "num_tokens": 348981766.0, + "step": 13487 + }, + { + "epoch": 1.481221172853064, + "grad_norm": 2.1242613792419434, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7116682529449463, + "num_tokens": 349010130.0, + "step": 13488 + }, + { + "epoch": 1.4813309905556775, + "grad_norm": 2.2786433696746826, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6943966150283813, + "num_tokens": 349037423.0, + "step": 13489 + }, + { + "epoch": 1.4814408082582913, + "grad_norm": 2.2811548709869385, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7021883726119995, + "num_tokens": 349063582.0, + "step": 13490 + }, + { + "epoch": 1.4815506259609048, + "grad_norm": 1.9023663997650146, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7064947485923767, + "num_tokens": 349097443.0, + "step": 13491 + }, + { + "epoch": 1.4816604436635186, + "grad_norm": 2.1573562622070312, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7028670907020569, + "num_tokens": 349124836.0, + "step": 13492 + }, + { + "epoch": 1.4817702613661323, + "grad_norm": 2.4550766944885254, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.698972225189209, + "num_tokens": 349150018.0, + "step": 13493 + }, + { + "epoch": 1.4818800790687459, + "grad_norm": 2.0987648963928223, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7184228301048279, + "num_tokens": 349178029.0, + "step": 13494 + }, + { + "epoch": 1.4819898967713596, + "grad_norm": 2.4452884197235107, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7056140899658203, + "num_tokens": 349201132.0, + "step": 13495 + }, + { + "epoch": 1.4820997144739732, + "grad_norm": 2.4682490825653076, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7319669723510742, + "num_tokens": 349222121.0, + "step": 13496 + }, + { + "epoch": 1.482209532176587, + "grad_norm": 2.141756057739258, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7381678819656372, + "num_tokens": 349250114.0, + "step": 13497 + }, + { + "epoch": 1.4823193498792007, + "grad_norm": 2.032961845397949, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7079908847808838, + "num_tokens": 349278670.0, + "step": 13498 + }, + { + "epoch": 1.4824291675818142, + "grad_norm": 2.2851109504699707, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7082186937332153, + "num_tokens": 349304750.0, + "step": 13499 + }, + { + "epoch": 1.4825389852844277, + "grad_norm": 2.183168649673462, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7346256971359253, + "num_tokens": 349332424.0, + "step": 13500 + }, + { + "epoch": 1.4826488029870415, + "grad_norm": 2.1967556476593018, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7060656547546387, + "num_tokens": 349359251.0, + "step": 13501 + }, + { + "epoch": 1.4827586206896552, + "grad_norm": 2.106987714767456, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7023100852966309, + "num_tokens": 349390802.0, + "step": 13502 + }, + { + "epoch": 1.4828684383922688, + "grad_norm": 2.4221231937408447, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7375573515892029, + "num_tokens": 349411462.0, + "step": 13503 + }, + { + "epoch": 1.4829782560948825, + "grad_norm": 2.0917654037475586, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7513304948806763, + "num_tokens": 349437804.0, + "step": 13504 + }, + { + "epoch": 1.483088073797496, + "grad_norm": 2.440232038497925, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7272312641143799, + "num_tokens": 349461433.0, + "step": 13505 + }, + { + "epoch": 1.4831978915001098, + "grad_norm": 2.101836919784546, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.69606614112854, + "num_tokens": 349492785.0, + "step": 13506 + }, + { + "epoch": 1.4833077092027236, + "grad_norm": 2.519989490509033, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7204493284225464, + "num_tokens": 349514255.0, + "step": 13507 + }, + { + "epoch": 1.4834175269053371, + "grad_norm": 2.2262420654296875, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7211959362030029, + "num_tokens": 349540250.0, + "step": 13508 + }, + { + "epoch": 1.4835273446079509, + "grad_norm": 2.9377224445343018, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7334069013595581, + "num_tokens": 349555718.0, + "step": 13509 + }, + { + "epoch": 1.4836371623105644, + "grad_norm": 2.277778387069702, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7255232334136963, + "num_tokens": 349581064.0, + "step": 13510 + }, + { + "epoch": 1.4837469800131782, + "grad_norm": 2.143601655960083, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7034038305282593, + "num_tokens": 349610750.0, + "step": 13511 + }, + { + "epoch": 1.4838567977157917, + "grad_norm": 2.477658271789551, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7237253189086914, + "num_tokens": 349631750.0, + "step": 13512 + }, + { + "epoch": 1.4839666154184055, + "grad_norm": 2.1933751106262207, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.706074595451355, + "num_tokens": 349660878.0, + "step": 13513 + }, + { + "epoch": 1.484076433121019, + "grad_norm": 2.304133415222168, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.732287585735321, + "num_tokens": 349684386.0, + "step": 13514 + }, + { + "epoch": 1.4841862508236328, + "grad_norm": 2.042527198791504, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6923627853393555, + "num_tokens": 349713907.0, + "step": 13515 + }, + { + "epoch": 1.4842960685262465, + "grad_norm": 1.9904779195785522, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7164853811264038, + "num_tokens": 349744778.0, + "step": 13516 + }, + { + "epoch": 1.48440588622886, + "grad_norm": 2.3938984870910645, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7530161142349243, + "num_tokens": 349767661.0, + "step": 13517 + }, + { + "epoch": 1.4845157039314738, + "grad_norm": 2.253772020339966, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7057682275772095, + "num_tokens": 349794478.0, + "step": 13518 + }, + { + "epoch": 1.4846255216340873, + "grad_norm": 2.345998525619507, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7088203430175781, + "num_tokens": 349819844.0, + "step": 13519 + }, + { + "epoch": 1.484735339336701, + "grad_norm": 2.1670775413513184, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7222846746444702, + "num_tokens": 349847165.0, + "step": 13520 + }, + { + "epoch": 1.4848451570393149, + "grad_norm": 2.1482784748077393, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.702622652053833, + "num_tokens": 349874515.0, + "step": 13521 + }, + { + "epoch": 1.4849549747419284, + "grad_norm": 2.250927686691284, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.723178505897522, + "num_tokens": 349899695.0, + "step": 13522 + }, + { + "epoch": 1.485064792444542, + "grad_norm": 1.9867491722106934, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7068362236022949, + "num_tokens": 349930611.0, + "step": 13523 + }, + { + "epoch": 1.4851746101471557, + "grad_norm": 2.159540891647339, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7052586674690247, + "num_tokens": 349958000.0, + "step": 13524 + }, + { + "epoch": 1.4852844278497694, + "grad_norm": 2.146885871887207, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7144849896430969, + "num_tokens": 349984827.0, + "step": 13525 + }, + { + "epoch": 1.485394245552383, + "grad_norm": 2.3685355186462402, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7209415435791016, + "num_tokens": 350008778.0, + "step": 13526 + }, + { + "epoch": 1.4855040632549967, + "grad_norm": 2.2071542739868164, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7202510833740234, + "num_tokens": 350034827.0, + "step": 13527 + }, + { + "epoch": 1.4856138809576103, + "grad_norm": 2.1320393085479736, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7316696643829346, + "num_tokens": 350062547.0, + "step": 13528 + }, + { + "epoch": 1.485723698660224, + "grad_norm": 2.2182061672210693, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7234360575675964, + "num_tokens": 350089143.0, + "step": 13529 + }, + { + "epoch": 1.4858335163628378, + "grad_norm": 2.273939609527588, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7073655128479004, + "num_tokens": 350113252.0, + "step": 13530 + }, + { + "epoch": 1.4859433340654513, + "grad_norm": 1.9570342302322388, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7148152589797974, + "num_tokens": 350145085.0, + "step": 13531 + }, + { + "epoch": 1.486053151768065, + "grad_norm": 2.477142095565796, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.730372965335846, + "num_tokens": 350167183.0, + "step": 13532 + }, + { + "epoch": 1.4861629694706786, + "grad_norm": 2.3282687664031982, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.727566659450531, + "num_tokens": 350189984.0, + "step": 13533 + }, + { + "epoch": 1.4862727871732924, + "grad_norm": 2.228306770324707, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7133957147598267, + "num_tokens": 350216411.0, + "step": 13534 + }, + { + "epoch": 1.4863826048759061, + "grad_norm": 1.9605498313903809, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7243384122848511, + "num_tokens": 350247603.0, + "step": 13535 + }, + { + "epoch": 1.4864924225785197, + "grad_norm": 2.2625508308410645, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7317692041397095, + "num_tokens": 350273919.0, + "step": 13536 + }, + { + "epoch": 1.4866022402811332, + "grad_norm": 2.1577560901641846, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.718092143535614, + "num_tokens": 350300025.0, + "step": 13537 + }, + { + "epoch": 1.486712057983747, + "grad_norm": 2.507718086242676, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7371660470962524, + "num_tokens": 350323919.0, + "step": 13538 + }, + { + "epoch": 1.4868218756863607, + "grad_norm": 2.3905065059661865, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7201941013336182, + "num_tokens": 350349168.0, + "step": 13539 + }, + { + "epoch": 1.4869316933889742, + "grad_norm": 2.5681095123291016, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7272129058837891, + "num_tokens": 350370466.0, + "step": 13540 + }, + { + "epoch": 1.487041511091588, + "grad_norm": 2.3992679119110107, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7308570146560669, + "num_tokens": 350392292.0, + "step": 13541 + }, + { + "epoch": 1.4871513287942015, + "grad_norm": 2.2881133556365967, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6923125982284546, + "num_tokens": 350418165.0, + "step": 13542 + }, + { + "epoch": 1.4872611464968153, + "grad_norm": 2.4727320671081543, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7202627062797546, + "num_tokens": 350438677.0, + "step": 13543 + }, + { + "epoch": 1.487370964199429, + "grad_norm": 2.312037944793701, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7179216742515564, + "num_tokens": 350463147.0, + "step": 13544 + }, + { + "epoch": 1.4874807819020426, + "grad_norm": 2.0121426582336426, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7032918930053711, + "num_tokens": 350493784.0, + "step": 13545 + }, + { + "epoch": 1.4875905996046563, + "grad_norm": 2.5017306804656982, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7163960933685303, + "num_tokens": 350515693.0, + "step": 13546 + }, + { + "epoch": 1.4877004173072699, + "grad_norm": 1.9755533933639526, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7227298021316528, + "num_tokens": 350549384.0, + "step": 13547 + }, + { + "epoch": 1.4878102350098836, + "grad_norm": 2.105877161026001, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6979305744171143, + "num_tokens": 350580496.0, + "step": 13548 + }, + { + "epoch": 1.4879200527124974, + "grad_norm": 2.4971370697021484, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7218526601791382, + "num_tokens": 350602962.0, + "step": 13549 + }, + { + "epoch": 1.488029870415111, + "grad_norm": 2.2147181034088135, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7161248922348022, + "num_tokens": 350628895.0, + "step": 13550 + }, + { + "epoch": 1.4881396881177245, + "grad_norm": 2.648660659790039, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7515131235122681, + "num_tokens": 350646645.0, + "step": 13551 + }, + { + "epoch": 1.4882495058203382, + "grad_norm": 2.354968547821045, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7095995545387268, + "num_tokens": 350671701.0, + "step": 13552 + }, + { + "epoch": 1.488359323522952, + "grad_norm": 1.9214199781417847, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7100076675415039, + "num_tokens": 350703873.0, + "step": 13553 + }, + { + "epoch": 1.4884691412255655, + "grad_norm": 2.078376531600952, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7120534777641296, + "num_tokens": 350733475.0, + "step": 13554 + }, + { + "epoch": 1.4885789589281793, + "grad_norm": 2.180525541305542, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7451031804084778, + "num_tokens": 350758833.0, + "step": 13555 + }, + { + "epoch": 1.4886887766307928, + "grad_norm": 2.473808765411377, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7181087136268616, + "num_tokens": 350780400.0, + "step": 13556 + }, + { + "epoch": 1.4887985943334066, + "grad_norm": 2.0316145420074463, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7359872460365295, + "num_tokens": 350808141.0, + "step": 13557 + }, + { + "epoch": 1.4889084120360203, + "grad_norm": 2.2735798358917236, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7214617729187012, + "num_tokens": 350835002.0, + "step": 13558 + }, + { + "epoch": 1.4890182297386338, + "grad_norm": 2.3934876918792725, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7609933614730835, + "num_tokens": 350856990.0, + "step": 13559 + }, + { + "epoch": 1.4891280474412476, + "grad_norm": 2.221250295639038, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.709301233291626, + "num_tokens": 350886291.0, + "step": 13560 + }, + { + "epoch": 1.4892378651438611, + "grad_norm": 2.077951669692993, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7174781560897827, + "num_tokens": 350914005.0, + "step": 13561 + }, + { + "epoch": 1.489347682846475, + "grad_norm": 2.2565441131591797, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.6931653618812561, + "num_tokens": 350941507.0, + "step": 13562 + }, + { + "epoch": 1.4894575005490884, + "grad_norm": 2.664039134979248, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.736614465713501, + "num_tokens": 350960208.0, + "step": 13563 + }, + { + "epoch": 1.4895673182517022, + "grad_norm": 2.166663885116577, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6905649900436401, + "num_tokens": 350988155.0, + "step": 13564 + }, + { + "epoch": 1.4896771359543157, + "grad_norm": 2.233126401901245, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.727523922920227, + "num_tokens": 351013036.0, + "step": 13565 + }, + { + "epoch": 1.4897869536569295, + "grad_norm": 2.3620612621307373, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7285448908805847, + "num_tokens": 351036133.0, + "step": 13566 + }, + { + "epoch": 1.4898967713595432, + "grad_norm": 1.9524304866790771, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7225509285926819, + "num_tokens": 351069401.0, + "step": 13567 + }, + { + "epoch": 1.4900065890621568, + "grad_norm": 2.040126323699951, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.6977320313453674, + "num_tokens": 351102336.0, + "step": 13568 + }, + { + "epoch": 1.4901164067647705, + "grad_norm": 2.753084897994995, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7345867156982422, + "num_tokens": 351120748.0, + "step": 13569 + }, + { + "epoch": 1.490226224467384, + "grad_norm": 2.569958209991455, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7193375825881958, + "num_tokens": 351140821.0, + "step": 13570 + }, + { + "epoch": 1.4903360421699978, + "grad_norm": 1.9540283679962158, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.6973435282707214, + "num_tokens": 351170688.0, + "step": 13571 + }, + { + "epoch": 1.4904458598726116, + "grad_norm": 2.238023281097412, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7129493951797485, + "num_tokens": 351195904.0, + "step": 13572 + }, + { + "epoch": 1.4905556775752251, + "grad_norm": 2.4176785945892334, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7200747728347778, + "num_tokens": 351218093.0, + "step": 13573 + }, + { + "epoch": 1.4906654952778389, + "grad_norm": 2.1958539485931396, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.710814356803894, + "num_tokens": 351243442.0, + "step": 13574 + }, + { + "epoch": 1.4907753129804524, + "grad_norm": 2.5866808891296387, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6990666389465332, + "num_tokens": 351264376.0, + "step": 13575 + }, + { + "epoch": 1.4908851306830662, + "grad_norm": 2.687882423400879, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7524639368057251, + "num_tokens": 351282166.0, + "step": 13576 + }, + { + "epoch": 1.4909949483856797, + "grad_norm": 2.1472694873809814, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7039642333984375, + "num_tokens": 351310966.0, + "step": 13577 + }, + { + "epoch": 1.4911047660882935, + "grad_norm": 2.194010019302368, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7296860814094543, + "num_tokens": 351338849.0, + "step": 13578 + }, + { + "epoch": 1.491214583790907, + "grad_norm": 2.2817819118499756, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7200690507888794, + "num_tokens": 351364478.0, + "step": 13579 + }, + { + "epoch": 1.4913244014935207, + "grad_norm": 2.311936378479004, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7172356247901917, + "num_tokens": 351389319.0, + "step": 13580 + }, + { + "epoch": 1.4914342191961345, + "grad_norm": 2.192974328994751, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7175509929656982, + "num_tokens": 351413848.0, + "step": 13581 + }, + { + "epoch": 1.491544036898748, + "grad_norm": 2.0270144939422607, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.7008810043334961, + "num_tokens": 351443166.0, + "step": 13582 + }, + { + "epoch": 1.4916538546013618, + "grad_norm": 2.7159881591796875, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7226113677024841, + "num_tokens": 351462274.0, + "step": 13583 + }, + { + "epoch": 1.4917636723039753, + "grad_norm": 2.3924238681793213, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7270532250404358, + "num_tokens": 351487010.0, + "step": 13584 + }, + { + "epoch": 1.491873490006589, + "grad_norm": 2.2283999919891357, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7426491975784302, + "num_tokens": 351514148.0, + "step": 13585 + }, + { + "epoch": 1.4919833077092028, + "grad_norm": 1.9659544229507446, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6899818181991577, + "num_tokens": 351548859.0, + "step": 13586 + }, + { + "epoch": 1.4920931254118164, + "grad_norm": 2.5100460052490234, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.748380184173584, + "num_tokens": 351568450.0, + "step": 13587 + }, + { + "epoch": 1.49220294311443, + "grad_norm": 2.4009034633636475, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.721710205078125, + "num_tokens": 351590179.0, + "step": 13588 + }, + { + "epoch": 1.4923127608170437, + "grad_norm": 2.034374713897705, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7357650995254517, + "num_tokens": 351617740.0, + "step": 13589 + }, + { + "epoch": 1.4924225785196574, + "grad_norm": 2.1300885677337646, + "learning_rate": 1e-06, + "loss": 1.0463, + "mean_token_accuracy": 0.6899468302726746, + "num_tokens": 351649879.0, + "step": 13590 + }, + { + "epoch": 1.492532396222271, + "grad_norm": 2.527836561203003, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7139781713485718, + "num_tokens": 351671227.0, + "step": 13591 + }, + { + "epoch": 1.4926422139248847, + "grad_norm": 2.4824390411376953, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7183101773262024, + "num_tokens": 351693296.0, + "step": 13592 + }, + { + "epoch": 1.4927520316274983, + "grad_norm": 2.6271414756774902, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7751797437667847, + "num_tokens": 351711924.0, + "step": 13593 + }, + { + "epoch": 1.492861849330112, + "grad_norm": 2.2531259059906006, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7307777404785156, + "num_tokens": 351737262.0, + "step": 13594 + }, + { + "epoch": 1.4929716670327258, + "grad_norm": 2.2750396728515625, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7304776906967163, + "num_tokens": 351760464.0, + "step": 13595 + }, + { + "epoch": 1.4930814847353393, + "grad_norm": 2.288698434829712, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.6974945664405823, + "num_tokens": 351784671.0, + "step": 13596 + }, + { + "epoch": 1.493191302437953, + "grad_norm": 2.210695505142212, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7060726881027222, + "num_tokens": 351813130.0, + "step": 13597 + }, + { + "epoch": 1.4933011201405666, + "grad_norm": 2.508364677429199, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7110859751701355, + "num_tokens": 351836354.0, + "step": 13598 + }, + { + "epoch": 1.4934109378431804, + "grad_norm": 2.1686995029449463, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7334840297698975, + "num_tokens": 351864756.0, + "step": 13599 + }, + { + "epoch": 1.493520755545794, + "grad_norm": 1.880936622619629, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.701457679271698, + "num_tokens": 351900574.0, + "step": 13600 + }, + { + "epoch": 1.4936305732484076, + "grad_norm": 2.385978937149048, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7329299449920654, + "num_tokens": 351922710.0, + "step": 13601 + }, + { + "epoch": 1.4937403909510212, + "grad_norm": 2.201519250869751, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7245571613311768, + "num_tokens": 351949225.0, + "step": 13602 + }, + { + "epoch": 1.493850208653635, + "grad_norm": 2.2475011348724365, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7137726545333862, + "num_tokens": 351974234.0, + "step": 13603 + }, + { + "epoch": 1.4939600263562487, + "grad_norm": 2.3430240154266357, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7139315605163574, + "num_tokens": 351996602.0, + "step": 13604 + }, + { + "epoch": 1.4940698440588622, + "grad_norm": 2.3421437740325928, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7116565704345703, + "num_tokens": 352020711.0, + "step": 13605 + }, + { + "epoch": 1.494179661761476, + "grad_norm": 2.071812629699707, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6944859623908997, + "num_tokens": 352052927.0, + "step": 13606 + }, + { + "epoch": 1.4942894794640895, + "grad_norm": 2.3754520416259766, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7179052829742432, + "num_tokens": 352077195.0, + "step": 13607 + }, + { + "epoch": 1.4943992971667033, + "grad_norm": 2.01157808303833, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7067162394523621, + "num_tokens": 352106228.0, + "step": 13608 + }, + { + "epoch": 1.494509114869317, + "grad_norm": 2.06046199798584, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7152673006057739, + "num_tokens": 352133965.0, + "step": 13609 + }, + { + "epoch": 1.4946189325719306, + "grad_norm": 2.35939621925354, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7113443613052368, + "num_tokens": 352158760.0, + "step": 13610 + }, + { + "epoch": 1.4947287502745443, + "grad_norm": 2.8326332569122314, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.728535532951355, + "num_tokens": 352177506.0, + "step": 13611 + }, + { + "epoch": 1.4948385679771579, + "grad_norm": 2.6084132194519043, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7229055166244507, + "num_tokens": 352198833.0, + "step": 13612 + }, + { + "epoch": 1.4949483856797716, + "grad_norm": 2.286856174468994, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7413100004196167, + "num_tokens": 352220545.0, + "step": 13613 + }, + { + "epoch": 1.4950582033823854, + "grad_norm": 2.0431129932403564, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7098087072372437, + "num_tokens": 352251267.0, + "step": 13614 + }, + { + "epoch": 1.495168021084999, + "grad_norm": 2.287367105484009, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7019413113594055, + "num_tokens": 352277812.0, + "step": 13615 + }, + { + "epoch": 1.4952778387876124, + "grad_norm": 2.388432502746582, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.712157130241394, + "num_tokens": 352302730.0, + "step": 13616 + }, + { + "epoch": 1.4953876564902262, + "grad_norm": 2.557232141494751, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.709165096282959, + "num_tokens": 352323936.0, + "step": 13617 + }, + { + "epoch": 1.49549747419284, + "grad_norm": 2.1531128883361816, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7106274962425232, + "num_tokens": 352353122.0, + "step": 13618 + }, + { + "epoch": 1.4956072918954535, + "grad_norm": 2.340437412261963, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7145689725875854, + "num_tokens": 352379397.0, + "step": 13619 + }, + { + "epoch": 1.4957171095980673, + "grad_norm": 2.240504741668701, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7277847528457642, + "num_tokens": 352405394.0, + "step": 13620 + }, + { + "epoch": 1.4958269273006808, + "grad_norm": 2.356067180633545, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7345404624938965, + "num_tokens": 352427832.0, + "step": 13621 + }, + { + "epoch": 1.4959367450032945, + "grad_norm": 2.2623093128204346, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7322685122489929, + "num_tokens": 352451288.0, + "step": 13622 + }, + { + "epoch": 1.4960465627059083, + "grad_norm": 2.0394163131713867, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7213193774223328, + "num_tokens": 352480422.0, + "step": 13623 + }, + { + "epoch": 1.4961563804085218, + "grad_norm": 2.4897923469543457, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7173315286636353, + "num_tokens": 352502417.0, + "step": 13624 + }, + { + "epoch": 1.4962661981111356, + "grad_norm": 2.040299415588379, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6915367245674133, + "num_tokens": 352534885.0, + "step": 13625 + }, + { + "epoch": 1.4963760158137491, + "grad_norm": 2.0201125144958496, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7276670336723328, + "num_tokens": 352564471.0, + "step": 13626 + }, + { + "epoch": 1.4964858335163629, + "grad_norm": 2.369204521179199, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.697359561920166, + "num_tokens": 352586736.0, + "step": 13627 + }, + { + "epoch": 1.4965956512189764, + "grad_norm": 2.071262836456299, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7231259346008301, + "num_tokens": 352615469.0, + "step": 13628 + }, + { + "epoch": 1.4967054689215902, + "grad_norm": 2.0988526344299316, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.721351683139801, + "num_tokens": 352643181.0, + "step": 13629 + }, + { + "epoch": 1.4968152866242037, + "grad_norm": 2.1104681491851807, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6867094039916992, + "num_tokens": 352673504.0, + "step": 13630 + }, + { + "epoch": 1.4969251043268175, + "grad_norm": 2.0538995265960693, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7276020050048828, + "num_tokens": 352701778.0, + "step": 13631 + }, + { + "epoch": 1.4970349220294312, + "grad_norm": 2.241755962371826, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7175658941268921, + "num_tokens": 352728066.0, + "step": 13632 + }, + { + "epoch": 1.4971447397320448, + "grad_norm": 2.2609164714813232, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.711479902267456, + "num_tokens": 352752048.0, + "step": 13633 + }, + { + "epoch": 1.4972545574346585, + "grad_norm": 1.902132272720337, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.6966010928153992, + "num_tokens": 352786096.0, + "step": 13634 + }, + { + "epoch": 1.497364375137272, + "grad_norm": 2.187448501586914, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7374817728996277, + "num_tokens": 352812176.0, + "step": 13635 + }, + { + "epoch": 1.4974741928398858, + "grad_norm": 2.3116886615753174, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7195767760276794, + "num_tokens": 352835227.0, + "step": 13636 + }, + { + "epoch": 1.4975840105424996, + "grad_norm": 2.192183017730713, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.722515881061554, + "num_tokens": 352861196.0, + "step": 13637 + }, + { + "epoch": 1.497693828245113, + "grad_norm": 2.4313912391662598, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7656323313713074, + "num_tokens": 352882435.0, + "step": 13638 + }, + { + "epoch": 1.4978036459477266, + "grad_norm": 2.289370536804199, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7004745006561279, + "num_tokens": 352907753.0, + "step": 13639 + }, + { + "epoch": 1.4979134636503404, + "grad_norm": 2.4564261436462402, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.749699056148529, + "num_tokens": 352929689.0, + "step": 13640 + }, + { + "epoch": 1.4980232813529542, + "grad_norm": 2.62951397895813, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7270255088806152, + "num_tokens": 352950689.0, + "step": 13641 + }, + { + "epoch": 1.4981330990555677, + "grad_norm": 2.117013931274414, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7224538326263428, + "num_tokens": 352978013.0, + "step": 13642 + }, + { + "epoch": 1.4982429167581814, + "grad_norm": 2.110924482345581, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7202986478805542, + "num_tokens": 353007336.0, + "step": 13643 + }, + { + "epoch": 1.498352734460795, + "grad_norm": 2.2694993019104004, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7272436022758484, + "num_tokens": 353032226.0, + "step": 13644 + }, + { + "epoch": 1.4984625521634087, + "grad_norm": 2.4000110626220703, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7378216981887817, + "num_tokens": 353054202.0, + "step": 13645 + }, + { + "epoch": 1.4985723698660225, + "grad_norm": 2.3604772090911865, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7256650328636169, + "num_tokens": 353077624.0, + "step": 13646 + }, + { + "epoch": 1.498682187568636, + "grad_norm": 2.145169734954834, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7231098413467407, + "num_tokens": 353105436.0, + "step": 13647 + }, + { + "epoch": 1.4987920052712498, + "grad_norm": 2.14304518699646, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7056992053985596, + "num_tokens": 353133795.0, + "step": 13648 + }, + { + "epoch": 1.4989018229738633, + "grad_norm": 2.9993348121643066, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7433114051818848, + "num_tokens": 353153980.0, + "step": 13649 + }, + { + "epoch": 1.499011640676477, + "grad_norm": 2.235579252243042, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7260043621063232, + "num_tokens": 353180072.0, + "step": 13650 + }, + { + "epoch": 1.4991214583790908, + "grad_norm": 2.2401301860809326, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7055667042732239, + "num_tokens": 353209854.0, + "step": 13651 + }, + { + "epoch": 1.4992312760817044, + "grad_norm": 2.8481264114379883, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7300138473510742, + "num_tokens": 353227772.0, + "step": 13652 + }, + { + "epoch": 1.499341093784318, + "grad_norm": 2.4809043407440186, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.750711977481842, + "num_tokens": 353248130.0, + "step": 13653 + }, + { + "epoch": 1.4994509114869317, + "grad_norm": 2.7281887531280518, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7158451676368713, + "num_tokens": 353268005.0, + "step": 13654 + }, + { + "epoch": 1.4995607291895454, + "grad_norm": 2.2645351886749268, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.736029863357544, + "num_tokens": 353293531.0, + "step": 13655 + }, + { + "epoch": 1.499670546892159, + "grad_norm": 2.0943450927734375, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.6983410716056824, + "num_tokens": 353323734.0, + "step": 13656 + }, + { + "epoch": 1.4997803645947727, + "grad_norm": 2.46018123626709, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7085832953453064, + "num_tokens": 353345463.0, + "step": 13657 + }, + { + "epoch": 1.4998901822973862, + "grad_norm": 2.3632688522338867, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7149505615234375, + "num_tokens": 353369808.0, + "step": 13658 + }, + { + "epoch": 1.5, + "grad_norm": 2.2904226779937744, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7482516169548035, + "num_tokens": 353393328.0, + "step": 13659 + }, + { + "epoch": 1.5001098177026138, + "grad_norm": 2.449411392211914, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7215858697891235, + "num_tokens": 353417908.0, + "step": 13660 + }, + { + "epoch": 1.5002196354052273, + "grad_norm": 2.264676570892334, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7193809747695923, + "num_tokens": 353443856.0, + "step": 13661 + }, + { + "epoch": 1.5003294531078408, + "grad_norm": 2.169938087463379, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7242006659507751, + "num_tokens": 353469890.0, + "step": 13662 + }, + { + "epoch": 1.5004392708104546, + "grad_norm": 2.238180637359619, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.703106164932251, + "num_tokens": 353496955.0, + "step": 13663 + }, + { + "epoch": 1.5005490885130683, + "grad_norm": 2.0318360328674316, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7084665298461914, + "num_tokens": 353527810.0, + "step": 13664 + }, + { + "epoch": 1.500658906215682, + "grad_norm": 2.1756105422973633, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7085095643997192, + "num_tokens": 353554731.0, + "step": 13665 + }, + { + "epoch": 1.5007687239182956, + "grad_norm": 2.009476900100708, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7046388387680054, + "num_tokens": 353586332.0, + "step": 13666 + }, + { + "epoch": 1.5008785416209092, + "grad_norm": 2.152064800262451, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.700310468673706, + "num_tokens": 353612790.0, + "step": 13667 + }, + { + "epoch": 1.500988359323523, + "grad_norm": 2.094295024871826, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7168400287628174, + "num_tokens": 353640385.0, + "step": 13668 + }, + { + "epoch": 1.5010981770261367, + "grad_norm": 2.294590473175049, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7039219737052917, + "num_tokens": 353665468.0, + "step": 13669 + }, + { + "epoch": 1.5012079947287504, + "grad_norm": 2.2374253273010254, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7343559861183167, + "num_tokens": 353690606.0, + "step": 13670 + }, + { + "epoch": 1.501317812431364, + "grad_norm": 2.394744396209717, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7272769212722778, + "num_tokens": 353713806.0, + "step": 13671 + }, + { + "epoch": 1.5014276301339775, + "grad_norm": 2.2637922763824463, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7273277044296265, + "num_tokens": 353738924.0, + "step": 13672 + }, + { + "epoch": 1.5015374478365913, + "grad_norm": 2.245297431945801, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7141598463058472, + "num_tokens": 353763454.0, + "step": 13673 + }, + { + "epoch": 1.501647265539205, + "grad_norm": 2.2468202114105225, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.741795003414154, + "num_tokens": 353787635.0, + "step": 13674 + }, + { + "epoch": 1.5017570832418186, + "grad_norm": 2.161160707473755, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7176636457443237, + "num_tokens": 353815796.0, + "step": 13675 + }, + { + "epoch": 1.501866900944432, + "grad_norm": 2.237095832824707, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7454360723495483, + "num_tokens": 353840383.0, + "step": 13676 + }, + { + "epoch": 1.5019767186470458, + "grad_norm": 2.1942107677459717, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7191044092178345, + "num_tokens": 353866233.0, + "step": 13677 + }, + { + "epoch": 1.5020865363496596, + "grad_norm": 2.0583434104919434, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7138593792915344, + "num_tokens": 353894026.0, + "step": 13678 + }, + { + "epoch": 1.5021963540522734, + "grad_norm": 2.0564076900482178, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6948531866073608, + "num_tokens": 353923991.0, + "step": 13679 + }, + { + "epoch": 1.502306171754887, + "grad_norm": 2.0753486156463623, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7286759614944458, + "num_tokens": 353954798.0, + "step": 13680 + }, + { + "epoch": 1.5024159894575004, + "grad_norm": 2.6725692749023438, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.725471019744873, + "num_tokens": 353975069.0, + "step": 13681 + }, + { + "epoch": 1.5025258071601142, + "grad_norm": 2.6242990493774414, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7438764572143555, + "num_tokens": 353994106.0, + "step": 13682 + }, + { + "epoch": 1.502635624862728, + "grad_norm": 2.335418462753296, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7010365724563599, + "num_tokens": 354018331.0, + "step": 13683 + }, + { + "epoch": 1.5027454425653415, + "grad_norm": 2.5370543003082275, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7368167638778687, + "num_tokens": 354040243.0, + "step": 13684 + }, + { + "epoch": 1.5028552602679552, + "grad_norm": 2.358729600906372, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7269665002822876, + "num_tokens": 354064205.0, + "step": 13685 + }, + { + "epoch": 1.5029650779705688, + "grad_norm": 2.3652284145355225, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7123095989227295, + "num_tokens": 354089638.0, + "step": 13686 + }, + { + "epoch": 1.5030748956731825, + "grad_norm": 2.50496244430542, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7307982444763184, + "num_tokens": 354113131.0, + "step": 13687 + }, + { + "epoch": 1.5031847133757963, + "grad_norm": 2.3584582805633545, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7313461303710938, + "num_tokens": 354137117.0, + "step": 13688 + }, + { + "epoch": 1.5032945310784098, + "grad_norm": 2.3301637172698975, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7380399107933044, + "num_tokens": 354161436.0, + "step": 13689 + }, + { + "epoch": 1.5034043487810234, + "grad_norm": 2.2613511085510254, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7134093642234802, + "num_tokens": 354186678.0, + "step": 13690 + }, + { + "epoch": 1.5035141664836371, + "grad_norm": 2.032590866088867, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7365731596946716, + "num_tokens": 354214855.0, + "step": 13691 + }, + { + "epoch": 1.5036239841862509, + "grad_norm": 2.082364320755005, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6812334060668945, + "num_tokens": 354247092.0, + "step": 13692 + }, + { + "epoch": 1.5037338018888646, + "grad_norm": 2.4642045497894287, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7386280298233032, + "num_tokens": 354268746.0, + "step": 13693 + }, + { + "epoch": 1.5038436195914782, + "grad_norm": 2.133544683456421, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7216722369194031, + "num_tokens": 354296795.0, + "step": 13694 + }, + { + "epoch": 1.5039534372940917, + "grad_norm": 2.055879592895508, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.728039562702179, + "num_tokens": 354325884.0, + "step": 13695 + }, + { + "epoch": 1.5040632549967055, + "grad_norm": 2.35650634765625, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.729271411895752, + "num_tokens": 354348878.0, + "step": 13696 + }, + { + "epoch": 1.5041730726993192, + "grad_norm": 2.305863857269287, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7485643625259399, + "num_tokens": 354373386.0, + "step": 13697 + }, + { + "epoch": 1.5042828904019327, + "grad_norm": 2.1509530544281006, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7212492227554321, + "num_tokens": 354399582.0, + "step": 13698 + }, + { + "epoch": 1.5043927081045465, + "grad_norm": 2.173772096633911, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.685130774974823, + "num_tokens": 354427759.0, + "step": 13699 + }, + { + "epoch": 1.50450252580716, + "grad_norm": 2.4650585651397705, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7004014849662781, + "num_tokens": 354450286.0, + "step": 13700 + }, + { + "epoch": 1.5046123435097738, + "grad_norm": 2.2717599868774414, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7072269916534424, + "num_tokens": 354476107.0, + "step": 13701 + }, + { + "epoch": 1.5047221612123876, + "grad_norm": 2.1115612983703613, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6961622834205627, + "num_tokens": 354505040.0, + "step": 13702 + }, + { + "epoch": 1.504831978915001, + "grad_norm": 2.1418559551239014, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7142806649208069, + "num_tokens": 354534336.0, + "step": 13703 + }, + { + "epoch": 1.5049417966176146, + "grad_norm": 2.277024745941162, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7181747555732727, + "num_tokens": 354559588.0, + "step": 13704 + }, + { + "epoch": 1.5050516143202284, + "grad_norm": 2.296882152557373, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7155314087867737, + "num_tokens": 354583966.0, + "step": 13705 + }, + { + "epoch": 1.5051614320228421, + "grad_norm": 2.280306100845337, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.722880482673645, + "num_tokens": 354608592.0, + "step": 13706 + }, + { + "epoch": 1.505271249725456, + "grad_norm": 2.404188871383667, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7326765060424805, + "num_tokens": 354631381.0, + "step": 13707 + }, + { + "epoch": 1.5053810674280694, + "grad_norm": 2.3165392875671387, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7221388816833496, + "num_tokens": 354655687.0, + "step": 13708 + }, + { + "epoch": 1.505490885130683, + "grad_norm": 2.2745044231414795, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7106898427009583, + "num_tokens": 354680390.0, + "step": 13709 + }, + { + "epoch": 1.5056007028332967, + "grad_norm": 2.2426536083221436, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7006487846374512, + "num_tokens": 354706633.0, + "step": 13710 + }, + { + "epoch": 1.5057105205359105, + "grad_norm": 2.400881052017212, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7208049297332764, + "num_tokens": 354730625.0, + "step": 13711 + }, + { + "epoch": 1.505820338238524, + "grad_norm": 2.75429368019104, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.741381049156189, + "num_tokens": 354747197.0, + "step": 13712 + }, + { + "epoch": 1.5059301559411375, + "grad_norm": 2.3240885734558105, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7257694602012634, + "num_tokens": 354771713.0, + "step": 13713 + }, + { + "epoch": 1.5060399736437513, + "grad_norm": 2.2600204944610596, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7246567606925964, + "num_tokens": 354799653.0, + "step": 13714 + }, + { + "epoch": 1.506149791346365, + "grad_norm": 2.188056707382202, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7082394957542419, + "num_tokens": 354827490.0, + "step": 13715 + }, + { + "epoch": 1.5062596090489788, + "grad_norm": 2.3584516048431396, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7098751664161682, + "num_tokens": 354854128.0, + "step": 13716 + }, + { + "epoch": 1.5063694267515924, + "grad_norm": 2.2692036628723145, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7154318690299988, + "num_tokens": 354879134.0, + "step": 13717 + }, + { + "epoch": 1.506479244454206, + "grad_norm": 2.49432635307312, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7144426703453064, + "num_tokens": 354902971.0, + "step": 13718 + }, + { + "epoch": 1.5065890621568196, + "grad_norm": 2.3513455390930176, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7305881977081299, + "num_tokens": 354926622.0, + "step": 13719 + }, + { + "epoch": 1.5066988798594334, + "grad_norm": 2.2299575805664062, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7220582962036133, + "num_tokens": 354950837.0, + "step": 13720 + }, + { + "epoch": 1.5068086975620472, + "grad_norm": 2.2046027183532715, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7006804943084717, + "num_tokens": 354978412.0, + "step": 13721 + }, + { + "epoch": 1.5069185152646607, + "grad_norm": 2.373383045196533, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7263838648796082, + "num_tokens": 355003622.0, + "step": 13722 + }, + { + "epoch": 1.5070283329672742, + "grad_norm": 2.604257345199585, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7201780080795288, + "num_tokens": 355023549.0, + "step": 13723 + }, + { + "epoch": 1.507138150669888, + "grad_norm": 2.2753965854644775, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.70919269323349, + "num_tokens": 355048551.0, + "step": 13724 + }, + { + "epoch": 1.5072479683725017, + "grad_norm": 2.2799010276794434, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7180969715118408, + "num_tokens": 355075336.0, + "step": 13725 + }, + { + "epoch": 1.5073577860751153, + "grad_norm": 2.055722951889038, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7127411365509033, + "num_tokens": 355105548.0, + "step": 13726 + }, + { + "epoch": 1.5074676037777288, + "grad_norm": 2.105118751525879, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7304386496543884, + "num_tokens": 355133482.0, + "step": 13727 + }, + { + "epoch": 1.5075774214803426, + "grad_norm": 2.3373947143554688, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7120286226272583, + "num_tokens": 355158234.0, + "step": 13728 + }, + { + "epoch": 1.5076872391829563, + "grad_norm": 2.1606764793395996, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7162068486213684, + "num_tokens": 355184906.0, + "step": 13729 + }, + { + "epoch": 1.50779705688557, + "grad_norm": 2.4002277851104736, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7013090252876282, + "num_tokens": 355206627.0, + "step": 13730 + }, + { + "epoch": 1.5079068745881836, + "grad_norm": 1.943812608718872, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7105937004089355, + "num_tokens": 355236100.0, + "step": 13731 + }, + { + "epoch": 1.5080166922907972, + "grad_norm": 1.9465835094451904, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6915199160575867, + "num_tokens": 355270428.0, + "step": 13732 + }, + { + "epoch": 1.508126509993411, + "grad_norm": 1.9812140464782715, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7128310203552246, + "num_tokens": 355303500.0, + "step": 13733 + }, + { + "epoch": 1.5082363276960247, + "grad_norm": 2.110097646713257, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7089154124259949, + "num_tokens": 355333276.0, + "step": 13734 + }, + { + "epoch": 1.5083461453986382, + "grad_norm": 2.3419315814971924, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7353054285049438, + "num_tokens": 355356832.0, + "step": 13735 + }, + { + "epoch": 1.508455963101252, + "grad_norm": 2.437593936920166, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.722493588924408, + "num_tokens": 355379912.0, + "step": 13736 + }, + { + "epoch": 1.5085657808038655, + "grad_norm": 2.2255935668945312, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7024459838867188, + "num_tokens": 355408113.0, + "step": 13737 + }, + { + "epoch": 1.5086755985064793, + "grad_norm": 2.3237128257751465, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7335249781608582, + "num_tokens": 355432286.0, + "step": 13738 + }, + { + "epoch": 1.508785416209093, + "grad_norm": 2.4072983264923096, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7011318206787109, + "num_tokens": 355457847.0, + "step": 13739 + }, + { + "epoch": 1.5088952339117065, + "grad_norm": 2.2628188133239746, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7280611991882324, + "num_tokens": 355483148.0, + "step": 13740 + }, + { + "epoch": 1.50900505161432, + "grad_norm": 2.126659631729126, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7301174402236938, + "num_tokens": 355510583.0, + "step": 13741 + }, + { + "epoch": 1.5091148693169338, + "grad_norm": 2.313194513320923, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7434973120689392, + "num_tokens": 355535520.0, + "step": 13742 + }, + { + "epoch": 1.5092246870195476, + "grad_norm": 2.113236904144287, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7203347086906433, + "num_tokens": 355562645.0, + "step": 13743 + }, + { + "epoch": 1.5093345047221614, + "grad_norm": 2.134646415710449, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7078162431716919, + "num_tokens": 355590352.0, + "step": 13744 + }, + { + "epoch": 1.5094443224247749, + "grad_norm": 2.306814193725586, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7039986848831177, + "num_tokens": 355616447.0, + "step": 13745 + }, + { + "epoch": 1.5095541401273884, + "grad_norm": 2.187063217163086, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7066757678985596, + "num_tokens": 355643921.0, + "step": 13746 + }, + { + "epoch": 1.5096639578300022, + "grad_norm": 2.502441644668579, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7356414794921875, + "num_tokens": 355663688.0, + "step": 13747 + }, + { + "epoch": 1.509773775532616, + "grad_norm": 2.7290761470794678, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7186248302459717, + "num_tokens": 355684287.0, + "step": 13748 + }, + { + "epoch": 1.5098835932352295, + "grad_norm": 2.1323211193084717, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7062769532203674, + "num_tokens": 355714256.0, + "step": 13749 + }, + { + "epoch": 1.5099934109378432, + "grad_norm": 2.1666386127471924, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7144657373428345, + "num_tokens": 355741462.0, + "step": 13750 + }, + { + "epoch": 1.5101032286404568, + "grad_norm": 2.408144474029541, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7447423934936523, + "num_tokens": 355762296.0, + "step": 13751 + }, + { + "epoch": 1.5102130463430705, + "grad_norm": 2.080366611480713, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.699148416519165, + "num_tokens": 355792887.0, + "step": 13752 + }, + { + "epoch": 1.5103228640456843, + "grad_norm": 2.144155263900757, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7129662036895752, + "num_tokens": 355821379.0, + "step": 13753 + }, + { + "epoch": 1.5104326817482978, + "grad_norm": 2.426990509033203, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7270373106002808, + "num_tokens": 355843447.0, + "step": 13754 + }, + { + "epoch": 1.5105424994509113, + "grad_norm": 2.3678698539733887, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7135075330734253, + "num_tokens": 355867998.0, + "step": 13755 + }, + { + "epoch": 1.510652317153525, + "grad_norm": 2.391695976257324, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7383483648300171, + "num_tokens": 355888455.0, + "step": 13756 + }, + { + "epoch": 1.5107621348561389, + "grad_norm": 2.163917064666748, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7399219274520874, + "num_tokens": 355913597.0, + "step": 13757 + }, + { + "epoch": 1.5108719525587526, + "grad_norm": 2.2035257816314697, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7145338654518127, + "num_tokens": 355939107.0, + "step": 13758 + }, + { + "epoch": 1.5109817702613662, + "grad_norm": 1.9755725860595703, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7193003296852112, + "num_tokens": 355969652.0, + "step": 13759 + }, + { + "epoch": 1.5110915879639797, + "grad_norm": 2.2694740295410156, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7406022548675537, + "num_tokens": 355995923.0, + "step": 13760 + }, + { + "epoch": 1.5112014056665934, + "grad_norm": 2.2742600440979004, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7163219451904297, + "num_tokens": 356020220.0, + "step": 13761 + }, + { + "epoch": 1.5113112233692072, + "grad_norm": 2.1200082302093506, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7082885503768921, + "num_tokens": 356047911.0, + "step": 13762 + }, + { + "epoch": 1.5114210410718207, + "grad_norm": 2.1408581733703613, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6900894045829773, + "num_tokens": 356074462.0, + "step": 13763 + }, + { + "epoch": 1.5115308587744345, + "grad_norm": 2.0938501358032227, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7311527729034424, + "num_tokens": 356101105.0, + "step": 13764 + }, + { + "epoch": 1.511640676477048, + "grad_norm": 2.2463624477386475, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.701012372970581, + "num_tokens": 356128051.0, + "step": 13765 + }, + { + "epoch": 1.5117504941796618, + "grad_norm": 2.4187119007110596, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.72813481092453, + "num_tokens": 356151505.0, + "step": 13766 + }, + { + "epoch": 1.5118603118822755, + "grad_norm": 2.3392019271850586, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7228822112083435, + "num_tokens": 356177332.0, + "step": 13767 + }, + { + "epoch": 1.511970129584889, + "grad_norm": 2.0396645069122314, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7473410367965698, + "num_tokens": 356205591.0, + "step": 13768 + }, + { + "epoch": 1.5120799472875026, + "grad_norm": 2.399080514907837, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7459598779678345, + "num_tokens": 356226094.0, + "step": 13769 + }, + { + "epoch": 1.5121897649901164, + "grad_norm": 2.248490810394287, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6941235065460205, + "num_tokens": 356252556.0, + "step": 13770 + }, + { + "epoch": 1.5122995826927301, + "grad_norm": 2.2467200756073, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7400444149971008, + "num_tokens": 356276592.0, + "step": 13771 + }, + { + "epoch": 1.5124094003953439, + "grad_norm": 2.445802927017212, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7230315208435059, + "num_tokens": 356298457.0, + "step": 13772 + }, + { + "epoch": 1.5125192180979574, + "grad_norm": 2.217966318130493, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7128087282180786, + "num_tokens": 356324337.0, + "step": 13773 + }, + { + "epoch": 1.512629035800571, + "grad_norm": 2.8649590015411377, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7265344858169556, + "num_tokens": 356345853.0, + "step": 13774 + }, + { + "epoch": 1.5127388535031847, + "grad_norm": 2.095414161682129, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6843026280403137, + "num_tokens": 356379117.0, + "step": 13775 + }, + { + "epoch": 1.5128486712057985, + "grad_norm": 2.303675413131714, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.717707097530365, + "num_tokens": 356405071.0, + "step": 13776 + }, + { + "epoch": 1.512958488908412, + "grad_norm": 2.4952027797698975, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7232104539871216, + "num_tokens": 356425984.0, + "step": 13777 + }, + { + "epoch": 1.5130683066110255, + "grad_norm": 2.289956569671631, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7257505059242249, + "num_tokens": 356448982.0, + "step": 13778 + }, + { + "epoch": 1.5131781243136393, + "grad_norm": 2.495952606201172, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7324918508529663, + "num_tokens": 356468840.0, + "step": 13779 + }, + { + "epoch": 1.513287942016253, + "grad_norm": 2.1494460105895996, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6964505910873413, + "num_tokens": 356496916.0, + "step": 13780 + }, + { + "epoch": 1.5133977597188668, + "grad_norm": 2.3503689765930176, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7404271960258484, + "num_tokens": 356519639.0, + "step": 13781 + }, + { + "epoch": 1.5135075774214803, + "grad_norm": 2.3712260723114014, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7118194103240967, + "num_tokens": 356543945.0, + "step": 13782 + }, + { + "epoch": 1.5136173951240939, + "grad_norm": 2.1641032695770264, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7120618224143982, + "num_tokens": 356571028.0, + "step": 13783 + }, + { + "epoch": 1.5137272128267076, + "grad_norm": 2.1438119411468506, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.720605731010437, + "num_tokens": 356600256.0, + "step": 13784 + }, + { + "epoch": 1.5138370305293214, + "grad_norm": 2.516322374343872, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7056660652160645, + "num_tokens": 356625344.0, + "step": 13785 + }, + { + "epoch": 1.5139468482319351, + "grad_norm": 2.1659910678863525, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7265978455543518, + "num_tokens": 356650057.0, + "step": 13786 + }, + { + "epoch": 1.5140566659345487, + "grad_norm": 2.261493444442749, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7215293645858765, + "num_tokens": 356675194.0, + "step": 13787 + }, + { + "epoch": 1.5141664836371622, + "grad_norm": 2.3855433464050293, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7332452535629272, + "num_tokens": 356698600.0, + "step": 13788 + }, + { + "epoch": 1.514276301339776, + "grad_norm": 2.036146879196167, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7058373093605042, + "num_tokens": 356729849.0, + "step": 13789 + }, + { + "epoch": 1.5143861190423897, + "grad_norm": 2.308483839035034, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7267143726348877, + "num_tokens": 356755212.0, + "step": 13790 + }, + { + "epoch": 1.5144959367450033, + "grad_norm": 1.9650232791900635, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.7027087807655334, + "num_tokens": 356786954.0, + "step": 13791 + }, + { + "epoch": 1.5146057544476168, + "grad_norm": 2.28373646736145, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7059042453765869, + "num_tokens": 356811653.0, + "step": 13792 + }, + { + "epoch": 1.5147155721502306, + "grad_norm": 1.9912285804748535, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7155674695968628, + "num_tokens": 356841369.0, + "step": 13793 + }, + { + "epoch": 1.5148253898528443, + "grad_norm": 2.207171678543091, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7082862257957458, + "num_tokens": 356870485.0, + "step": 13794 + }, + { + "epoch": 1.514935207555458, + "grad_norm": 1.9436917304992676, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7043526768684387, + "num_tokens": 356904064.0, + "step": 13795 + }, + { + "epoch": 1.5150450252580716, + "grad_norm": 2.4710185527801514, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7372782230377197, + "num_tokens": 356924167.0, + "step": 13796 + }, + { + "epoch": 1.5151548429606851, + "grad_norm": 2.5463624000549316, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.744867205619812, + "num_tokens": 356944442.0, + "step": 13797 + }, + { + "epoch": 1.515264660663299, + "grad_norm": 2.673060178756714, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7159446477890015, + "num_tokens": 356963483.0, + "step": 13798 + }, + { + "epoch": 1.5153744783659127, + "grad_norm": 2.4078176021575928, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7063549160957336, + "num_tokens": 356986322.0, + "step": 13799 + }, + { + "epoch": 1.5154842960685262, + "grad_norm": 2.058109998703003, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6912801861763, + "num_tokens": 357019143.0, + "step": 13800 + }, + { + "epoch": 1.51559411377114, + "grad_norm": 2.427562713623047, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7250970005989075, + "num_tokens": 357041884.0, + "step": 13801 + }, + { + "epoch": 1.5157039314737535, + "grad_norm": 2.0679233074188232, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.705653727054596, + "num_tokens": 357072800.0, + "step": 13802 + }, + { + "epoch": 1.5158137491763672, + "grad_norm": 2.226574420928955, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.723077654838562, + "num_tokens": 357096559.0, + "step": 13803 + }, + { + "epoch": 1.515923566878981, + "grad_norm": 2.4284112453460693, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.6995253562927246, + "num_tokens": 357118220.0, + "step": 13804 + }, + { + "epoch": 1.5160333845815945, + "grad_norm": 2.0899229049682617, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6837650537490845, + "num_tokens": 357150262.0, + "step": 13805 + }, + { + "epoch": 1.516143202284208, + "grad_norm": 2.2059333324432373, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7291854023933411, + "num_tokens": 357176008.0, + "step": 13806 + }, + { + "epoch": 1.5162530199868218, + "grad_norm": 2.258058547973633, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7076504230499268, + "num_tokens": 357202813.0, + "step": 13807 + }, + { + "epoch": 1.5163628376894356, + "grad_norm": 2.2030577659606934, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7236162424087524, + "num_tokens": 357229265.0, + "step": 13808 + }, + { + "epoch": 1.5164726553920493, + "grad_norm": 2.1286532878875732, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7352877855300903, + "num_tokens": 357256720.0, + "step": 13809 + }, + { + "epoch": 1.5165824730946629, + "grad_norm": 2.246136426925659, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7186558246612549, + "num_tokens": 357281438.0, + "step": 13810 + }, + { + "epoch": 1.5166922907972764, + "grad_norm": 2.035487413406372, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7200756072998047, + "num_tokens": 357309993.0, + "step": 13811 + }, + { + "epoch": 1.5168021084998902, + "grad_norm": 2.149374485015869, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7043197751045227, + "num_tokens": 357338261.0, + "step": 13812 + }, + { + "epoch": 1.516911926202504, + "grad_norm": 2.225152015686035, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7217469215393066, + "num_tokens": 357364164.0, + "step": 13813 + }, + { + "epoch": 1.5170217439051175, + "grad_norm": 2.1798276901245117, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7333977818489075, + "num_tokens": 357390889.0, + "step": 13814 + }, + { + "epoch": 1.5171315616077312, + "grad_norm": 2.2134978771209717, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7117960453033447, + "num_tokens": 357417326.0, + "step": 13815 + }, + { + "epoch": 1.5172413793103448, + "grad_norm": 2.6404125690460205, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7325382232666016, + "num_tokens": 357437130.0, + "step": 13816 + }, + { + "epoch": 1.5173511970129585, + "grad_norm": 2.270517349243164, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7245237231254578, + "num_tokens": 357462575.0, + "step": 13817 + }, + { + "epoch": 1.5174610147155723, + "grad_norm": 2.2960939407348633, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7350804805755615, + "num_tokens": 357485898.0, + "step": 13818 + }, + { + "epoch": 1.5175708324181858, + "grad_norm": 2.4685516357421875, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7209984064102173, + "num_tokens": 357508517.0, + "step": 13819 + }, + { + "epoch": 1.5176806501207993, + "grad_norm": 2.1916277408599854, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7315807342529297, + "num_tokens": 357535443.0, + "step": 13820 + }, + { + "epoch": 1.517790467823413, + "grad_norm": 2.0920190811157227, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7147741913795471, + "num_tokens": 357565210.0, + "step": 13821 + }, + { + "epoch": 1.5179002855260268, + "grad_norm": 2.3587028980255127, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7114654779434204, + "num_tokens": 357589676.0, + "step": 13822 + }, + { + "epoch": 1.5180101032286406, + "grad_norm": 2.09336519241333, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7380026578903198, + "num_tokens": 357616481.0, + "step": 13823 + }, + { + "epoch": 1.5181199209312541, + "grad_norm": 2.3689613342285156, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7579644322395325, + "num_tokens": 357637844.0, + "step": 13824 + }, + { + "epoch": 1.5182297386338677, + "grad_norm": 2.059641122817993, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7175467014312744, + "num_tokens": 357665617.0, + "step": 13825 + }, + { + "epoch": 1.5183395563364814, + "grad_norm": 2.076572895050049, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7184383869171143, + "num_tokens": 357693346.0, + "step": 13826 + }, + { + "epoch": 1.5184493740390952, + "grad_norm": 2.322042226791382, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7183794975280762, + "num_tokens": 357716822.0, + "step": 13827 + }, + { + "epoch": 1.5185591917417087, + "grad_norm": 2.1743171215057373, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7234907746315002, + "num_tokens": 357743016.0, + "step": 13828 + }, + { + "epoch": 1.5186690094443223, + "grad_norm": 2.348451852798462, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7120251655578613, + "num_tokens": 357767385.0, + "step": 13829 + }, + { + "epoch": 1.518778827146936, + "grad_norm": 2.225308895111084, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7101942300796509, + "num_tokens": 357794899.0, + "step": 13830 + }, + { + "epoch": 1.5188886448495498, + "grad_norm": 2.345710039138794, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7352727055549622, + "num_tokens": 357817416.0, + "step": 13831 + }, + { + "epoch": 1.5189984625521635, + "grad_norm": 2.243842124938965, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7077246904373169, + "num_tokens": 357843749.0, + "step": 13832 + }, + { + "epoch": 1.519108280254777, + "grad_norm": 2.39445424079895, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7281486988067627, + "num_tokens": 357867032.0, + "step": 13833 + }, + { + "epoch": 1.5192180979573906, + "grad_norm": 2.393221616744995, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7486506700515747, + "num_tokens": 357888379.0, + "step": 13834 + }, + { + "epoch": 1.5193279156600044, + "grad_norm": 2.348389148712158, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.715538501739502, + "num_tokens": 357911911.0, + "step": 13835 + }, + { + "epoch": 1.5194377333626181, + "grad_norm": 2.652783155441284, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7593048810958862, + "num_tokens": 357929136.0, + "step": 13836 + }, + { + "epoch": 1.5195475510652319, + "grad_norm": 2.1595327854156494, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.721455454826355, + "num_tokens": 357955890.0, + "step": 13837 + }, + { + "epoch": 1.5196573687678454, + "grad_norm": 2.4251184463500977, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7267652153968811, + "num_tokens": 357978799.0, + "step": 13838 + }, + { + "epoch": 1.519767186470459, + "grad_norm": 2.743861436843872, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7204666137695312, + "num_tokens": 357997097.0, + "step": 13839 + }, + { + "epoch": 1.5198770041730727, + "grad_norm": 2.1670196056365967, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7155622243881226, + "num_tokens": 358023895.0, + "step": 13840 + }, + { + "epoch": 1.5199868218756865, + "grad_norm": 2.0313496589660645, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7074622511863708, + "num_tokens": 358055221.0, + "step": 13841 + }, + { + "epoch": 1.5200966395783, + "grad_norm": 1.9449315071105957, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7257665395736694, + "num_tokens": 358086249.0, + "step": 13842 + }, + { + "epoch": 1.5202064572809135, + "grad_norm": 2.371265411376953, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.732792854309082, + "num_tokens": 358108163.0, + "step": 13843 + }, + { + "epoch": 1.5203162749835273, + "grad_norm": 2.1759660243988037, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7101470828056335, + "num_tokens": 358136746.0, + "step": 13844 + }, + { + "epoch": 1.520426092686141, + "grad_norm": 2.270465612411499, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7170680165290833, + "num_tokens": 358161247.0, + "step": 13845 + }, + { + "epoch": 1.5205359103887548, + "grad_norm": 2.3211588859558105, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7086694240570068, + "num_tokens": 358185895.0, + "step": 13846 + }, + { + "epoch": 1.5206457280913683, + "grad_norm": 2.3595752716064453, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7370613217353821, + "num_tokens": 358207700.0, + "step": 13847 + }, + { + "epoch": 1.5207555457939819, + "grad_norm": 2.469761848449707, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7295606136322021, + "num_tokens": 358229003.0, + "step": 13848 + }, + { + "epoch": 1.5208653634965956, + "grad_norm": 2.1565816402435303, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7256627082824707, + "num_tokens": 358254281.0, + "step": 13849 + }, + { + "epoch": 1.5209751811992094, + "grad_norm": 2.3211889266967773, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6997368335723877, + "num_tokens": 358280670.0, + "step": 13850 + }, + { + "epoch": 1.5210849989018231, + "grad_norm": 2.023932933807373, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.7001752257347107, + "num_tokens": 358310030.0, + "step": 13851 + }, + { + "epoch": 1.5211948166044367, + "grad_norm": 2.3557991981506348, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7063819766044617, + "num_tokens": 358335096.0, + "step": 13852 + }, + { + "epoch": 1.5213046343070502, + "grad_norm": 2.3408727645874023, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7284829616546631, + "num_tokens": 358356406.0, + "step": 13853 + }, + { + "epoch": 1.521414452009664, + "grad_norm": 2.505528211593628, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7326398491859436, + "num_tokens": 358377855.0, + "step": 13854 + }, + { + "epoch": 1.5215242697122777, + "grad_norm": 2.171666383743286, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7403981685638428, + "num_tokens": 358405613.0, + "step": 13855 + }, + { + "epoch": 1.5216340874148913, + "grad_norm": 2.052030563354492, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6876506805419922, + "num_tokens": 358434728.0, + "step": 13856 + }, + { + "epoch": 1.5217439051175048, + "grad_norm": 2.3259437084198, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7257820963859558, + "num_tokens": 358458558.0, + "step": 13857 + }, + { + "epoch": 1.5218537228201185, + "grad_norm": 2.6575489044189453, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7592180371284485, + "num_tokens": 358475837.0, + "step": 13858 + }, + { + "epoch": 1.5219635405227323, + "grad_norm": 2.3553671836853027, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7329044938087463, + "num_tokens": 358500531.0, + "step": 13859 + }, + { + "epoch": 1.522073358225346, + "grad_norm": 2.70082688331604, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7371538281440735, + "num_tokens": 358519057.0, + "step": 13860 + }, + { + "epoch": 1.5221831759279596, + "grad_norm": 2.2567543983459473, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.720129132270813, + "num_tokens": 358543699.0, + "step": 13861 + }, + { + "epoch": 1.5222929936305731, + "grad_norm": 2.546858310699463, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7276418209075928, + "num_tokens": 358564328.0, + "step": 13862 + }, + { + "epoch": 1.522402811333187, + "grad_norm": 2.2918920516967773, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7399266958236694, + "num_tokens": 358587141.0, + "step": 13863 + }, + { + "epoch": 1.5225126290358006, + "grad_norm": 2.0355706214904785, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7265578508377075, + "num_tokens": 358616127.0, + "step": 13864 + }, + { + "epoch": 1.5226224467384142, + "grad_norm": 2.4268152713775635, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7145472764968872, + "num_tokens": 358639200.0, + "step": 13865 + }, + { + "epoch": 1.522732264441028, + "grad_norm": 2.415160894393921, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7138690948486328, + "num_tokens": 358666300.0, + "step": 13866 + }, + { + "epoch": 1.5228420821436415, + "grad_norm": 2.0414488315582275, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7045980095863342, + "num_tokens": 358696624.0, + "step": 13867 + }, + { + "epoch": 1.5229518998462552, + "grad_norm": 2.058788299560547, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7101690769195557, + "num_tokens": 358725844.0, + "step": 13868 + }, + { + "epoch": 1.523061717548869, + "grad_norm": 2.4723620414733887, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7133677005767822, + "num_tokens": 358746996.0, + "step": 13869 + }, + { + "epoch": 1.5231715352514825, + "grad_norm": 2.2038497924804688, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7274227738380432, + "num_tokens": 358772315.0, + "step": 13870 + }, + { + "epoch": 1.523281352954096, + "grad_norm": 2.165374517440796, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7199659943580627, + "num_tokens": 358798304.0, + "step": 13871 + }, + { + "epoch": 1.5233911706567098, + "grad_norm": 2.194408655166626, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7222400903701782, + "num_tokens": 358823430.0, + "step": 13872 + }, + { + "epoch": 1.5235009883593236, + "grad_norm": 2.598536729812622, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7084434628486633, + "num_tokens": 358845496.0, + "step": 13873 + }, + { + "epoch": 1.5236108060619373, + "grad_norm": 2.0636656284332275, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7269592881202698, + "num_tokens": 358874770.0, + "step": 13874 + }, + { + "epoch": 1.5237206237645509, + "grad_norm": 2.241248607635498, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7259032726287842, + "num_tokens": 358898948.0, + "step": 13875 + }, + { + "epoch": 1.5238304414671644, + "grad_norm": 2.1709201335906982, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7404192686080933, + "num_tokens": 358925509.0, + "step": 13876 + }, + { + "epoch": 1.5239402591697782, + "grad_norm": 2.0124642848968506, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7157316207885742, + "num_tokens": 358956547.0, + "step": 13877 + }, + { + "epoch": 1.524050076872392, + "grad_norm": 2.026735544204712, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7222367525100708, + "num_tokens": 358985992.0, + "step": 13878 + }, + { + "epoch": 1.5241598945750054, + "grad_norm": 2.24061918258667, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7173870801925659, + "num_tokens": 359011171.0, + "step": 13879 + }, + { + "epoch": 1.5242697122776192, + "grad_norm": 2.7501144409179688, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7499337792396545, + "num_tokens": 359029728.0, + "step": 13880 + }, + { + "epoch": 1.5243795299802327, + "grad_norm": 2.494976043701172, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7088894248008728, + "num_tokens": 359051188.0, + "step": 13881 + }, + { + "epoch": 1.5244893476828465, + "grad_norm": 2.2371511459350586, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7221852540969849, + "num_tokens": 359075009.0, + "step": 13882 + }, + { + "epoch": 1.5245991653854603, + "grad_norm": 2.49149489402771, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7435346245765686, + "num_tokens": 359096328.0, + "step": 13883 + }, + { + "epoch": 1.5247089830880738, + "grad_norm": 2.4328696727752686, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7325252294540405, + "num_tokens": 359120262.0, + "step": 13884 + }, + { + "epoch": 1.5248188007906873, + "grad_norm": 2.474745035171509, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.725013792514801, + "num_tokens": 359140696.0, + "step": 13885 + }, + { + "epoch": 1.524928618493301, + "grad_norm": 2.3217551708221436, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7105396389961243, + "num_tokens": 359164468.0, + "step": 13886 + }, + { + "epoch": 1.5250384361959148, + "grad_norm": 2.033038854598999, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7033374905586243, + "num_tokens": 359195179.0, + "step": 13887 + }, + { + "epoch": 1.5251482538985286, + "grad_norm": 2.259697198867798, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6975162625312805, + "num_tokens": 359222376.0, + "step": 13888 + }, + { + "epoch": 1.5252580716011421, + "grad_norm": 2.1915385723114014, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6959573030471802, + "num_tokens": 359251589.0, + "step": 13889 + }, + { + "epoch": 1.5253678893037557, + "grad_norm": 2.0030853748321533, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7050066590309143, + "num_tokens": 359283194.0, + "step": 13890 + }, + { + "epoch": 1.5254777070063694, + "grad_norm": 2.2807536125183105, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7328892946243286, + "num_tokens": 359306658.0, + "step": 13891 + }, + { + "epoch": 1.5255875247089832, + "grad_norm": 2.049335241317749, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.697145938873291, + "num_tokens": 359338558.0, + "step": 13892 + }, + { + "epoch": 1.5256973424115967, + "grad_norm": 2.121138334274292, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7215847969055176, + "num_tokens": 359365423.0, + "step": 13893 + }, + { + "epoch": 1.5258071601142102, + "grad_norm": 2.5826995372772217, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7037011981010437, + "num_tokens": 359387069.0, + "step": 13894 + }, + { + "epoch": 1.525916977816824, + "grad_norm": 2.2386977672576904, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7153048515319824, + "num_tokens": 359412170.0, + "step": 13895 + }, + { + "epoch": 1.5260267955194378, + "grad_norm": 2.100013256072998, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.6976625919342041, + "num_tokens": 359441555.0, + "step": 13896 + }, + { + "epoch": 1.5261366132220515, + "grad_norm": 2.476470708847046, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7471951246261597, + "num_tokens": 359463418.0, + "step": 13897 + }, + { + "epoch": 1.526246430924665, + "grad_norm": 2.062908887863159, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7254839539527893, + "num_tokens": 359493384.0, + "step": 13898 + }, + { + "epoch": 1.5263562486272786, + "grad_norm": 1.9863288402557373, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7245602011680603, + "num_tokens": 359525207.0, + "step": 13899 + }, + { + "epoch": 1.5264660663298923, + "grad_norm": 2.128830909729004, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.719537079334259, + "num_tokens": 359554399.0, + "step": 13900 + }, + { + "epoch": 1.526575884032506, + "grad_norm": 2.4884791374206543, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.728489100933075, + "num_tokens": 359574869.0, + "step": 13901 + }, + { + "epoch": 1.5266857017351199, + "grad_norm": 2.351543426513672, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7179690599441528, + "num_tokens": 359598923.0, + "step": 13902 + }, + { + "epoch": 1.5267955194377334, + "grad_norm": 2.02376389503479, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7048525810241699, + "num_tokens": 359633756.0, + "step": 13903 + }, + { + "epoch": 1.526905337140347, + "grad_norm": 2.461632490158081, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7239104509353638, + "num_tokens": 359655317.0, + "step": 13904 + }, + { + "epoch": 1.5270151548429607, + "grad_norm": 2.3719518184661865, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6930841207504272, + "num_tokens": 359679187.0, + "step": 13905 + }, + { + "epoch": 1.5271249725455744, + "grad_norm": 2.392667770385742, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.724988579750061, + "num_tokens": 359701225.0, + "step": 13906 + }, + { + "epoch": 1.527234790248188, + "grad_norm": 2.108933210372925, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7260306477546692, + "num_tokens": 359731034.0, + "step": 13907 + }, + { + "epoch": 1.5273446079508015, + "grad_norm": 2.1381659507751465, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7050713300704956, + "num_tokens": 359759267.0, + "step": 13908 + }, + { + "epoch": 1.5274544256534153, + "grad_norm": 2.2967827320098877, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7249165177345276, + "num_tokens": 359783366.0, + "step": 13909 + }, + { + "epoch": 1.527564243356029, + "grad_norm": 2.0851035118103027, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7004960179328918, + "num_tokens": 359813623.0, + "step": 13910 + }, + { + "epoch": 1.5276740610586428, + "grad_norm": 2.12265682220459, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7375403046607971, + "num_tokens": 359839622.0, + "step": 13911 + }, + { + "epoch": 1.5277838787612563, + "grad_norm": 2.5566482543945312, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.719585120677948, + "num_tokens": 359861511.0, + "step": 13912 + }, + { + "epoch": 1.5278936964638699, + "grad_norm": 2.5920042991638184, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7104922533035278, + "num_tokens": 359882319.0, + "step": 13913 + }, + { + "epoch": 1.5280035141664836, + "grad_norm": 2.220290184020996, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7148092985153198, + "num_tokens": 359907346.0, + "step": 13914 + }, + { + "epoch": 1.5281133318690974, + "grad_norm": 2.081850290298462, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7268410325050354, + "num_tokens": 359937308.0, + "step": 13915 + }, + { + "epoch": 1.5282231495717111, + "grad_norm": 1.92599356174469, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7219963073730469, + "num_tokens": 359969915.0, + "step": 13916 + }, + { + "epoch": 1.5283329672743247, + "grad_norm": 2.363755464553833, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7238324880599976, + "num_tokens": 359992266.0, + "step": 13917 + }, + { + "epoch": 1.5284427849769382, + "grad_norm": 2.408315420150757, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7325432896614075, + "num_tokens": 360014096.0, + "step": 13918 + }, + { + "epoch": 1.528552602679552, + "grad_norm": 2.041522264480591, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.6980481743812561, + "num_tokens": 360044076.0, + "step": 13919 + }, + { + "epoch": 1.5286624203821657, + "grad_norm": 2.011207342147827, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7193992137908936, + "num_tokens": 360073970.0, + "step": 13920 + }, + { + "epoch": 1.5287722380847792, + "grad_norm": 2.1812314987182617, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7248004078865051, + "num_tokens": 360099803.0, + "step": 13921 + }, + { + "epoch": 1.5288820557873928, + "grad_norm": 2.203672170639038, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7164082527160645, + "num_tokens": 360124665.0, + "step": 13922 + }, + { + "epoch": 1.5289918734900065, + "grad_norm": 2.0501227378845215, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6974462866783142, + "num_tokens": 360154624.0, + "step": 13923 + }, + { + "epoch": 1.5291016911926203, + "grad_norm": 2.150317430496216, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7263662815093994, + "num_tokens": 360181063.0, + "step": 13924 + }, + { + "epoch": 1.529211508895234, + "grad_norm": 2.0052757263183594, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7121667861938477, + "num_tokens": 360209406.0, + "step": 13925 + }, + { + "epoch": 1.5293213265978476, + "grad_norm": 2.4312663078308105, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7358675003051758, + "num_tokens": 360231664.0, + "step": 13926 + }, + { + "epoch": 1.5294311443004611, + "grad_norm": 2.1261556148529053, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7244303226470947, + "num_tokens": 360259971.0, + "step": 13927 + }, + { + "epoch": 1.5295409620030749, + "grad_norm": 2.339369773864746, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7318669557571411, + "num_tokens": 360283538.0, + "step": 13928 + }, + { + "epoch": 1.5296507797056886, + "grad_norm": 2.104593276977539, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7015626430511475, + "num_tokens": 360311061.0, + "step": 13929 + }, + { + "epoch": 1.5297605974083022, + "grad_norm": 2.821782112121582, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7415376305580139, + "num_tokens": 360327539.0, + "step": 13930 + }, + { + "epoch": 1.529870415110916, + "grad_norm": 2.257639169692993, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7162309885025024, + "num_tokens": 360352858.0, + "step": 13931 + }, + { + "epoch": 1.5299802328135295, + "grad_norm": 2.099518299102783, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7132186889648438, + "num_tokens": 360383213.0, + "step": 13932 + }, + { + "epoch": 1.5300900505161432, + "grad_norm": 2.2246053218841553, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7105118632316589, + "num_tokens": 360411710.0, + "step": 13933 + }, + { + "epoch": 1.530199868218757, + "grad_norm": 2.3986618518829346, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7105839252471924, + "num_tokens": 360434777.0, + "step": 13934 + }, + { + "epoch": 1.5303096859213705, + "grad_norm": 1.9997917413711548, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6900572776794434, + "num_tokens": 360467550.0, + "step": 13935 + }, + { + "epoch": 1.530419503623984, + "grad_norm": 2.2422165870666504, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7043623924255371, + "num_tokens": 360494011.0, + "step": 13936 + }, + { + "epoch": 1.5305293213265978, + "grad_norm": 2.1307218074798584, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7193783521652222, + "num_tokens": 360521569.0, + "step": 13937 + }, + { + "epoch": 1.5306391390292116, + "grad_norm": 2.267435073852539, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7214433550834656, + "num_tokens": 360548911.0, + "step": 13938 + }, + { + "epoch": 1.5307489567318253, + "grad_norm": 2.2744908332824707, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7090033888816833, + "num_tokens": 360574672.0, + "step": 13939 + }, + { + "epoch": 1.5308587744344389, + "grad_norm": 2.284998655319214, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.723811149597168, + "num_tokens": 360599959.0, + "step": 13940 + }, + { + "epoch": 1.5309685921370524, + "grad_norm": 2.197683572769165, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7189246416091919, + "num_tokens": 360626176.0, + "step": 13941 + }, + { + "epoch": 1.5310784098396661, + "grad_norm": 2.180601119995117, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7278186082839966, + "num_tokens": 360650766.0, + "step": 13942 + }, + { + "epoch": 1.53118822754228, + "grad_norm": 2.3707351684570312, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7089091539382935, + "num_tokens": 360674516.0, + "step": 13943 + }, + { + "epoch": 1.5312980452448934, + "grad_norm": 2.002363443374634, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7090991735458374, + "num_tokens": 360702725.0, + "step": 13944 + }, + { + "epoch": 1.5314078629475072, + "grad_norm": 2.2900843620300293, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7078627347946167, + "num_tokens": 360727098.0, + "step": 13945 + }, + { + "epoch": 1.5315176806501207, + "grad_norm": 2.195725202560425, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7198162078857422, + "num_tokens": 360755180.0, + "step": 13946 + }, + { + "epoch": 1.5316274983527345, + "grad_norm": 2.308147668838501, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7351050972938538, + "num_tokens": 360779517.0, + "step": 13947 + }, + { + "epoch": 1.5317373160553482, + "grad_norm": 2.441096782684326, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7178243398666382, + "num_tokens": 360799708.0, + "step": 13948 + }, + { + "epoch": 1.5318471337579618, + "grad_norm": 2.269320011138916, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7037661671638489, + "num_tokens": 360825254.0, + "step": 13949 + }, + { + "epoch": 1.5319569514605753, + "grad_norm": 2.5484156608581543, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7174263000488281, + "num_tokens": 360844500.0, + "step": 13950 + }, + { + "epoch": 1.532066769163189, + "grad_norm": 2.604147434234619, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7145252227783203, + "num_tokens": 360864445.0, + "step": 13951 + }, + { + "epoch": 1.5321765868658028, + "grad_norm": 2.3478610515594482, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7432268261909485, + "num_tokens": 360887670.0, + "step": 13952 + }, + { + "epoch": 1.5322864045684166, + "grad_norm": 2.3087360858917236, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7390804290771484, + "num_tokens": 360911609.0, + "step": 13953 + }, + { + "epoch": 1.5323962222710301, + "grad_norm": 2.3420987129211426, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7194753885269165, + "num_tokens": 360935516.0, + "step": 13954 + }, + { + "epoch": 1.5325060399736437, + "grad_norm": 2.539682388305664, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.714557945728302, + "num_tokens": 360957172.0, + "step": 13955 + }, + { + "epoch": 1.5326158576762574, + "grad_norm": 2.1223742961883545, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7428054809570312, + "num_tokens": 360984712.0, + "step": 13956 + }, + { + "epoch": 1.5327256753788712, + "grad_norm": 2.187623977661133, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.6995266675949097, + "num_tokens": 361013040.0, + "step": 13957 + }, + { + "epoch": 1.5328354930814847, + "grad_norm": 2.0423388481140137, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7012720108032227, + "num_tokens": 361045049.0, + "step": 13958 + }, + { + "epoch": 1.5329453107840982, + "grad_norm": 2.1689329147338867, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6948740482330322, + "num_tokens": 361074053.0, + "step": 13959 + }, + { + "epoch": 1.533055128486712, + "grad_norm": 2.164695978164673, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6892725229263306, + "num_tokens": 361104096.0, + "step": 13960 + }, + { + "epoch": 1.5331649461893258, + "grad_norm": 2.3192710876464844, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.723980188369751, + "num_tokens": 361127413.0, + "step": 13961 + }, + { + "epoch": 1.5332747638919395, + "grad_norm": 2.0385961532592773, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7109428644180298, + "num_tokens": 361156264.0, + "step": 13962 + }, + { + "epoch": 1.533384581594553, + "grad_norm": 2.4127490520477295, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7164725065231323, + "num_tokens": 361179506.0, + "step": 13963 + }, + { + "epoch": 1.5334943992971666, + "grad_norm": 1.9951372146606445, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7098550200462341, + "num_tokens": 361208503.0, + "step": 13964 + }, + { + "epoch": 1.5336042169997803, + "grad_norm": 2.082153797149658, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7094857692718506, + "num_tokens": 361236807.0, + "step": 13965 + }, + { + "epoch": 1.533714034702394, + "grad_norm": 2.1505465507507324, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7433146238327026, + "num_tokens": 361260891.0, + "step": 13966 + }, + { + "epoch": 1.5338238524050078, + "grad_norm": 2.5401241779327393, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7526246905326843, + "num_tokens": 361280470.0, + "step": 13967 + }, + { + "epoch": 1.5339336701076214, + "grad_norm": 2.0354416370391846, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7169133424758911, + "num_tokens": 361308667.0, + "step": 13968 + }, + { + "epoch": 1.534043487810235, + "grad_norm": 2.258526086807251, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7289869785308838, + "num_tokens": 361333878.0, + "step": 13969 + }, + { + "epoch": 1.5341533055128487, + "grad_norm": 2.241046905517578, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7018522024154663, + "num_tokens": 361360833.0, + "step": 13970 + }, + { + "epoch": 1.5342631232154624, + "grad_norm": 2.239453077316284, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7113944292068481, + "num_tokens": 361388075.0, + "step": 13971 + }, + { + "epoch": 1.534372940918076, + "grad_norm": 2.3102378845214844, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.726982593536377, + "num_tokens": 361413148.0, + "step": 13972 + }, + { + "epoch": 1.5344827586206895, + "grad_norm": 2.2812228202819824, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7122434973716736, + "num_tokens": 361441051.0, + "step": 13973 + }, + { + "epoch": 1.5345925763233033, + "grad_norm": 2.3870766162872314, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7283481359481812, + "num_tokens": 361464033.0, + "step": 13974 + }, + { + "epoch": 1.534702394025917, + "grad_norm": 2.1782326698303223, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7245080471038818, + "num_tokens": 361490130.0, + "step": 13975 + }, + { + "epoch": 1.5348122117285308, + "grad_norm": 2.078786611557007, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7331066131591797, + "num_tokens": 361516786.0, + "step": 13976 + }, + { + "epoch": 1.5349220294311443, + "grad_norm": 2.29742693901062, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7347633838653564, + "num_tokens": 361541102.0, + "step": 13977 + }, + { + "epoch": 1.5350318471337578, + "grad_norm": 2.519761562347412, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.705794095993042, + "num_tokens": 361564405.0, + "step": 13978 + }, + { + "epoch": 1.5351416648363716, + "grad_norm": 2.3557727336883545, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7376264333724976, + "num_tokens": 361587221.0, + "step": 13979 + }, + { + "epoch": 1.5352514825389854, + "grad_norm": 2.143378496170044, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6906792521476746, + "num_tokens": 361617470.0, + "step": 13980 + }, + { + "epoch": 1.535361300241599, + "grad_norm": 2.4052093029022217, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7076327800750732, + "num_tokens": 361641579.0, + "step": 13981 + }, + { + "epoch": 1.5354711179442126, + "grad_norm": 2.6821951866149902, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7202175855636597, + "num_tokens": 361660463.0, + "step": 13982 + }, + { + "epoch": 1.5355809356468262, + "grad_norm": 2.218878984451294, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7159228920936584, + "num_tokens": 361687467.0, + "step": 13983 + }, + { + "epoch": 1.53569075334944, + "grad_norm": 2.31309175491333, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7323285341262817, + "num_tokens": 361711970.0, + "step": 13984 + }, + { + "epoch": 1.5358005710520537, + "grad_norm": 2.49772047996521, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7223438024520874, + "num_tokens": 361734584.0, + "step": 13985 + }, + { + "epoch": 1.5359103887546672, + "grad_norm": 2.0738744735717773, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7143296003341675, + "num_tokens": 361764865.0, + "step": 13986 + }, + { + "epoch": 1.5360202064572808, + "grad_norm": 2.0633254051208496, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7123953700065613, + "num_tokens": 361794556.0, + "step": 13987 + }, + { + "epoch": 1.5361300241598945, + "grad_norm": 2.3587839603424072, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7220320701599121, + "num_tokens": 361820100.0, + "step": 13988 + }, + { + "epoch": 1.5362398418625083, + "grad_norm": 2.163764238357544, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7107934951782227, + "num_tokens": 361847555.0, + "step": 13989 + }, + { + "epoch": 1.536349659565122, + "grad_norm": 2.2602598667144775, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7254853844642639, + "num_tokens": 361873202.0, + "step": 13990 + }, + { + "epoch": 1.5364594772677356, + "grad_norm": 2.056555986404419, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7286799550056458, + "num_tokens": 361900641.0, + "step": 13991 + }, + { + "epoch": 1.536569294970349, + "grad_norm": 2.539398431777954, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7206093668937683, + "num_tokens": 361920081.0, + "step": 13992 + }, + { + "epoch": 1.5366791126729629, + "grad_norm": 1.9726107120513916, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7097646594047546, + "num_tokens": 361951476.0, + "step": 13993 + }, + { + "epoch": 1.5367889303755766, + "grad_norm": 2.205730438232422, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7148634791374207, + "num_tokens": 361978693.0, + "step": 13994 + }, + { + "epoch": 1.5368987480781902, + "grad_norm": 2.47503924369812, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.6839447021484375, + "num_tokens": 362001740.0, + "step": 13995 + }, + { + "epoch": 1.537008565780804, + "grad_norm": 2.3618264198303223, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7178117036819458, + "num_tokens": 362024253.0, + "step": 13996 + }, + { + "epoch": 1.5371183834834174, + "grad_norm": 2.1192080974578857, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7126811146736145, + "num_tokens": 362052889.0, + "step": 13997 + }, + { + "epoch": 1.5372282011860312, + "grad_norm": 2.072115182876587, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.742879331111908, + "num_tokens": 362081419.0, + "step": 13998 + }, + { + "epoch": 1.537338018888645, + "grad_norm": 2.7438576221466064, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7339153289794922, + "num_tokens": 362101127.0, + "step": 13999 + }, + { + "epoch": 1.5374478365912585, + "grad_norm": 2.1653575897216797, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7485114932060242, + "num_tokens": 362129145.0, + "step": 14000 + }, + { + "epoch": 1.537557654293872, + "grad_norm": 2.6083011627197266, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7347415089607239, + "num_tokens": 362148434.0, + "step": 14001 + }, + { + "epoch": 1.5376674719964858, + "grad_norm": 2.240122079849243, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7068828344345093, + "num_tokens": 362177949.0, + "step": 14002 + }, + { + "epoch": 1.5377772896990995, + "grad_norm": 2.2875895500183105, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7026700377464294, + "num_tokens": 362202940.0, + "step": 14003 + }, + { + "epoch": 1.5378871074017133, + "grad_norm": 2.2950141429901123, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.719465970993042, + "num_tokens": 362227009.0, + "step": 14004 + }, + { + "epoch": 1.5379969251043268, + "grad_norm": 2.2895140647888184, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7332800626754761, + "num_tokens": 362251623.0, + "step": 14005 + }, + { + "epoch": 1.5381067428069404, + "grad_norm": 2.0859384536743164, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6876024007797241, + "num_tokens": 362282130.0, + "step": 14006 + }, + { + "epoch": 1.5382165605095541, + "grad_norm": 2.1962928771972656, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.697595477104187, + "num_tokens": 362309947.0, + "step": 14007 + }, + { + "epoch": 1.5383263782121679, + "grad_norm": 2.5478012561798096, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7252804040908813, + "num_tokens": 362331758.0, + "step": 14008 + }, + { + "epoch": 1.5384361959147814, + "grad_norm": 2.2837374210357666, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7156072854995728, + "num_tokens": 362355589.0, + "step": 14009 + }, + { + "epoch": 1.538546013617395, + "grad_norm": 2.3360724449157715, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7341153025627136, + "num_tokens": 362378441.0, + "step": 14010 + }, + { + "epoch": 1.5386558313200087, + "grad_norm": 2.121093988418579, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7199331521987915, + "num_tokens": 362406264.0, + "step": 14011 + }, + { + "epoch": 1.5387656490226225, + "grad_norm": 2.321772575378418, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7276268601417542, + "num_tokens": 362429698.0, + "step": 14012 + }, + { + "epoch": 1.5388754667252362, + "grad_norm": 2.375525951385498, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7358239889144897, + "num_tokens": 362451815.0, + "step": 14013 + }, + { + "epoch": 1.5389852844278498, + "grad_norm": 2.4262502193450928, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7072758674621582, + "num_tokens": 362475655.0, + "step": 14014 + }, + { + "epoch": 1.5390951021304633, + "grad_norm": 1.9820154905319214, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6979918479919434, + "num_tokens": 362509816.0, + "step": 14015 + }, + { + "epoch": 1.539204919833077, + "grad_norm": 2.1750807762145996, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7280959486961365, + "num_tokens": 362536931.0, + "step": 14016 + }, + { + "epoch": 1.5393147375356908, + "grad_norm": 2.4570462703704834, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7289638519287109, + "num_tokens": 362558784.0, + "step": 14017 + }, + { + "epoch": 1.5394245552383046, + "grad_norm": 2.4355993270874023, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.75437992811203, + "num_tokens": 362579862.0, + "step": 14018 + }, + { + "epoch": 1.539534372940918, + "grad_norm": 2.229546308517456, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7418746948242188, + "num_tokens": 362604486.0, + "step": 14019 + }, + { + "epoch": 1.5396441906435316, + "grad_norm": 2.036378860473633, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7294459939002991, + "num_tokens": 362632912.0, + "step": 14020 + }, + { + "epoch": 1.5397540083461454, + "grad_norm": 2.173578977584839, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7347660064697266, + "num_tokens": 362658011.0, + "step": 14021 + }, + { + "epoch": 1.5398638260487592, + "grad_norm": 2.511535406112671, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7092622518539429, + "num_tokens": 362678574.0, + "step": 14022 + }, + { + "epoch": 1.5399736437513727, + "grad_norm": 2.113980770111084, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7044897675514221, + "num_tokens": 362707936.0, + "step": 14023 + }, + { + "epoch": 1.5400834614539862, + "grad_norm": 2.2390801906585693, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7005813121795654, + "num_tokens": 362736487.0, + "step": 14024 + }, + { + "epoch": 1.5401932791566, + "grad_norm": 2.4003405570983887, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7128645777702332, + "num_tokens": 362759878.0, + "step": 14025 + }, + { + "epoch": 1.5403030968592137, + "grad_norm": 2.04567551612854, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7026066184043884, + "num_tokens": 362788417.0, + "step": 14026 + }, + { + "epoch": 1.5404129145618275, + "grad_norm": 2.0370566844940186, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7222827672958374, + "num_tokens": 362817928.0, + "step": 14027 + }, + { + "epoch": 1.540522732264441, + "grad_norm": 2.1318652629852295, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6890207529067993, + "num_tokens": 362846569.0, + "step": 14028 + }, + { + "epoch": 1.5406325499670546, + "grad_norm": 2.242486000061035, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7364665269851685, + "num_tokens": 362872228.0, + "step": 14029 + }, + { + "epoch": 1.5407423676696683, + "grad_norm": 2.316042184829712, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7337247133255005, + "num_tokens": 362894173.0, + "step": 14030 + }, + { + "epoch": 1.540852185372282, + "grad_norm": 2.216510534286499, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.716497540473938, + "num_tokens": 362920259.0, + "step": 14031 + }, + { + "epoch": 1.5409620030748958, + "grad_norm": 2.607567071914673, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7457823157310486, + "num_tokens": 362938237.0, + "step": 14032 + }, + { + "epoch": 1.5410718207775094, + "grad_norm": 2.4254496097564697, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7222797870635986, + "num_tokens": 362961984.0, + "step": 14033 + }, + { + "epoch": 1.541181638480123, + "grad_norm": 2.571232557296753, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.6965371966362, + "num_tokens": 362985064.0, + "step": 14034 + }, + { + "epoch": 1.5412914561827367, + "grad_norm": 2.3184332847595215, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7131274342536926, + "num_tokens": 363010611.0, + "step": 14035 + }, + { + "epoch": 1.5414012738853504, + "grad_norm": 2.425933837890625, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7318683862686157, + "num_tokens": 363032997.0, + "step": 14036 + }, + { + "epoch": 1.541511091587964, + "grad_norm": 2.155092239379883, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7017216682434082, + "num_tokens": 363060709.0, + "step": 14037 + }, + { + "epoch": 1.5416209092905775, + "grad_norm": 2.220034599304199, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7143675088882446, + "num_tokens": 363087751.0, + "step": 14038 + }, + { + "epoch": 1.5417307269931912, + "grad_norm": 2.490814685821533, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7170999050140381, + "num_tokens": 363109821.0, + "step": 14039 + }, + { + "epoch": 1.541840544695805, + "grad_norm": 2.0984630584716797, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7056006789207458, + "num_tokens": 363140215.0, + "step": 14040 + }, + { + "epoch": 1.5419503623984188, + "grad_norm": 2.264071464538574, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7130116820335388, + "num_tokens": 363164361.0, + "step": 14041 + }, + { + "epoch": 1.5420601801010323, + "grad_norm": 1.9656589031219482, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7043781876564026, + "num_tokens": 363199268.0, + "step": 14042 + }, + { + "epoch": 1.5421699978036458, + "grad_norm": 2.127261161804199, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7022685408592224, + "num_tokens": 363229026.0, + "step": 14043 + }, + { + "epoch": 1.5422798155062596, + "grad_norm": 2.1720314025878906, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7285343408584595, + "num_tokens": 363257039.0, + "step": 14044 + }, + { + "epoch": 1.5423896332088733, + "grad_norm": 2.2150938510894775, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7051650285720825, + "num_tokens": 363282409.0, + "step": 14045 + }, + { + "epoch": 1.5424994509114869, + "grad_norm": 2.2165088653564453, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7109510898590088, + "num_tokens": 363311899.0, + "step": 14046 + }, + { + "epoch": 1.5426092686141006, + "grad_norm": 2.1671595573425293, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7253965735435486, + "num_tokens": 363337782.0, + "step": 14047 + }, + { + "epoch": 1.5427190863167142, + "grad_norm": 2.3267102241516113, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7325475811958313, + "num_tokens": 363361218.0, + "step": 14048 + }, + { + "epoch": 1.542828904019328, + "grad_norm": 1.9774363040924072, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.730388879776001, + "num_tokens": 363390987.0, + "step": 14049 + }, + { + "epoch": 1.5429387217219417, + "grad_norm": 1.8194799423217773, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7383650541305542, + "num_tokens": 363423598.0, + "step": 14050 + }, + { + "epoch": 1.5430485394245552, + "grad_norm": 2.3580214977264404, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7294527292251587, + "num_tokens": 363445154.0, + "step": 14051 + }, + { + "epoch": 1.5431583571271688, + "grad_norm": 2.01289439201355, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7031185626983643, + "num_tokens": 363475840.0, + "step": 14052 + }, + { + "epoch": 1.5432681748297825, + "grad_norm": 2.321772575378418, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7391237020492554, + "num_tokens": 363497933.0, + "step": 14053 + }, + { + "epoch": 1.5433779925323963, + "grad_norm": 2.090092897415161, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7164956331253052, + "num_tokens": 363524024.0, + "step": 14054 + }, + { + "epoch": 1.54348781023501, + "grad_norm": 2.278787612915039, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7188748121261597, + "num_tokens": 363549265.0, + "step": 14055 + }, + { + "epoch": 1.5435976279376236, + "grad_norm": 2.5029056072235107, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.724078357219696, + "num_tokens": 363571016.0, + "step": 14056 + }, + { + "epoch": 1.543707445640237, + "grad_norm": 2.062136650085449, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6966274976730347, + "num_tokens": 363600326.0, + "step": 14057 + }, + { + "epoch": 1.5438172633428509, + "grad_norm": 2.1838643550872803, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7163974642753601, + "num_tokens": 363626848.0, + "step": 14058 + }, + { + "epoch": 1.5439270810454646, + "grad_norm": 2.060059070587158, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7197645306587219, + "num_tokens": 363656209.0, + "step": 14059 + }, + { + "epoch": 1.5440368987480781, + "grad_norm": 2.1379477977752686, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7183484435081482, + "num_tokens": 363683720.0, + "step": 14060 + }, + { + "epoch": 1.544146716450692, + "grad_norm": 2.376535415649414, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7175178527832031, + "num_tokens": 363705783.0, + "step": 14061 + }, + { + "epoch": 1.5442565341533054, + "grad_norm": 2.3041627407073975, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7202460169792175, + "num_tokens": 363729198.0, + "step": 14062 + }, + { + "epoch": 1.5443663518559192, + "grad_norm": 2.0684850215911865, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7369658946990967, + "num_tokens": 363757236.0, + "step": 14063 + }, + { + "epoch": 1.544476169558533, + "grad_norm": 2.202059507369995, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6893855929374695, + "num_tokens": 363784799.0, + "step": 14064 + }, + { + "epoch": 1.5445859872611465, + "grad_norm": 2.4265358448028564, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7261572480201721, + "num_tokens": 363806497.0, + "step": 14065 + }, + { + "epoch": 1.54469580496376, + "grad_norm": 2.241013288497925, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7016657590866089, + "num_tokens": 363833430.0, + "step": 14066 + }, + { + "epoch": 1.5448056226663738, + "grad_norm": 2.4056649208068848, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.709754467010498, + "num_tokens": 363856361.0, + "step": 14067 + }, + { + "epoch": 1.5449154403689875, + "grad_norm": 2.23746395111084, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7274813055992126, + "num_tokens": 363881557.0, + "step": 14068 + }, + { + "epoch": 1.5450252580716013, + "grad_norm": 2.645439624786377, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7166286110877991, + "num_tokens": 363901078.0, + "step": 14069 + }, + { + "epoch": 1.5451350757742148, + "grad_norm": 2.4075138568878174, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7422857284545898, + "num_tokens": 363923541.0, + "step": 14070 + }, + { + "epoch": 1.5452448934768284, + "grad_norm": 2.300154447555542, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.717606246471405, + "num_tokens": 363947757.0, + "step": 14071 + }, + { + "epoch": 1.5453547111794421, + "grad_norm": 2.4973809719085693, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7098870277404785, + "num_tokens": 363969680.0, + "step": 14072 + }, + { + "epoch": 1.5454645288820559, + "grad_norm": 2.251729965209961, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6980395913124084, + "num_tokens": 363995485.0, + "step": 14073 + }, + { + "epoch": 1.5455743465846694, + "grad_norm": 2.4453721046447754, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7098979949951172, + "num_tokens": 364017312.0, + "step": 14074 + }, + { + "epoch": 1.545684164287283, + "grad_norm": 2.201624631881714, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7150849103927612, + "num_tokens": 364043938.0, + "step": 14075 + }, + { + "epoch": 1.5457939819898967, + "grad_norm": 2.3303720951080322, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7387369871139526, + "num_tokens": 364066453.0, + "step": 14076 + }, + { + "epoch": 1.5459037996925105, + "grad_norm": 2.3481552600860596, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7007260322570801, + "num_tokens": 364089828.0, + "step": 14077 + }, + { + "epoch": 1.5460136173951242, + "grad_norm": 2.2208690643310547, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7213477492332458, + "num_tokens": 364117049.0, + "step": 14078 + }, + { + "epoch": 1.5461234350977378, + "grad_norm": 2.3068478107452393, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7228427529335022, + "num_tokens": 364141653.0, + "step": 14079 + }, + { + "epoch": 1.5462332528003513, + "grad_norm": 2.2213027477264404, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7460034489631653, + "num_tokens": 364167485.0, + "step": 14080 + }, + { + "epoch": 1.546343070502965, + "grad_norm": 2.8276846408843994, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7518252730369568, + "num_tokens": 364183903.0, + "step": 14081 + }, + { + "epoch": 1.5464528882055788, + "grad_norm": 2.1935625076293945, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7064602971076965, + "num_tokens": 364210196.0, + "step": 14082 + }, + { + "epoch": 1.5465627059081926, + "grad_norm": 2.235121965408325, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7311128973960876, + "num_tokens": 364235182.0, + "step": 14083 + }, + { + "epoch": 1.546672523610806, + "grad_norm": 2.167788505554199, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7171369791030884, + "num_tokens": 364260900.0, + "step": 14084 + }, + { + "epoch": 1.5467823413134196, + "grad_norm": 2.2183473110198975, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6988962292671204, + "num_tokens": 364288023.0, + "step": 14085 + }, + { + "epoch": 1.5468921590160334, + "grad_norm": 1.9578970670700073, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7154629826545715, + "num_tokens": 364318840.0, + "step": 14086 + }, + { + "epoch": 1.5470019767186471, + "grad_norm": 2.1293716430664062, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7171759605407715, + "num_tokens": 364346546.0, + "step": 14087 + }, + { + "epoch": 1.5471117944212607, + "grad_norm": 2.690880298614502, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.731363832950592, + "num_tokens": 364367853.0, + "step": 14088 + }, + { + "epoch": 1.5472216121238742, + "grad_norm": 2.05021333694458, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7122626304626465, + "num_tokens": 364395855.0, + "step": 14089 + }, + { + "epoch": 1.547331429826488, + "grad_norm": 2.289752960205078, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6980156898498535, + "num_tokens": 364421719.0, + "step": 14090 + }, + { + "epoch": 1.5474412475291017, + "grad_norm": 2.55914044380188, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.734585702419281, + "num_tokens": 364444366.0, + "step": 14091 + }, + { + "epoch": 1.5475510652317155, + "grad_norm": 2.3984804153442383, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7113759517669678, + "num_tokens": 364467061.0, + "step": 14092 + }, + { + "epoch": 1.547660882934329, + "grad_norm": 2.437199115753174, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7309673428535461, + "num_tokens": 364488964.0, + "step": 14093 + }, + { + "epoch": 1.5477707006369426, + "grad_norm": 2.396951913833618, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7147526741027832, + "num_tokens": 364511490.0, + "step": 14094 + }, + { + "epoch": 1.5478805183395563, + "grad_norm": 2.3519182205200195, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7239036560058594, + "num_tokens": 364537702.0, + "step": 14095 + }, + { + "epoch": 1.54799033604217, + "grad_norm": 2.0145175457000732, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7235535979270935, + "num_tokens": 364566065.0, + "step": 14096 + }, + { + "epoch": 1.5481001537447838, + "grad_norm": 2.263226270675659, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.724332332611084, + "num_tokens": 364591670.0, + "step": 14097 + }, + { + "epoch": 1.5482099714473974, + "grad_norm": 1.9580928087234497, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7306066751480103, + "num_tokens": 364622290.0, + "step": 14098 + }, + { + "epoch": 1.548319789150011, + "grad_norm": 2.4903318881988525, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7245383858680725, + "num_tokens": 364643694.0, + "step": 14099 + }, + { + "epoch": 1.5484296068526247, + "grad_norm": 2.1942086219787598, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7142224311828613, + "num_tokens": 364671334.0, + "step": 14100 + }, + { + "epoch": 1.5485394245552384, + "grad_norm": 2.35917329788208, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7155123353004456, + "num_tokens": 364694191.0, + "step": 14101 + }, + { + "epoch": 1.548649242257852, + "grad_norm": 2.3605408668518066, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7121302485466003, + "num_tokens": 364718287.0, + "step": 14102 + }, + { + "epoch": 1.5487590599604655, + "grad_norm": 2.307698965072632, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.747113049030304, + "num_tokens": 364742007.0, + "step": 14103 + }, + { + "epoch": 1.5488688776630792, + "grad_norm": 2.0455222129821777, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7137869000434875, + "num_tokens": 364772320.0, + "step": 14104 + }, + { + "epoch": 1.548978695365693, + "grad_norm": 2.412170886993408, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7241230010986328, + "num_tokens": 364794693.0, + "step": 14105 + }, + { + "epoch": 1.5490885130683067, + "grad_norm": 2.225208044052124, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7136609554290771, + "num_tokens": 364821681.0, + "step": 14106 + }, + { + "epoch": 1.5491983307709203, + "grad_norm": 1.9569029808044434, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6927998065948486, + "num_tokens": 364854392.0, + "step": 14107 + }, + { + "epoch": 1.5493081484735338, + "grad_norm": 2.189406633377075, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.72376549243927, + "num_tokens": 364882341.0, + "step": 14108 + }, + { + "epoch": 1.5494179661761476, + "grad_norm": 2.337176561355591, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7125915884971619, + "num_tokens": 364905926.0, + "step": 14109 + }, + { + "epoch": 1.5495277838787613, + "grad_norm": 2.0609121322631836, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7035508751869202, + "num_tokens": 364934990.0, + "step": 14110 + }, + { + "epoch": 1.5496376015813749, + "grad_norm": 2.3323140144348145, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7104595303535461, + "num_tokens": 364960789.0, + "step": 14111 + }, + { + "epoch": 1.5497474192839886, + "grad_norm": 2.0876386165618896, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7100798487663269, + "num_tokens": 364986802.0, + "step": 14112 + }, + { + "epoch": 1.5498572369866022, + "grad_norm": 2.2488667964935303, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7153566479682922, + "num_tokens": 365012297.0, + "step": 14113 + }, + { + "epoch": 1.549967054689216, + "grad_norm": 2.185744047164917, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7334524393081665, + "num_tokens": 365037997.0, + "step": 14114 + }, + { + "epoch": 1.5500768723918297, + "grad_norm": 2.639315366744995, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7490525245666504, + "num_tokens": 365058239.0, + "step": 14115 + }, + { + "epoch": 1.5501866900944432, + "grad_norm": 2.194524049758911, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7071965932846069, + "num_tokens": 365086407.0, + "step": 14116 + }, + { + "epoch": 1.5502965077970567, + "grad_norm": 2.4534823894500732, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7387371063232422, + "num_tokens": 365109176.0, + "step": 14117 + }, + { + "epoch": 1.5504063254996705, + "grad_norm": 2.4360196590423584, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.723268985748291, + "num_tokens": 365131615.0, + "step": 14118 + }, + { + "epoch": 1.5505161432022843, + "grad_norm": 2.34171724319458, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.7027958035469055, + "num_tokens": 365158257.0, + "step": 14119 + }, + { + "epoch": 1.550625960904898, + "grad_norm": 2.772700071334839, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7547612190246582, + "num_tokens": 365175178.0, + "step": 14120 + }, + { + "epoch": 1.5507357786075116, + "grad_norm": 2.4296751022338867, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7273182272911072, + "num_tokens": 365197933.0, + "step": 14121 + }, + { + "epoch": 1.550845596310125, + "grad_norm": 2.0466043949127197, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.705610990524292, + "num_tokens": 365233179.0, + "step": 14122 + }, + { + "epoch": 1.5509554140127388, + "grad_norm": 2.144895315170288, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7096049189567566, + "num_tokens": 365259249.0, + "step": 14123 + }, + { + "epoch": 1.5510652317153526, + "grad_norm": 2.3350608348846436, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7219754457473755, + "num_tokens": 365282260.0, + "step": 14124 + }, + { + "epoch": 1.5511750494179661, + "grad_norm": 2.1815009117126465, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.71709144115448, + "num_tokens": 365308282.0, + "step": 14125 + }, + { + "epoch": 1.55128486712058, + "grad_norm": 2.1478652954101562, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7240794897079468, + "num_tokens": 365335380.0, + "step": 14126 + }, + { + "epoch": 1.5513946848231934, + "grad_norm": 2.2028939723968506, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7159218788146973, + "num_tokens": 365360887.0, + "step": 14127 + }, + { + "epoch": 1.5515045025258072, + "grad_norm": 2.5297129154205322, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7422057390213013, + "num_tokens": 365380668.0, + "step": 14128 + }, + { + "epoch": 1.551614320228421, + "grad_norm": 2.1057722568511963, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.728434681892395, + "num_tokens": 365407323.0, + "step": 14129 + }, + { + "epoch": 1.5517241379310345, + "grad_norm": 2.758425712585449, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7407492995262146, + "num_tokens": 365426765.0, + "step": 14130 + }, + { + "epoch": 1.551833955633648, + "grad_norm": 2.140528440475464, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7228984832763672, + "num_tokens": 365452730.0, + "step": 14131 + }, + { + "epoch": 1.5519437733362618, + "grad_norm": 2.216893434524536, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7264429330825806, + "num_tokens": 365478368.0, + "step": 14132 + }, + { + "epoch": 1.5520535910388755, + "grad_norm": 2.3359079360961914, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7149505615234375, + "num_tokens": 365502634.0, + "step": 14133 + }, + { + "epoch": 1.5521634087414893, + "grad_norm": 2.532846689224243, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7301293611526489, + "num_tokens": 365523409.0, + "step": 14134 + }, + { + "epoch": 1.5522732264441028, + "grad_norm": 2.154578685760498, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7231053113937378, + "num_tokens": 365550306.0, + "step": 14135 + }, + { + "epoch": 1.5523830441467164, + "grad_norm": 2.171740770339966, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7453364133834839, + "num_tokens": 365575175.0, + "step": 14136 + }, + { + "epoch": 1.55249286184933, + "grad_norm": 2.7530324459075928, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7501131892204285, + "num_tokens": 365592589.0, + "step": 14137 + }, + { + "epoch": 1.5526026795519439, + "grad_norm": 2.417963981628418, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7291433811187744, + "num_tokens": 365615450.0, + "step": 14138 + }, + { + "epoch": 1.5527124972545574, + "grad_norm": 2.2146971225738525, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7215273380279541, + "num_tokens": 365643137.0, + "step": 14139 + }, + { + "epoch": 1.552822314957171, + "grad_norm": 2.078850507736206, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7246001958847046, + "num_tokens": 365670653.0, + "step": 14140 + }, + { + "epoch": 1.5529321326597847, + "grad_norm": 2.099876880645752, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7009801268577576, + "num_tokens": 365699304.0, + "step": 14141 + }, + { + "epoch": 1.5530419503623984, + "grad_norm": 2.180889129638672, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7027373313903809, + "num_tokens": 365727050.0, + "step": 14142 + }, + { + "epoch": 1.5531517680650122, + "grad_norm": 2.377690315246582, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7200254201889038, + "num_tokens": 365752438.0, + "step": 14143 + }, + { + "epoch": 1.5532615857676257, + "grad_norm": 2.291571855545044, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7174253463745117, + "num_tokens": 365775100.0, + "step": 14144 + }, + { + "epoch": 1.5533714034702393, + "grad_norm": 2.143934726715088, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6888445615768433, + "num_tokens": 365804574.0, + "step": 14145 + }, + { + "epoch": 1.553481221172853, + "grad_norm": 2.4836432933807373, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7029726505279541, + "num_tokens": 365827418.0, + "step": 14146 + }, + { + "epoch": 1.5535910388754668, + "grad_norm": 2.4627068042755127, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7541751861572266, + "num_tokens": 365847419.0, + "step": 14147 + }, + { + "epoch": 1.5537008565780805, + "grad_norm": 2.1735424995422363, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7235974073410034, + "num_tokens": 365874118.0, + "step": 14148 + }, + { + "epoch": 1.553810674280694, + "grad_norm": 2.091974973678589, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7127754092216492, + "num_tokens": 365902006.0, + "step": 14149 + }, + { + "epoch": 1.5539204919833076, + "grad_norm": 2.4543991088867188, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7204478979110718, + "num_tokens": 365922732.0, + "step": 14150 + }, + { + "epoch": 1.5540303096859214, + "grad_norm": 2.1730895042419434, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6962485313415527, + "num_tokens": 365951237.0, + "step": 14151 + }, + { + "epoch": 1.5541401273885351, + "grad_norm": 2.2297356128692627, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7231184244155884, + "num_tokens": 365976298.0, + "step": 14152 + }, + { + "epoch": 1.5542499450911487, + "grad_norm": 2.2109193801879883, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7020478248596191, + "num_tokens": 366003615.0, + "step": 14153 + }, + { + "epoch": 1.5543597627937622, + "grad_norm": 2.342733144760132, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7151106595993042, + "num_tokens": 366028463.0, + "step": 14154 + }, + { + "epoch": 1.554469580496376, + "grad_norm": 2.2546298503875732, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.713379979133606, + "num_tokens": 366054708.0, + "step": 14155 + }, + { + "epoch": 1.5545793981989897, + "grad_norm": 2.0607569217681885, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7203805446624756, + "num_tokens": 366084130.0, + "step": 14156 + }, + { + "epoch": 1.5546892159016035, + "grad_norm": 2.522599220275879, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7425920963287354, + "num_tokens": 366105798.0, + "step": 14157 + }, + { + "epoch": 1.554799033604217, + "grad_norm": 1.8875067234039307, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7061306834220886, + "num_tokens": 366138830.0, + "step": 14158 + }, + { + "epoch": 1.5549088513068305, + "grad_norm": 2.2827305793762207, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7306034564971924, + "num_tokens": 366164338.0, + "step": 14159 + }, + { + "epoch": 1.5550186690094443, + "grad_norm": 2.391655206680298, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7191638946533203, + "num_tokens": 366188310.0, + "step": 14160 + }, + { + "epoch": 1.555128486712058, + "grad_norm": 2.0876660346984863, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7295870184898376, + "num_tokens": 366218560.0, + "step": 14161 + }, + { + "epoch": 1.5552383044146716, + "grad_norm": 2.4806599617004395, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.726770281791687, + "num_tokens": 366239808.0, + "step": 14162 + }, + { + "epoch": 1.5553481221172853, + "grad_norm": 2.257009983062744, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7332906723022461, + "num_tokens": 366265356.0, + "step": 14163 + }, + { + "epoch": 1.5554579398198989, + "grad_norm": 2.1228888034820557, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.718507707118988, + "num_tokens": 366294312.0, + "step": 14164 + }, + { + "epoch": 1.5555677575225126, + "grad_norm": 2.5250084400177, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7129234075546265, + "num_tokens": 366315345.0, + "step": 14165 + }, + { + "epoch": 1.5556775752251264, + "grad_norm": 2.2889552116394043, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7091343402862549, + "num_tokens": 366338596.0, + "step": 14166 + }, + { + "epoch": 1.55578739292774, + "grad_norm": 2.432039976119995, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7456890940666199, + "num_tokens": 366360721.0, + "step": 14167 + }, + { + "epoch": 1.5558972106303535, + "grad_norm": 2.44641375541687, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7353450059890747, + "num_tokens": 366381748.0, + "step": 14168 + }, + { + "epoch": 1.5560070283329672, + "grad_norm": 2.1703968048095703, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7227493524551392, + "num_tokens": 366408008.0, + "step": 14169 + }, + { + "epoch": 1.556116846035581, + "grad_norm": 2.3507866859436035, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7229931354522705, + "num_tokens": 366431701.0, + "step": 14170 + }, + { + "epoch": 1.5562266637381947, + "grad_norm": 2.141503095626831, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7278398275375366, + "num_tokens": 366457522.0, + "step": 14171 + }, + { + "epoch": 1.5563364814408083, + "grad_norm": 2.196016311645508, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7375848889350891, + "num_tokens": 366481980.0, + "step": 14172 + }, + { + "epoch": 1.5564462991434218, + "grad_norm": 2.4311399459838867, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.6994631290435791, + "num_tokens": 366504694.0, + "step": 14173 + }, + { + "epoch": 1.5565561168460356, + "grad_norm": 2.4237678050994873, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.719139575958252, + "num_tokens": 366526182.0, + "step": 14174 + }, + { + "epoch": 1.5566659345486493, + "grad_norm": 2.0509700775146484, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.729642391204834, + "num_tokens": 366555008.0, + "step": 14175 + }, + { + "epoch": 1.5567757522512629, + "grad_norm": 2.238438844680786, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7252024412155151, + "num_tokens": 366580796.0, + "step": 14176 + }, + { + "epoch": 1.5568855699538766, + "grad_norm": 2.2886407375335693, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7137288451194763, + "num_tokens": 366606322.0, + "step": 14177 + }, + { + "epoch": 1.5569953876564901, + "grad_norm": 2.344050884246826, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7222244143486023, + "num_tokens": 366629526.0, + "step": 14178 + }, + { + "epoch": 1.557105205359104, + "grad_norm": 2.400110960006714, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.735450029373169, + "num_tokens": 366650121.0, + "step": 14179 + }, + { + "epoch": 1.5572150230617177, + "grad_norm": 2.4671590328216553, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7196968793869019, + "num_tokens": 366673256.0, + "step": 14180 + }, + { + "epoch": 1.5573248407643312, + "grad_norm": 2.0919086933135986, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7154235243797302, + "num_tokens": 366703456.0, + "step": 14181 + }, + { + "epoch": 1.5574346584669447, + "grad_norm": 2.068324565887451, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.704859733581543, + "num_tokens": 366734352.0, + "step": 14182 + }, + { + "epoch": 1.5575444761695585, + "grad_norm": 2.2126986980438232, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7233939170837402, + "num_tokens": 366761205.0, + "step": 14183 + }, + { + "epoch": 1.5576542938721722, + "grad_norm": 2.2484076023101807, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7252733111381531, + "num_tokens": 366785165.0, + "step": 14184 + }, + { + "epoch": 1.557764111574786, + "grad_norm": 2.232891082763672, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7250068783760071, + "num_tokens": 366809717.0, + "step": 14185 + }, + { + "epoch": 1.5578739292773995, + "grad_norm": 2.244732618331909, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.746903121471405, + "num_tokens": 366832449.0, + "step": 14186 + }, + { + "epoch": 1.557983746980013, + "grad_norm": 2.3837125301361084, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7259158492088318, + "num_tokens": 366857295.0, + "step": 14187 + }, + { + "epoch": 1.5580935646826268, + "grad_norm": 2.1749465465545654, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6923756003379822, + "num_tokens": 366885833.0, + "step": 14188 + }, + { + "epoch": 1.5582033823852406, + "grad_norm": 2.0967929363250732, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7140178680419922, + "num_tokens": 366912767.0, + "step": 14189 + }, + { + "epoch": 1.5583132000878541, + "grad_norm": 2.412363052368164, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7169280648231506, + "num_tokens": 366935518.0, + "step": 14190 + }, + { + "epoch": 1.5584230177904677, + "grad_norm": 2.221961259841919, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7222087383270264, + "num_tokens": 366960821.0, + "step": 14191 + }, + { + "epoch": 1.5585328354930814, + "grad_norm": 1.8953951597213745, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.695722222328186, + "num_tokens": 366996506.0, + "step": 14192 + }, + { + "epoch": 1.5586426531956952, + "grad_norm": 2.3792378902435303, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7123891115188599, + "num_tokens": 367021721.0, + "step": 14193 + }, + { + "epoch": 1.558752470898309, + "grad_norm": 2.262280225753784, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7286511063575745, + "num_tokens": 367047808.0, + "step": 14194 + }, + { + "epoch": 1.5588622886009225, + "grad_norm": 2.2970027923583984, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7040656208992004, + "num_tokens": 367073388.0, + "step": 14195 + }, + { + "epoch": 1.558972106303536, + "grad_norm": 2.218573570251465, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.704808235168457, + "num_tokens": 367101125.0, + "step": 14196 + }, + { + "epoch": 1.5590819240061498, + "grad_norm": 2.382366180419922, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.704902172088623, + "num_tokens": 367126849.0, + "step": 14197 + }, + { + "epoch": 1.5591917417087635, + "grad_norm": 2.3319191932678223, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7142230272293091, + "num_tokens": 367150035.0, + "step": 14198 + }, + { + "epoch": 1.5593015594113773, + "grad_norm": 2.357128858566284, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7409533262252808, + "num_tokens": 367171590.0, + "step": 14199 + }, + { + "epoch": 1.5594113771139908, + "grad_norm": 1.974814534187317, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7312730550765991, + "num_tokens": 367199356.0, + "step": 14200 + }, + { + "epoch": 1.5595211948166043, + "grad_norm": 2.035914659500122, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7011677026748657, + "num_tokens": 367228959.0, + "step": 14201 + }, + { + "epoch": 1.559631012519218, + "grad_norm": 2.0854551792144775, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7189656496047974, + "num_tokens": 367256230.0, + "step": 14202 + }, + { + "epoch": 1.5597408302218319, + "grad_norm": 2.020402193069458, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7084676623344421, + "num_tokens": 367287741.0, + "step": 14203 + }, + { + "epoch": 1.5598506479244454, + "grad_norm": 2.414515256881714, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7363016605377197, + "num_tokens": 367310656.0, + "step": 14204 + }, + { + "epoch": 1.559960465627059, + "grad_norm": 2.2054786682128906, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7222470045089722, + "num_tokens": 367337462.0, + "step": 14205 + }, + { + "epoch": 1.5600702833296727, + "grad_norm": 2.2077834606170654, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7287764549255371, + "num_tokens": 367363685.0, + "step": 14206 + }, + { + "epoch": 1.5601801010322864, + "grad_norm": 2.541944742202759, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7199273705482483, + "num_tokens": 367386120.0, + "step": 14207 + }, + { + "epoch": 1.5602899187349002, + "grad_norm": 2.089628219604492, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6886814832687378, + "num_tokens": 367417056.0, + "step": 14208 + }, + { + "epoch": 1.5603997364375137, + "grad_norm": 2.176126480102539, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7226755619049072, + "num_tokens": 367442342.0, + "step": 14209 + }, + { + "epoch": 1.5605095541401273, + "grad_norm": 1.9614813327789307, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7212120294570923, + "num_tokens": 367472690.0, + "step": 14210 + }, + { + "epoch": 1.560619371842741, + "grad_norm": 2.2675609588623047, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7106317281723022, + "num_tokens": 367498412.0, + "step": 14211 + }, + { + "epoch": 1.5607291895453548, + "grad_norm": 2.1130154132843018, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7342039346694946, + "num_tokens": 367525955.0, + "step": 14212 + }, + { + "epoch": 1.5608390072479685, + "grad_norm": 2.0488975048065186, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7346062660217285, + "num_tokens": 367554794.0, + "step": 14213 + }, + { + "epoch": 1.560948824950582, + "grad_norm": 2.026606321334839, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7108136415481567, + "num_tokens": 367586005.0, + "step": 14214 + }, + { + "epoch": 1.5610586426531956, + "grad_norm": 2.1662580966949463, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7298957705497742, + "num_tokens": 367612133.0, + "step": 14215 + }, + { + "epoch": 1.5611684603558094, + "grad_norm": 2.4363043308258057, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7264402508735657, + "num_tokens": 367633180.0, + "step": 14216 + }, + { + "epoch": 1.5612782780584231, + "grad_norm": 2.1644694805145264, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7082715034484863, + "num_tokens": 367663191.0, + "step": 14217 + }, + { + "epoch": 1.5613880957610367, + "grad_norm": 2.2652173042297363, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7100520133972168, + "num_tokens": 367689478.0, + "step": 14218 + }, + { + "epoch": 1.5614979134636502, + "grad_norm": 2.1880435943603516, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7400959134101868, + "num_tokens": 367713709.0, + "step": 14219 + }, + { + "epoch": 1.561607731166264, + "grad_norm": 2.0204405784606934, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7019261717796326, + "num_tokens": 367744479.0, + "step": 14220 + }, + { + "epoch": 1.5617175488688777, + "grad_norm": 2.640510082244873, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7376617193222046, + "num_tokens": 367765403.0, + "step": 14221 + }, + { + "epoch": 1.5618273665714915, + "grad_norm": 2.1581292152404785, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7279844284057617, + "num_tokens": 367792333.0, + "step": 14222 + }, + { + "epoch": 1.561937184274105, + "grad_norm": 2.1601223945617676, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7258549928665161, + "num_tokens": 367817925.0, + "step": 14223 + }, + { + "epoch": 1.5620470019767185, + "grad_norm": 2.0785820484161377, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7201229929924011, + "num_tokens": 367846064.0, + "step": 14224 + }, + { + "epoch": 1.5621568196793323, + "grad_norm": 2.411288261413574, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7330091595649719, + "num_tokens": 367871089.0, + "step": 14225 + }, + { + "epoch": 1.562266637381946, + "grad_norm": 2.7547550201416016, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7287728190422058, + "num_tokens": 367891372.0, + "step": 14226 + }, + { + "epoch": 1.5623764550845596, + "grad_norm": 2.351384401321411, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7291228175163269, + "num_tokens": 367914436.0, + "step": 14227 + }, + { + "epoch": 1.5624862727871733, + "grad_norm": 2.7236056327819824, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.7488444447517395, + "num_tokens": 367931973.0, + "step": 14228 + }, + { + "epoch": 1.5625960904897869, + "grad_norm": 2.139125347137451, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.725141167640686, + "num_tokens": 367957829.0, + "step": 14229 + }, + { + "epoch": 1.5627059081924006, + "grad_norm": 2.3690578937530518, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7002449631690979, + "num_tokens": 367981594.0, + "step": 14230 + }, + { + "epoch": 1.5628157258950144, + "grad_norm": 2.0117080211639404, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7212613821029663, + "num_tokens": 368012875.0, + "step": 14231 + }, + { + "epoch": 1.562925543597628, + "grad_norm": 2.209038019180298, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7257230281829834, + "num_tokens": 368038115.0, + "step": 14232 + }, + { + "epoch": 1.5630353613002415, + "grad_norm": 2.242168426513672, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.6987466812133789, + "num_tokens": 368064082.0, + "step": 14233 + }, + { + "epoch": 1.5631451790028552, + "grad_norm": 2.4940967559814453, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7239588499069214, + "num_tokens": 368085281.0, + "step": 14234 + }, + { + "epoch": 1.563254996705469, + "grad_norm": 2.3458547592163086, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7383720278739929, + "num_tokens": 368108808.0, + "step": 14235 + }, + { + "epoch": 1.5633648144080827, + "grad_norm": 2.5317254066467285, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7577759027481079, + "num_tokens": 368127261.0, + "step": 14236 + }, + { + "epoch": 1.5634746321106963, + "grad_norm": 2.0955209732055664, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.746389627456665, + "num_tokens": 368152865.0, + "step": 14237 + }, + { + "epoch": 1.5635844498133098, + "grad_norm": 1.963661551475525, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7066690921783447, + "num_tokens": 368183795.0, + "step": 14238 + }, + { + "epoch": 1.5636942675159236, + "grad_norm": 2.258988857269287, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7074441909790039, + "num_tokens": 368209576.0, + "step": 14239 + }, + { + "epoch": 1.5638040852185373, + "grad_norm": 1.99605393409729, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7256472110748291, + "num_tokens": 368240376.0, + "step": 14240 + }, + { + "epoch": 1.5639139029211508, + "grad_norm": 1.9897428750991821, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7223487496376038, + "num_tokens": 368269971.0, + "step": 14241 + }, + { + "epoch": 1.5640237206237646, + "grad_norm": 2.8215279579162598, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7306774258613586, + "num_tokens": 368288264.0, + "step": 14242 + }, + { + "epoch": 1.5641335383263781, + "grad_norm": 2.287698745727539, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7322059273719788, + "num_tokens": 368313174.0, + "step": 14243 + }, + { + "epoch": 1.564243356028992, + "grad_norm": 2.2507128715515137, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7237333655357361, + "num_tokens": 368338117.0, + "step": 14244 + }, + { + "epoch": 1.5643531737316057, + "grad_norm": 2.162123441696167, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.73089998960495, + "num_tokens": 368363618.0, + "step": 14245 + }, + { + "epoch": 1.5644629914342192, + "grad_norm": 2.5955350399017334, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.70185387134552, + "num_tokens": 368384049.0, + "step": 14246 + }, + { + "epoch": 1.5645728091368327, + "grad_norm": 2.591485023498535, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7399830222129822, + "num_tokens": 368404667.0, + "step": 14247 + }, + { + "epoch": 1.5646826268394465, + "grad_norm": 2.3159945011138916, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7228726148605347, + "num_tokens": 368429066.0, + "step": 14248 + }, + { + "epoch": 1.5647924445420602, + "grad_norm": 2.2512216567993164, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7089367508888245, + "num_tokens": 368455380.0, + "step": 14249 + }, + { + "epoch": 1.564902262244674, + "grad_norm": 2.6706502437591553, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7326814532279968, + "num_tokens": 368474136.0, + "step": 14250 + }, + { + "epoch": 1.5650120799472875, + "grad_norm": 1.9363946914672852, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7080080509185791, + "num_tokens": 368505615.0, + "step": 14251 + }, + { + "epoch": 1.565121897649901, + "grad_norm": 2.2890400886535645, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7070909738540649, + "num_tokens": 368530575.0, + "step": 14252 + }, + { + "epoch": 1.5652317153525148, + "grad_norm": 2.557849168777466, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7290067672729492, + "num_tokens": 368550127.0, + "step": 14253 + }, + { + "epoch": 1.5653415330551286, + "grad_norm": 2.526642084121704, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7340183258056641, + "num_tokens": 368570110.0, + "step": 14254 + }, + { + "epoch": 1.565451350757742, + "grad_norm": 2.246572494506836, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7549414038658142, + "num_tokens": 368593773.0, + "step": 14255 + }, + { + "epoch": 1.5655611684603556, + "grad_norm": 2.2173972129821777, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7394055128097534, + "num_tokens": 368620602.0, + "step": 14256 + }, + { + "epoch": 1.5656709861629694, + "grad_norm": 2.3657126426696777, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7288573980331421, + "num_tokens": 368645461.0, + "step": 14257 + }, + { + "epoch": 1.5657808038655832, + "grad_norm": 2.201760768890381, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7061591148376465, + "num_tokens": 368671850.0, + "step": 14258 + }, + { + "epoch": 1.565890621568197, + "grad_norm": 2.2196903228759766, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7281166315078735, + "num_tokens": 368697958.0, + "step": 14259 + }, + { + "epoch": 1.5660004392708105, + "grad_norm": 2.1475281715393066, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.71319580078125, + "num_tokens": 368725058.0, + "step": 14260 + }, + { + "epoch": 1.566110256973424, + "grad_norm": 2.2448391914367676, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7054791450500488, + "num_tokens": 368750045.0, + "step": 14261 + }, + { + "epoch": 1.5662200746760377, + "grad_norm": 2.1855545043945312, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7082347869873047, + "num_tokens": 368773970.0, + "step": 14262 + }, + { + "epoch": 1.5663298923786515, + "grad_norm": 2.318568229675293, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7286611795425415, + "num_tokens": 368797353.0, + "step": 14263 + }, + { + "epoch": 1.5664397100812653, + "grad_norm": 2.172574758529663, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.718738317489624, + "num_tokens": 368822956.0, + "step": 14264 + }, + { + "epoch": 1.5665495277838788, + "grad_norm": 2.605501890182495, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7012662887573242, + "num_tokens": 368844391.0, + "step": 14265 + }, + { + "epoch": 1.5666593454864923, + "grad_norm": 2.11091947555542, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7278562188148499, + "num_tokens": 368871579.0, + "step": 14266 + }, + { + "epoch": 1.566769163189106, + "grad_norm": 2.4397239685058594, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7674991488456726, + "num_tokens": 368892953.0, + "step": 14267 + }, + { + "epoch": 1.5668789808917198, + "grad_norm": 2.343158006668091, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7350571155548096, + "num_tokens": 368915633.0, + "step": 14268 + }, + { + "epoch": 1.5669887985943334, + "grad_norm": 2.4268314838409424, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7501671314239502, + "num_tokens": 368937851.0, + "step": 14269 + }, + { + "epoch": 1.567098616296947, + "grad_norm": 2.3888511657714844, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7302908897399902, + "num_tokens": 368959935.0, + "step": 14270 + }, + { + "epoch": 1.5672084339995607, + "grad_norm": 2.342472791671753, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6968507170677185, + "num_tokens": 368983887.0, + "step": 14271 + }, + { + "epoch": 1.5673182517021744, + "grad_norm": 2.427886724472046, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7097803354263306, + "num_tokens": 369007119.0, + "step": 14272 + }, + { + "epoch": 1.5674280694047882, + "grad_norm": 1.9866863489151, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7166582345962524, + "num_tokens": 369038244.0, + "step": 14273 + }, + { + "epoch": 1.5675378871074017, + "grad_norm": 2.25276780128479, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7208305597305298, + "num_tokens": 369064040.0, + "step": 14274 + }, + { + "epoch": 1.5676477048100153, + "grad_norm": 1.861689805984497, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7253431677818298, + "num_tokens": 369099983.0, + "step": 14275 + }, + { + "epoch": 1.567757522512629, + "grad_norm": 2.3356080055236816, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7195007801055908, + "num_tokens": 369124905.0, + "step": 14276 + }, + { + "epoch": 1.5678673402152428, + "grad_norm": 2.158907890319824, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7080121040344238, + "num_tokens": 369152553.0, + "step": 14277 + }, + { + "epoch": 1.5679771579178565, + "grad_norm": 2.5164988040924072, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7384721636772156, + "num_tokens": 369173755.0, + "step": 14278 + }, + { + "epoch": 1.56808697562047, + "grad_norm": 1.9282833337783813, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7283409833908081, + "num_tokens": 369204835.0, + "step": 14279 + }, + { + "epoch": 1.5681967933230836, + "grad_norm": 2.3203418254852295, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7032953500747681, + "num_tokens": 369229517.0, + "step": 14280 + }, + { + "epoch": 1.5683066110256974, + "grad_norm": 2.3344314098358154, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7093291282653809, + "num_tokens": 369251994.0, + "step": 14281 + }, + { + "epoch": 1.568416428728311, + "grad_norm": 2.1796815395355225, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7091613411903381, + "num_tokens": 369278656.0, + "step": 14282 + }, + { + "epoch": 1.5685262464309246, + "grad_norm": 2.2328097820281982, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7125499248504639, + "num_tokens": 369307060.0, + "step": 14283 + }, + { + "epoch": 1.5686360641335382, + "grad_norm": 2.0803134441375732, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7092305421829224, + "num_tokens": 369334937.0, + "step": 14284 + }, + { + "epoch": 1.568745881836152, + "grad_norm": 2.414698839187622, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7189540266990662, + "num_tokens": 369359214.0, + "step": 14285 + }, + { + "epoch": 1.5688556995387657, + "grad_norm": 2.1459505558013916, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7302930951118469, + "num_tokens": 369387553.0, + "step": 14286 + }, + { + "epoch": 1.5689655172413794, + "grad_norm": 2.2675485610961914, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7252365350723267, + "num_tokens": 369411021.0, + "step": 14287 + }, + { + "epoch": 1.569075334943993, + "grad_norm": 2.4153032302856445, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7528600692749023, + "num_tokens": 369433815.0, + "step": 14288 + }, + { + "epoch": 1.5691851526466065, + "grad_norm": 2.3868019580841064, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7226734161376953, + "num_tokens": 369456333.0, + "step": 14289 + }, + { + "epoch": 1.5692949703492203, + "grad_norm": 2.330049753189087, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7214934825897217, + "num_tokens": 369479927.0, + "step": 14290 + }, + { + "epoch": 1.569404788051834, + "grad_norm": 2.097834587097168, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7112607955932617, + "num_tokens": 369509725.0, + "step": 14291 + }, + { + "epoch": 1.5695146057544476, + "grad_norm": 2.272347927093506, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7268958687782288, + "num_tokens": 369533129.0, + "step": 14292 + }, + { + "epoch": 1.5696244234570613, + "grad_norm": 1.8859847784042358, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.706554651260376, + "num_tokens": 369566005.0, + "step": 14293 + }, + { + "epoch": 1.5697342411596749, + "grad_norm": 2.202153444290161, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7070716619491577, + "num_tokens": 369594485.0, + "step": 14294 + }, + { + "epoch": 1.5698440588622886, + "grad_norm": 2.540045738220215, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7264614105224609, + "num_tokens": 369616813.0, + "step": 14295 + }, + { + "epoch": 1.5699538765649024, + "grad_norm": 2.2983591556549072, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7119817733764648, + "num_tokens": 369644035.0, + "step": 14296 + }, + { + "epoch": 1.570063694267516, + "grad_norm": 2.234442949295044, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7167250514030457, + "num_tokens": 369671675.0, + "step": 14297 + }, + { + "epoch": 1.5701735119701294, + "grad_norm": 2.213898181915283, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.738179087638855, + "num_tokens": 369697141.0, + "step": 14298 + }, + { + "epoch": 1.5702833296727432, + "grad_norm": 2.164818525314331, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7214071154594421, + "num_tokens": 369723377.0, + "step": 14299 + }, + { + "epoch": 1.570393147375357, + "grad_norm": 2.466066360473633, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7163573503494263, + "num_tokens": 369746629.0, + "step": 14300 + }, + { + "epoch": 1.5705029650779707, + "grad_norm": 2.207160234451294, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7251057624816895, + "num_tokens": 369772301.0, + "step": 14301 + }, + { + "epoch": 1.5706127827805842, + "grad_norm": 2.436234712600708, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7211781144142151, + "num_tokens": 369794383.0, + "step": 14302 + }, + { + "epoch": 1.5707226004831978, + "grad_norm": 2.301424026489258, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7238737344741821, + "num_tokens": 369820624.0, + "step": 14303 + }, + { + "epoch": 1.5708324181858115, + "grad_norm": 2.2052199840545654, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.710229754447937, + "num_tokens": 369847660.0, + "step": 14304 + }, + { + "epoch": 1.5709422358884253, + "grad_norm": 2.4196598529815674, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7274006605148315, + "num_tokens": 369869638.0, + "step": 14305 + }, + { + "epoch": 1.5710520535910388, + "grad_norm": 2.1163065433502197, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7467454671859741, + "num_tokens": 369894309.0, + "step": 14306 + }, + { + "epoch": 1.5711618712936526, + "grad_norm": 2.2217748165130615, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7233068346977234, + "num_tokens": 369921955.0, + "step": 14307 + }, + { + "epoch": 1.5712716889962661, + "grad_norm": 2.211322069168091, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7170709371566772, + "num_tokens": 369947219.0, + "step": 14308 + }, + { + "epoch": 1.5713815066988799, + "grad_norm": 2.2577097415924072, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7467820644378662, + "num_tokens": 369972667.0, + "step": 14309 + }, + { + "epoch": 1.5714913244014936, + "grad_norm": 2.2888967990875244, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7279411554336548, + "num_tokens": 369998788.0, + "step": 14310 + }, + { + "epoch": 1.5716011421041072, + "grad_norm": 1.9357811212539673, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7568474411964417, + "num_tokens": 370029169.0, + "step": 14311 + }, + { + "epoch": 1.5717109598067207, + "grad_norm": 2.3057243824005127, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7198207378387451, + "num_tokens": 370053808.0, + "step": 14312 + }, + { + "epoch": 1.5718207775093345, + "grad_norm": 2.2898688316345215, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7209616899490356, + "num_tokens": 370078890.0, + "step": 14313 + }, + { + "epoch": 1.5719305952119482, + "grad_norm": 2.334036350250244, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.685194730758667, + "num_tokens": 370104070.0, + "step": 14314 + }, + { + "epoch": 1.572040412914562, + "grad_norm": 2.2696175575256348, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7285006046295166, + "num_tokens": 370128170.0, + "step": 14315 + }, + { + "epoch": 1.5721502306171755, + "grad_norm": 1.9695215225219727, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7204299569129944, + "num_tokens": 370159853.0, + "step": 14316 + }, + { + "epoch": 1.572260048319789, + "grad_norm": 2.059678554534912, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7254421710968018, + "num_tokens": 370188294.0, + "step": 14317 + }, + { + "epoch": 1.5723698660224028, + "grad_norm": 2.2743704319000244, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6982462406158447, + "num_tokens": 370213186.0, + "step": 14318 + }, + { + "epoch": 1.5724796837250166, + "grad_norm": 2.018808603286743, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7261383533477783, + "num_tokens": 370242167.0, + "step": 14319 + }, + { + "epoch": 1.57258950142763, + "grad_norm": 1.9113619327545166, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.722459077835083, + "num_tokens": 370275462.0, + "step": 14320 + }, + { + "epoch": 1.5726993191302436, + "grad_norm": 2.144275665283203, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7209742069244385, + "num_tokens": 370301238.0, + "step": 14321 + }, + { + "epoch": 1.5728091368328574, + "grad_norm": 2.388030529022217, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7269577383995056, + "num_tokens": 370322420.0, + "step": 14322 + }, + { + "epoch": 1.5729189545354711, + "grad_norm": 2.2331812381744385, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6914195418357849, + "num_tokens": 370350201.0, + "step": 14323 + }, + { + "epoch": 1.573028772238085, + "grad_norm": 2.154892683029175, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.6990065574645996, + "num_tokens": 370379132.0, + "step": 14324 + }, + { + "epoch": 1.5731385899406984, + "grad_norm": 2.3103678226470947, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6989752054214478, + "num_tokens": 370403156.0, + "step": 14325 + }, + { + "epoch": 1.573248407643312, + "grad_norm": 2.341784954071045, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.725951611995697, + "num_tokens": 370425779.0, + "step": 14326 + }, + { + "epoch": 1.5733582253459257, + "grad_norm": 2.0957424640655518, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7216210961341858, + "num_tokens": 370456410.0, + "step": 14327 + }, + { + "epoch": 1.5734680430485395, + "grad_norm": 2.1622283458709717, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7346062064170837, + "num_tokens": 370482807.0, + "step": 14328 + }, + { + "epoch": 1.5735778607511532, + "grad_norm": 2.3175973892211914, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7236414551734924, + "num_tokens": 370504454.0, + "step": 14329 + }, + { + "epoch": 1.5736876784537668, + "grad_norm": 2.116706609725952, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7071443796157837, + "num_tokens": 370531949.0, + "step": 14330 + }, + { + "epoch": 1.5737974961563803, + "grad_norm": 2.184112548828125, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7329318523406982, + "num_tokens": 370557677.0, + "step": 14331 + }, + { + "epoch": 1.573907313858994, + "grad_norm": 2.2947099208831787, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6986714005470276, + "num_tokens": 370584060.0, + "step": 14332 + }, + { + "epoch": 1.5740171315616078, + "grad_norm": 2.1796796321868896, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7282053828239441, + "num_tokens": 370608927.0, + "step": 14333 + }, + { + "epoch": 1.5741269492642214, + "grad_norm": 2.2529494762420654, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7226278781890869, + "num_tokens": 370634430.0, + "step": 14334 + }, + { + "epoch": 1.574236766966835, + "grad_norm": 2.3974497318267822, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7299834489822388, + "num_tokens": 370655568.0, + "step": 14335 + }, + { + "epoch": 1.5743465846694487, + "grad_norm": 2.241312265396118, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7140501141548157, + "num_tokens": 370681602.0, + "step": 14336 + }, + { + "epoch": 1.5744564023720624, + "grad_norm": 1.9230105876922607, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7194482088088989, + "num_tokens": 370713455.0, + "step": 14337 + }, + { + "epoch": 1.5745662200746762, + "grad_norm": 2.286515712738037, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.716534435749054, + "num_tokens": 370739938.0, + "step": 14338 + }, + { + "epoch": 1.5746760377772897, + "grad_norm": 2.3320939540863037, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7183676958084106, + "num_tokens": 370763405.0, + "step": 14339 + }, + { + "epoch": 1.5747858554799032, + "grad_norm": 2.0275895595550537, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6879147887229919, + "num_tokens": 370795013.0, + "step": 14340 + }, + { + "epoch": 1.574895673182517, + "grad_norm": 2.0503756999969482, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.716810405254364, + "num_tokens": 370821450.0, + "step": 14341 + }, + { + "epoch": 1.5750054908851308, + "grad_norm": 2.298063039779663, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7213439345359802, + "num_tokens": 370846368.0, + "step": 14342 + }, + { + "epoch": 1.5751153085877443, + "grad_norm": 2.236104965209961, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7275073528289795, + "num_tokens": 370872383.0, + "step": 14343 + }, + { + "epoch": 1.575225126290358, + "grad_norm": 2.2066800594329834, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7028670310974121, + "num_tokens": 370901133.0, + "step": 14344 + }, + { + "epoch": 1.5753349439929716, + "grad_norm": 2.343289613723755, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7001556158065796, + "num_tokens": 370926893.0, + "step": 14345 + }, + { + "epoch": 1.5754447616955853, + "grad_norm": 2.2322397232055664, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.726111650466919, + "num_tokens": 370953501.0, + "step": 14346 + }, + { + "epoch": 1.575554579398199, + "grad_norm": 2.017108917236328, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7298605442047119, + "num_tokens": 370983747.0, + "step": 14347 + }, + { + "epoch": 1.5756643971008126, + "grad_norm": 2.0442428588867188, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.717440664768219, + "num_tokens": 371012900.0, + "step": 14348 + }, + { + "epoch": 1.5757742148034262, + "grad_norm": 2.307626485824585, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7174904346466064, + "num_tokens": 371037220.0, + "step": 14349 + }, + { + "epoch": 1.57588403250604, + "grad_norm": 2.4606735706329346, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7146629095077515, + "num_tokens": 371059267.0, + "step": 14350 + }, + { + "epoch": 1.5759938502086537, + "grad_norm": 2.0726547241210938, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6875900626182556, + "num_tokens": 371090398.0, + "step": 14351 + }, + { + "epoch": 1.5761036679112674, + "grad_norm": 2.1464614868164062, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7022465467453003, + "num_tokens": 371119403.0, + "step": 14352 + }, + { + "epoch": 1.576213485613881, + "grad_norm": 1.9064089059829712, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7094965577125549, + "num_tokens": 371154075.0, + "step": 14353 + }, + { + "epoch": 1.5763233033164945, + "grad_norm": 2.448427677154541, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7224774956703186, + "num_tokens": 371176912.0, + "step": 14354 + }, + { + "epoch": 1.5764331210191083, + "grad_norm": 2.140293836593628, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7133153676986694, + "num_tokens": 371202272.0, + "step": 14355 + }, + { + "epoch": 1.576542938721722, + "grad_norm": 2.2652294635772705, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7209780216217041, + "num_tokens": 371227268.0, + "step": 14356 + }, + { + "epoch": 1.5766527564243356, + "grad_norm": 2.385592222213745, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.73316890001297, + "num_tokens": 371252192.0, + "step": 14357 + }, + { + "epoch": 1.5767625741269493, + "grad_norm": 2.3874645233154297, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7233394384384155, + "num_tokens": 371274291.0, + "step": 14358 + }, + { + "epoch": 1.5768723918295628, + "grad_norm": 2.056001901626587, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6995949745178223, + "num_tokens": 371304575.0, + "step": 14359 + }, + { + "epoch": 1.5769822095321766, + "grad_norm": 2.0824637413024902, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7306288480758667, + "num_tokens": 371333681.0, + "step": 14360 + }, + { + "epoch": 1.5770920272347904, + "grad_norm": 2.022869825363159, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7135341167449951, + "num_tokens": 371367450.0, + "step": 14361 + }, + { + "epoch": 1.577201844937404, + "grad_norm": 2.182114839553833, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7148797512054443, + "num_tokens": 371393930.0, + "step": 14362 + }, + { + "epoch": 1.5773116626400174, + "grad_norm": 2.4170572757720947, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.71962970495224, + "num_tokens": 371415346.0, + "step": 14363 + }, + { + "epoch": 1.5774214803426312, + "grad_norm": 2.438838481903076, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7077694535255432, + "num_tokens": 371436946.0, + "step": 14364 + }, + { + "epoch": 1.577531298045245, + "grad_norm": 2.1109743118286133, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7121882438659668, + "num_tokens": 371466547.0, + "step": 14365 + }, + { + "epoch": 1.5776411157478587, + "grad_norm": 1.9953982830047607, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7275938391685486, + "num_tokens": 371498664.0, + "step": 14366 + }, + { + "epoch": 1.5777509334504722, + "grad_norm": 2.0808279514312744, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7271842956542969, + "num_tokens": 371526832.0, + "step": 14367 + }, + { + "epoch": 1.5778607511530858, + "grad_norm": 2.254457950592041, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7231982946395874, + "num_tokens": 371556697.0, + "step": 14368 + }, + { + "epoch": 1.5779705688556995, + "grad_norm": 1.907649278640747, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.704020619392395, + "num_tokens": 371587747.0, + "step": 14369 + }, + { + "epoch": 1.5780803865583133, + "grad_norm": 2.3090109825134277, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.716720461845398, + "num_tokens": 371611011.0, + "step": 14370 + }, + { + "epoch": 1.5781902042609268, + "grad_norm": 2.2115976810455322, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7158981561660767, + "num_tokens": 371636310.0, + "step": 14371 + }, + { + "epoch": 1.5783000219635406, + "grad_norm": 2.1109790802001953, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7126355767250061, + "num_tokens": 371664235.0, + "step": 14372 + }, + { + "epoch": 1.5784098396661541, + "grad_norm": 2.3145911693573, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7145429849624634, + "num_tokens": 371689834.0, + "step": 14373 + }, + { + "epoch": 1.5785196573687679, + "grad_norm": 1.899943470954895, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7072948217391968, + "num_tokens": 371725226.0, + "step": 14374 + }, + { + "epoch": 1.5786294750713816, + "grad_norm": 2.402118682861328, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7411606907844543, + "num_tokens": 371748342.0, + "step": 14375 + }, + { + "epoch": 1.5787392927739952, + "grad_norm": 2.2475526332855225, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.721105694770813, + "num_tokens": 371771183.0, + "step": 14376 + }, + { + "epoch": 1.5788491104766087, + "grad_norm": 2.3035123348236084, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7160907983779907, + "num_tokens": 371797106.0, + "step": 14377 + }, + { + "epoch": 1.5789589281792225, + "grad_norm": 2.2224652767181396, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7040261030197144, + "num_tokens": 371824608.0, + "step": 14378 + }, + { + "epoch": 1.5790687458818362, + "grad_norm": 2.3117434978485107, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7159287929534912, + "num_tokens": 371849493.0, + "step": 14379 + }, + { + "epoch": 1.57917856358445, + "grad_norm": 2.0372421741485596, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7262799739837646, + "num_tokens": 371877672.0, + "step": 14380 + }, + { + "epoch": 1.5792883812870635, + "grad_norm": 2.104555606842041, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6832167506217957, + "num_tokens": 371909281.0, + "step": 14381 + }, + { + "epoch": 1.579398198989677, + "grad_norm": 2.397521734237671, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7125275135040283, + "num_tokens": 371932126.0, + "step": 14382 + }, + { + "epoch": 1.5795080166922908, + "grad_norm": 2.545771598815918, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7422116994857788, + "num_tokens": 371952465.0, + "step": 14383 + }, + { + "epoch": 1.5796178343949046, + "grad_norm": 2.798837661743164, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7388643026351929, + "num_tokens": 371971853.0, + "step": 14384 + }, + { + "epoch": 1.579727652097518, + "grad_norm": 2.2003939151763916, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6965895295143127, + "num_tokens": 371999889.0, + "step": 14385 + }, + { + "epoch": 1.5798374698001316, + "grad_norm": 2.0032083988189697, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7032876014709473, + "num_tokens": 372031413.0, + "step": 14386 + }, + { + "epoch": 1.5799472875027454, + "grad_norm": 2.2366113662719727, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7299497127532959, + "num_tokens": 372055163.0, + "step": 14387 + }, + { + "epoch": 1.5800571052053591, + "grad_norm": 2.201009511947632, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7082661390304565, + "num_tokens": 372081104.0, + "step": 14388 + }, + { + "epoch": 1.580166922907973, + "grad_norm": 2.383101463317871, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7287085056304932, + "num_tokens": 372105152.0, + "step": 14389 + }, + { + "epoch": 1.5802767406105864, + "grad_norm": 2.132789134979248, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7148385643959045, + "num_tokens": 372134673.0, + "step": 14390 + }, + { + "epoch": 1.5803865583132, + "grad_norm": 2.613227367401123, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7216265201568604, + "num_tokens": 372157655.0, + "step": 14391 + }, + { + "epoch": 1.5804963760158137, + "grad_norm": 2.15598726272583, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7330442667007446, + "num_tokens": 372181746.0, + "step": 14392 + }, + { + "epoch": 1.5806061937184275, + "grad_norm": 2.215860366821289, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7057293057441711, + "num_tokens": 372209356.0, + "step": 14393 + }, + { + "epoch": 1.5807160114210412, + "grad_norm": 2.380333423614502, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.697172999382019, + "num_tokens": 372233574.0, + "step": 14394 + }, + { + "epoch": 1.5808258291236548, + "grad_norm": 2.2543063163757324, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7047399878501892, + "num_tokens": 372261343.0, + "step": 14395 + }, + { + "epoch": 1.5809356468262683, + "grad_norm": 2.3905029296875, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7165210247039795, + "num_tokens": 372284609.0, + "step": 14396 + }, + { + "epoch": 1.581045464528882, + "grad_norm": 2.264728307723999, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7249155044555664, + "num_tokens": 372309407.0, + "step": 14397 + }, + { + "epoch": 1.5811552822314958, + "grad_norm": 2.368722438812256, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7310059666633606, + "num_tokens": 372330438.0, + "step": 14398 + }, + { + "epoch": 1.5812650999341094, + "grad_norm": 2.2901837825775146, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7253296375274658, + "num_tokens": 372358343.0, + "step": 14399 + }, + { + "epoch": 1.5813749176367229, + "grad_norm": 2.534400463104248, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7447637319564819, + "num_tokens": 372377730.0, + "step": 14400 + }, + { + "epoch": 1.5814847353393366, + "grad_norm": 1.9923224449157715, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7072310447692871, + "num_tokens": 372410573.0, + "step": 14401 + }, + { + "epoch": 1.5815945530419504, + "grad_norm": 2.119609832763672, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7011798620223999, + "num_tokens": 372438940.0, + "step": 14402 + }, + { + "epoch": 1.5817043707445642, + "grad_norm": 2.322847366333008, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.709223747253418, + "num_tokens": 372462985.0, + "step": 14403 + }, + { + "epoch": 1.5818141884471777, + "grad_norm": 2.3086321353912354, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.716250479221344, + "num_tokens": 372488502.0, + "step": 14404 + }, + { + "epoch": 1.5819240061497912, + "grad_norm": 2.2719180583953857, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7115927934646606, + "num_tokens": 372514893.0, + "step": 14405 + }, + { + "epoch": 1.582033823852405, + "grad_norm": 2.375108003616333, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7148791551589966, + "num_tokens": 372538957.0, + "step": 14406 + }, + { + "epoch": 1.5821436415550187, + "grad_norm": 2.0539276599884033, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7046793699264526, + "num_tokens": 372568819.0, + "step": 14407 + }, + { + "epoch": 1.5822534592576323, + "grad_norm": 2.1646413803100586, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6960345506668091, + "num_tokens": 372599091.0, + "step": 14408 + }, + { + "epoch": 1.582363276960246, + "grad_norm": 2.153000593185425, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7400462031364441, + "num_tokens": 372627719.0, + "step": 14409 + }, + { + "epoch": 1.5824730946628596, + "grad_norm": 2.4749605655670166, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7060266137123108, + "num_tokens": 372650903.0, + "step": 14410 + }, + { + "epoch": 1.5825829123654733, + "grad_norm": 2.3092281818389893, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7150968909263611, + "num_tokens": 372678165.0, + "step": 14411 + }, + { + "epoch": 1.582692730068087, + "grad_norm": 1.9050629138946533, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.704012393951416, + "num_tokens": 372713150.0, + "step": 14412 + }, + { + "epoch": 1.5828025477707006, + "grad_norm": 2.1929664611816406, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7115877866744995, + "num_tokens": 372739846.0, + "step": 14413 + }, + { + "epoch": 1.5829123654733142, + "grad_norm": 2.249824047088623, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7101092338562012, + "num_tokens": 372765815.0, + "step": 14414 + }, + { + "epoch": 1.583022183175928, + "grad_norm": 2.0164339542388916, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.6996772289276123, + "num_tokens": 372797191.0, + "step": 14415 + }, + { + "epoch": 1.5831320008785417, + "grad_norm": 2.1340553760528564, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7275277972221375, + "num_tokens": 372824195.0, + "step": 14416 + }, + { + "epoch": 1.5832418185811554, + "grad_norm": 2.410247564315796, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.716738760471344, + "num_tokens": 372847841.0, + "step": 14417 + }, + { + "epoch": 1.583351636283769, + "grad_norm": 2.526644229888916, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7272984981536865, + "num_tokens": 372867002.0, + "step": 14418 + }, + { + "epoch": 1.5834614539863825, + "grad_norm": 2.2793126106262207, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7150638103485107, + "num_tokens": 372892138.0, + "step": 14419 + }, + { + "epoch": 1.5835712716889963, + "grad_norm": 2.132995843887329, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7163142561912537, + "num_tokens": 372917333.0, + "step": 14420 + }, + { + "epoch": 1.58368108939161, + "grad_norm": 2.258906602859497, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7280632257461548, + "num_tokens": 372941779.0, + "step": 14421 + }, + { + "epoch": 1.5837909070942235, + "grad_norm": 2.4397928714752197, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7190341949462891, + "num_tokens": 372962844.0, + "step": 14422 + }, + { + "epoch": 1.5839007247968373, + "grad_norm": 2.171875, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7072172164916992, + "num_tokens": 372989413.0, + "step": 14423 + }, + { + "epoch": 1.5840105424994508, + "grad_norm": 2.345254421234131, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7229636907577515, + "num_tokens": 373014865.0, + "step": 14424 + }, + { + "epoch": 1.5841203602020646, + "grad_norm": 2.1568663120269775, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7038846611976624, + "num_tokens": 373043360.0, + "step": 14425 + }, + { + "epoch": 1.5842301779046783, + "grad_norm": 2.0866072177886963, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7129014134407043, + "num_tokens": 373071402.0, + "step": 14426 + }, + { + "epoch": 1.5843399956072919, + "grad_norm": 2.4993743896484375, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7166039943695068, + "num_tokens": 373092339.0, + "step": 14427 + }, + { + "epoch": 1.5844498133099054, + "grad_norm": 2.449343681335449, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7147229909896851, + "num_tokens": 373113931.0, + "step": 14428 + }, + { + "epoch": 1.5845596310125192, + "grad_norm": 2.314602851867676, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7338860034942627, + "num_tokens": 373137454.0, + "step": 14429 + }, + { + "epoch": 1.584669448715133, + "grad_norm": 2.1650121212005615, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7016690969467163, + "num_tokens": 373163741.0, + "step": 14430 + }, + { + "epoch": 1.5847792664177467, + "grad_norm": 2.1617305278778076, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7128362655639648, + "num_tokens": 373191778.0, + "step": 14431 + }, + { + "epoch": 1.5848890841203602, + "grad_norm": 2.3411059379577637, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7100170850753784, + "num_tokens": 373215503.0, + "step": 14432 + }, + { + "epoch": 1.5849989018229738, + "grad_norm": 2.3651344776153564, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7213497161865234, + "num_tokens": 373239727.0, + "step": 14433 + }, + { + "epoch": 1.5851087195255875, + "grad_norm": 2.1795966625213623, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6905235648155212, + "num_tokens": 373268032.0, + "step": 14434 + }, + { + "epoch": 1.5852185372282013, + "grad_norm": 2.167205810546875, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7146327495574951, + "num_tokens": 373295587.0, + "step": 14435 + }, + { + "epoch": 1.5853283549308148, + "grad_norm": 2.529007911682129, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7132964730262756, + "num_tokens": 373316872.0, + "step": 14436 + }, + { + "epoch": 1.5854381726334283, + "grad_norm": 2.4287824630737305, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7379419803619385, + "num_tokens": 373337786.0, + "step": 14437 + }, + { + "epoch": 1.585547990336042, + "grad_norm": 2.482570171356201, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7191416621208191, + "num_tokens": 373359915.0, + "step": 14438 + }, + { + "epoch": 1.5856578080386559, + "grad_norm": 2.3139379024505615, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7204281687736511, + "num_tokens": 373384787.0, + "step": 14439 + }, + { + "epoch": 1.5857676257412696, + "grad_norm": 2.134382724761963, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7246061563491821, + "num_tokens": 373414612.0, + "step": 14440 + }, + { + "epoch": 1.5858774434438832, + "grad_norm": 2.3160340785980225, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7213735580444336, + "num_tokens": 373439049.0, + "step": 14441 + }, + { + "epoch": 1.5859872611464967, + "grad_norm": 1.9813761711120605, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6904758214950562, + "num_tokens": 373474120.0, + "step": 14442 + }, + { + "epoch": 1.5860970788491104, + "grad_norm": 2.3731231689453125, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7411564588546753, + "num_tokens": 373497659.0, + "step": 14443 + }, + { + "epoch": 1.5862068965517242, + "grad_norm": 2.2157702445983887, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7378750443458557, + "num_tokens": 373522404.0, + "step": 14444 + }, + { + "epoch": 1.586316714254338, + "grad_norm": 2.1544313430786133, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7348600029945374, + "num_tokens": 373549575.0, + "step": 14445 + }, + { + "epoch": 1.5864265319569515, + "grad_norm": 2.3923351764678955, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7415788173675537, + "num_tokens": 373572238.0, + "step": 14446 + }, + { + "epoch": 1.586536349659565, + "grad_norm": 2.360804796218872, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7189269065856934, + "num_tokens": 373597048.0, + "step": 14447 + }, + { + "epoch": 1.5866461673621788, + "grad_norm": 2.3273415565490723, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.720028817653656, + "num_tokens": 373620945.0, + "step": 14448 + }, + { + "epoch": 1.5867559850647925, + "grad_norm": 2.409139394760132, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7156404256820679, + "num_tokens": 373644170.0, + "step": 14449 + }, + { + "epoch": 1.586865802767406, + "grad_norm": 2.4983272552490234, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7158215641975403, + "num_tokens": 373665553.0, + "step": 14450 + }, + { + "epoch": 1.5869756204700196, + "grad_norm": 2.2335755825042725, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7117190957069397, + "num_tokens": 373693133.0, + "step": 14451 + }, + { + "epoch": 1.5870854381726334, + "grad_norm": 2.314038038253784, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7188578844070435, + "num_tokens": 373716781.0, + "step": 14452 + }, + { + "epoch": 1.5871952558752471, + "grad_norm": 2.3728461265563965, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7232523560523987, + "num_tokens": 373742232.0, + "step": 14453 + }, + { + "epoch": 1.5873050735778609, + "grad_norm": 2.212169647216797, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7164844870567322, + "num_tokens": 373768072.0, + "step": 14454 + }, + { + "epoch": 1.5874148912804744, + "grad_norm": 2.170907497406006, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7244417667388916, + "num_tokens": 373794979.0, + "step": 14455 + }, + { + "epoch": 1.587524708983088, + "grad_norm": 2.364154577255249, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7271456122398376, + "num_tokens": 373817546.0, + "step": 14456 + }, + { + "epoch": 1.5876345266857017, + "grad_norm": 2.1533384323120117, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.72010338306427, + "num_tokens": 373845958.0, + "step": 14457 + }, + { + "epoch": 1.5877443443883155, + "grad_norm": 2.402853012084961, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7314044237136841, + "num_tokens": 373869453.0, + "step": 14458 + }, + { + "epoch": 1.5878541620909292, + "grad_norm": 2.248581647872925, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7175816297531128, + "num_tokens": 373894136.0, + "step": 14459 + }, + { + "epoch": 1.5879639797935428, + "grad_norm": 2.549194097518921, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7156447172164917, + "num_tokens": 373914939.0, + "step": 14460 + }, + { + "epoch": 1.5880737974961563, + "grad_norm": 2.1507246494293213, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6930721998214722, + "num_tokens": 373942048.0, + "step": 14461 + }, + { + "epoch": 1.58818361519877, + "grad_norm": 2.146953582763672, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7195366621017456, + "num_tokens": 373971014.0, + "step": 14462 + }, + { + "epoch": 1.5882934329013838, + "grad_norm": 2.25783109664917, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7607859373092651, + "num_tokens": 373994789.0, + "step": 14463 + }, + { + "epoch": 1.5884032506039973, + "grad_norm": 2.3367760181427, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7276324033737183, + "num_tokens": 374016747.0, + "step": 14464 + }, + { + "epoch": 1.5885130683066109, + "grad_norm": 2.470568895339966, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7297227382659912, + "num_tokens": 374039477.0, + "step": 14465 + }, + { + "epoch": 1.5886228860092246, + "grad_norm": 2.1103901863098145, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.709916353225708, + "num_tokens": 374068549.0, + "step": 14466 + }, + { + "epoch": 1.5887327037118384, + "grad_norm": 2.5252676010131836, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.7339901924133301, + "num_tokens": 374090115.0, + "step": 14467 + }, + { + "epoch": 1.5888425214144521, + "grad_norm": 2.2955243587493896, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7104017734527588, + "num_tokens": 374114433.0, + "step": 14468 + }, + { + "epoch": 1.5889523391170657, + "grad_norm": 2.142235279083252, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7137055397033691, + "num_tokens": 374144386.0, + "step": 14469 + }, + { + "epoch": 1.5890621568196792, + "grad_norm": 2.390531063079834, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.733113169670105, + "num_tokens": 374166726.0, + "step": 14470 + }, + { + "epoch": 1.589171974522293, + "grad_norm": 2.0380966663360596, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7152379155158997, + "num_tokens": 374197369.0, + "step": 14471 + }, + { + "epoch": 1.5892817922249067, + "grad_norm": 2.3625409603118896, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7135156393051147, + "num_tokens": 374220021.0, + "step": 14472 + }, + { + "epoch": 1.5893916099275203, + "grad_norm": 2.021885633468628, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7053300738334656, + "num_tokens": 374252238.0, + "step": 14473 + }, + { + "epoch": 1.589501427630134, + "grad_norm": 2.174670934677124, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7173839807510376, + "num_tokens": 374279155.0, + "step": 14474 + }, + { + "epoch": 1.5896112453327476, + "grad_norm": 2.1015748977661133, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7190448045730591, + "num_tokens": 374308634.0, + "step": 14475 + }, + { + "epoch": 1.5897210630353613, + "grad_norm": 2.429218053817749, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7202197313308716, + "num_tokens": 374331359.0, + "step": 14476 + }, + { + "epoch": 1.589830880737975, + "grad_norm": 2.5068979263305664, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.6964705586433411, + "num_tokens": 374355995.0, + "step": 14477 + }, + { + "epoch": 1.5899406984405886, + "grad_norm": 2.414407253265381, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7214559316635132, + "num_tokens": 374377534.0, + "step": 14478 + }, + { + "epoch": 1.5900505161432021, + "grad_norm": 2.226827621459961, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7246683835983276, + "num_tokens": 374403849.0, + "step": 14479 + }, + { + "epoch": 1.590160333845816, + "grad_norm": 2.2218077182769775, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7375674843788147, + "num_tokens": 374429684.0, + "step": 14480 + }, + { + "epoch": 1.5902701515484297, + "grad_norm": 2.185359001159668, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7280474305152893, + "num_tokens": 374455969.0, + "step": 14481 + }, + { + "epoch": 1.5903799692510434, + "grad_norm": 2.2686123847961426, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7217820286750793, + "num_tokens": 374481316.0, + "step": 14482 + }, + { + "epoch": 1.590489786953657, + "grad_norm": 2.1960573196411133, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7057260274887085, + "num_tokens": 374505819.0, + "step": 14483 + }, + { + "epoch": 1.5905996046562705, + "grad_norm": 2.1348066329956055, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7551620602607727, + "num_tokens": 374529943.0, + "step": 14484 + }, + { + "epoch": 1.5907094223588842, + "grad_norm": 2.5906529426574707, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.727950930595398, + "num_tokens": 374551720.0, + "step": 14485 + }, + { + "epoch": 1.590819240061498, + "grad_norm": 2.0810441970825195, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7118411064147949, + "num_tokens": 374580820.0, + "step": 14486 + }, + { + "epoch": 1.5909290577641115, + "grad_norm": 2.4595680236816406, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7272729873657227, + "num_tokens": 374603548.0, + "step": 14487 + }, + { + "epoch": 1.5910388754667253, + "grad_norm": 2.2385401725769043, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7110493183135986, + "num_tokens": 374629400.0, + "step": 14488 + }, + { + "epoch": 1.5911486931693388, + "grad_norm": 2.3729355335235596, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7177139520645142, + "num_tokens": 374651432.0, + "step": 14489 + }, + { + "epoch": 1.5912585108719526, + "grad_norm": 1.934103012084961, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7133416533470154, + "num_tokens": 374684310.0, + "step": 14490 + }, + { + "epoch": 1.5913683285745663, + "grad_norm": 2.2598142623901367, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7073477506637573, + "num_tokens": 374709123.0, + "step": 14491 + }, + { + "epoch": 1.5914781462771799, + "grad_norm": 2.1989378929138184, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7191143035888672, + "num_tokens": 374737090.0, + "step": 14492 + }, + { + "epoch": 1.5915879639797934, + "grad_norm": 2.2254011631011963, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7194482088088989, + "num_tokens": 374762696.0, + "step": 14493 + }, + { + "epoch": 1.5916977816824072, + "grad_norm": 2.388033866882324, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7253412008285522, + "num_tokens": 374787539.0, + "step": 14494 + }, + { + "epoch": 1.591807599385021, + "grad_norm": 2.083552360534668, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7101601958274841, + "num_tokens": 374815548.0, + "step": 14495 + }, + { + "epoch": 1.5919174170876347, + "grad_norm": 2.1956634521484375, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7077405452728271, + "num_tokens": 374842855.0, + "step": 14496 + }, + { + "epoch": 1.5920272347902482, + "grad_norm": 1.953529953956604, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7122284173965454, + "num_tokens": 374874927.0, + "step": 14497 + }, + { + "epoch": 1.5921370524928617, + "grad_norm": 2.4839913845062256, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7434912919998169, + "num_tokens": 374895034.0, + "step": 14498 + }, + { + "epoch": 1.5922468701954755, + "grad_norm": 2.3181066513061523, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.710503101348877, + "num_tokens": 374919496.0, + "step": 14499 + }, + { + "epoch": 1.5923566878980893, + "grad_norm": 2.364919424057007, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7140012979507446, + "num_tokens": 374943244.0, + "step": 14500 + }, + { + "epoch": 1.5924665056007028, + "grad_norm": 2.161707878112793, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7140288949012756, + "num_tokens": 374971146.0, + "step": 14501 + }, + { + "epoch": 1.5925763233033163, + "grad_norm": 2.1255297660827637, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.725425124168396, + "num_tokens": 374997752.0, + "step": 14502 + }, + { + "epoch": 1.59268614100593, + "grad_norm": 2.2017886638641357, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7046661972999573, + "num_tokens": 375025536.0, + "step": 14503 + }, + { + "epoch": 1.5927959587085438, + "grad_norm": 2.3684959411621094, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7243894338607788, + "num_tokens": 375049295.0, + "step": 14504 + }, + { + "epoch": 1.5929057764111576, + "grad_norm": 2.2282474040985107, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7344253659248352, + "num_tokens": 375073356.0, + "step": 14505 + }, + { + "epoch": 1.5930155941137711, + "grad_norm": 2.320596933364868, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7342986464500427, + "num_tokens": 375097520.0, + "step": 14506 + }, + { + "epoch": 1.5931254118163847, + "grad_norm": 2.129913568496704, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7185670733451843, + "num_tokens": 375124237.0, + "step": 14507 + }, + { + "epoch": 1.5932352295189984, + "grad_norm": 2.263587236404419, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7478635311126709, + "num_tokens": 375148049.0, + "step": 14508 + }, + { + "epoch": 1.5933450472216122, + "grad_norm": 2.101208448410034, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7177621722221375, + "num_tokens": 375177807.0, + "step": 14509 + }, + { + "epoch": 1.593454864924226, + "grad_norm": 2.260735034942627, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7370854616165161, + "num_tokens": 375201645.0, + "step": 14510 + }, + { + "epoch": 1.5935646826268395, + "grad_norm": 2.0687267780303955, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7123470902442932, + "num_tokens": 375229559.0, + "step": 14511 + }, + { + "epoch": 1.593674500329453, + "grad_norm": 2.0497496128082275, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.712460994720459, + "num_tokens": 375259333.0, + "step": 14512 + }, + { + "epoch": 1.5937843180320668, + "grad_norm": 2.0856211185455322, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.696198582649231, + "num_tokens": 375288550.0, + "step": 14513 + }, + { + "epoch": 1.5938941357346805, + "grad_norm": 1.914654016494751, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7344210743904114, + "num_tokens": 375320715.0, + "step": 14514 + }, + { + "epoch": 1.594003953437294, + "grad_norm": 2.1681325435638428, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6936448812484741, + "num_tokens": 375350974.0, + "step": 14515 + }, + { + "epoch": 1.5941137711399076, + "grad_norm": 2.062443256378174, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7085825800895691, + "num_tokens": 375380124.0, + "step": 14516 + }, + { + "epoch": 1.5942235888425214, + "grad_norm": 2.2994675636291504, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7396979331970215, + "num_tokens": 375403688.0, + "step": 14517 + }, + { + "epoch": 1.5943334065451351, + "grad_norm": 2.4698574542999268, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7296785116195679, + "num_tokens": 375424204.0, + "step": 14518 + }, + { + "epoch": 1.5944432242477489, + "grad_norm": 2.0630931854248047, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6987561583518982, + "num_tokens": 375454822.0, + "step": 14519 + }, + { + "epoch": 1.5945530419503624, + "grad_norm": 2.662022829055786, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7116681337356567, + "num_tokens": 375481295.0, + "step": 14520 + }, + { + "epoch": 1.594662859652976, + "grad_norm": 2.250774621963501, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7278624773025513, + "num_tokens": 375505553.0, + "step": 14521 + }, + { + "epoch": 1.5947726773555897, + "grad_norm": 2.3371994495391846, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7215638756752014, + "num_tokens": 375529493.0, + "step": 14522 + }, + { + "epoch": 1.5948824950582035, + "grad_norm": 2.017331838607788, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.718018651008606, + "num_tokens": 375560012.0, + "step": 14523 + }, + { + "epoch": 1.5949923127608172, + "grad_norm": 2.2948737144470215, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7225424647331238, + "num_tokens": 375584353.0, + "step": 14524 + }, + { + "epoch": 1.5951021304634307, + "grad_norm": 2.1773505210876465, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7283334732055664, + "num_tokens": 375610895.0, + "step": 14525 + }, + { + "epoch": 1.5952119481660443, + "grad_norm": 2.446533679962158, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7295395135879517, + "num_tokens": 375632870.0, + "step": 14526 + }, + { + "epoch": 1.595321765868658, + "grad_norm": 2.243783950805664, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7089120745658875, + "num_tokens": 375658316.0, + "step": 14527 + }, + { + "epoch": 1.5954315835712718, + "grad_norm": 2.187589645385742, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7125875353813171, + "num_tokens": 375684916.0, + "step": 14528 + }, + { + "epoch": 1.5955414012738853, + "grad_norm": 2.1983683109283447, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7167685031890869, + "num_tokens": 375711341.0, + "step": 14529 + }, + { + "epoch": 1.5956512189764989, + "grad_norm": 2.224583864212036, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7006133794784546, + "num_tokens": 375738416.0, + "step": 14530 + }, + { + "epoch": 1.5957610366791126, + "grad_norm": 2.3839259147644043, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7436539530754089, + "num_tokens": 375760713.0, + "step": 14531 + }, + { + "epoch": 1.5958708543817264, + "grad_norm": 2.175288200378418, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7435227632522583, + "num_tokens": 375786840.0, + "step": 14532 + }, + { + "epoch": 1.5959806720843401, + "grad_norm": 2.355529546737671, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7179462909698486, + "num_tokens": 375811614.0, + "step": 14533 + }, + { + "epoch": 1.5960904897869537, + "grad_norm": 2.2166037559509277, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7187671661376953, + "num_tokens": 375838020.0, + "step": 14534 + }, + { + "epoch": 1.5962003074895672, + "grad_norm": 2.358889102935791, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7191858291625977, + "num_tokens": 375861796.0, + "step": 14535 + }, + { + "epoch": 1.596310125192181, + "grad_norm": 2.2901439666748047, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7220419049263, + "num_tokens": 375887156.0, + "step": 14536 + }, + { + "epoch": 1.5964199428947947, + "grad_norm": 2.192901611328125, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7383564710617065, + "num_tokens": 375911343.0, + "step": 14537 + }, + { + "epoch": 1.5965297605974083, + "grad_norm": 2.3327701091766357, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6978658437728882, + "num_tokens": 375936167.0, + "step": 14538 + }, + { + "epoch": 1.596639578300022, + "grad_norm": 1.9614211320877075, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7162173390388489, + "num_tokens": 375967462.0, + "step": 14539 + }, + { + "epoch": 1.5967493960026355, + "grad_norm": 2.5355660915374756, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7305291891098022, + "num_tokens": 375990868.0, + "step": 14540 + }, + { + "epoch": 1.5968592137052493, + "grad_norm": 2.305534601211548, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7047760486602783, + "num_tokens": 376016493.0, + "step": 14541 + }, + { + "epoch": 1.596969031407863, + "grad_norm": 2.513204574584961, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7202700972557068, + "num_tokens": 376039072.0, + "step": 14542 + }, + { + "epoch": 1.5970788491104766, + "grad_norm": 2.1250877380371094, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7228330969810486, + "num_tokens": 376067183.0, + "step": 14543 + }, + { + "epoch": 1.5971886668130901, + "grad_norm": 2.348498821258545, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7299441695213318, + "num_tokens": 376091007.0, + "step": 14544 + }, + { + "epoch": 1.5972984845157039, + "grad_norm": 2.536430597305298, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7420976161956787, + "num_tokens": 376110858.0, + "step": 14545 + }, + { + "epoch": 1.5974083022183176, + "grad_norm": 2.419382095336914, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7454521656036377, + "num_tokens": 376131065.0, + "step": 14546 + }, + { + "epoch": 1.5975181199209314, + "grad_norm": 2.258598566055298, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7150946855545044, + "num_tokens": 376156691.0, + "step": 14547 + }, + { + "epoch": 1.597627937623545, + "grad_norm": 2.324784278869629, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7103332877159119, + "num_tokens": 376180701.0, + "step": 14548 + }, + { + "epoch": 1.5977377553261585, + "grad_norm": 2.1509454250335693, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7100847959518433, + "num_tokens": 376209886.0, + "step": 14549 + }, + { + "epoch": 1.5978475730287722, + "grad_norm": 2.2359020709991455, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7347816228866577, + "num_tokens": 376234719.0, + "step": 14550 + }, + { + "epoch": 1.597957390731386, + "grad_norm": 2.4977059364318848, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.729924201965332, + "num_tokens": 376255215.0, + "step": 14551 + }, + { + "epoch": 1.5980672084339995, + "grad_norm": 2.194204568862915, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7188414931297302, + "num_tokens": 376280985.0, + "step": 14552 + }, + { + "epoch": 1.5981770261366133, + "grad_norm": 2.4070565700531006, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7217861413955688, + "num_tokens": 376305600.0, + "step": 14553 + }, + { + "epoch": 1.5982868438392268, + "grad_norm": 2.4796836376190186, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7162625789642334, + "num_tokens": 376329147.0, + "step": 14554 + }, + { + "epoch": 1.5983966615418406, + "grad_norm": 2.247723340988159, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7173287868499756, + "num_tokens": 376353320.0, + "step": 14555 + }, + { + "epoch": 1.5985064792444543, + "grad_norm": 2.1589977741241455, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7045223712921143, + "num_tokens": 376380025.0, + "step": 14556 + }, + { + "epoch": 1.5986162969470679, + "grad_norm": 2.0006752014160156, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7117091417312622, + "num_tokens": 376411452.0, + "step": 14557 + }, + { + "epoch": 1.5987261146496814, + "grad_norm": 1.9792749881744385, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7045078277587891, + "num_tokens": 376444923.0, + "step": 14558 + }, + { + "epoch": 1.5988359323522952, + "grad_norm": 2.3925280570983887, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7160173058509827, + "num_tokens": 376467832.0, + "step": 14559 + }, + { + "epoch": 1.598945750054909, + "grad_norm": 2.2454516887664795, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7170261144638062, + "num_tokens": 376492742.0, + "step": 14560 + }, + { + "epoch": 1.5990555677575227, + "grad_norm": 1.9947141408920288, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7205934524536133, + "num_tokens": 376525201.0, + "step": 14561 + }, + { + "epoch": 1.5991653854601362, + "grad_norm": 2.394958972930908, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7326755523681641, + "num_tokens": 376549642.0, + "step": 14562 + }, + { + "epoch": 1.5992752031627497, + "grad_norm": 2.8065996170043945, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7144753932952881, + "num_tokens": 376568606.0, + "step": 14563 + }, + { + "epoch": 1.5993850208653635, + "grad_norm": 2.3306312561035156, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7034180164337158, + "num_tokens": 376593160.0, + "step": 14564 + }, + { + "epoch": 1.5994948385679773, + "grad_norm": 2.356412172317505, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7407633066177368, + "num_tokens": 376615368.0, + "step": 14565 + }, + { + "epoch": 1.5996046562705908, + "grad_norm": 2.3609633445739746, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.6959051489830017, + "num_tokens": 376639264.0, + "step": 14566 + }, + { + "epoch": 1.5997144739732043, + "grad_norm": 2.0542681217193604, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7067921757698059, + "num_tokens": 376669428.0, + "step": 14567 + }, + { + "epoch": 1.599824291675818, + "grad_norm": 2.1413519382476807, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7384063601493835, + "num_tokens": 376694719.0, + "step": 14568 + }, + { + "epoch": 1.5999341093784318, + "grad_norm": 2.3472020626068115, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7170169353485107, + "num_tokens": 376718948.0, + "step": 14569 + }, + { + "epoch": 1.6000439270810456, + "grad_norm": 2.1578104496002197, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7200263738632202, + "num_tokens": 376747031.0, + "step": 14570 + }, + { + "epoch": 1.6001537447836591, + "grad_norm": 2.39436674118042, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7346310019493103, + "num_tokens": 376770942.0, + "step": 14571 + }, + { + "epoch": 1.6002635624862727, + "grad_norm": 2.131751775741577, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7235603332519531, + "num_tokens": 376800528.0, + "step": 14572 + }, + { + "epoch": 1.6003733801888864, + "grad_norm": 1.940760612487793, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7148541212081909, + "num_tokens": 376833529.0, + "step": 14573 + }, + { + "epoch": 1.6004831978915002, + "grad_norm": 2.2533986568450928, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7250909209251404, + "num_tokens": 376856645.0, + "step": 14574 + }, + { + "epoch": 1.600593015594114, + "grad_norm": 2.065432071685791, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7226144075393677, + "num_tokens": 376888611.0, + "step": 14575 + }, + { + "epoch": 1.6007028332967275, + "grad_norm": 2.256253719329834, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7189761996269226, + "num_tokens": 376912936.0, + "step": 14576 + }, + { + "epoch": 1.600812650999341, + "grad_norm": 2.258456230163574, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.708069384098053, + "num_tokens": 376937269.0, + "step": 14577 + }, + { + "epoch": 1.6009224687019548, + "grad_norm": 2.32106876373291, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7246830463409424, + "num_tokens": 376960702.0, + "step": 14578 + }, + { + "epoch": 1.6010322864045685, + "grad_norm": 2.102630615234375, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7252184748649597, + "num_tokens": 376988619.0, + "step": 14579 + }, + { + "epoch": 1.601142104107182, + "grad_norm": 1.9880845546722412, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6886186599731445, + "num_tokens": 377020953.0, + "step": 14580 + }, + { + "epoch": 1.6012519218097956, + "grad_norm": 2.3182730674743652, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7272887825965881, + "num_tokens": 377044028.0, + "step": 14581 + }, + { + "epoch": 1.6013617395124093, + "grad_norm": 2.1778671741485596, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7133300304412842, + "num_tokens": 377076356.0, + "step": 14582 + }, + { + "epoch": 1.601471557215023, + "grad_norm": 2.2829651832580566, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7252706289291382, + "num_tokens": 377100286.0, + "step": 14583 + }, + { + "epoch": 1.6015813749176369, + "grad_norm": 2.082385540008545, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7281537055969238, + "num_tokens": 377129006.0, + "step": 14584 + }, + { + "epoch": 1.6016911926202504, + "grad_norm": 2.7613067626953125, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7403215169906616, + "num_tokens": 377147317.0, + "step": 14585 + }, + { + "epoch": 1.601801010322864, + "grad_norm": 2.2063257694244385, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7104260921478271, + "num_tokens": 377174880.0, + "step": 14586 + }, + { + "epoch": 1.6019108280254777, + "grad_norm": 2.418095827102661, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.720008134841919, + "num_tokens": 377197324.0, + "step": 14587 + }, + { + "epoch": 1.6020206457280914, + "grad_norm": 2.5113863945007324, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7147811651229858, + "num_tokens": 377218568.0, + "step": 14588 + }, + { + "epoch": 1.602130463430705, + "grad_norm": 2.2915232181549072, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7208243012428284, + "num_tokens": 377242710.0, + "step": 14589 + }, + { + "epoch": 1.6022402811333187, + "grad_norm": 1.884214162826538, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7168769240379333, + "num_tokens": 377282627.0, + "step": 14590 + }, + { + "epoch": 1.6023500988359323, + "grad_norm": 2.450885772705078, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7324283123016357, + "num_tokens": 377304607.0, + "step": 14591 + }, + { + "epoch": 1.602459916538546, + "grad_norm": 2.6608147621154785, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7398678064346313, + "num_tokens": 377324519.0, + "step": 14592 + }, + { + "epoch": 1.6025697342411598, + "grad_norm": 1.8461798429489136, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6898313760757446, + "num_tokens": 377360265.0, + "step": 14593 + }, + { + "epoch": 1.6026795519437733, + "grad_norm": 2.794752836227417, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7201023101806641, + "num_tokens": 377377975.0, + "step": 14594 + }, + { + "epoch": 1.6027893696463869, + "grad_norm": 2.068333625793457, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7242271900177002, + "num_tokens": 377407738.0, + "step": 14595 + }, + { + "epoch": 1.6028991873490006, + "grad_norm": 2.35382342338562, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7220845222473145, + "num_tokens": 377429190.0, + "step": 14596 + }, + { + "epoch": 1.6030090050516144, + "grad_norm": 2.3629019260406494, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7414067983627319, + "num_tokens": 377451639.0, + "step": 14597 + }, + { + "epoch": 1.6031188227542281, + "grad_norm": 2.3578174114227295, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7110638618469238, + "num_tokens": 377477193.0, + "step": 14598 + }, + { + "epoch": 1.6032286404568417, + "grad_norm": 2.412874460220337, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7227498292922974, + "num_tokens": 377503375.0, + "step": 14599 + }, + { + "epoch": 1.6033384581594552, + "grad_norm": 2.519590377807617, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7283757328987122, + "num_tokens": 377523637.0, + "step": 14600 + }, + { + "epoch": 1.603448275862069, + "grad_norm": 2.193476438522339, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6999528408050537, + "num_tokens": 377553443.0, + "step": 14601 + }, + { + "epoch": 1.6035580935646827, + "grad_norm": 2.001021146774292, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7545818090438843, + "num_tokens": 377582274.0, + "step": 14602 + }, + { + "epoch": 1.6036679112672962, + "grad_norm": 2.443516731262207, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7103926539421082, + "num_tokens": 377604082.0, + "step": 14603 + }, + { + "epoch": 1.60377772896991, + "grad_norm": 2.114453077316284, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7392263412475586, + "num_tokens": 377629865.0, + "step": 14604 + }, + { + "epoch": 1.6038875466725235, + "grad_norm": 2.3779735565185547, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7405614256858826, + "num_tokens": 377651745.0, + "step": 14605 + }, + { + "epoch": 1.6039973643751373, + "grad_norm": 1.9946469068527222, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7018139362335205, + "num_tokens": 377683487.0, + "step": 14606 + }, + { + "epoch": 1.604107182077751, + "grad_norm": 2.3123779296875, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.6995931267738342, + "num_tokens": 377708915.0, + "step": 14607 + }, + { + "epoch": 1.6042169997803646, + "grad_norm": 2.1502254009246826, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7198336720466614, + "num_tokens": 377733857.0, + "step": 14608 + }, + { + "epoch": 1.6043268174829781, + "grad_norm": 2.5883750915527344, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7466146945953369, + "num_tokens": 377752263.0, + "step": 14609 + }, + { + "epoch": 1.6044366351855919, + "grad_norm": 2.288194179534912, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7196632623672485, + "num_tokens": 377775949.0, + "step": 14610 + }, + { + "epoch": 1.6045464528882056, + "grad_norm": 2.465158224105835, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7367804050445557, + "num_tokens": 377797937.0, + "step": 14611 + }, + { + "epoch": 1.6046562705908194, + "grad_norm": 2.148479700088501, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7229725122451782, + "num_tokens": 377828982.0, + "step": 14612 + }, + { + "epoch": 1.604766088293433, + "grad_norm": 1.9119577407836914, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7040395736694336, + "num_tokens": 377863360.0, + "step": 14613 + }, + { + "epoch": 1.6048759059960465, + "grad_norm": 2.151325225830078, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7025192975997925, + "num_tokens": 377891909.0, + "step": 14614 + }, + { + "epoch": 1.6049857236986602, + "grad_norm": 2.2864370346069336, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7205213308334351, + "num_tokens": 377916869.0, + "step": 14615 + }, + { + "epoch": 1.605095541401274, + "grad_norm": 2.337745428085327, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6910744309425354, + "num_tokens": 377941425.0, + "step": 14616 + }, + { + "epoch": 1.6052053591038875, + "grad_norm": 2.2276155948638916, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7353323698043823, + "num_tokens": 377966567.0, + "step": 14617 + }, + { + "epoch": 1.605315176806501, + "grad_norm": 2.284738779067993, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.6958544254302979, + "num_tokens": 377993994.0, + "step": 14618 + }, + { + "epoch": 1.6054249945091148, + "grad_norm": 2.3347535133361816, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.723300576210022, + "num_tokens": 378018392.0, + "step": 14619 + }, + { + "epoch": 1.6055348122117286, + "grad_norm": 2.556335687637329, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7581120133399963, + "num_tokens": 378038143.0, + "step": 14620 + }, + { + "epoch": 1.6056446299143423, + "grad_norm": 2.288235664367676, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7109943628311157, + "num_tokens": 378063431.0, + "step": 14621 + }, + { + "epoch": 1.6057544476169558, + "grad_norm": 2.2677419185638428, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.745997428894043, + "num_tokens": 378086320.0, + "step": 14622 + }, + { + "epoch": 1.6058642653195694, + "grad_norm": 2.627709150314331, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7178969979286194, + "num_tokens": 378107634.0, + "step": 14623 + }, + { + "epoch": 1.6059740830221831, + "grad_norm": 2.3332064151763916, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7231473922729492, + "num_tokens": 378134425.0, + "step": 14624 + }, + { + "epoch": 1.606083900724797, + "grad_norm": 2.0159785747528076, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.6993563175201416, + "num_tokens": 378163654.0, + "step": 14625 + }, + { + "epoch": 1.6061937184274107, + "grad_norm": 2.1632068157196045, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7212742567062378, + "num_tokens": 378190317.0, + "step": 14626 + }, + { + "epoch": 1.6063035361300242, + "grad_norm": 2.1875691413879395, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6918727159500122, + "num_tokens": 378219657.0, + "step": 14627 + }, + { + "epoch": 1.6064133538326377, + "grad_norm": 2.5917065143585205, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7263104915618896, + "num_tokens": 378239882.0, + "step": 14628 + }, + { + "epoch": 1.6065231715352515, + "grad_norm": 2.184580087661743, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6951588988304138, + "num_tokens": 378267995.0, + "step": 14629 + }, + { + "epoch": 1.6066329892378652, + "grad_norm": 2.3541667461395264, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7131491899490356, + "num_tokens": 378291931.0, + "step": 14630 + }, + { + "epoch": 1.6067428069404788, + "grad_norm": 2.365382671356201, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7132940292358398, + "num_tokens": 378315418.0, + "step": 14631 + }, + { + "epoch": 1.6068526246430923, + "grad_norm": 2.3521764278411865, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7273669838905334, + "num_tokens": 378336854.0, + "step": 14632 + }, + { + "epoch": 1.606962442345706, + "grad_norm": 2.0625083446502686, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7161214351654053, + "num_tokens": 378365320.0, + "step": 14633 + }, + { + "epoch": 1.6070722600483198, + "grad_norm": 2.6198177337646484, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7284871339797974, + "num_tokens": 378387196.0, + "step": 14634 + }, + { + "epoch": 1.6071820777509336, + "grad_norm": 2.623669147491455, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7364786863327026, + "num_tokens": 378406099.0, + "step": 14635 + }, + { + "epoch": 1.6072918954535471, + "grad_norm": 2.048943519592285, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6965137720108032, + "num_tokens": 378434516.0, + "step": 14636 + }, + { + "epoch": 1.6074017131561606, + "grad_norm": 2.040131092071533, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7352393269538879, + "num_tokens": 378463211.0, + "step": 14637 + }, + { + "epoch": 1.6075115308587744, + "grad_norm": 2.371479034423828, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7084159255027771, + "num_tokens": 378487206.0, + "step": 14638 + }, + { + "epoch": 1.6076213485613882, + "grad_norm": 2.316932439804077, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.720154881477356, + "num_tokens": 378509810.0, + "step": 14639 + }, + { + "epoch": 1.607731166264002, + "grad_norm": 2.2862186431884766, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7113045454025269, + "num_tokens": 378535236.0, + "step": 14640 + }, + { + "epoch": 1.6078409839666155, + "grad_norm": 2.1280858516693115, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7181383371353149, + "num_tokens": 378564805.0, + "step": 14641 + }, + { + "epoch": 1.607950801669229, + "grad_norm": 2.24770188331604, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7158291339874268, + "num_tokens": 378590294.0, + "step": 14642 + }, + { + "epoch": 1.6080606193718427, + "grad_norm": 2.252725601196289, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.6971174478530884, + "num_tokens": 378616209.0, + "step": 14643 + }, + { + "epoch": 1.6081704370744565, + "grad_norm": 2.2284507751464844, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7056809663772583, + "num_tokens": 378642574.0, + "step": 14644 + }, + { + "epoch": 1.60828025477707, + "grad_norm": 2.2153968811035156, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7164062261581421, + "num_tokens": 378667975.0, + "step": 14645 + }, + { + "epoch": 1.6083900724796836, + "grad_norm": 2.1372475624084473, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7317242622375488, + "num_tokens": 378697179.0, + "step": 14646 + }, + { + "epoch": 1.6084998901822973, + "grad_norm": 2.5770888328552246, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7258618474006653, + "num_tokens": 378718268.0, + "step": 14647 + }, + { + "epoch": 1.608609707884911, + "grad_norm": 2.228198766708374, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7149132490158081, + "num_tokens": 378745182.0, + "step": 14648 + }, + { + "epoch": 1.6087195255875248, + "grad_norm": 2.290961265563965, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7361512780189514, + "num_tokens": 378770063.0, + "step": 14649 + }, + { + "epoch": 1.6088293432901384, + "grad_norm": 2.253063440322876, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.732452929019928, + "num_tokens": 378794104.0, + "step": 14650 + }, + { + "epoch": 1.608939160992752, + "grad_norm": 2.175633668899536, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7180328965187073, + "num_tokens": 378820354.0, + "step": 14651 + }, + { + "epoch": 1.6090489786953657, + "grad_norm": 2.2409112453460693, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7166473865509033, + "num_tokens": 378847049.0, + "step": 14652 + }, + { + "epoch": 1.6091587963979794, + "grad_norm": 2.261531352996826, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7368030548095703, + "num_tokens": 378871601.0, + "step": 14653 + }, + { + "epoch": 1.609268614100593, + "grad_norm": 2.238262414932251, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7336170673370361, + "num_tokens": 378895536.0, + "step": 14654 + }, + { + "epoch": 1.6093784318032067, + "grad_norm": 2.1290132999420166, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7180606722831726, + "num_tokens": 378923499.0, + "step": 14655 + }, + { + "epoch": 1.6094882495058203, + "grad_norm": 2.050819158554077, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7109118700027466, + "num_tokens": 378952595.0, + "step": 14656 + }, + { + "epoch": 1.609598067208434, + "grad_norm": 2.109858989715576, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7030242085456848, + "num_tokens": 378981185.0, + "step": 14657 + }, + { + "epoch": 1.6097078849110478, + "grad_norm": 2.211437225341797, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7109098434448242, + "num_tokens": 379009117.0, + "step": 14658 + }, + { + "epoch": 1.6098177026136613, + "grad_norm": 2.6131467819213867, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7365570068359375, + "num_tokens": 379028374.0, + "step": 14659 + }, + { + "epoch": 1.6099275203162748, + "grad_norm": 2.019810438156128, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.697445273399353, + "num_tokens": 379058265.0, + "step": 14660 + }, + { + "epoch": 1.6100373380188886, + "grad_norm": 2.445507526397705, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7153893113136292, + "num_tokens": 379078961.0, + "step": 14661 + }, + { + "epoch": 1.6101471557215024, + "grad_norm": 2.332197904586792, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7236841320991516, + "num_tokens": 379104371.0, + "step": 14662 + }, + { + "epoch": 1.6102569734241161, + "grad_norm": 2.481832981109619, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7300544381141663, + "num_tokens": 379125368.0, + "step": 14663 + }, + { + "epoch": 1.6103667911267296, + "grad_norm": 2.214561939239502, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7167008519172668, + "num_tokens": 379148661.0, + "step": 14664 + }, + { + "epoch": 1.6104766088293432, + "grad_norm": 2.014848232269287, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7066223621368408, + "num_tokens": 379180087.0, + "step": 14665 + }, + { + "epoch": 1.610586426531957, + "grad_norm": 2.1743335723876953, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6947556734085083, + "num_tokens": 379208009.0, + "step": 14666 + }, + { + "epoch": 1.6106962442345707, + "grad_norm": 2.410783529281616, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7180854082107544, + "num_tokens": 379229757.0, + "step": 14667 + }, + { + "epoch": 1.6108060619371842, + "grad_norm": 2.3543174266815186, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7343382835388184, + "num_tokens": 379252466.0, + "step": 14668 + }, + { + "epoch": 1.610915879639798, + "grad_norm": 2.139336109161377, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.729667067527771, + "num_tokens": 379281949.0, + "step": 14669 + }, + { + "epoch": 1.6110256973424115, + "grad_norm": 2.0682361125946045, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.696835458278656, + "num_tokens": 379314100.0, + "step": 14670 + }, + { + "epoch": 1.6111355150450253, + "grad_norm": 2.185042381286621, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7221331596374512, + "num_tokens": 379341840.0, + "step": 14671 + }, + { + "epoch": 1.611245332747639, + "grad_norm": 2.1061859130859375, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7292114496231079, + "num_tokens": 379368847.0, + "step": 14672 + }, + { + "epoch": 1.6113551504502526, + "grad_norm": 2.380880117416382, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7384265661239624, + "num_tokens": 379391704.0, + "step": 14673 + }, + { + "epoch": 1.611464968152866, + "grad_norm": 2.4128520488739014, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7306751012802124, + "num_tokens": 379415996.0, + "step": 14674 + }, + { + "epoch": 1.6115747858554799, + "grad_norm": 2.025205373764038, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.727579653263092, + "num_tokens": 379445398.0, + "step": 14675 + }, + { + "epoch": 1.6116846035580936, + "grad_norm": 2.1336758136749268, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7236063480377197, + "num_tokens": 379473108.0, + "step": 14676 + }, + { + "epoch": 1.6117944212607074, + "grad_norm": 2.3104164600372314, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7287595272064209, + "num_tokens": 379497755.0, + "step": 14677 + }, + { + "epoch": 1.611904238963321, + "grad_norm": 2.4763288497924805, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7285128831863403, + "num_tokens": 379520189.0, + "step": 14678 + }, + { + "epoch": 1.6120140566659344, + "grad_norm": 2.4393324851989746, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7161785364151001, + "num_tokens": 379543320.0, + "step": 14679 + }, + { + "epoch": 1.6121238743685482, + "grad_norm": 2.2438297271728516, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7118682265281677, + "num_tokens": 379566938.0, + "step": 14680 + }, + { + "epoch": 1.612233692071162, + "grad_norm": 2.268054723739624, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7191329002380371, + "num_tokens": 379591567.0, + "step": 14681 + }, + { + "epoch": 1.6123435097737755, + "grad_norm": 2.170330286026001, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7317476868629456, + "num_tokens": 379617964.0, + "step": 14682 + }, + { + "epoch": 1.612453327476389, + "grad_norm": 2.136404514312744, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7169535160064697, + "num_tokens": 379645750.0, + "step": 14683 + }, + { + "epoch": 1.6125631451790028, + "grad_norm": 2.2731401920318604, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7397489547729492, + "num_tokens": 379669620.0, + "step": 14684 + }, + { + "epoch": 1.6126729628816165, + "grad_norm": 2.239778995513916, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7166513800621033, + "num_tokens": 379695559.0, + "step": 14685 + }, + { + "epoch": 1.6127827805842303, + "grad_norm": 2.395207166671753, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7199701070785522, + "num_tokens": 379719060.0, + "step": 14686 + }, + { + "epoch": 1.6128925982868438, + "grad_norm": 2.3910930156707764, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7039006948471069, + "num_tokens": 379746301.0, + "step": 14687 + }, + { + "epoch": 1.6130024159894574, + "grad_norm": 2.107661247253418, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7064265012741089, + "num_tokens": 379774412.0, + "step": 14688 + }, + { + "epoch": 1.6131122336920711, + "grad_norm": 2.1242587566375732, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.69926518201828, + "num_tokens": 379802433.0, + "step": 14689 + }, + { + "epoch": 1.6132220513946849, + "grad_norm": 1.949874758720398, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7128391265869141, + "num_tokens": 379834583.0, + "step": 14690 + }, + { + "epoch": 1.6133318690972986, + "grad_norm": 2.704525947570801, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.723118782043457, + "num_tokens": 379854369.0, + "step": 14691 + }, + { + "epoch": 1.6134416867999122, + "grad_norm": 2.240563154220581, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7200254201889038, + "num_tokens": 379880685.0, + "step": 14692 + }, + { + "epoch": 1.6135515045025257, + "grad_norm": 2.139836311340332, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.689130961894989, + "num_tokens": 379910731.0, + "step": 14693 + }, + { + "epoch": 1.6136613222051395, + "grad_norm": 2.2779462337493896, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7303507328033447, + "num_tokens": 379934116.0, + "step": 14694 + }, + { + "epoch": 1.6137711399077532, + "grad_norm": 2.2469441890716553, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7292605042457581, + "num_tokens": 379960684.0, + "step": 14695 + }, + { + "epoch": 1.6138809576103668, + "grad_norm": 2.2504844665527344, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7090470790863037, + "num_tokens": 379987816.0, + "step": 14696 + }, + { + "epoch": 1.6139907753129803, + "grad_norm": 2.0388739109039307, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7185643911361694, + "num_tokens": 380017144.0, + "step": 14697 + }, + { + "epoch": 1.614100593015594, + "grad_norm": 2.4696383476257324, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7271893620491028, + "num_tokens": 380039497.0, + "step": 14698 + }, + { + "epoch": 1.6142104107182078, + "grad_norm": 2.111778974533081, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7170634865760803, + "num_tokens": 380068009.0, + "step": 14699 + }, + { + "epoch": 1.6143202284208216, + "grad_norm": 2.045413017272949, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7062421441078186, + "num_tokens": 380101244.0, + "step": 14700 + }, + { + "epoch": 1.614430046123435, + "grad_norm": 2.144325017929077, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7341365814208984, + "num_tokens": 380129242.0, + "step": 14701 + }, + { + "epoch": 1.6145398638260486, + "grad_norm": 2.4838640689849854, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7044999003410339, + "num_tokens": 380151774.0, + "step": 14702 + }, + { + "epoch": 1.6146496815286624, + "grad_norm": 2.060811758041382, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7341646552085876, + "num_tokens": 380178855.0, + "step": 14703 + }, + { + "epoch": 1.6147594992312762, + "grad_norm": 2.3813278675079346, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7223337292671204, + "num_tokens": 380200391.0, + "step": 14704 + }, + { + "epoch": 1.61486931693389, + "grad_norm": 2.0629165172576904, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7009667158126831, + "num_tokens": 380231561.0, + "step": 14705 + }, + { + "epoch": 1.6149791346365034, + "grad_norm": 2.185783863067627, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7351353764533997, + "num_tokens": 380255188.0, + "step": 14706 + }, + { + "epoch": 1.615088952339117, + "grad_norm": 2.4277472496032715, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7255533933639526, + "num_tokens": 380278009.0, + "step": 14707 + }, + { + "epoch": 1.6151987700417307, + "grad_norm": 1.9450932741165161, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.6975115537643433, + "num_tokens": 380311382.0, + "step": 14708 + }, + { + "epoch": 1.6153085877443445, + "grad_norm": 1.8186125755310059, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6866788864135742, + "num_tokens": 380348900.0, + "step": 14709 + }, + { + "epoch": 1.615418405446958, + "grad_norm": 2.2474942207336426, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7383454442024231, + "num_tokens": 380373273.0, + "step": 14710 + }, + { + "epoch": 1.6155282231495716, + "grad_norm": 2.013808250427246, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7261662483215332, + "num_tokens": 380404588.0, + "step": 14711 + }, + { + "epoch": 1.6156380408521853, + "grad_norm": 2.0379655361175537, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7028701305389404, + "num_tokens": 380435705.0, + "step": 14712 + }, + { + "epoch": 1.615747858554799, + "grad_norm": 2.338254451751709, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7094974517822266, + "num_tokens": 380459583.0, + "step": 14713 + }, + { + "epoch": 1.6158576762574128, + "grad_norm": 2.3565444946289062, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7475996017456055, + "num_tokens": 380483648.0, + "step": 14714 + }, + { + "epoch": 1.6159674939600264, + "grad_norm": 2.230553150177002, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.7019292116165161, + "num_tokens": 380510835.0, + "step": 14715 + }, + { + "epoch": 1.61607731166264, + "grad_norm": 2.4412457942962646, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7063845992088318, + "num_tokens": 380534479.0, + "step": 14716 + }, + { + "epoch": 1.6161871293652537, + "grad_norm": 2.266435146331787, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7149859666824341, + "num_tokens": 380558985.0, + "step": 14717 + }, + { + "epoch": 1.6162969470678674, + "grad_norm": 2.449331760406494, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7123109102249146, + "num_tokens": 380582954.0, + "step": 14718 + }, + { + "epoch": 1.616406764770481, + "grad_norm": 2.1734871864318848, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7097137570381165, + "num_tokens": 380609904.0, + "step": 14719 + }, + { + "epoch": 1.6165165824730947, + "grad_norm": 2.473284959793091, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7170286774635315, + "num_tokens": 380633853.0, + "step": 14720 + }, + { + "epoch": 1.6166264001757082, + "grad_norm": 2.356154203414917, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7162002325057983, + "num_tokens": 380658581.0, + "step": 14721 + }, + { + "epoch": 1.616736217878322, + "grad_norm": 2.166125774383545, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7339243292808533, + "num_tokens": 380684309.0, + "step": 14722 + }, + { + "epoch": 1.6168460355809358, + "grad_norm": 2.198040008544922, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7101185321807861, + "num_tokens": 380714010.0, + "step": 14723 + }, + { + "epoch": 1.6169558532835493, + "grad_norm": 2.2179272174835205, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7654694318771362, + "num_tokens": 380736775.0, + "step": 14724 + }, + { + "epoch": 1.6170656709861628, + "grad_norm": 2.445509910583496, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7418577075004578, + "num_tokens": 380758733.0, + "step": 14725 + }, + { + "epoch": 1.6171754886887766, + "grad_norm": 2.3884482383728027, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7330419421195984, + "num_tokens": 380781061.0, + "step": 14726 + }, + { + "epoch": 1.6172853063913903, + "grad_norm": 2.559326648712158, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7485605478286743, + "num_tokens": 380800537.0, + "step": 14727 + }, + { + "epoch": 1.617395124094004, + "grad_norm": 2.511162281036377, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7121874094009399, + "num_tokens": 380822551.0, + "step": 14728 + }, + { + "epoch": 1.6175049417966176, + "grad_norm": 2.1958725452423096, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6933337450027466, + "num_tokens": 380850508.0, + "step": 14729 + }, + { + "epoch": 1.6176147594992312, + "grad_norm": 2.2884902954101562, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7166515588760376, + "num_tokens": 380873596.0, + "step": 14730 + }, + { + "epoch": 1.617724577201845, + "grad_norm": 2.1447246074676514, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.715962290763855, + "num_tokens": 380900552.0, + "step": 14731 + }, + { + "epoch": 1.6178343949044587, + "grad_norm": 2.1696226596832275, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7254738807678223, + "num_tokens": 380927255.0, + "step": 14732 + }, + { + "epoch": 1.6179442126070722, + "grad_norm": 2.2146999835968018, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7028907537460327, + "num_tokens": 380952323.0, + "step": 14733 + }, + { + "epoch": 1.618054030309686, + "grad_norm": 2.4747204780578613, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6929503679275513, + "num_tokens": 380978041.0, + "step": 14734 + }, + { + "epoch": 1.6181638480122995, + "grad_norm": 2.3933801651000977, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7452032566070557, + "num_tokens": 380998558.0, + "step": 14735 + }, + { + "epoch": 1.6182736657149133, + "grad_norm": 2.0297420024871826, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7038780450820923, + "num_tokens": 381029442.0, + "step": 14736 + }, + { + "epoch": 1.618383483417527, + "grad_norm": 2.379034996032715, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7351782321929932, + "num_tokens": 381053022.0, + "step": 14737 + }, + { + "epoch": 1.6184933011201406, + "grad_norm": 2.4058430194854736, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.7012406587600708, + "num_tokens": 381078358.0, + "step": 14738 + }, + { + "epoch": 1.618603118822754, + "grad_norm": 2.161207675933838, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7195546627044678, + "num_tokens": 381105573.0, + "step": 14739 + }, + { + "epoch": 1.6187129365253679, + "grad_norm": 2.0604166984558105, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7135772705078125, + "num_tokens": 381136566.0, + "step": 14740 + }, + { + "epoch": 1.6188227542279816, + "grad_norm": 2.39725923538208, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7179083824157715, + "num_tokens": 381159490.0, + "step": 14741 + }, + { + "epoch": 1.6189325719305954, + "grad_norm": 2.1489875316619873, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7408225536346436, + "num_tokens": 381184986.0, + "step": 14742 + }, + { + "epoch": 1.619042389633209, + "grad_norm": 2.133267641067505, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.723190188407898, + "num_tokens": 381210746.0, + "step": 14743 + }, + { + "epoch": 1.6191522073358224, + "grad_norm": 1.9771274328231812, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7106679677963257, + "num_tokens": 381241557.0, + "step": 14744 + }, + { + "epoch": 1.6192620250384362, + "grad_norm": 2.253979206085205, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7470757365226746, + "num_tokens": 381266486.0, + "step": 14745 + }, + { + "epoch": 1.61937184274105, + "grad_norm": 2.1833300590515137, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7323874235153198, + "num_tokens": 381292874.0, + "step": 14746 + }, + { + "epoch": 1.6194816604436635, + "grad_norm": 2.073890447616577, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6957870721817017, + "num_tokens": 381323286.0, + "step": 14747 + }, + { + "epoch": 1.619591478146277, + "grad_norm": 2.3466548919677734, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7313821315765381, + "num_tokens": 381344594.0, + "step": 14748 + }, + { + "epoch": 1.6197012958488908, + "grad_norm": 2.215144395828247, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7205881476402283, + "num_tokens": 381371462.0, + "step": 14749 + }, + { + "epoch": 1.6198111135515045, + "grad_norm": 2.296485185623169, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7004678249359131, + "num_tokens": 381397383.0, + "step": 14750 + }, + { + "epoch": 1.6199209312541183, + "grad_norm": 2.0585386753082275, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7216882705688477, + "num_tokens": 381426112.0, + "step": 14751 + }, + { + "epoch": 1.6200307489567318, + "grad_norm": 2.2843196392059326, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7114197015762329, + "num_tokens": 381452303.0, + "step": 14752 + }, + { + "epoch": 1.6201405666593454, + "grad_norm": 2.3692331314086914, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.70975261926651, + "num_tokens": 381476583.0, + "step": 14753 + }, + { + "epoch": 1.6202503843619591, + "grad_norm": 2.1192827224731445, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7073042392730713, + "num_tokens": 381509030.0, + "step": 14754 + }, + { + "epoch": 1.6203602020645729, + "grad_norm": 2.496408462524414, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.750504732131958, + "num_tokens": 381529557.0, + "step": 14755 + }, + { + "epoch": 1.6204700197671866, + "grad_norm": 2.0531532764434814, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7012572288513184, + "num_tokens": 381558586.0, + "step": 14756 + }, + { + "epoch": 1.6205798374698002, + "grad_norm": 2.1082170009613037, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6970592737197876, + "num_tokens": 381587177.0, + "step": 14757 + }, + { + "epoch": 1.6206896551724137, + "grad_norm": 2.014256715774536, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7073370218276978, + "num_tokens": 381617923.0, + "step": 14758 + }, + { + "epoch": 1.6207994728750275, + "grad_norm": 2.1131742000579834, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7057266235351562, + "num_tokens": 381647988.0, + "step": 14759 + }, + { + "epoch": 1.6209092905776412, + "grad_norm": 2.3307981491088867, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7121239900588989, + "num_tokens": 381671221.0, + "step": 14760 + }, + { + "epoch": 1.6210191082802548, + "grad_norm": 2.245473861694336, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6992804408073425, + "num_tokens": 381698408.0, + "step": 14761 + }, + { + "epoch": 1.6211289259828683, + "grad_norm": 2.657453775405884, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7272589206695557, + "num_tokens": 381717742.0, + "step": 14762 + }, + { + "epoch": 1.621238743685482, + "grad_norm": 2.087313652038574, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7242400646209717, + "num_tokens": 381748107.0, + "step": 14763 + }, + { + "epoch": 1.6213485613880958, + "grad_norm": 2.3717334270477295, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7296165227890015, + "num_tokens": 381770856.0, + "step": 14764 + }, + { + "epoch": 1.6214583790907096, + "grad_norm": 2.0680911540985107, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7187964916229248, + "num_tokens": 381799450.0, + "step": 14765 + }, + { + "epoch": 1.621568196793323, + "grad_norm": 2.204923152923584, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7126022577285767, + "num_tokens": 381825078.0, + "step": 14766 + }, + { + "epoch": 1.6216780144959366, + "grad_norm": 2.154022693634033, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.709902822971344, + "num_tokens": 381853370.0, + "step": 14767 + }, + { + "epoch": 1.6217878321985504, + "grad_norm": 2.0923590660095215, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7247200012207031, + "num_tokens": 381880615.0, + "step": 14768 + }, + { + "epoch": 1.6218976499011641, + "grad_norm": 2.3541247844696045, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7178001999855042, + "num_tokens": 381904268.0, + "step": 14769 + }, + { + "epoch": 1.6220074676037777, + "grad_norm": 2.271373748779297, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7133803367614746, + "num_tokens": 381928517.0, + "step": 14770 + }, + { + "epoch": 1.6221172853063914, + "grad_norm": 2.4180150032043457, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7180202603340149, + "num_tokens": 381951430.0, + "step": 14771 + }, + { + "epoch": 1.622227103009005, + "grad_norm": 2.2743802070617676, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7212939858436584, + "num_tokens": 381976894.0, + "step": 14772 + }, + { + "epoch": 1.6223369207116187, + "grad_norm": 2.303098440170288, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7372255921363831, + "num_tokens": 382002349.0, + "step": 14773 + }, + { + "epoch": 1.6224467384142325, + "grad_norm": 2.4604008197784424, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7135610580444336, + "num_tokens": 382025309.0, + "step": 14774 + }, + { + "epoch": 1.622556556116846, + "grad_norm": 2.4188332557678223, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7118935585021973, + "num_tokens": 382047531.0, + "step": 14775 + }, + { + "epoch": 1.6226663738194596, + "grad_norm": 2.453265905380249, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6991325616836548, + "num_tokens": 382070872.0, + "step": 14776 + }, + { + "epoch": 1.6227761915220733, + "grad_norm": 2.0356087684631348, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7202401161193848, + "num_tokens": 382100230.0, + "step": 14777 + }, + { + "epoch": 1.622886009224687, + "grad_norm": 2.008141279220581, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7479404807090759, + "num_tokens": 382129822.0, + "step": 14778 + }, + { + "epoch": 1.6229958269273008, + "grad_norm": 2.2231814861297607, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7375315427780151, + "num_tokens": 382155243.0, + "step": 14779 + }, + { + "epoch": 1.6231056446299144, + "grad_norm": 2.261197328567505, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7369108200073242, + "num_tokens": 382179428.0, + "step": 14780 + }, + { + "epoch": 1.623215462332528, + "grad_norm": 2.217043876647949, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.71565181016922, + "num_tokens": 382205016.0, + "step": 14781 + }, + { + "epoch": 1.6233252800351416, + "grad_norm": 2.0008506774902344, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.6990740895271301, + "num_tokens": 382236986.0, + "step": 14782 + }, + { + "epoch": 1.6234350977377554, + "grad_norm": 2.031238555908203, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7054393291473389, + "num_tokens": 382268483.0, + "step": 14783 + }, + { + "epoch": 1.623544915440369, + "grad_norm": 2.320327043533325, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7159607410430908, + "num_tokens": 382294735.0, + "step": 14784 + }, + { + "epoch": 1.6236547331429827, + "grad_norm": 2.205073356628418, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7494328022003174, + "num_tokens": 382319364.0, + "step": 14785 + }, + { + "epoch": 1.6237645508455962, + "grad_norm": 2.2113401889801025, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7092791199684143, + "num_tokens": 382343621.0, + "step": 14786 + }, + { + "epoch": 1.62387436854821, + "grad_norm": 2.1960997581481934, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7095446586608887, + "num_tokens": 382371121.0, + "step": 14787 + }, + { + "epoch": 1.6239841862508237, + "grad_norm": 2.249805450439453, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7152395248413086, + "num_tokens": 382397503.0, + "step": 14788 + }, + { + "epoch": 1.6240940039534373, + "grad_norm": 2.1842763423919678, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6997673511505127, + "num_tokens": 382424574.0, + "step": 14789 + }, + { + "epoch": 1.6242038216560508, + "grad_norm": 2.11236310005188, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7121506929397583, + "num_tokens": 382450711.0, + "step": 14790 + }, + { + "epoch": 1.6243136393586646, + "grad_norm": 2.230008125305176, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7220553755760193, + "num_tokens": 382476821.0, + "step": 14791 + }, + { + "epoch": 1.6244234570612783, + "grad_norm": 2.1958398818969727, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7172322273254395, + "num_tokens": 382503988.0, + "step": 14792 + }, + { + "epoch": 1.624533274763892, + "grad_norm": 2.0201404094696045, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.711117148399353, + "num_tokens": 382533558.0, + "step": 14793 + }, + { + "epoch": 1.6246430924665056, + "grad_norm": 2.428798198699951, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7160547375679016, + "num_tokens": 382557276.0, + "step": 14794 + }, + { + "epoch": 1.6247529101691192, + "grad_norm": 2.3369436264038086, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7066495418548584, + "num_tokens": 382582157.0, + "step": 14795 + }, + { + "epoch": 1.624862727871733, + "grad_norm": 2.064154863357544, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6993533968925476, + "num_tokens": 382612789.0, + "step": 14796 + }, + { + "epoch": 1.6249725455743467, + "grad_norm": 1.9639700651168823, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7014747262001038, + "num_tokens": 382646381.0, + "step": 14797 + }, + { + "epoch": 1.6250823632769602, + "grad_norm": 2.316972494125366, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7174444794654846, + "num_tokens": 382670294.0, + "step": 14798 + }, + { + "epoch": 1.6251921809795737, + "grad_norm": 2.2445836067199707, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7289680242538452, + "num_tokens": 382694925.0, + "step": 14799 + }, + { + "epoch": 1.6253019986821875, + "grad_norm": 2.0689778327941895, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7036902904510498, + "num_tokens": 382724352.0, + "step": 14800 + }, + { + "epoch": 1.6254118163848013, + "grad_norm": 2.171779155731201, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7231258153915405, + "num_tokens": 382751825.0, + "step": 14801 + }, + { + "epoch": 1.625521634087415, + "grad_norm": 2.279569387435913, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7213692665100098, + "num_tokens": 382777700.0, + "step": 14802 + }, + { + "epoch": 1.6256314517900285, + "grad_norm": 2.356945276260376, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7146425247192383, + "num_tokens": 382801147.0, + "step": 14803 + }, + { + "epoch": 1.625741269492642, + "grad_norm": 2.421344041824341, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7082024812698364, + "num_tokens": 382823777.0, + "step": 14804 + }, + { + "epoch": 1.6258510871952558, + "grad_norm": 2.1344401836395264, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7147494554519653, + "num_tokens": 382851424.0, + "step": 14805 + }, + { + "epoch": 1.6259609048978696, + "grad_norm": 2.1710433959960938, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7325891852378845, + "num_tokens": 382879282.0, + "step": 14806 + }, + { + "epoch": 1.6260707226004834, + "grad_norm": 2.389880895614624, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7105470895767212, + "num_tokens": 382905436.0, + "step": 14807 + }, + { + "epoch": 1.6261805403030969, + "grad_norm": 2.169055461883545, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7149313688278198, + "num_tokens": 382933757.0, + "step": 14808 + }, + { + "epoch": 1.6262903580057104, + "grad_norm": 2.454655408859253, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7125592231750488, + "num_tokens": 382956239.0, + "step": 14809 + }, + { + "epoch": 1.6264001757083242, + "grad_norm": 2.009355306625366, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7210597991943359, + "num_tokens": 382988165.0, + "step": 14810 + }, + { + "epoch": 1.626509993410938, + "grad_norm": 2.3520472049713135, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7168161869049072, + "num_tokens": 383014012.0, + "step": 14811 + }, + { + "epoch": 1.6266198111135515, + "grad_norm": 2.3120338916778564, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7091078162193298, + "num_tokens": 383039558.0, + "step": 14812 + }, + { + "epoch": 1.626729628816165, + "grad_norm": 2.180196523666382, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7088596820831299, + "num_tokens": 383068124.0, + "step": 14813 + }, + { + "epoch": 1.6268394465187788, + "grad_norm": 2.3642525672912598, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.727652907371521, + "num_tokens": 383091127.0, + "step": 14814 + }, + { + "epoch": 1.6269492642213925, + "grad_norm": 2.2108445167541504, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7222646474838257, + "num_tokens": 383117348.0, + "step": 14815 + }, + { + "epoch": 1.6270590819240063, + "grad_norm": 2.1703267097473145, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.726828396320343, + "num_tokens": 383144594.0, + "step": 14816 + }, + { + "epoch": 1.6271688996266198, + "grad_norm": 2.2309083938598633, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7081741094589233, + "num_tokens": 383173211.0, + "step": 14817 + }, + { + "epoch": 1.6272787173292333, + "grad_norm": 2.3178324699401855, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7102029919624329, + "num_tokens": 383196574.0, + "step": 14818 + }, + { + "epoch": 1.627388535031847, + "grad_norm": 2.138280153274536, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7126557230949402, + "num_tokens": 383225233.0, + "step": 14819 + }, + { + "epoch": 1.6274983527344609, + "grad_norm": 2.466150999069214, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7272751331329346, + "num_tokens": 383246301.0, + "step": 14820 + }, + { + "epoch": 1.6276081704370746, + "grad_norm": 2.0616536140441895, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7117077112197876, + "num_tokens": 383276684.0, + "step": 14821 + }, + { + "epoch": 1.6277179881396882, + "grad_norm": 2.4520936012268066, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7348071336746216, + "num_tokens": 383298194.0, + "step": 14822 + }, + { + "epoch": 1.6278278058423017, + "grad_norm": 2.4099631309509277, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7489620447158813, + "num_tokens": 383319361.0, + "step": 14823 + }, + { + "epoch": 1.6279376235449154, + "grad_norm": 1.9572994709014893, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7247942686080933, + "num_tokens": 383349452.0, + "step": 14824 + }, + { + "epoch": 1.6280474412475292, + "grad_norm": 2.2207746505737305, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7082564830780029, + "num_tokens": 383380037.0, + "step": 14825 + }, + { + "epoch": 1.6281572589501427, + "grad_norm": 2.2006301879882812, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7382095456123352, + "num_tokens": 383405249.0, + "step": 14826 + }, + { + "epoch": 1.6282670766527563, + "grad_norm": 2.2209227085113525, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7034310102462769, + "num_tokens": 383430728.0, + "step": 14827 + }, + { + "epoch": 1.62837689435537, + "grad_norm": 2.02138352394104, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7164762020111084, + "num_tokens": 383460877.0, + "step": 14828 + }, + { + "epoch": 1.6284867120579838, + "grad_norm": 2.446892023086548, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7271345853805542, + "num_tokens": 383484181.0, + "step": 14829 + }, + { + "epoch": 1.6285965297605975, + "grad_norm": 2.072904109954834, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.71661376953125, + "num_tokens": 383513323.0, + "step": 14830 + }, + { + "epoch": 1.628706347463211, + "grad_norm": 2.1336069107055664, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7411000728607178, + "num_tokens": 383538472.0, + "step": 14831 + }, + { + "epoch": 1.6288161651658246, + "grad_norm": 2.257349967956543, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.733552098274231, + "num_tokens": 383563054.0, + "step": 14832 + }, + { + "epoch": 1.6289259828684384, + "grad_norm": 2.4452273845672607, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7151987552642822, + "num_tokens": 383586167.0, + "step": 14833 + }, + { + "epoch": 1.6290358005710521, + "grad_norm": 2.3056817054748535, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7400556802749634, + "num_tokens": 383609904.0, + "step": 14834 + }, + { + "epoch": 1.6291456182736657, + "grad_norm": 2.4019362926483154, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7226681709289551, + "num_tokens": 383631601.0, + "step": 14835 + }, + { + "epoch": 1.6292554359762794, + "grad_norm": 2.383187770843506, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7352817058563232, + "num_tokens": 383653670.0, + "step": 14836 + }, + { + "epoch": 1.629365253678893, + "grad_norm": 2.100273370742798, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7045193910598755, + "num_tokens": 383683590.0, + "step": 14837 + }, + { + "epoch": 1.6294750713815067, + "grad_norm": 1.978963017463684, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7326207160949707, + "num_tokens": 383716512.0, + "step": 14838 + }, + { + "epoch": 1.6295848890841205, + "grad_norm": 2.0812129974365234, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.6990525126457214, + "num_tokens": 383748673.0, + "step": 14839 + }, + { + "epoch": 1.629694706786734, + "grad_norm": 2.05137300491333, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7368482947349548, + "num_tokens": 383775985.0, + "step": 14840 + }, + { + "epoch": 1.6298045244893475, + "grad_norm": 2.1941823959350586, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7100088000297546, + "num_tokens": 383803403.0, + "step": 14841 + }, + { + "epoch": 1.6299143421919613, + "grad_norm": 2.153430700302124, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7405462265014648, + "num_tokens": 383829954.0, + "step": 14842 + }, + { + "epoch": 1.630024159894575, + "grad_norm": 2.3517847061157227, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7195282578468323, + "num_tokens": 383852254.0, + "step": 14843 + }, + { + "epoch": 1.6301339775971888, + "grad_norm": 2.5033576488494873, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7122753262519836, + "num_tokens": 383873300.0, + "step": 14844 + }, + { + "epoch": 1.6302437952998023, + "grad_norm": 2.2446634769439697, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7291499972343445, + "num_tokens": 383898671.0, + "step": 14845 + }, + { + "epoch": 1.6303536130024159, + "grad_norm": 2.262958526611328, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7212874293327332, + "num_tokens": 383924445.0, + "step": 14846 + }, + { + "epoch": 1.6304634307050296, + "grad_norm": 2.092334747314453, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7220221757888794, + "num_tokens": 383952449.0, + "step": 14847 + }, + { + "epoch": 1.6305732484076434, + "grad_norm": 2.37678599357605, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7225958108901978, + "num_tokens": 383976174.0, + "step": 14848 + }, + { + "epoch": 1.630683066110257, + "grad_norm": 2.1985278129577637, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7230874300003052, + "num_tokens": 384003191.0, + "step": 14849 + }, + { + "epoch": 1.6307928838128707, + "grad_norm": 2.1260104179382324, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7178328037261963, + "num_tokens": 384030783.0, + "step": 14850 + }, + { + "epoch": 1.6309027015154842, + "grad_norm": 2.0418078899383545, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6962595582008362, + "num_tokens": 384061920.0, + "step": 14851 + }, + { + "epoch": 1.631012519218098, + "grad_norm": 2.301745891571045, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7356750965118408, + "num_tokens": 384086460.0, + "step": 14852 + }, + { + "epoch": 1.6311223369207117, + "grad_norm": 2.349823474884033, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7309375405311584, + "num_tokens": 384110573.0, + "step": 14853 + }, + { + "epoch": 1.6312321546233253, + "grad_norm": 2.376882553100586, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7270828485488892, + "num_tokens": 384135820.0, + "step": 14854 + }, + { + "epoch": 1.6313419723259388, + "grad_norm": 2.1292974948883057, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7161628603935242, + "num_tokens": 384161870.0, + "step": 14855 + }, + { + "epoch": 1.6314517900285526, + "grad_norm": 2.3545970916748047, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7161767482757568, + "num_tokens": 384185167.0, + "step": 14856 + }, + { + "epoch": 1.6315616077311663, + "grad_norm": 2.249692916870117, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7353545427322388, + "num_tokens": 384209478.0, + "step": 14857 + }, + { + "epoch": 1.63167142543378, + "grad_norm": 2.1048691272735596, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7153685092926025, + "num_tokens": 384238523.0, + "step": 14858 + }, + { + "epoch": 1.6317812431363936, + "grad_norm": 2.1373941898345947, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7335233092308044, + "num_tokens": 384264608.0, + "step": 14859 + }, + { + "epoch": 1.6318910608390071, + "grad_norm": 2.176487445831299, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7176476716995239, + "num_tokens": 384292304.0, + "step": 14860 + }, + { + "epoch": 1.632000878541621, + "grad_norm": 2.1349611282348633, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6874558925628662, + "num_tokens": 384322313.0, + "step": 14861 + }, + { + "epoch": 1.6321106962442347, + "grad_norm": 2.0669522285461426, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7075881361961365, + "num_tokens": 384351098.0, + "step": 14862 + }, + { + "epoch": 1.6322205139468482, + "grad_norm": 2.273141860961914, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7188913822174072, + "num_tokens": 384376359.0, + "step": 14863 + }, + { + "epoch": 1.6323303316494617, + "grad_norm": 2.0762405395507812, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7193012237548828, + "num_tokens": 384404628.0, + "step": 14864 + }, + { + "epoch": 1.6324401493520755, + "grad_norm": 2.218878984451294, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.708919107913971, + "num_tokens": 384431541.0, + "step": 14865 + }, + { + "epoch": 1.6325499670546892, + "grad_norm": 2.1626734733581543, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7010166645050049, + "num_tokens": 384460588.0, + "step": 14866 + }, + { + "epoch": 1.632659784757303, + "grad_norm": 2.5148749351501465, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7337068915367126, + "num_tokens": 384481212.0, + "step": 14867 + }, + { + "epoch": 1.6327696024599165, + "grad_norm": 2.1415274143218994, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7454078197479248, + "num_tokens": 384507741.0, + "step": 14868 + }, + { + "epoch": 1.63287942016253, + "grad_norm": 2.158921718597412, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7247289419174194, + "num_tokens": 384532661.0, + "step": 14869 + }, + { + "epoch": 1.6329892378651438, + "grad_norm": 2.163511276245117, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.690765380859375, + "num_tokens": 384560894.0, + "step": 14870 + }, + { + "epoch": 1.6330990555677576, + "grad_norm": 2.111262559890747, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7059389352798462, + "num_tokens": 384588914.0, + "step": 14871 + }, + { + "epoch": 1.6332088732703713, + "grad_norm": 2.2561068534851074, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7069637775421143, + "num_tokens": 384616629.0, + "step": 14872 + }, + { + "epoch": 1.6333186909729849, + "grad_norm": 2.0755748748779297, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7077707052230835, + "num_tokens": 384647781.0, + "step": 14873 + }, + { + "epoch": 1.6334285086755984, + "grad_norm": 2.456430196762085, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7090645432472229, + "num_tokens": 384670700.0, + "step": 14874 + }, + { + "epoch": 1.6335383263782122, + "grad_norm": 2.5881388187408447, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7302275896072388, + "num_tokens": 384689528.0, + "step": 14875 + }, + { + "epoch": 1.633648144080826, + "grad_norm": 1.9330703020095825, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7009189128875732, + "num_tokens": 384723911.0, + "step": 14876 + }, + { + "epoch": 1.6337579617834395, + "grad_norm": 2.146543264389038, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7191620469093323, + "num_tokens": 384750167.0, + "step": 14877 + }, + { + "epoch": 1.633867779486053, + "grad_norm": 2.110994338989258, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7625933885574341, + "num_tokens": 384775381.0, + "step": 14878 + }, + { + "epoch": 1.6339775971886668, + "grad_norm": 2.4980597496032715, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7261921763420105, + "num_tokens": 384795819.0, + "step": 14879 + }, + { + "epoch": 1.6340874148912805, + "grad_norm": 2.313655376434326, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7173524498939514, + "num_tokens": 384818800.0, + "step": 14880 + }, + { + "epoch": 1.6341972325938943, + "grad_norm": 2.228184700012207, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7088090777397156, + "num_tokens": 384844704.0, + "step": 14881 + }, + { + "epoch": 1.6343070502965078, + "grad_norm": 2.258207321166992, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7183966636657715, + "num_tokens": 384871538.0, + "step": 14882 + }, + { + "epoch": 1.6344168679991213, + "grad_norm": 2.2345223426818848, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7050231695175171, + "num_tokens": 384899619.0, + "step": 14883 + }, + { + "epoch": 1.634526685701735, + "grad_norm": 2.206733465194702, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7187961339950562, + "num_tokens": 384927967.0, + "step": 14884 + }, + { + "epoch": 1.6346365034043489, + "grad_norm": 2.5156164169311523, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7412900328636169, + "num_tokens": 384950049.0, + "step": 14885 + }, + { + "epoch": 1.6347463211069626, + "grad_norm": 2.348548173904419, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7200108766555786, + "num_tokens": 384974721.0, + "step": 14886 + }, + { + "epoch": 1.6348561388095761, + "grad_norm": 2.549459934234619, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7263142466545105, + "num_tokens": 384997463.0, + "step": 14887 + }, + { + "epoch": 1.6349659565121897, + "grad_norm": 2.8446238040924072, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7321311235427856, + "num_tokens": 385015660.0, + "step": 14888 + }, + { + "epoch": 1.6350757742148034, + "grad_norm": 2.3357605934143066, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7311368584632874, + "num_tokens": 385040549.0, + "step": 14889 + }, + { + "epoch": 1.6351855919174172, + "grad_norm": 2.403416872024536, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7557708621025085, + "num_tokens": 385061746.0, + "step": 14890 + }, + { + "epoch": 1.6352954096200307, + "grad_norm": 2.5772032737731934, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7136452198028564, + "num_tokens": 385082821.0, + "step": 14891 + }, + { + "epoch": 1.6354052273226443, + "grad_norm": 1.95186448097229, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7053861618041992, + "num_tokens": 385115248.0, + "step": 14892 + }, + { + "epoch": 1.635515045025258, + "grad_norm": 2.3613553047180176, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7244426012039185, + "num_tokens": 385139636.0, + "step": 14893 + }, + { + "epoch": 1.6356248627278718, + "grad_norm": 2.2968909740448, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.716109037399292, + "num_tokens": 385164473.0, + "step": 14894 + }, + { + "epoch": 1.6357346804304855, + "grad_norm": 2.162524938583374, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6969982385635376, + "num_tokens": 385192452.0, + "step": 14895 + }, + { + "epoch": 1.635844498133099, + "grad_norm": 2.194856882095337, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7267760038375854, + "num_tokens": 385215938.0, + "step": 14896 + }, + { + "epoch": 1.6359543158357126, + "grad_norm": 2.2948508262634277, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7292611598968506, + "num_tokens": 385239699.0, + "step": 14897 + }, + { + "epoch": 1.6360641335383264, + "grad_norm": 2.0382614135742188, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6970887184143066, + "num_tokens": 385271485.0, + "step": 14898 + }, + { + "epoch": 1.6361739512409401, + "grad_norm": 2.5187911987304688, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7129586935043335, + "num_tokens": 385292962.0, + "step": 14899 + }, + { + "epoch": 1.6362837689435537, + "grad_norm": 2.285012722015381, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7215391993522644, + "num_tokens": 385318331.0, + "step": 14900 + }, + { + "epoch": 1.6363935866461674, + "grad_norm": 2.4146547317504883, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7166883945465088, + "num_tokens": 385342660.0, + "step": 14901 + }, + { + "epoch": 1.636503404348781, + "grad_norm": 2.1156680583953857, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7239286303520203, + "num_tokens": 385371735.0, + "step": 14902 + }, + { + "epoch": 1.6366132220513947, + "grad_norm": 2.2637312412261963, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7317507863044739, + "num_tokens": 385396048.0, + "step": 14903 + }, + { + "epoch": 1.6367230397540085, + "grad_norm": 2.2699942588806152, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.714497983455658, + "num_tokens": 385421690.0, + "step": 14904 + }, + { + "epoch": 1.636832857456622, + "grad_norm": 2.4197580814361572, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.718669593334198, + "num_tokens": 385444381.0, + "step": 14905 + }, + { + "epoch": 1.6369426751592355, + "grad_norm": 2.2163333892822266, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7262856960296631, + "num_tokens": 385470342.0, + "step": 14906 + }, + { + "epoch": 1.6370524928618493, + "grad_norm": 2.29327392578125, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7397549152374268, + "num_tokens": 385493926.0, + "step": 14907 + }, + { + "epoch": 1.637162310564463, + "grad_norm": 2.2741618156433105, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.723494291305542, + "num_tokens": 385520209.0, + "step": 14908 + }, + { + "epoch": 1.6372721282670768, + "grad_norm": 2.2685537338256836, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7191830277442932, + "num_tokens": 385546552.0, + "step": 14909 + }, + { + "epoch": 1.6373819459696903, + "grad_norm": 2.160055637359619, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7057304382324219, + "num_tokens": 385573161.0, + "step": 14910 + }, + { + "epoch": 1.6374917636723039, + "grad_norm": 2.260835647583008, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7597138285636902, + "num_tokens": 385597104.0, + "step": 14911 + }, + { + "epoch": 1.6376015813749176, + "grad_norm": 2.313300848007202, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.734724223613739, + "num_tokens": 385619874.0, + "step": 14912 + }, + { + "epoch": 1.6377113990775314, + "grad_norm": 2.3371963500976562, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.6973333358764648, + "num_tokens": 385646407.0, + "step": 14913 + }, + { + "epoch": 1.637821216780145, + "grad_norm": 2.3670341968536377, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7189512252807617, + "num_tokens": 385671319.0, + "step": 14914 + }, + { + "epoch": 1.6379310344827587, + "grad_norm": 2.156885862350464, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6983333826065063, + "num_tokens": 385701247.0, + "step": 14915 + }, + { + "epoch": 1.6380408521853722, + "grad_norm": 2.064777374267578, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6969271898269653, + "num_tokens": 385732322.0, + "step": 14916 + }, + { + "epoch": 1.638150669887986, + "grad_norm": 2.491323947906494, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.726648211479187, + "num_tokens": 385754702.0, + "step": 14917 + }, + { + "epoch": 1.6382604875905997, + "grad_norm": 2.02876353263855, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7062408328056335, + "num_tokens": 385785464.0, + "step": 14918 + }, + { + "epoch": 1.6383703052932133, + "grad_norm": 2.3134095668792725, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7246056795120239, + "num_tokens": 385810503.0, + "step": 14919 + }, + { + "epoch": 1.6384801229958268, + "grad_norm": 2.66748309135437, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7340453267097473, + "num_tokens": 385829758.0, + "step": 14920 + }, + { + "epoch": 1.6385899406984406, + "grad_norm": 2.660628080368042, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7259880304336548, + "num_tokens": 385849368.0, + "step": 14921 + }, + { + "epoch": 1.6386997584010543, + "grad_norm": 2.0994203090667725, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7137529850006104, + "num_tokens": 385876430.0, + "step": 14922 + }, + { + "epoch": 1.638809576103668, + "grad_norm": 2.3441717624664307, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7319909334182739, + "num_tokens": 385899573.0, + "step": 14923 + }, + { + "epoch": 1.6389193938062816, + "grad_norm": 2.3916168212890625, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7270830869674683, + "num_tokens": 385922626.0, + "step": 14924 + }, + { + "epoch": 1.6390292115088951, + "grad_norm": 2.212585926055908, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7267826795578003, + "num_tokens": 385947999.0, + "step": 14925 + }, + { + "epoch": 1.639139029211509, + "grad_norm": 2.612070322036743, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7256483435630798, + "num_tokens": 385968647.0, + "step": 14926 + }, + { + "epoch": 1.6392488469141226, + "grad_norm": 2.5980355739593506, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7375214099884033, + "num_tokens": 385988918.0, + "step": 14927 + }, + { + "epoch": 1.6393586646167362, + "grad_norm": 2.2459397315979004, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7233256101608276, + "num_tokens": 386014473.0, + "step": 14928 + }, + { + "epoch": 1.6394684823193497, + "grad_norm": 1.9768956899642944, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7193621397018433, + "num_tokens": 386044083.0, + "step": 14929 + }, + { + "epoch": 1.6395783000219635, + "grad_norm": 2.444209337234497, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.720154881477356, + "num_tokens": 386066503.0, + "step": 14930 + }, + { + "epoch": 1.6396881177245772, + "grad_norm": 1.8967069387435913, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7062144875526428, + "num_tokens": 386100803.0, + "step": 14931 + }, + { + "epoch": 1.639797935427191, + "grad_norm": 1.9748649597167969, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7079989910125732, + "num_tokens": 386133291.0, + "step": 14932 + }, + { + "epoch": 1.6399077531298045, + "grad_norm": 2.003838539123535, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7003734111785889, + "num_tokens": 386165398.0, + "step": 14933 + }, + { + "epoch": 1.640017570832418, + "grad_norm": 2.0258800983428955, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7063983678817749, + "num_tokens": 386196459.0, + "step": 14934 + }, + { + "epoch": 1.6401273885350318, + "grad_norm": 2.3649935722351074, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7073268890380859, + "num_tokens": 386221217.0, + "step": 14935 + }, + { + "epoch": 1.6402372062376456, + "grad_norm": 2.3797245025634766, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7234838008880615, + "num_tokens": 386244454.0, + "step": 14936 + }, + { + "epoch": 1.6403470239402593, + "grad_norm": 2.3439745903015137, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7142632603645325, + "num_tokens": 386271152.0, + "step": 14937 + }, + { + "epoch": 1.6404568416428729, + "grad_norm": 2.2356181144714355, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7084318995475769, + "num_tokens": 386296627.0, + "step": 14938 + }, + { + "epoch": 1.6405666593454864, + "grad_norm": 2.1707730293273926, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7289478182792664, + "num_tokens": 386322958.0, + "step": 14939 + }, + { + "epoch": 1.6406764770481002, + "grad_norm": 2.2988815307617188, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7333957552909851, + "num_tokens": 386345027.0, + "step": 14940 + }, + { + "epoch": 1.640786294750714, + "grad_norm": 2.1004858016967773, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7250707149505615, + "num_tokens": 386375754.0, + "step": 14941 + }, + { + "epoch": 1.6408961124533274, + "grad_norm": 2.2503116130828857, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.704863965511322, + "num_tokens": 386405499.0, + "step": 14942 + }, + { + "epoch": 1.641005930155941, + "grad_norm": 2.098445177078247, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7272354364395142, + "num_tokens": 386435079.0, + "step": 14943 + }, + { + "epoch": 1.6411157478585547, + "grad_norm": 2.181133508682251, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7146751284599304, + "num_tokens": 386461966.0, + "step": 14944 + }, + { + "epoch": 1.6412255655611685, + "grad_norm": 2.202632427215576, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7258579134941101, + "num_tokens": 386487258.0, + "step": 14945 + }, + { + "epoch": 1.6413353832637823, + "grad_norm": 2.4295430183410645, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7105907201766968, + "num_tokens": 386511094.0, + "step": 14946 + }, + { + "epoch": 1.6414452009663958, + "grad_norm": 2.1816751956939697, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7092889547348022, + "num_tokens": 386537644.0, + "step": 14947 + }, + { + "epoch": 1.6415550186690093, + "grad_norm": 2.3974123001098633, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7093405723571777, + "num_tokens": 386563414.0, + "step": 14948 + }, + { + "epoch": 1.641664836371623, + "grad_norm": 2.3042938709259033, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7061978578567505, + "num_tokens": 386588261.0, + "step": 14949 + }, + { + "epoch": 1.6417746540742368, + "grad_norm": 2.0992650985717773, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7251788973808289, + "num_tokens": 386618288.0, + "step": 14950 + }, + { + "epoch": 1.6418844717768504, + "grad_norm": 2.3821566104888916, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7129848003387451, + "num_tokens": 386641683.0, + "step": 14951 + }, + { + "epoch": 1.6419942894794641, + "grad_norm": 2.1692416667938232, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7345294952392578, + "num_tokens": 386669054.0, + "step": 14952 + }, + { + "epoch": 1.6421041071820777, + "grad_norm": 2.095404863357544, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7263364195823669, + "num_tokens": 386698654.0, + "step": 14953 + }, + { + "epoch": 1.6422139248846914, + "grad_norm": 2.2282493114471436, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7095969319343567, + "num_tokens": 386725386.0, + "step": 14954 + }, + { + "epoch": 1.6423237425873052, + "grad_norm": 2.301724433898926, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7221702337265015, + "num_tokens": 386748385.0, + "step": 14955 + }, + { + "epoch": 1.6424335602899187, + "grad_norm": 2.1665825843811035, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7258681654930115, + "num_tokens": 386775453.0, + "step": 14956 + }, + { + "epoch": 1.6425433779925322, + "grad_norm": 2.166944980621338, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7312014698982239, + "num_tokens": 386802118.0, + "step": 14957 + }, + { + "epoch": 1.642653195695146, + "grad_norm": 2.206064224243164, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6966449022293091, + "num_tokens": 386833173.0, + "step": 14958 + }, + { + "epoch": 1.6427630133977598, + "grad_norm": 2.209479331970215, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7075718641281128, + "num_tokens": 386858346.0, + "step": 14959 + }, + { + "epoch": 1.6428728311003735, + "grad_norm": 2.226288318634033, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.720611035823822, + "num_tokens": 386886388.0, + "step": 14960 + }, + { + "epoch": 1.642982648802987, + "grad_norm": 2.551466464996338, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7408018708229065, + "num_tokens": 386906007.0, + "step": 14961 + }, + { + "epoch": 1.6430924665056006, + "grad_norm": 2.1079113483428955, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7095901966094971, + "num_tokens": 386933428.0, + "step": 14962 + }, + { + "epoch": 1.6432022842082143, + "grad_norm": 2.1069834232330322, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7056812047958374, + "num_tokens": 386964853.0, + "step": 14963 + }, + { + "epoch": 1.643312101910828, + "grad_norm": 2.053354263305664, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7058811187744141, + "num_tokens": 386993593.0, + "step": 14964 + }, + { + "epoch": 1.6434219196134416, + "grad_norm": 2.3279738426208496, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7094441652297974, + "num_tokens": 387017013.0, + "step": 14965 + }, + { + "epoch": 1.6435317373160554, + "grad_norm": 2.458789348602295, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7401171922683716, + "num_tokens": 387037142.0, + "step": 14966 + }, + { + "epoch": 1.643641555018669, + "grad_norm": 2.4856648445129395, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7333598136901855, + "num_tokens": 387058297.0, + "step": 14967 + }, + { + "epoch": 1.6437513727212827, + "grad_norm": 2.4126017093658447, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7231128215789795, + "num_tokens": 387079291.0, + "step": 14968 + }, + { + "epoch": 1.6438611904238964, + "grad_norm": 2.525965690612793, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7121806144714355, + "num_tokens": 387100506.0, + "step": 14969 + }, + { + "epoch": 1.64397100812651, + "grad_norm": 2.178018808364868, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7067390084266663, + "num_tokens": 387128650.0, + "step": 14970 + }, + { + "epoch": 1.6440808258291235, + "grad_norm": 2.4384946823120117, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7230490446090698, + "num_tokens": 387151054.0, + "step": 14971 + }, + { + "epoch": 1.6441906435317373, + "grad_norm": 2.43009614944458, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7241001129150391, + "num_tokens": 387173317.0, + "step": 14972 + }, + { + "epoch": 1.644300461234351, + "grad_norm": 2.456156015396118, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7093103528022766, + "num_tokens": 387195398.0, + "step": 14973 + }, + { + "epoch": 1.6444102789369648, + "grad_norm": 2.5434560775756836, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7275363206863403, + "num_tokens": 387216806.0, + "step": 14974 + }, + { + "epoch": 1.6445200966395783, + "grad_norm": 2.1259613037109375, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7087822556495667, + "num_tokens": 387244637.0, + "step": 14975 + }, + { + "epoch": 1.6446299143421919, + "grad_norm": 2.2001304626464844, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.718789279460907, + "num_tokens": 387271443.0, + "step": 14976 + }, + { + "epoch": 1.6447397320448056, + "grad_norm": 2.113276481628418, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7047317028045654, + "num_tokens": 387300901.0, + "step": 14977 + }, + { + "epoch": 1.6448495497474194, + "grad_norm": 2.825787305831909, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7456551194190979, + "num_tokens": 387319865.0, + "step": 14978 + }, + { + "epoch": 1.644959367450033, + "grad_norm": 2.282150983810425, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7144324779510498, + "num_tokens": 387343059.0, + "step": 14979 + }, + { + "epoch": 1.6450691851526464, + "grad_norm": 2.1083037853240967, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.723691463470459, + "num_tokens": 387372396.0, + "step": 14980 + }, + { + "epoch": 1.6451790028552602, + "grad_norm": 1.8693493604660034, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6812224388122559, + "num_tokens": 387405565.0, + "step": 14981 + }, + { + "epoch": 1.645288820557874, + "grad_norm": 2.507873296737671, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7405184507369995, + "num_tokens": 387425517.0, + "step": 14982 + }, + { + "epoch": 1.6453986382604877, + "grad_norm": 2.547403335571289, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7424824833869934, + "num_tokens": 387445500.0, + "step": 14983 + }, + { + "epoch": 1.6455084559631012, + "grad_norm": 2.579979419708252, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7338257431983948, + "num_tokens": 387464702.0, + "step": 14984 + }, + { + "epoch": 1.6456182736657148, + "grad_norm": 2.257023811340332, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7104320526123047, + "num_tokens": 387489147.0, + "step": 14985 + }, + { + "epoch": 1.6457280913683285, + "grad_norm": 2.039263963699341, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.708836019039154, + "num_tokens": 387516359.0, + "step": 14986 + }, + { + "epoch": 1.6458379090709423, + "grad_norm": 2.3678476810455322, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7185971736907959, + "num_tokens": 387539169.0, + "step": 14987 + }, + { + "epoch": 1.645947726773556, + "grad_norm": 2.4476795196533203, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.744096040725708, + "num_tokens": 387561416.0, + "step": 14988 + }, + { + "epoch": 1.6460575444761696, + "grad_norm": 2.1560728549957275, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6950024366378784, + "num_tokens": 387588704.0, + "step": 14989 + }, + { + "epoch": 1.6461673621787831, + "grad_norm": 2.575197458267212, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7187764644622803, + "num_tokens": 387609671.0, + "step": 14990 + }, + { + "epoch": 1.6462771798813969, + "grad_norm": 1.9791730642318726, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.6993128657341003, + "num_tokens": 387644406.0, + "step": 14991 + }, + { + "epoch": 1.6463869975840106, + "grad_norm": 1.9546763896942139, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7197889089584351, + "num_tokens": 387675385.0, + "step": 14992 + }, + { + "epoch": 1.6464968152866242, + "grad_norm": 2.0688276290893555, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7302342653274536, + "num_tokens": 387702653.0, + "step": 14993 + }, + { + "epoch": 1.6466066329892377, + "grad_norm": 2.308445692062378, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7408549785614014, + "num_tokens": 387726744.0, + "step": 14994 + }, + { + "epoch": 1.6467164506918515, + "grad_norm": 2.3859050273895264, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7282106876373291, + "num_tokens": 387749051.0, + "step": 14995 + }, + { + "epoch": 1.6468262683944652, + "grad_norm": 2.2816548347473145, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7121522426605225, + "num_tokens": 387773399.0, + "step": 14996 + }, + { + "epoch": 1.646936086097079, + "grad_norm": 2.3184707164764404, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7243044972419739, + "num_tokens": 387798562.0, + "step": 14997 + }, + { + "epoch": 1.6470459037996925, + "grad_norm": 2.304264783859253, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7151190042495728, + "num_tokens": 387825123.0, + "step": 14998 + }, + { + "epoch": 1.647155721502306, + "grad_norm": 2.062678575515747, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.707530677318573, + "num_tokens": 387855785.0, + "step": 14999 + }, + { + "epoch": 1.6472655392049198, + "grad_norm": 2.1270036697387695, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.714152991771698, + "num_tokens": 387885037.0, + "step": 15000 + }, + { + "epoch": 1.6473753569075336, + "grad_norm": 2.7192039489746094, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7375026941299438, + "num_tokens": 387903513.0, + "step": 15001 + }, + { + "epoch": 1.6474851746101473, + "grad_norm": 1.9758048057556152, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7010442018508911, + "num_tokens": 387935721.0, + "step": 15002 + }, + { + "epoch": 1.6475949923127609, + "grad_norm": 2.176182985305786, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7252271175384521, + "num_tokens": 387960680.0, + "step": 15003 + }, + { + "epoch": 1.6477048100153744, + "grad_norm": 2.15427827835083, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7007302045822144, + "num_tokens": 387990376.0, + "step": 15004 + }, + { + "epoch": 1.6478146277179881, + "grad_norm": 2.305849552154541, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7311639785766602, + "num_tokens": 388013280.0, + "step": 15005 + }, + { + "epoch": 1.647924445420602, + "grad_norm": 2.104008436203003, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7040914297103882, + "num_tokens": 388044228.0, + "step": 15006 + }, + { + "epoch": 1.6480342631232154, + "grad_norm": 2.1549174785614014, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7197353839874268, + "num_tokens": 388069822.0, + "step": 15007 + }, + { + "epoch": 1.648144080825829, + "grad_norm": 1.9744372367858887, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7049230337142944, + "num_tokens": 388102749.0, + "step": 15008 + }, + { + "epoch": 1.6482538985284427, + "grad_norm": 2.606612205505371, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7372775673866272, + "num_tokens": 388122017.0, + "step": 15009 + }, + { + "epoch": 1.6483637162310565, + "grad_norm": 2.440788984298706, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7091047763824463, + "num_tokens": 388146361.0, + "step": 15010 + }, + { + "epoch": 1.6484735339336702, + "grad_norm": 2.3743834495544434, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.6980446577072144, + "num_tokens": 388170621.0, + "step": 15011 + }, + { + "epoch": 1.6485833516362838, + "grad_norm": 2.1832046508789062, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7236831784248352, + "num_tokens": 388198618.0, + "step": 15012 + }, + { + "epoch": 1.6486931693388973, + "grad_norm": 2.248976469039917, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7103468179702759, + "num_tokens": 388224203.0, + "step": 15013 + }, + { + "epoch": 1.648802987041511, + "grad_norm": 2.1195640563964844, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6956064105033875, + "num_tokens": 388256049.0, + "step": 15014 + }, + { + "epoch": 1.6489128047441248, + "grad_norm": 2.323249578475952, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7230336666107178, + "num_tokens": 388280606.0, + "step": 15015 + }, + { + "epoch": 1.6490226224467384, + "grad_norm": 2.4533112049102783, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7057682871818542, + "num_tokens": 388305205.0, + "step": 15016 + }, + { + "epoch": 1.6491324401493521, + "grad_norm": 2.1259777545928955, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7152925133705139, + "num_tokens": 388335252.0, + "step": 15017 + }, + { + "epoch": 1.6492422578519657, + "grad_norm": 2.5187973976135254, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7305237054824829, + "num_tokens": 388358098.0, + "step": 15018 + }, + { + "epoch": 1.6493520755545794, + "grad_norm": 2.279928684234619, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7449846267700195, + "num_tokens": 388381800.0, + "step": 15019 + }, + { + "epoch": 1.6494618932571932, + "grad_norm": 2.265639543533325, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7054134607315063, + "num_tokens": 388406651.0, + "step": 15020 + }, + { + "epoch": 1.6495717109598067, + "grad_norm": 2.3554534912109375, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7484563589096069, + "num_tokens": 388431031.0, + "step": 15021 + }, + { + "epoch": 1.6496815286624202, + "grad_norm": 2.151779890060425, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.7029335498809814, + "num_tokens": 388459428.0, + "step": 15022 + }, + { + "epoch": 1.649791346365034, + "grad_norm": 2.1013879776000977, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7292925119400024, + "num_tokens": 388488077.0, + "step": 15023 + }, + { + "epoch": 1.6499011640676478, + "grad_norm": 2.141936779022217, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7214999198913574, + "num_tokens": 388516556.0, + "step": 15024 + }, + { + "epoch": 1.6500109817702615, + "grad_norm": 2.387707233428955, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7075828909873962, + "num_tokens": 388542273.0, + "step": 15025 + }, + { + "epoch": 1.650120799472875, + "grad_norm": 2.197763442993164, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7223718166351318, + "num_tokens": 388568517.0, + "step": 15026 + }, + { + "epoch": 1.6502306171754886, + "grad_norm": 2.347008228302002, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7248514294624329, + "num_tokens": 388590410.0, + "step": 15027 + }, + { + "epoch": 1.6503404348781023, + "grad_norm": 2.2605926990509033, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7466256022453308, + "num_tokens": 388614421.0, + "step": 15028 + }, + { + "epoch": 1.650450252580716, + "grad_norm": 1.9992960691452026, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.734285831451416, + "num_tokens": 388643445.0, + "step": 15029 + }, + { + "epoch": 1.6505600702833296, + "grad_norm": 2.2312443256378174, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7007749080657959, + "num_tokens": 388669637.0, + "step": 15030 + }, + { + "epoch": 1.6506698879859434, + "grad_norm": 2.351646661758423, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7300337553024292, + "num_tokens": 388692316.0, + "step": 15031 + }, + { + "epoch": 1.650779705688557, + "grad_norm": 2.443434715270996, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7381433248519897, + "num_tokens": 388714361.0, + "step": 15032 + }, + { + "epoch": 1.6508895233911707, + "grad_norm": 2.137315511703491, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.713421106338501, + "num_tokens": 388742386.0, + "step": 15033 + }, + { + "epoch": 1.6509993410937844, + "grad_norm": 2.4542429447174072, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7114137411117554, + "num_tokens": 388765011.0, + "step": 15034 + }, + { + "epoch": 1.651109158796398, + "grad_norm": 2.3060829639434814, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7106326222419739, + "num_tokens": 388790939.0, + "step": 15035 + }, + { + "epoch": 1.6512189764990115, + "grad_norm": 2.2889785766601562, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7210648059844971, + "num_tokens": 388816139.0, + "step": 15036 + }, + { + "epoch": 1.6513287942016253, + "grad_norm": 2.35489559173584, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.719315767288208, + "num_tokens": 388841709.0, + "step": 15037 + }, + { + "epoch": 1.651438611904239, + "grad_norm": 2.3439698219299316, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7122510075569153, + "num_tokens": 388868760.0, + "step": 15038 + }, + { + "epoch": 1.6515484296068528, + "grad_norm": 2.397733211517334, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7293422222137451, + "num_tokens": 388889984.0, + "step": 15039 + }, + { + "epoch": 1.6516582473094663, + "grad_norm": 2.140568733215332, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7237364649772644, + "num_tokens": 388915538.0, + "step": 15040 + }, + { + "epoch": 1.6517680650120798, + "grad_norm": 2.499945640563965, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7409734725952148, + "num_tokens": 388935827.0, + "step": 15041 + }, + { + "epoch": 1.6518778827146936, + "grad_norm": 2.4047791957855225, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7161242961883545, + "num_tokens": 388959234.0, + "step": 15042 + }, + { + "epoch": 1.6519877004173074, + "grad_norm": 2.2533140182495117, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7106038331985474, + "num_tokens": 388985059.0, + "step": 15043 + }, + { + "epoch": 1.652097518119921, + "grad_norm": 2.4071969985961914, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.696864128112793, + "num_tokens": 389008956.0, + "step": 15044 + }, + { + "epoch": 1.6522073358225344, + "grad_norm": 2.2826271057128906, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.703760027885437, + "num_tokens": 389035428.0, + "step": 15045 + }, + { + "epoch": 1.6523171535251482, + "grad_norm": 2.319260597229004, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7175241708755493, + "num_tokens": 389060170.0, + "step": 15046 + }, + { + "epoch": 1.652426971227762, + "grad_norm": 2.1676666736602783, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7059764862060547, + "num_tokens": 389087000.0, + "step": 15047 + }, + { + "epoch": 1.6525367889303757, + "grad_norm": 2.372828483581543, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7044613361358643, + "num_tokens": 389110650.0, + "step": 15048 + }, + { + "epoch": 1.6526466066329892, + "grad_norm": 2.5085067749023438, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.728081226348877, + "num_tokens": 389131713.0, + "step": 15049 + }, + { + "epoch": 1.6527564243356028, + "grad_norm": 2.3315141201019287, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7012310028076172, + "num_tokens": 389155984.0, + "step": 15050 + }, + { + "epoch": 1.6528662420382165, + "grad_norm": 2.594442129135132, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7209333181381226, + "num_tokens": 389175024.0, + "step": 15051 + }, + { + "epoch": 1.6529760597408303, + "grad_norm": 2.369152545928955, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7233144044876099, + "num_tokens": 389198396.0, + "step": 15052 + }, + { + "epoch": 1.653085877443444, + "grad_norm": 2.208200216293335, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7171735763549805, + "num_tokens": 389224011.0, + "step": 15053 + }, + { + "epoch": 1.6531956951460576, + "grad_norm": 2.095444440841675, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7262295484542847, + "num_tokens": 389250356.0, + "step": 15054 + }, + { + "epoch": 1.653305512848671, + "grad_norm": 2.26444149017334, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7256970405578613, + "num_tokens": 389274008.0, + "step": 15055 + }, + { + "epoch": 1.6534153305512849, + "grad_norm": 2.397334098815918, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6884198188781738, + "num_tokens": 389297790.0, + "step": 15056 + }, + { + "epoch": 1.6535251482538986, + "grad_norm": 1.919696569442749, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7051038146018982, + "num_tokens": 389329161.0, + "step": 15057 + }, + { + "epoch": 1.6536349659565122, + "grad_norm": 2.1537539958953857, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7284737229347229, + "num_tokens": 389355768.0, + "step": 15058 + }, + { + "epoch": 1.6537447836591257, + "grad_norm": 1.9652159214019775, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7156607508659363, + "num_tokens": 389386013.0, + "step": 15059 + }, + { + "epoch": 1.6538546013617395, + "grad_norm": 2.2464449405670166, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7071437835693359, + "num_tokens": 389412209.0, + "step": 15060 + }, + { + "epoch": 1.6539644190643532, + "grad_norm": 1.8882864713668823, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7008386254310608, + "num_tokens": 389446399.0, + "step": 15061 + }, + { + "epoch": 1.654074236766967, + "grad_norm": 2.1513290405273438, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7581268548965454, + "num_tokens": 389471148.0, + "step": 15062 + }, + { + "epoch": 1.6541840544695805, + "grad_norm": 1.861725091934204, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.6982291340827942, + "num_tokens": 389508397.0, + "step": 15063 + }, + { + "epoch": 1.654293872172194, + "grad_norm": 2.2439498901367188, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7133104205131531, + "num_tokens": 389535562.0, + "step": 15064 + }, + { + "epoch": 1.6544036898748078, + "grad_norm": 2.191857099533081, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.736557126045227, + "num_tokens": 389563004.0, + "step": 15065 + }, + { + "epoch": 1.6545135075774215, + "grad_norm": 2.316366195678711, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7582080960273743, + "num_tokens": 389586088.0, + "step": 15066 + }, + { + "epoch": 1.6546233252800353, + "grad_norm": 2.1988070011138916, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7344644069671631, + "num_tokens": 389612021.0, + "step": 15067 + }, + { + "epoch": 1.6547331429826488, + "grad_norm": 1.9487965106964111, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7236981391906738, + "num_tokens": 389642765.0, + "step": 15068 + }, + { + "epoch": 1.6548429606852624, + "grad_norm": 2.277724504470825, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7396849989891052, + "num_tokens": 389665514.0, + "step": 15069 + }, + { + "epoch": 1.6549527783878761, + "grad_norm": 2.2092180252075195, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7098968029022217, + "num_tokens": 389692152.0, + "step": 15070 + }, + { + "epoch": 1.65506259609049, + "grad_norm": 2.3889429569244385, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7239276170730591, + "num_tokens": 389717493.0, + "step": 15071 + }, + { + "epoch": 1.6551724137931034, + "grad_norm": 2.360788583755493, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7104551792144775, + "num_tokens": 389741451.0, + "step": 15072 + }, + { + "epoch": 1.655282231495717, + "grad_norm": 2.3299567699432373, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7194388508796692, + "num_tokens": 389767007.0, + "step": 15073 + }, + { + "epoch": 1.6553920491983307, + "grad_norm": 2.6081385612487793, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7007128000259399, + "num_tokens": 389788434.0, + "step": 15074 + }, + { + "epoch": 1.6555018669009445, + "grad_norm": 2.155967950820923, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7033395767211914, + "num_tokens": 389815287.0, + "step": 15075 + }, + { + "epoch": 1.6556116846035582, + "grad_norm": 2.455690383911133, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7408038377761841, + "num_tokens": 389836107.0, + "step": 15076 + }, + { + "epoch": 1.6557215023061718, + "grad_norm": 2.233553647994995, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7289669513702393, + "num_tokens": 389860566.0, + "step": 15077 + }, + { + "epoch": 1.6558313200087853, + "grad_norm": 1.9481843709945679, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7303189039230347, + "num_tokens": 389891930.0, + "step": 15078 + }, + { + "epoch": 1.655941137711399, + "grad_norm": 2.3255295753479004, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7103089094161987, + "num_tokens": 389916670.0, + "step": 15079 + }, + { + "epoch": 1.6560509554140128, + "grad_norm": 2.086153507232666, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6985325813293457, + "num_tokens": 389945835.0, + "step": 15080 + }, + { + "epoch": 1.6561607731166264, + "grad_norm": 2.1365067958831787, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7556427121162415, + "num_tokens": 389972301.0, + "step": 15081 + }, + { + "epoch": 1.65627059081924, + "grad_norm": 2.6081883907318115, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7485733032226562, + "num_tokens": 389992002.0, + "step": 15082 + }, + { + "epoch": 1.6563804085218536, + "grad_norm": 2.306755781173706, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7008939981460571, + "num_tokens": 390018505.0, + "step": 15083 + }, + { + "epoch": 1.6564902262244674, + "grad_norm": 2.4595677852630615, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7304494380950928, + "num_tokens": 390040048.0, + "step": 15084 + }, + { + "epoch": 1.6566000439270812, + "grad_norm": 2.101379156112671, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.709073543548584, + "num_tokens": 390068920.0, + "step": 15085 + }, + { + "epoch": 1.6567098616296947, + "grad_norm": 2.348841905593872, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.6961181163787842, + "num_tokens": 390095080.0, + "step": 15086 + }, + { + "epoch": 1.6568196793323082, + "grad_norm": 2.0866456031799316, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7267836928367615, + "num_tokens": 390123902.0, + "step": 15087 + }, + { + "epoch": 1.656929497034922, + "grad_norm": 2.2063047885894775, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7050294876098633, + "num_tokens": 390149633.0, + "step": 15088 + }, + { + "epoch": 1.6570393147375357, + "grad_norm": 2.4729695320129395, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.7020676732063293, + "num_tokens": 390173020.0, + "step": 15089 + }, + { + "epoch": 1.6571491324401495, + "grad_norm": 2.31679105758667, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7133809328079224, + "num_tokens": 390198722.0, + "step": 15090 + }, + { + "epoch": 1.657258950142763, + "grad_norm": 2.224132776260376, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.729874312877655, + "num_tokens": 390226112.0, + "step": 15091 + }, + { + "epoch": 1.6573687678453766, + "grad_norm": 2.224076271057129, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7369153499603271, + "num_tokens": 390250826.0, + "step": 15092 + }, + { + "epoch": 1.6574785855479903, + "grad_norm": 2.3685247898101807, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7277078032493591, + "num_tokens": 390273521.0, + "step": 15093 + }, + { + "epoch": 1.657588403250604, + "grad_norm": 2.277031421661377, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7255249619483948, + "num_tokens": 390300333.0, + "step": 15094 + }, + { + "epoch": 1.6576982209532176, + "grad_norm": 2.187633991241455, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7493748068809509, + "num_tokens": 390327090.0, + "step": 15095 + }, + { + "epoch": 1.6578080386558314, + "grad_norm": 2.3581702709198, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7052252292633057, + "num_tokens": 390352112.0, + "step": 15096 + }, + { + "epoch": 1.657917856358445, + "grad_norm": 2.171293020248413, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7380315065383911, + "num_tokens": 390379128.0, + "step": 15097 + }, + { + "epoch": 1.6580276740610587, + "grad_norm": 2.372885227203369, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7192263603210449, + "num_tokens": 390402702.0, + "step": 15098 + }, + { + "epoch": 1.6581374917636724, + "grad_norm": 2.6079041957855225, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7090331315994263, + "num_tokens": 390425482.0, + "step": 15099 + }, + { + "epoch": 1.658247309466286, + "grad_norm": 2.3684000968933105, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7238485217094421, + "num_tokens": 390448680.0, + "step": 15100 + }, + { + "epoch": 1.6583571271688995, + "grad_norm": 2.157146692276001, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6973599195480347, + "num_tokens": 390475776.0, + "step": 15101 + }, + { + "epoch": 1.6584669448715132, + "grad_norm": 2.3114850521087646, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7198559045791626, + "num_tokens": 390499965.0, + "step": 15102 + }, + { + "epoch": 1.658576762574127, + "grad_norm": 2.279378890991211, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7284423112869263, + "num_tokens": 390525663.0, + "step": 15103 + }, + { + "epoch": 1.6586865802767408, + "grad_norm": 2.3508739471435547, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7075874209403992, + "num_tokens": 390549509.0, + "step": 15104 + }, + { + "epoch": 1.6587963979793543, + "grad_norm": 2.2082314491271973, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.6993670463562012, + "num_tokens": 390577154.0, + "step": 15105 + }, + { + "epoch": 1.6589062156819678, + "grad_norm": 2.0955443382263184, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7126709222793579, + "num_tokens": 390606950.0, + "step": 15106 + }, + { + "epoch": 1.6590160333845816, + "grad_norm": 2.3914170265197754, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.7064774036407471, + "num_tokens": 390630059.0, + "step": 15107 + }, + { + "epoch": 1.6591258510871953, + "grad_norm": 2.140695810317993, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7187448740005493, + "num_tokens": 390656855.0, + "step": 15108 + }, + { + "epoch": 1.6592356687898089, + "grad_norm": 2.393397808074951, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7449471950531006, + "num_tokens": 390681132.0, + "step": 15109 + }, + { + "epoch": 1.6593454864924224, + "grad_norm": 2.125269889831543, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7181223034858704, + "num_tokens": 390710201.0, + "step": 15110 + }, + { + "epoch": 1.6594553041950362, + "grad_norm": 2.2002601623535156, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7395808696746826, + "num_tokens": 390735305.0, + "step": 15111 + }, + { + "epoch": 1.65956512189765, + "grad_norm": 2.480409860610962, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7191590666770935, + "num_tokens": 390760313.0, + "step": 15112 + }, + { + "epoch": 1.6596749396002637, + "grad_norm": 2.3260531425476074, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7252089977264404, + "num_tokens": 390782847.0, + "step": 15113 + }, + { + "epoch": 1.6597847573028772, + "grad_norm": 2.313640594482422, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7157762050628662, + "num_tokens": 390808668.0, + "step": 15114 + }, + { + "epoch": 1.6598945750054908, + "grad_norm": 2.2698471546173096, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7112683057785034, + "num_tokens": 390835133.0, + "step": 15115 + }, + { + "epoch": 1.6600043927081045, + "grad_norm": 2.2143168449401855, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7279375791549683, + "num_tokens": 390860844.0, + "step": 15116 + }, + { + "epoch": 1.6601142104107183, + "grad_norm": 2.617886543273926, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.719927191734314, + "num_tokens": 390881447.0, + "step": 15117 + }, + { + "epoch": 1.660224028113332, + "grad_norm": 1.9900197982788086, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7164430618286133, + "num_tokens": 390914311.0, + "step": 15118 + }, + { + "epoch": 1.6603338458159456, + "grad_norm": 2.460195541381836, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7424283027648926, + "num_tokens": 390936130.0, + "step": 15119 + }, + { + "epoch": 1.660443663518559, + "grad_norm": 2.034254550933838, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7331634163856506, + "num_tokens": 390963900.0, + "step": 15120 + }, + { + "epoch": 1.6605534812211729, + "grad_norm": 2.154447555541992, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7419291138648987, + "num_tokens": 390988318.0, + "step": 15121 + }, + { + "epoch": 1.6606632989237866, + "grad_norm": 2.3311479091644287, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7174526453018188, + "num_tokens": 391013088.0, + "step": 15122 + }, + { + "epoch": 1.6607731166264001, + "grad_norm": 2.2856900691986084, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7132325172424316, + "num_tokens": 391038533.0, + "step": 15123 + }, + { + "epoch": 1.6608829343290137, + "grad_norm": 2.304141044616699, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7164931297302246, + "num_tokens": 391063810.0, + "step": 15124 + }, + { + "epoch": 1.6609927520316274, + "grad_norm": 2.2547171115875244, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7035806179046631, + "num_tokens": 391089542.0, + "step": 15125 + }, + { + "epoch": 1.6611025697342412, + "grad_norm": 2.2484116554260254, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7427905797958374, + "num_tokens": 391113833.0, + "step": 15126 + }, + { + "epoch": 1.661212387436855, + "grad_norm": 2.2728493213653564, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7279647588729858, + "num_tokens": 391138425.0, + "step": 15127 + }, + { + "epoch": 1.6613222051394685, + "grad_norm": 2.364201784133911, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7417190670967102, + "num_tokens": 391161315.0, + "step": 15128 + }, + { + "epoch": 1.661432022842082, + "grad_norm": 2.2494232654571533, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7292574644088745, + "num_tokens": 391185436.0, + "step": 15129 + }, + { + "epoch": 1.6615418405446958, + "grad_norm": 2.3629181385040283, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7250003218650818, + "num_tokens": 391209066.0, + "step": 15130 + }, + { + "epoch": 1.6616516582473095, + "grad_norm": 2.565715789794922, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7221775054931641, + "num_tokens": 391229470.0, + "step": 15131 + }, + { + "epoch": 1.661761475949923, + "grad_norm": 2.014218330383301, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7180176973342896, + "num_tokens": 391259599.0, + "step": 15132 + }, + { + "epoch": 1.6618712936525368, + "grad_norm": 2.289132833480835, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7292383909225464, + "num_tokens": 391284759.0, + "step": 15133 + }, + { + "epoch": 1.6619811113551504, + "grad_norm": 2.280735969543457, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7235801219940186, + "num_tokens": 391310964.0, + "step": 15134 + }, + { + "epoch": 1.6620909290577641, + "grad_norm": 2.2625999450683594, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.726992130279541, + "num_tokens": 391336718.0, + "step": 15135 + }, + { + "epoch": 1.6622007467603779, + "grad_norm": 2.193129301071167, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7074943780899048, + "num_tokens": 391363522.0, + "step": 15136 + }, + { + "epoch": 1.6623105644629914, + "grad_norm": 2.153353452682495, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.722938060760498, + "num_tokens": 391390172.0, + "step": 15137 + }, + { + "epoch": 1.662420382165605, + "grad_norm": 2.177030086517334, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7114646434783936, + "num_tokens": 391416941.0, + "step": 15138 + }, + { + "epoch": 1.6625301998682187, + "grad_norm": 2.0595929622650146, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7052656412124634, + "num_tokens": 391444599.0, + "step": 15139 + }, + { + "epoch": 1.6626400175708325, + "grad_norm": 2.1729962825775146, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7054060697555542, + "num_tokens": 391471679.0, + "step": 15140 + }, + { + "epoch": 1.6627498352734462, + "grad_norm": 2.0913236141204834, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7142882347106934, + "num_tokens": 391500427.0, + "step": 15141 + }, + { + "epoch": 1.6628596529760598, + "grad_norm": 2.1797592639923096, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7493422627449036, + "num_tokens": 391526531.0, + "step": 15142 + }, + { + "epoch": 1.6629694706786733, + "grad_norm": 2.1920430660247803, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7472430467605591, + "num_tokens": 391549471.0, + "step": 15143 + }, + { + "epoch": 1.663079288381287, + "grad_norm": 2.3128912448883057, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7279537916183472, + "num_tokens": 391572180.0, + "step": 15144 + }, + { + "epoch": 1.6631891060839008, + "grad_norm": 2.202895164489746, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7229623794555664, + "num_tokens": 391598234.0, + "step": 15145 + }, + { + "epoch": 1.6632989237865143, + "grad_norm": 2.083503007888794, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.70357745885849, + "num_tokens": 391627804.0, + "step": 15146 + }, + { + "epoch": 1.663408741489128, + "grad_norm": 2.6245157718658447, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7314052581787109, + "num_tokens": 391650198.0, + "step": 15147 + }, + { + "epoch": 1.6635185591917416, + "grad_norm": 2.1244096755981445, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7211424112319946, + "num_tokens": 391679974.0, + "step": 15148 + }, + { + "epoch": 1.6636283768943554, + "grad_norm": 2.229599714279175, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7155347466468811, + "num_tokens": 391706421.0, + "step": 15149 + }, + { + "epoch": 1.6637381945969691, + "grad_norm": 2.0166594982147217, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7143954038619995, + "num_tokens": 391737789.0, + "step": 15150 + }, + { + "epoch": 1.6638480122995827, + "grad_norm": 2.01389741897583, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7317651510238647, + "num_tokens": 391767656.0, + "step": 15151 + }, + { + "epoch": 1.6639578300021962, + "grad_norm": 2.3446290493011475, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7271201014518738, + "num_tokens": 391788356.0, + "step": 15152 + }, + { + "epoch": 1.66406764770481, + "grad_norm": 2.100205183029175, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7159385681152344, + "num_tokens": 391817753.0, + "step": 15153 + }, + { + "epoch": 1.6641774654074237, + "grad_norm": 2.269946575164795, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7223876714706421, + "num_tokens": 391842091.0, + "step": 15154 + }, + { + "epoch": 1.6642872831100375, + "grad_norm": 2.1580324172973633, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7067948579788208, + "num_tokens": 391870270.0, + "step": 15155 + }, + { + "epoch": 1.664397100812651, + "grad_norm": 1.9120101928710938, + "learning_rate": 1e-06, + "loss": 1.0696, + "mean_token_accuracy": 0.6803395748138428, + "num_tokens": 391903803.0, + "step": 15156 + }, + { + "epoch": 1.6645069185152646, + "grad_norm": 2.3439507484436035, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7276518940925598, + "num_tokens": 391926968.0, + "step": 15157 + }, + { + "epoch": 1.6646167362178783, + "grad_norm": 2.210373878479004, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7056710720062256, + "num_tokens": 391954503.0, + "step": 15158 + }, + { + "epoch": 1.664726553920492, + "grad_norm": 2.5248208045959473, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7188248634338379, + "num_tokens": 391976220.0, + "step": 15159 + }, + { + "epoch": 1.6648363716231056, + "grad_norm": 2.144516706466675, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.716019868850708, + "num_tokens": 392003237.0, + "step": 15160 + }, + { + "epoch": 1.6649461893257194, + "grad_norm": 2.0027036666870117, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.722081184387207, + "num_tokens": 392032988.0, + "step": 15161 + }, + { + "epoch": 1.665056007028333, + "grad_norm": 2.2404301166534424, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7383345365524292, + "num_tokens": 392057474.0, + "step": 15162 + }, + { + "epoch": 1.6651658247309467, + "grad_norm": 2.1952545642852783, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7200272083282471, + "num_tokens": 392084909.0, + "step": 15163 + }, + { + "epoch": 1.6652756424335604, + "grad_norm": 2.1924827098846436, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7062058448791504, + "num_tokens": 392111931.0, + "step": 15164 + }, + { + "epoch": 1.665385460136174, + "grad_norm": 2.1602349281311035, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7323507070541382, + "num_tokens": 392138413.0, + "step": 15165 + }, + { + "epoch": 1.6654952778387875, + "grad_norm": 2.234529733657837, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7254235148429871, + "num_tokens": 392164030.0, + "step": 15166 + }, + { + "epoch": 1.6656050955414012, + "grad_norm": 2.2184066772460938, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7185790538787842, + "num_tokens": 392189971.0, + "step": 15167 + }, + { + "epoch": 1.665714913244015, + "grad_norm": 2.1161818504333496, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7233473062515259, + "num_tokens": 392217168.0, + "step": 15168 + }, + { + "epoch": 1.6658247309466288, + "grad_norm": 2.1367204189300537, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7182531952857971, + "num_tokens": 392245303.0, + "step": 15169 + }, + { + "epoch": 1.6659345486492423, + "grad_norm": 2.1827051639556885, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7318320870399475, + "num_tokens": 392270791.0, + "step": 15170 + }, + { + "epoch": 1.6660443663518558, + "grad_norm": 2.305229663848877, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7166599035263062, + "num_tokens": 392295622.0, + "step": 15171 + }, + { + "epoch": 1.6661541840544696, + "grad_norm": 2.2125461101531982, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.710667610168457, + "num_tokens": 392322785.0, + "step": 15172 + }, + { + "epoch": 1.6662640017570833, + "grad_norm": 2.6017885208129883, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7288771867752075, + "num_tokens": 392343448.0, + "step": 15173 + }, + { + "epoch": 1.6663738194596969, + "grad_norm": 2.6691551208496094, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7334225177764893, + "num_tokens": 392362200.0, + "step": 15174 + }, + { + "epoch": 1.6664836371623104, + "grad_norm": 2.0712907314300537, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7182981371879578, + "num_tokens": 392390515.0, + "step": 15175 + }, + { + "epoch": 1.6665934548649242, + "grad_norm": 2.181572675704956, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7199759483337402, + "num_tokens": 392416679.0, + "step": 15176 + }, + { + "epoch": 1.666703272567538, + "grad_norm": 2.401731014251709, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7161853909492493, + "num_tokens": 392441347.0, + "step": 15177 + }, + { + "epoch": 1.6668130902701517, + "grad_norm": 2.0725648403167725, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7104051113128662, + "num_tokens": 392472060.0, + "step": 15178 + }, + { + "epoch": 1.6669229079727652, + "grad_norm": 2.3171122074127197, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7011300325393677, + "num_tokens": 392494595.0, + "step": 15179 + }, + { + "epoch": 1.6670327256753787, + "grad_norm": 2.307372808456421, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7226638793945312, + "num_tokens": 392521890.0, + "step": 15180 + }, + { + "epoch": 1.6671425433779925, + "grad_norm": 2.404040813446045, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7179237604141235, + "num_tokens": 392543064.0, + "step": 15181 + }, + { + "epoch": 1.6672523610806063, + "grad_norm": 2.360499620437622, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7298313975334167, + "num_tokens": 392565624.0, + "step": 15182 + }, + { + "epoch": 1.66736217878322, + "grad_norm": 2.3975179195404053, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7337958216667175, + "num_tokens": 392588081.0, + "step": 15183 + }, + { + "epoch": 1.6674719964858336, + "grad_norm": 2.1929755210876465, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7236218452453613, + "num_tokens": 392615830.0, + "step": 15184 + }, + { + "epoch": 1.667581814188447, + "grad_norm": 2.0669596195220947, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7344881296157837, + "num_tokens": 392643642.0, + "step": 15185 + }, + { + "epoch": 1.6676916318910608, + "grad_norm": 2.225189685821533, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7147079110145569, + "num_tokens": 392668842.0, + "step": 15186 + }, + { + "epoch": 1.6678014495936746, + "grad_norm": 2.3940858840942383, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7341693639755249, + "num_tokens": 392691727.0, + "step": 15187 + }, + { + "epoch": 1.6679112672962881, + "grad_norm": 2.1928906440734863, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7167845964431763, + "num_tokens": 392718206.0, + "step": 15188 + }, + { + "epoch": 1.6680210849989017, + "grad_norm": 1.9071859121322632, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7190170288085938, + "num_tokens": 392749720.0, + "step": 15189 + }, + { + "epoch": 1.6681309027015154, + "grad_norm": 2.2449557781219482, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7042137980461121, + "num_tokens": 392775450.0, + "step": 15190 + }, + { + "epoch": 1.6682407204041292, + "grad_norm": 2.1401724815368652, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7108010053634644, + "num_tokens": 392805725.0, + "step": 15191 + }, + { + "epoch": 1.668350538106743, + "grad_norm": 2.3212316036224365, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7120965719223022, + "num_tokens": 392828935.0, + "step": 15192 + }, + { + "epoch": 1.6684603558093565, + "grad_norm": 2.3204457759857178, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7334953546524048, + "num_tokens": 392852371.0, + "step": 15193 + }, + { + "epoch": 1.66857017351197, + "grad_norm": 2.2792468070983887, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.722567081451416, + "num_tokens": 392876515.0, + "step": 15194 + }, + { + "epoch": 1.6686799912145838, + "grad_norm": 2.021589756011963, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7121620178222656, + "num_tokens": 392909674.0, + "step": 15195 + }, + { + "epoch": 1.6687898089171975, + "grad_norm": 1.9919946193695068, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7049100399017334, + "num_tokens": 392942641.0, + "step": 15196 + }, + { + "epoch": 1.668899626619811, + "grad_norm": 1.8826016187667847, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7063337564468384, + "num_tokens": 392976520.0, + "step": 15197 + }, + { + "epoch": 1.6690094443224248, + "grad_norm": 2.440763473510742, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7106850743293762, + "num_tokens": 392999207.0, + "step": 15198 + }, + { + "epoch": 1.6691192620250384, + "grad_norm": 2.008146047592163, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7121233940124512, + "num_tokens": 393029558.0, + "step": 15199 + }, + { + "epoch": 1.669229079727652, + "grad_norm": 2.2530710697174072, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7533416152000427, + "num_tokens": 393051392.0, + "step": 15200 + }, + { + "epoch": 1.6693388974302659, + "grad_norm": 2.2381997108459473, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7478728890419006, + "num_tokens": 393076234.0, + "step": 15201 + }, + { + "epoch": 1.6694487151328794, + "grad_norm": 2.295311689376831, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7417572736740112, + "num_tokens": 393099892.0, + "step": 15202 + }, + { + "epoch": 1.669558532835493, + "grad_norm": 2.1134486198425293, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7128473520278931, + "num_tokens": 393128596.0, + "step": 15203 + }, + { + "epoch": 1.6696683505381067, + "grad_norm": 2.048757553100586, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7128844857215881, + "num_tokens": 393159301.0, + "step": 15204 + }, + { + "epoch": 1.6697781682407205, + "grad_norm": 2.4352941513061523, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7269391417503357, + "num_tokens": 393181250.0, + "step": 15205 + }, + { + "epoch": 1.6698879859433342, + "grad_norm": 2.1782283782958984, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7054738998413086, + "num_tokens": 393208502.0, + "step": 15206 + }, + { + "epoch": 1.6699978036459477, + "grad_norm": 2.1310107707977295, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7209455370903015, + "num_tokens": 393234328.0, + "step": 15207 + }, + { + "epoch": 1.6701076213485613, + "grad_norm": 2.373640537261963, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.721744179725647, + "num_tokens": 393257177.0, + "step": 15208 + }, + { + "epoch": 1.670217439051175, + "grad_norm": 2.286271333694458, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6930068731307983, + "num_tokens": 393283471.0, + "step": 15209 + }, + { + "epoch": 1.6703272567537888, + "grad_norm": 2.430419921875, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7337460517883301, + "num_tokens": 393305955.0, + "step": 15210 + }, + { + "epoch": 1.6704370744564023, + "grad_norm": 2.1850054264068604, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7220618724822998, + "num_tokens": 393333716.0, + "step": 15211 + }, + { + "epoch": 1.670546892159016, + "grad_norm": 2.0774707794189453, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7294285297393799, + "num_tokens": 393362875.0, + "step": 15212 + }, + { + "epoch": 1.6706567098616296, + "grad_norm": 2.477269411087036, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7247347831726074, + "num_tokens": 393383648.0, + "step": 15213 + }, + { + "epoch": 1.6707665275642434, + "grad_norm": 2.2162458896636963, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.715266227722168, + "num_tokens": 393410602.0, + "step": 15214 + }, + { + "epoch": 1.6708763452668571, + "grad_norm": 2.3654539585113525, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7272598743438721, + "num_tokens": 393433729.0, + "step": 15215 + }, + { + "epoch": 1.6709861629694707, + "grad_norm": 2.336566686630249, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.714592456817627, + "num_tokens": 393457823.0, + "step": 15216 + }, + { + "epoch": 1.6710959806720842, + "grad_norm": 2.0604374408721924, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.704861044883728, + "num_tokens": 393487977.0, + "step": 15217 + }, + { + "epoch": 1.671205798374698, + "grad_norm": 2.0885367393493652, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6897175312042236, + "num_tokens": 393520805.0, + "step": 15218 + }, + { + "epoch": 1.6713156160773117, + "grad_norm": 2.0408785343170166, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7083070874214172, + "num_tokens": 393550646.0, + "step": 15219 + }, + { + "epoch": 1.6714254337799255, + "grad_norm": 2.336031913757324, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7367457151412964, + "num_tokens": 393572080.0, + "step": 15220 + }, + { + "epoch": 1.671535251482539, + "grad_norm": 2.21130108833313, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7090688347816467, + "num_tokens": 393599908.0, + "step": 15221 + }, + { + "epoch": 1.6716450691851525, + "grad_norm": 2.3563385009765625, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7320282459259033, + "num_tokens": 393622182.0, + "step": 15222 + }, + { + "epoch": 1.6717548868877663, + "grad_norm": 2.1812727451324463, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7213718891143799, + "num_tokens": 393647899.0, + "step": 15223 + }, + { + "epoch": 1.67186470459038, + "grad_norm": 2.368715524673462, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7341791987419128, + "num_tokens": 393671432.0, + "step": 15224 + }, + { + "epoch": 1.6719745222929936, + "grad_norm": 2.2878565788269043, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7326468825340271, + "num_tokens": 393695873.0, + "step": 15225 + }, + { + "epoch": 1.6720843399956071, + "grad_norm": 2.3041436672210693, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7240803241729736, + "num_tokens": 393721029.0, + "step": 15226 + }, + { + "epoch": 1.6721941576982209, + "grad_norm": 2.048672914505005, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7164773941040039, + "num_tokens": 393751345.0, + "step": 15227 + }, + { + "epoch": 1.6723039754008346, + "grad_norm": 2.1931304931640625, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.712951123714447, + "num_tokens": 393777302.0, + "step": 15228 + }, + { + "epoch": 1.6724137931034484, + "grad_norm": 2.0946438312530518, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.6999547481536865, + "num_tokens": 393805391.0, + "step": 15229 + }, + { + "epoch": 1.672523610806062, + "grad_norm": 2.6392979621887207, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7088654041290283, + "num_tokens": 393825653.0, + "step": 15230 + }, + { + "epoch": 1.6726334285086755, + "grad_norm": 2.1684999465942383, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7043917179107666, + "num_tokens": 393852131.0, + "step": 15231 + }, + { + "epoch": 1.6727432462112892, + "grad_norm": 2.114682674407959, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7214112281799316, + "num_tokens": 393879563.0, + "step": 15232 + }, + { + "epoch": 1.672853063913903, + "grad_norm": 2.0565099716186523, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7099165916442871, + "num_tokens": 393908384.0, + "step": 15233 + }, + { + "epoch": 1.6729628816165167, + "grad_norm": 2.5335116386413574, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7189856767654419, + "num_tokens": 393929500.0, + "step": 15234 + }, + { + "epoch": 1.6730726993191303, + "grad_norm": 2.2611563205718994, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.747654139995575, + "num_tokens": 393952337.0, + "step": 15235 + }, + { + "epoch": 1.6731825170217438, + "grad_norm": 2.17852520942688, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.703954815864563, + "num_tokens": 393978801.0, + "step": 15236 + }, + { + "epoch": 1.6732923347243576, + "grad_norm": 2.303889036178589, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7130619287490845, + "num_tokens": 394006803.0, + "step": 15237 + }, + { + "epoch": 1.6734021524269713, + "grad_norm": 2.2394800186157227, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7109838724136353, + "num_tokens": 394031596.0, + "step": 15238 + }, + { + "epoch": 1.6735119701295849, + "grad_norm": 2.5750696659088135, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7390210628509521, + "num_tokens": 394053285.0, + "step": 15239 + }, + { + "epoch": 1.6736217878321984, + "grad_norm": 2.069498300552368, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7337040901184082, + "num_tokens": 394081476.0, + "step": 15240 + }, + { + "epoch": 1.6737316055348122, + "grad_norm": 2.0391883850097656, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7224672436714172, + "num_tokens": 394109698.0, + "step": 15241 + }, + { + "epoch": 1.673841423237426, + "grad_norm": 2.2577171325683594, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7548283338546753, + "num_tokens": 394133480.0, + "step": 15242 + }, + { + "epoch": 1.6739512409400397, + "grad_norm": 2.300579071044922, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7305735349655151, + "num_tokens": 394159412.0, + "step": 15243 + }, + { + "epoch": 1.6740610586426532, + "grad_norm": 2.215568780899048, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7128986716270447, + "num_tokens": 394185616.0, + "step": 15244 + }, + { + "epoch": 1.6741708763452667, + "grad_norm": 2.0668632984161377, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.730937123298645, + "num_tokens": 394214586.0, + "step": 15245 + }, + { + "epoch": 1.6742806940478805, + "grad_norm": 2.151611566543579, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7104542851448059, + "num_tokens": 394241550.0, + "step": 15246 + }, + { + "epoch": 1.6743905117504942, + "grad_norm": 2.326556921005249, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7326676845550537, + "num_tokens": 394266073.0, + "step": 15247 + }, + { + "epoch": 1.674500329453108, + "grad_norm": 2.3210325241088867, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7184762954711914, + "num_tokens": 394292595.0, + "step": 15248 + }, + { + "epoch": 1.6746101471557215, + "grad_norm": 2.059234857559204, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6949084401130676, + "num_tokens": 394323940.0, + "step": 15249 + }, + { + "epoch": 1.674719964858335, + "grad_norm": 2.3129169940948486, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.707534909248352, + "num_tokens": 394349290.0, + "step": 15250 + }, + { + "epoch": 1.6748297825609488, + "grad_norm": 2.6191914081573486, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.761370062828064, + "num_tokens": 394368900.0, + "step": 15251 + }, + { + "epoch": 1.6749396002635626, + "grad_norm": 2.388019561767578, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7249535322189331, + "num_tokens": 394390688.0, + "step": 15252 + }, + { + "epoch": 1.6750494179661761, + "grad_norm": 2.11072039604187, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7066362500190735, + "num_tokens": 394419230.0, + "step": 15253 + }, + { + "epoch": 1.6751592356687897, + "grad_norm": 1.9474767446517944, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6949037313461304, + "num_tokens": 394458588.0, + "step": 15254 + }, + { + "epoch": 1.6752690533714034, + "grad_norm": 2.2350337505340576, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7027187347412109, + "num_tokens": 394486642.0, + "step": 15255 + }, + { + "epoch": 1.6753788710740172, + "grad_norm": 2.2386415004730225, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7143654823303223, + "num_tokens": 394513361.0, + "step": 15256 + }, + { + "epoch": 1.675488688776631, + "grad_norm": 2.0956737995147705, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7195688486099243, + "num_tokens": 394539867.0, + "step": 15257 + }, + { + "epoch": 1.6755985064792445, + "grad_norm": 2.020780324935913, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7144114375114441, + "num_tokens": 394571484.0, + "step": 15258 + }, + { + "epoch": 1.675708324181858, + "grad_norm": 2.4000067710876465, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.752435564994812, + "num_tokens": 394592336.0, + "step": 15259 + }, + { + "epoch": 1.6758181418844718, + "grad_norm": 2.277825355529785, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7137866020202637, + "num_tokens": 394617743.0, + "step": 15260 + }, + { + "epoch": 1.6759279595870855, + "grad_norm": 2.284487009048462, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7281481027603149, + "num_tokens": 394642515.0, + "step": 15261 + }, + { + "epoch": 1.676037777289699, + "grad_norm": 2.325273275375366, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.714137077331543, + "num_tokens": 394665608.0, + "step": 15262 + }, + { + "epoch": 1.6761475949923128, + "grad_norm": 2.359696626663208, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7265059947967529, + "num_tokens": 394690613.0, + "step": 15263 + }, + { + "epoch": 1.6762574126949263, + "grad_norm": 2.1851892471313477, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.698119044303894, + "num_tokens": 394720380.0, + "step": 15264 + }, + { + "epoch": 1.67636723039754, + "grad_norm": 2.1199448108673096, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.71224045753479, + "num_tokens": 394750186.0, + "step": 15265 + }, + { + "epoch": 1.6764770481001539, + "grad_norm": 2.0814225673675537, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7180343866348267, + "num_tokens": 394780990.0, + "step": 15266 + }, + { + "epoch": 1.6765868658027674, + "grad_norm": 2.4198215007781982, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7428005933761597, + "num_tokens": 394802607.0, + "step": 15267 + }, + { + "epoch": 1.676696683505381, + "grad_norm": 2.4953103065490723, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7269150018692017, + "num_tokens": 394825914.0, + "step": 15268 + }, + { + "epoch": 1.6768065012079947, + "grad_norm": 2.0281543731689453, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7277653813362122, + "num_tokens": 394855125.0, + "step": 15269 + }, + { + "epoch": 1.6769163189106084, + "grad_norm": 2.229436159133911, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7002318501472473, + "num_tokens": 394883159.0, + "step": 15270 + }, + { + "epoch": 1.6770261366132222, + "grad_norm": 2.2487456798553467, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7033090591430664, + "num_tokens": 394909074.0, + "step": 15271 + }, + { + "epoch": 1.6771359543158357, + "grad_norm": 2.3222224712371826, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7183477878570557, + "num_tokens": 394932896.0, + "step": 15272 + }, + { + "epoch": 1.6772457720184493, + "grad_norm": 2.431182622909546, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7068801522254944, + "num_tokens": 394957311.0, + "step": 15273 + }, + { + "epoch": 1.677355589721063, + "grad_norm": 2.2468647956848145, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7046103477478027, + "num_tokens": 394981625.0, + "step": 15274 + }, + { + "epoch": 1.6774654074236768, + "grad_norm": 2.1166939735412598, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7150981426239014, + "num_tokens": 395011025.0, + "step": 15275 + }, + { + "epoch": 1.6775752251262903, + "grad_norm": 2.5682501792907715, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7173032760620117, + "num_tokens": 395031310.0, + "step": 15276 + }, + { + "epoch": 1.677685042828904, + "grad_norm": 1.8986339569091797, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7055126428604126, + "num_tokens": 395064109.0, + "step": 15277 + }, + { + "epoch": 1.6777948605315176, + "grad_norm": 2.1169450283050537, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7079153060913086, + "num_tokens": 395092223.0, + "step": 15278 + }, + { + "epoch": 1.6779046782341314, + "grad_norm": 2.0787272453308105, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7189361453056335, + "num_tokens": 395121767.0, + "step": 15279 + }, + { + "epoch": 1.6780144959367451, + "grad_norm": 2.3121390342712402, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7386026382446289, + "num_tokens": 395145604.0, + "step": 15280 + }, + { + "epoch": 1.6781243136393587, + "grad_norm": 2.142063617706299, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7187084555625916, + "num_tokens": 395172840.0, + "step": 15281 + }, + { + "epoch": 1.6782341313419722, + "grad_norm": 2.5728273391723633, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7725073099136353, + "num_tokens": 395192119.0, + "step": 15282 + }, + { + "epoch": 1.678343949044586, + "grad_norm": 2.131854772567749, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7049684524536133, + "num_tokens": 395220787.0, + "step": 15283 + }, + { + "epoch": 1.6784537667471997, + "grad_norm": 2.2471611499786377, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6981928944587708, + "num_tokens": 395246050.0, + "step": 15284 + }, + { + "epoch": 1.6785635844498135, + "grad_norm": 2.5003786087036133, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7284661531448364, + "num_tokens": 395267040.0, + "step": 15285 + }, + { + "epoch": 1.678673402152427, + "grad_norm": 2.2041447162628174, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7135351300239563, + "num_tokens": 395293994.0, + "step": 15286 + }, + { + "epoch": 1.6787832198550405, + "grad_norm": 2.350606679916382, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7234553098678589, + "num_tokens": 395319169.0, + "step": 15287 + }, + { + "epoch": 1.6788930375576543, + "grad_norm": 2.5690994262695312, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.715507447719574, + "num_tokens": 395341689.0, + "step": 15288 + }, + { + "epoch": 1.679002855260268, + "grad_norm": 2.131138324737549, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7025707960128784, + "num_tokens": 395369341.0, + "step": 15289 + }, + { + "epoch": 1.6791126729628816, + "grad_norm": 2.3240344524383545, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7202457189559937, + "num_tokens": 395391825.0, + "step": 15290 + }, + { + "epoch": 1.6792224906654951, + "grad_norm": 2.2500405311584473, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7164446115493774, + "num_tokens": 395416109.0, + "step": 15291 + }, + { + "epoch": 1.6793323083681089, + "grad_norm": 2.3487040996551514, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7028707265853882, + "num_tokens": 395440664.0, + "step": 15292 + }, + { + "epoch": 1.6794421260707226, + "grad_norm": 2.4687750339508057, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7427765130996704, + "num_tokens": 395460977.0, + "step": 15293 + }, + { + "epoch": 1.6795519437733364, + "grad_norm": 2.3841288089752197, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6954747438430786, + "num_tokens": 395486277.0, + "step": 15294 + }, + { + "epoch": 1.67966176147595, + "grad_norm": 2.207695722579956, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7234059572219849, + "num_tokens": 395511382.0, + "step": 15295 + }, + { + "epoch": 1.6797715791785635, + "grad_norm": 2.1881308555603027, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.739395797252655, + "num_tokens": 395535975.0, + "step": 15296 + }, + { + "epoch": 1.6798813968811772, + "grad_norm": 2.212740421295166, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7088690996170044, + "num_tokens": 395561926.0, + "step": 15297 + }, + { + "epoch": 1.679991214583791, + "grad_norm": 2.1228816509246826, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6947780847549438, + "num_tokens": 395589193.0, + "step": 15298 + }, + { + "epoch": 1.6801010322864047, + "grad_norm": 1.947333574295044, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7106490731239319, + "num_tokens": 395622093.0, + "step": 15299 + }, + { + "epoch": 1.6802108499890183, + "grad_norm": 2.0889177322387695, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7386142015457153, + "num_tokens": 395649011.0, + "step": 15300 + }, + { + "epoch": 1.6803206676916318, + "grad_norm": 2.306694507598877, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7290346622467041, + "num_tokens": 395673215.0, + "step": 15301 + }, + { + "epoch": 1.6804304853942456, + "grad_norm": 2.2155466079711914, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7056931853294373, + "num_tokens": 395699330.0, + "step": 15302 + }, + { + "epoch": 1.6805403030968593, + "grad_norm": 2.296422243118286, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7260265350341797, + "num_tokens": 395723199.0, + "step": 15303 + }, + { + "epoch": 1.6806501207994728, + "grad_norm": 2.2493948936462402, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.708719789981842, + "num_tokens": 395748689.0, + "step": 15304 + }, + { + "epoch": 1.6807599385020864, + "grad_norm": 2.2086501121520996, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7415739297866821, + "num_tokens": 395775038.0, + "step": 15305 + }, + { + "epoch": 1.6808697562047001, + "grad_norm": 2.022749662399292, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7490756511688232, + "num_tokens": 395803491.0, + "step": 15306 + }, + { + "epoch": 1.680979573907314, + "grad_norm": 2.155974864959717, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.732616662979126, + "num_tokens": 395829483.0, + "step": 15307 + }, + { + "epoch": 1.6810893916099277, + "grad_norm": 2.5392751693725586, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7198363542556763, + "num_tokens": 395849479.0, + "step": 15308 + }, + { + "epoch": 1.6811992093125412, + "grad_norm": 2.198847770690918, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6934312582015991, + "num_tokens": 395877741.0, + "step": 15309 + }, + { + "epoch": 1.6813090270151547, + "grad_norm": 2.3841781616210938, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7297735810279846, + "num_tokens": 395900303.0, + "step": 15310 + }, + { + "epoch": 1.6814188447177685, + "grad_norm": 2.1435089111328125, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7174780368804932, + "num_tokens": 395928543.0, + "step": 15311 + }, + { + "epoch": 1.6815286624203822, + "grad_norm": 2.1555302143096924, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7299680709838867, + "num_tokens": 395954915.0, + "step": 15312 + }, + { + "epoch": 1.681638480122996, + "grad_norm": 2.3457555770874023, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7479932308197021, + "num_tokens": 395977319.0, + "step": 15313 + }, + { + "epoch": 1.6817482978256095, + "grad_norm": 2.3224093914031982, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7165759801864624, + "num_tokens": 396000471.0, + "step": 15314 + }, + { + "epoch": 1.681858115528223, + "grad_norm": 2.0458953380584717, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.712749183177948, + "num_tokens": 396033624.0, + "step": 15315 + }, + { + "epoch": 1.6819679332308368, + "grad_norm": 2.2487189769744873, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7231765985488892, + "num_tokens": 396060174.0, + "step": 15316 + }, + { + "epoch": 1.6820777509334506, + "grad_norm": 2.59110426902771, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7113427519798279, + "num_tokens": 396081819.0, + "step": 15317 + }, + { + "epoch": 1.6821875686360641, + "grad_norm": 2.1494476795196533, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6793895959854126, + "num_tokens": 396112089.0, + "step": 15318 + }, + { + "epoch": 1.6822973863386776, + "grad_norm": 2.4346933364868164, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7297143936157227, + "num_tokens": 396134208.0, + "step": 15319 + }, + { + "epoch": 1.6824072040412914, + "grad_norm": 2.212671995162964, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6865382194519043, + "num_tokens": 396165284.0, + "step": 15320 + }, + { + "epoch": 1.6825170217439052, + "grad_norm": 2.5911991596221924, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7227287888526917, + "num_tokens": 396185911.0, + "step": 15321 + }, + { + "epoch": 1.682626839446519, + "grad_norm": 2.3620855808258057, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7202073335647583, + "num_tokens": 396208581.0, + "step": 15322 + }, + { + "epoch": 1.6827366571491325, + "grad_norm": 2.0286076068878174, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.739151120185852, + "num_tokens": 396235996.0, + "step": 15323 + }, + { + "epoch": 1.682846474851746, + "grad_norm": 2.3190202713012695, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7081872820854187, + "num_tokens": 396259865.0, + "step": 15324 + }, + { + "epoch": 1.6829562925543597, + "grad_norm": 2.1243889331817627, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7124840617179871, + "num_tokens": 396288236.0, + "step": 15325 + }, + { + "epoch": 1.6830661102569735, + "grad_norm": 2.27060866355896, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7290966510772705, + "num_tokens": 396310576.0, + "step": 15326 + }, + { + "epoch": 1.683175927959587, + "grad_norm": 2.356329917907715, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7200230956077576, + "num_tokens": 396334790.0, + "step": 15327 + }, + { + "epoch": 1.6832857456622008, + "grad_norm": 2.160050630569458, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7141924500465393, + "num_tokens": 396362040.0, + "step": 15328 + }, + { + "epoch": 1.6833955633648143, + "grad_norm": 2.4189417362213135, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.694245457649231, + "num_tokens": 396385734.0, + "step": 15329 + }, + { + "epoch": 1.683505381067428, + "grad_norm": 2.183037519454956, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7247387170791626, + "num_tokens": 396410754.0, + "step": 15330 + }, + { + "epoch": 1.6836151987700418, + "grad_norm": 2.041306495666504, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.6991442441940308, + "num_tokens": 396440831.0, + "step": 15331 + }, + { + "epoch": 1.6837250164726554, + "grad_norm": 2.269294500350952, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.6986991763114929, + "num_tokens": 396466952.0, + "step": 15332 + }, + { + "epoch": 1.683834834175269, + "grad_norm": 2.1356563568115234, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7238274812698364, + "num_tokens": 396496122.0, + "step": 15333 + }, + { + "epoch": 1.6839446518778827, + "grad_norm": 2.279242992401123, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7197436094284058, + "num_tokens": 396518979.0, + "step": 15334 + }, + { + "epoch": 1.6840544695804964, + "grad_norm": 2.1146278381347656, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7135790586471558, + "num_tokens": 396549076.0, + "step": 15335 + }, + { + "epoch": 1.6841642872831102, + "grad_norm": 2.0181033611297607, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7106893658638, + "num_tokens": 396579410.0, + "step": 15336 + }, + { + "epoch": 1.6842741049857237, + "grad_norm": 2.1894216537475586, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7245473861694336, + "num_tokens": 396605417.0, + "step": 15337 + }, + { + "epoch": 1.6843839226883373, + "grad_norm": 2.2923107147216797, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7228271961212158, + "num_tokens": 396629173.0, + "step": 15338 + }, + { + "epoch": 1.684493740390951, + "grad_norm": 2.1586527824401855, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7127362489700317, + "num_tokens": 396657129.0, + "step": 15339 + }, + { + "epoch": 1.6846035580935648, + "grad_norm": 2.462472677230835, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7372303605079651, + "num_tokens": 396678250.0, + "step": 15340 + }, + { + "epoch": 1.6847133757961783, + "grad_norm": 2.245328664779663, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.739516019821167, + "num_tokens": 396702668.0, + "step": 15341 + }, + { + "epoch": 1.684823193498792, + "grad_norm": 2.31001877784729, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7255635857582092, + "num_tokens": 396727824.0, + "step": 15342 + }, + { + "epoch": 1.6849330112014056, + "grad_norm": 2.134448528289795, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7177098393440247, + "num_tokens": 396756753.0, + "step": 15343 + }, + { + "epoch": 1.6850428289040194, + "grad_norm": 2.38132905960083, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7210261821746826, + "num_tokens": 396780860.0, + "step": 15344 + }, + { + "epoch": 1.685152646606633, + "grad_norm": 2.4375953674316406, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.70484459400177, + "num_tokens": 396804809.0, + "step": 15345 + }, + { + "epoch": 1.6852624643092466, + "grad_norm": 2.46278715133667, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7411876320838928, + "num_tokens": 396827140.0, + "step": 15346 + }, + { + "epoch": 1.6853722820118602, + "grad_norm": 2.331753969192505, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7150309085845947, + "num_tokens": 396851950.0, + "step": 15347 + }, + { + "epoch": 1.685482099714474, + "grad_norm": 2.2111289501190186, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7325390577316284, + "num_tokens": 396876087.0, + "step": 15348 + }, + { + "epoch": 1.6855919174170877, + "grad_norm": 2.0248613357543945, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7146416902542114, + "num_tokens": 396905161.0, + "step": 15349 + }, + { + "epoch": 1.6857017351197014, + "grad_norm": 2.379180431365967, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7210177183151245, + "num_tokens": 396928353.0, + "step": 15350 + }, + { + "epoch": 1.685811552822315, + "grad_norm": 2.2779672145843506, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7242928147315979, + "num_tokens": 396952381.0, + "step": 15351 + }, + { + "epoch": 1.6859213705249285, + "grad_norm": 2.1558825969696045, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6882156133651733, + "num_tokens": 396981463.0, + "step": 15352 + }, + { + "epoch": 1.6860311882275423, + "grad_norm": 2.169137716293335, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.690455436706543, + "num_tokens": 397008751.0, + "step": 15353 + }, + { + "epoch": 1.686141005930156, + "grad_norm": 2.505079984664917, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7216665744781494, + "num_tokens": 397030367.0, + "step": 15354 + }, + { + "epoch": 1.6862508236327696, + "grad_norm": 1.8794797658920288, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7149085998535156, + "num_tokens": 397063919.0, + "step": 15355 + }, + { + "epoch": 1.686360641335383, + "grad_norm": 2.1619346141815186, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.723393440246582, + "num_tokens": 397090232.0, + "step": 15356 + }, + { + "epoch": 1.6864704590379969, + "grad_norm": 2.381080389022827, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7316874265670776, + "num_tokens": 397111520.0, + "step": 15357 + }, + { + "epoch": 1.6865802767406106, + "grad_norm": 1.9242398738861084, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7005517482757568, + "num_tokens": 397144790.0, + "step": 15358 + }, + { + "epoch": 1.6866900944432244, + "grad_norm": 2.156190872192383, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7190066576004028, + "num_tokens": 397171378.0, + "step": 15359 + }, + { + "epoch": 1.686799912145838, + "grad_norm": 2.402768135070801, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7248899936676025, + "num_tokens": 397192336.0, + "step": 15360 + }, + { + "epoch": 1.6869097298484514, + "grad_norm": 2.2628870010375977, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7398403882980347, + "num_tokens": 397217011.0, + "step": 15361 + }, + { + "epoch": 1.6870195475510652, + "grad_norm": 2.5515382289886475, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7079343795776367, + "num_tokens": 397236937.0, + "step": 15362 + }, + { + "epoch": 1.687129365253679, + "grad_norm": 2.298211097717285, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7133256196975708, + "num_tokens": 397262375.0, + "step": 15363 + }, + { + "epoch": 1.6872391829562927, + "grad_norm": 2.236464500427246, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7286792993545532, + "num_tokens": 397287748.0, + "step": 15364 + }, + { + "epoch": 1.6873490006589063, + "grad_norm": 2.0455455780029297, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6998604536056519, + "num_tokens": 397318339.0, + "step": 15365 + }, + { + "epoch": 1.6874588183615198, + "grad_norm": 2.550111770629883, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7402772903442383, + "num_tokens": 397339436.0, + "step": 15366 + }, + { + "epoch": 1.6875686360641335, + "grad_norm": 2.2124478816986084, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7323702573776245, + "num_tokens": 397364289.0, + "step": 15367 + }, + { + "epoch": 1.6876784537667473, + "grad_norm": 2.1979188919067383, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.692023515701294, + "num_tokens": 397390891.0, + "step": 15368 + }, + { + "epoch": 1.6877882714693608, + "grad_norm": 2.2801432609558105, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.726682722568512, + "num_tokens": 397415729.0, + "step": 15369 + }, + { + "epoch": 1.6878980891719744, + "grad_norm": 2.0788300037384033, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.71546870470047, + "num_tokens": 397444613.0, + "step": 15370 + }, + { + "epoch": 1.6880079068745881, + "grad_norm": 2.239485502243042, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7244052886962891, + "num_tokens": 397470442.0, + "step": 15371 + }, + { + "epoch": 1.6881177245772019, + "grad_norm": 2.3580825328826904, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.734954833984375, + "num_tokens": 397493162.0, + "step": 15372 + }, + { + "epoch": 1.6882275422798156, + "grad_norm": 1.896620750427246, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7014935612678528, + "num_tokens": 397527873.0, + "step": 15373 + }, + { + "epoch": 1.6883373599824292, + "grad_norm": 2.3204338550567627, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.736339807510376, + "num_tokens": 397551717.0, + "step": 15374 + }, + { + "epoch": 1.6884471776850427, + "grad_norm": 2.2582755088806152, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7230880260467529, + "num_tokens": 397576725.0, + "step": 15375 + }, + { + "epoch": 1.6885569953876565, + "grad_norm": 2.4284348487854004, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7219380736351013, + "num_tokens": 397597859.0, + "step": 15376 + }, + { + "epoch": 1.6886668130902702, + "grad_norm": 2.326819896697998, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7210849523544312, + "num_tokens": 397620983.0, + "step": 15377 + }, + { + "epoch": 1.6887766307928838, + "grad_norm": 1.9389840364456177, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7274799346923828, + "num_tokens": 397654202.0, + "step": 15378 + }, + { + "epoch": 1.6888864484954975, + "grad_norm": 2.0168204307556152, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7353554964065552, + "num_tokens": 397684205.0, + "step": 15379 + }, + { + "epoch": 1.688996266198111, + "grad_norm": 2.357771873474121, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7223739624023438, + "num_tokens": 397709024.0, + "step": 15380 + }, + { + "epoch": 1.6891060839007248, + "grad_norm": 2.317817211151123, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7350032925605774, + "num_tokens": 397734698.0, + "step": 15381 + }, + { + "epoch": 1.6892159016033386, + "grad_norm": 2.4168083667755127, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7233293056488037, + "num_tokens": 397757587.0, + "step": 15382 + }, + { + "epoch": 1.689325719305952, + "grad_norm": 2.6249985694885254, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7360571026802063, + "num_tokens": 397776130.0, + "step": 15383 + }, + { + "epoch": 1.6894355370085656, + "grad_norm": 2.1863391399383545, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7239342331886292, + "num_tokens": 397801847.0, + "step": 15384 + }, + { + "epoch": 1.6895453547111794, + "grad_norm": 2.02583384513855, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7317162156105042, + "num_tokens": 397830107.0, + "step": 15385 + }, + { + "epoch": 1.6896551724137931, + "grad_norm": 2.209746837615967, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7090039253234863, + "num_tokens": 397856490.0, + "step": 15386 + }, + { + "epoch": 1.689764990116407, + "grad_norm": 2.0297257900238037, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7106962203979492, + "num_tokens": 397886156.0, + "step": 15387 + }, + { + "epoch": 1.6898748078190204, + "grad_norm": 2.2830398082733154, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7366511821746826, + "num_tokens": 397910565.0, + "step": 15388 + }, + { + "epoch": 1.689984625521634, + "grad_norm": 2.593186855316162, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7423409223556519, + "num_tokens": 397931254.0, + "step": 15389 + }, + { + "epoch": 1.6900944432242477, + "grad_norm": 2.0940234661102295, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7234748005867004, + "num_tokens": 397960044.0, + "step": 15390 + }, + { + "epoch": 1.6902042609268615, + "grad_norm": 2.305750608444214, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7078114748001099, + "num_tokens": 397988078.0, + "step": 15391 + }, + { + "epoch": 1.690314078629475, + "grad_norm": 2.2927587032318115, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7025725841522217, + "num_tokens": 398014185.0, + "step": 15392 + }, + { + "epoch": 1.6904238963320888, + "grad_norm": 2.125558853149414, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7198911309242249, + "num_tokens": 398041080.0, + "step": 15393 + }, + { + "epoch": 1.6905337140347023, + "grad_norm": 2.2756643295288086, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7253632545471191, + "num_tokens": 398066396.0, + "step": 15394 + }, + { + "epoch": 1.690643531737316, + "grad_norm": 2.3398241996765137, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7227402925491333, + "num_tokens": 398091231.0, + "step": 15395 + }, + { + "epoch": 1.6907533494399298, + "grad_norm": 2.33858060836792, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7200571298599243, + "num_tokens": 398113988.0, + "step": 15396 + }, + { + "epoch": 1.6908631671425434, + "grad_norm": 2.1619672775268555, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7053928971290588, + "num_tokens": 398139097.0, + "step": 15397 + }, + { + "epoch": 1.690972984845157, + "grad_norm": 2.217855930328369, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7332432270050049, + "num_tokens": 398166310.0, + "step": 15398 + }, + { + "epoch": 1.6910828025477707, + "grad_norm": 2.3276824951171875, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7313247919082642, + "num_tokens": 398188479.0, + "step": 15399 + }, + { + "epoch": 1.6911926202503844, + "grad_norm": 2.2036373615264893, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7220823764801025, + "num_tokens": 398214429.0, + "step": 15400 + }, + { + "epoch": 1.6913024379529982, + "grad_norm": 1.971209168434143, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7114193439483643, + "num_tokens": 398246406.0, + "step": 15401 + }, + { + "epoch": 1.6914122556556117, + "grad_norm": 2.134136915206909, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7169357538223267, + "num_tokens": 398273894.0, + "step": 15402 + }, + { + "epoch": 1.6915220733582252, + "grad_norm": 2.503498077392578, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7266599535942078, + "num_tokens": 398295936.0, + "step": 15403 + }, + { + "epoch": 1.691631891060839, + "grad_norm": 2.194605827331543, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7127223610877991, + "num_tokens": 398322470.0, + "step": 15404 + }, + { + "epoch": 1.6917417087634528, + "grad_norm": 2.3478832244873047, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.704152524471283, + "num_tokens": 398347501.0, + "step": 15405 + }, + { + "epoch": 1.6918515264660663, + "grad_norm": 2.110208034515381, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7165720462799072, + "num_tokens": 398377375.0, + "step": 15406 + }, + { + "epoch": 1.6919613441686798, + "grad_norm": 2.5489871501922607, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7297329306602478, + "num_tokens": 398397998.0, + "step": 15407 + }, + { + "epoch": 1.6920711618712936, + "grad_norm": 1.8589929342269897, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7129831910133362, + "num_tokens": 398433069.0, + "step": 15408 + }, + { + "epoch": 1.6921809795739073, + "grad_norm": 2.4431378841400146, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7314527034759521, + "num_tokens": 398455663.0, + "step": 15409 + }, + { + "epoch": 1.692290797276521, + "grad_norm": 2.3579044342041016, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7203103303909302, + "num_tokens": 398478737.0, + "step": 15410 + }, + { + "epoch": 1.6924006149791346, + "grad_norm": 2.092116355895996, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7293177843093872, + "num_tokens": 398506106.0, + "step": 15411 + }, + { + "epoch": 1.6925104326817482, + "grad_norm": 2.4334375858306885, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7489120960235596, + "num_tokens": 398526706.0, + "step": 15412 + }, + { + "epoch": 1.692620250384362, + "grad_norm": 2.0959126949310303, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.694021463394165, + "num_tokens": 398558623.0, + "step": 15413 + }, + { + "epoch": 1.6927300680869757, + "grad_norm": 2.5353164672851562, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7341950535774231, + "num_tokens": 398580334.0, + "step": 15414 + }, + { + "epoch": 1.6928398857895894, + "grad_norm": 2.5238144397735596, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7506837844848633, + "num_tokens": 398601379.0, + "step": 15415 + }, + { + "epoch": 1.692949703492203, + "grad_norm": 2.1430792808532715, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.741729199886322, + "num_tokens": 398629504.0, + "step": 15416 + }, + { + "epoch": 1.6930595211948165, + "grad_norm": 2.325045347213745, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6929947137832642, + "num_tokens": 398655967.0, + "step": 15417 + }, + { + "epoch": 1.6931693388974303, + "grad_norm": 2.4423747062683105, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7150537967681885, + "num_tokens": 398677510.0, + "step": 15418 + }, + { + "epoch": 1.693279156600044, + "grad_norm": 2.061356782913208, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7139077186584473, + "num_tokens": 398705020.0, + "step": 15419 + }, + { + "epoch": 1.6933889743026576, + "grad_norm": 2.389986276626587, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7197962999343872, + "num_tokens": 398730620.0, + "step": 15420 + }, + { + "epoch": 1.693498792005271, + "grad_norm": 2.3375184535980225, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.745047926902771, + "num_tokens": 398752590.0, + "step": 15421 + }, + { + "epoch": 1.6936086097078848, + "grad_norm": 2.2336554527282715, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7474956512451172, + "num_tokens": 398775873.0, + "step": 15422 + }, + { + "epoch": 1.6937184274104986, + "grad_norm": 2.3621466159820557, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7302873134613037, + "num_tokens": 398798566.0, + "step": 15423 + }, + { + "epoch": 1.6938282451131124, + "grad_norm": 2.3697216510772705, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7150185108184814, + "num_tokens": 398822146.0, + "step": 15424 + }, + { + "epoch": 1.693938062815726, + "grad_norm": 2.1103568077087402, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.6973577737808228, + "num_tokens": 398851007.0, + "step": 15425 + }, + { + "epoch": 1.6940478805183394, + "grad_norm": 2.3446097373962402, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7289830446243286, + "num_tokens": 398874032.0, + "step": 15426 + }, + { + "epoch": 1.6941576982209532, + "grad_norm": 1.9593855142593384, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7140020728111267, + "num_tokens": 398907392.0, + "step": 15427 + }, + { + "epoch": 1.694267515923567, + "grad_norm": 2.2865450382232666, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.7037512063980103, + "num_tokens": 398932488.0, + "step": 15428 + }, + { + "epoch": 1.6943773336261807, + "grad_norm": 2.2987799644470215, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7391005158424377, + "num_tokens": 398955318.0, + "step": 15429 + }, + { + "epoch": 1.6944871513287942, + "grad_norm": 2.733255624771118, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7429563999176025, + "num_tokens": 398973510.0, + "step": 15430 + }, + { + "epoch": 1.6945969690314078, + "grad_norm": 2.284987211227417, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7289427518844604, + "num_tokens": 398998123.0, + "step": 15431 + }, + { + "epoch": 1.6947067867340215, + "grad_norm": 2.350853681564331, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7300750613212585, + "num_tokens": 399021861.0, + "step": 15432 + }, + { + "epoch": 1.6948166044366353, + "grad_norm": 2.4813222885131836, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7372004389762878, + "num_tokens": 399042557.0, + "step": 15433 + }, + { + "epoch": 1.6949264221392488, + "grad_norm": 2.06952166557312, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7305035591125488, + "num_tokens": 399072586.0, + "step": 15434 + }, + { + "epoch": 1.6950362398418624, + "grad_norm": 2.540142297744751, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6991387009620667, + "num_tokens": 399099061.0, + "step": 15435 + }, + { + "epoch": 1.6951460575444761, + "grad_norm": 2.558666944503784, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.72337406873703, + "num_tokens": 399117783.0, + "step": 15436 + }, + { + "epoch": 1.6952558752470899, + "grad_norm": 2.338632106781006, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7227540612220764, + "num_tokens": 399141358.0, + "step": 15437 + }, + { + "epoch": 1.6953656929497036, + "grad_norm": 2.146179676055908, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.6971820592880249, + "num_tokens": 399169370.0, + "step": 15438 + }, + { + "epoch": 1.6954755106523172, + "grad_norm": 2.0246989727020264, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7266371250152588, + "num_tokens": 399197959.0, + "step": 15439 + }, + { + "epoch": 1.6955853283549307, + "grad_norm": 2.343006134033203, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.721264123916626, + "num_tokens": 399220990.0, + "step": 15440 + }, + { + "epoch": 1.6956951460575445, + "grad_norm": 2.258607864379883, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7093104124069214, + "num_tokens": 399246665.0, + "step": 15441 + }, + { + "epoch": 1.6958049637601582, + "grad_norm": 2.0145316123962402, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6967204213142395, + "num_tokens": 399276337.0, + "step": 15442 + }, + { + "epoch": 1.6959147814627717, + "grad_norm": 2.3697335720062256, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7285162210464478, + "num_tokens": 399298346.0, + "step": 15443 + }, + { + "epoch": 1.6960245991653855, + "grad_norm": 2.2454464435577393, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7099064588546753, + "num_tokens": 399323963.0, + "step": 15444 + }, + { + "epoch": 1.696134416867999, + "grad_norm": 2.564157724380493, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7243821024894714, + "num_tokens": 399344041.0, + "step": 15445 + }, + { + "epoch": 1.6962442345706128, + "grad_norm": 2.5241103172302246, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7180172204971313, + "num_tokens": 399365154.0, + "step": 15446 + }, + { + "epoch": 1.6963540522732266, + "grad_norm": 2.169764757156372, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7280097007751465, + "num_tokens": 399390134.0, + "step": 15447 + }, + { + "epoch": 1.69646386997584, + "grad_norm": 2.2567367553710938, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7227160930633545, + "num_tokens": 399414294.0, + "step": 15448 + }, + { + "epoch": 1.6965736876784536, + "grad_norm": 2.2892696857452393, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6939995884895325, + "num_tokens": 399442632.0, + "step": 15449 + }, + { + "epoch": 1.6966835053810674, + "grad_norm": 2.1514413356781006, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6874032020568848, + "num_tokens": 399471270.0, + "step": 15450 + }, + { + "epoch": 1.6967933230836811, + "grad_norm": 2.28304386138916, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.724416196346283, + "num_tokens": 399495752.0, + "step": 15451 + }, + { + "epoch": 1.696903140786295, + "grad_norm": 2.145291328430176, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7111634016036987, + "num_tokens": 399523581.0, + "step": 15452 + }, + { + "epoch": 1.6970129584889084, + "grad_norm": 2.176504611968994, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7098687887191772, + "num_tokens": 399550092.0, + "step": 15453 + }, + { + "epoch": 1.697122776191522, + "grad_norm": 1.9282448291778564, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7370350360870361, + "num_tokens": 399579278.0, + "step": 15454 + }, + { + "epoch": 1.6972325938941357, + "grad_norm": 1.8706014156341553, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7227380275726318, + "num_tokens": 399610000.0, + "step": 15455 + }, + { + "epoch": 1.6973424115967495, + "grad_norm": 2.169506072998047, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7362412214279175, + "num_tokens": 399635600.0, + "step": 15456 + }, + { + "epoch": 1.697452229299363, + "grad_norm": 2.3126697540283203, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7197046279907227, + "num_tokens": 399660664.0, + "step": 15457 + }, + { + "epoch": 1.6975620470019768, + "grad_norm": 2.312908411026001, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7035057544708252, + "num_tokens": 399687216.0, + "step": 15458 + }, + { + "epoch": 1.6976718647045903, + "grad_norm": 2.2501959800720215, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7006489038467407, + "num_tokens": 399711844.0, + "step": 15459 + }, + { + "epoch": 1.697781682407204, + "grad_norm": 1.9953553676605225, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6933824419975281, + "num_tokens": 399743060.0, + "step": 15460 + }, + { + "epoch": 1.6978915001098178, + "grad_norm": 2.163407325744629, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.706636905670166, + "num_tokens": 399769595.0, + "step": 15461 + }, + { + "epoch": 1.6980013178124314, + "grad_norm": 2.3126132488250732, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7175027132034302, + "num_tokens": 399794162.0, + "step": 15462 + }, + { + "epoch": 1.698111135515045, + "grad_norm": 2.286513566970825, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.6987048387527466, + "num_tokens": 399820012.0, + "step": 15463 + }, + { + "epoch": 1.6982209532176586, + "grad_norm": 2.1478726863861084, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7122926712036133, + "num_tokens": 399846566.0, + "step": 15464 + }, + { + "epoch": 1.6983307709202724, + "grad_norm": 2.1473517417907715, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7449985146522522, + "num_tokens": 399872193.0, + "step": 15465 + }, + { + "epoch": 1.6984405886228862, + "grad_norm": 2.1788082122802734, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7152572870254517, + "num_tokens": 399897730.0, + "step": 15466 + }, + { + "epoch": 1.6985504063254997, + "grad_norm": 2.156646728515625, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7070176601409912, + "num_tokens": 399926421.0, + "step": 15467 + }, + { + "epoch": 1.6986602240281132, + "grad_norm": 2.459048271179199, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7213022708892822, + "num_tokens": 399946819.0, + "step": 15468 + }, + { + "epoch": 1.698770041730727, + "grad_norm": 2.3671224117279053, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7296258211135864, + "num_tokens": 399970656.0, + "step": 15469 + }, + { + "epoch": 1.6988798594333407, + "grad_norm": 2.0861666202545166, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7204605937004089, + "num_tokens": 399998962.0, + "step": 15470 + }, + { + "epoch": 1.6989896771359543, + "grad_norm": 2.017195701599121, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7214422821998596, + "num_tokens": 400029421.0, + "step": 15471 + }, + { + "epoch": 1.6990994948385678, + "grad_norm": 2.3497121334075928, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.707588255405426, + "num_tokens": 400053607.0, + "step": 15472 + }, + { + "epoch": 1.6992093125411816, + "grad_norm": 2.0578935146331787, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7330900430679321, + "num_tokens": 400080343.0, + "step": 15473 + }, + { + "epoch": 1.6993191302437953, + "grad_norm": 2.0780961513519287, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7071925401687622, + "num_tokens": 400109965.0, + "step": 15474 + }, + { + "epoch": 1.699428947946409, + "grad_norm": 2.339240074157715, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.726033091545105, + "num_tokens": 400132559.0, + "step": 15475 + }, + { + "epoch": 1.6995387656490226, + "grad_norm": 2.0890090465545654, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7261282205581665, + "num_tokens": 400159520.0, + "step": 15476 + }, + { + "epoch": 1.6996485833516362, + "grad_norm": 2.3698008060455322, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7111604809761047, + "num_tokens": 400182574.0, + "step": 15477 + }, + { + "epoch": 1.69975840105425, + "grad_norm": 2.1224656105041504, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7124768495559692, + "num_tokens": 400212425.0, + "step": 15478 + }, + { + "epoch": 1.6998682187568637, + "grad_norm": 2.33979868888855, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7350614070892334, + "num_tokens": 400234508.0, + "step": 15479 + }, + { + "epoch": 1.6999780364594774, + "grad_norm": 2.120898485183716, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7267590165138245, + "num_tokens": 400261336.0, + "step": 15480 + }, + { + "epoch": 1.700087854162091, + "grad_norm": 2.334474563598633, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7183398008346558, + "num_tokens": 400284616.0, + "step": 15481 + }, + { + "epoch": 1.7001976718647045, + "grad_norm": 2.227170467376709, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7098406553268433, + "num_tokens": 400311134.0, + "step": 15482 + }, + { + "epoch": 1.7003074895673183, + "grad_norm": 2.3724172115325928, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7272859811782837, + "num_tokens": 400333231.0, + "step": 15483 + }, + { + "epoch": 1.700417307269932, + "grad_norm": 1.9831016063690186, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6934077739715576, + "num_tokens": 400366946.0, + "step": 15484 + }, + { + "epoch": 1.7005271249725455, + "grad_norm": 2.234973669052124, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7259302139282227, + "num_tokens": 400391020.0, + "step": 15485 + }, + { + "epoch": 1.700636942675159, + "grad_norm": 2.0736589431762695, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7138609886169434, + "num_tokens": 400420854.0, + "step": 15486 + }, + { + "epoch": 1.7007467603777728, + "grad_norm": 2.2699291706085205, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7301453948020935, + "num_tokens": 400445400.0, + "step": 15487 + }, + { + "epoch": 1.7008565780803866, + "grad_norm": 2.2393078804016113, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7035778760910034, + "num_tokens": 400472258.0, + "step": 15488 + }, + { + "epoch": 1.7009663957830004, + "grad_norm": 2.288736343383789, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.727195143699646, + "num_tokens": 400495777.0, + "step": 15489 + }, + { + "epoch": 1.7010762134856139, + "grad_norm": 2.3177390098571777, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7090088129043579, + "num_tokens": 400523789.0, + "step": 15490 + }, + { + "epoch": 1.7011860311882274, + "grad_norm": 2.3603506088256836, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7303300499916077, + "num_tokens": 400548564.0, + "step": 15491 + }, + { + "epoch": 1.7012958488908412, + "grad_norm": 2.26033616065979, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.710748016834259, + "num_tokens": 400572430.0, + "step": 15492 + }, + { + "epoch": 1.701405666593455, + "grad_norm": 2.1056275367736816, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7206249237060547, + "num_tokens": 400600268.0, + "step": 15493 + }, + { + "epoch": 1.7015154842960687, + "grad_norm": 2.3838624954223633, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7397757768630981, + "num_tokens": 400622347.0, + "step": 15494 + }, + { + "epoch": 1.7016253019986822, + "grad_norm": 2.7602181434631348, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7336771488189697, + "num_tokens": 400640295.0, + "step": 15495 + }, + { + "epoch": 1.7017351197012958, + "grad_norm": 2.7201790809631348, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7181838154792786, + "num_tokens": 400657890.0, + "step": 15496 + }, + { + "epoch": 1.7018449374039095, + "grad_norm": 2.2491040229797363, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7206799983978271, + "num_tokens": 400682174.0, + "step": 15497 + }, + { + "epoch": 1.7019547551065233, + "grad_norm": 2.2465708255767822, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7262651324272156, + "num_tokens": 400707258.0, + "step": 15498 + }, + { + "epoch": 1.7020645728091368, + "grad_norm": 2.1281533241271973, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6948659420013428, + "num_tokens": 400736267.0, + "step": 15499 + }, + { + "epoch": 1.7021743905117503, + "grad_norm": 2.533745050430298, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7171868085861206, + "num_tokens": 400758113.0, + "step": 15500 + }, + { + "epoch": 1.702284208214364, + "grad_norm": 2.56589937210083, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7377053499221802, + "num_tokens": 400779398.0, + "step": 15501 + }, + { + "epoch": 1.7023940259169779, + "grad_norm": 2.3168060779571533, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7294876575469971, + "num_tokens": 400803902.0, + "step": 15502 + }, + { + "epoch": 1.7025038436195916, + "grad_norm": 2.2288172245025635, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7180430293083191, + "num_tokens": 400828366.0, + "step": 15503 + }, + { + "epoch": 1.7026136613222052, + "grad_norm": 2.03250789642334, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7098003029823303, + "num_tokens": 400857298.0, + "step": 15504 + }, + { + "epoch": 1.7027234790248187, + "grad_norm": 2.574887752532959, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7263492345809937, + "num_tokens": 400878272.0, + "step": 15505 + }, + { + "epoch": 1.7028332967274324, + "grad_norm": 2.2758469581604004, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.717215895652771, + "num_tokens": 400904791.0, + "step": 15506 + }, + { + "epoch": 1.7029431144300462, + "grad_norm": 2.0308611392974854, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7429419755935669, + "num_tokens": 400931359.0, + "step": 15507 + }, + { + "epoch": 1.7030529321326597, + "grad_norm": 1.8584667444229126, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7068324685096741, + "num_tokens": 400968802.0, + "step": 15508 + }, + { + "epoch": 1.7031627498352735, + "grad_norm": 2.3418514728546143, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7142871022224426, + "num_tokens": 400993412.0, + "step": 15509 + }, + { + "epoch": 1.703272567537887, + "grad_norm": 2.4197325706481934, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.717911422252655, + "num_tokens": 401016039.0, + "step": 15510 + }, + { + "epoch": 1.7033823852405008, + "grad_norm": 2.382728099822998, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7312232255935669, + "num_tokens": 401037545.0, + "step": 15511 + }, + { + "epoch": 1.7034922029431145, + "grad_norm": 2.2023160457611084, + "learning_rate": 1e-06, + "loss": 1.04, + "mean_token_accuracy": 0.6821820735931396, + "num_tokens": 401064993.0, + "step": 15512 + }, + { + "epoch": 1.703602020645728, + "grad_norm": 2.1804847717285156, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7287672758102417, + "num_tokens": 401091084.0, + "step": 15513 + }, + { + "epoch": 1.7037118383483416, + "grad_norm": 1.9749960899353027, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6951923966407776, + "num_tokens": 401122728.0, + "step": 15514 + }, + { + "epoch": 1.7038216560509554, + "grad_norm": 2.6144845485687256, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7165296077728271, + "num_tokens": 401142675.0, + "step": 15515 + }, + { + "epoch": 1.7039314737535691, + "grad_norm": 2.113694190979004, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7067756056785583, + "num_tokens": 401170929.0, + "step": 15516 + }, + { + "epoch": 1.7040412914561829, + "grad_norm": 2.219644069671631, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.69953852891922, + "num_tokens": 401198278.0, + "step": 15517 + }, + { + "epoch": 1.7041511091587964, + "grad_norm": 2.148136615753174, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7198727130889893, + "num_tokens": 401225270.0, + "step": 15518 + }, + { + "epoch": 1.70426092686141, + "grad_norm": 2.1902384757995605, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7386633157730103, + "num_tokens": 401249881.0, + "step": 15519 + }, + { + "epoch": 1.7043707445640237, + "grad_norm": 2.3443002700805664, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7244366407394409, + "num_tokens": 401272146.0, + "step": 15520 + }, + { + "epoch": 1.7044805622666375, + "grad_norm": 2.4899535179138184, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7253230214118958, + "num_tokens": 401293937.0, + "step": 15521 + }, + { + "epoch": 1.704590379969251, + "grad_norm": 2.2568094730377197, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7354834079742432, + "num_tokens": 401317738.0, + "step": 15522 + }, + { + "epoch": 1.7047001976718648, + "grad_norm": 2.1220216751098633, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6836379766464233, + "num_tokens": 401348387.0, + "step": 15523 + }, + { + "epoch": 1.7048100153744783, + "grad_norm": 2.0021915435791016, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.719313383102417, + "num_tokens": 401381340.0, + "step": 15524 + }, + { + "epoch": 1.704919833077092, + "grad_norm": 2.4736056327819824, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7242907285690308, + "num_tokens": 401402595.0, + "step": 15525 + }, + { + "epoch": 1.7050296507797058, + "grad_norm": 2.345388412475586, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.730215311050415, + "num_tokens": 401425200.0, + "step": 15526 + }, + { + "epoch": 1.7051394684823193, + "grad_norm": 2.382067918777466, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7078335285186768, + "num_tokens": 401448836.0, + "step": 15527 + }, + { + "epoch": 1.7052492861849329, + "grad_norm": 2.597900390625, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7255633473396301, + "num_tokens": 401470107.0, + "step": 15528 + }, + { + "epoch": 1.7053591038875466, + "grad_norm": 2.1758601665496826, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7059710025787354, + "num_tokens": 401497024.0, + "step": 15529 + }, + { + "epoch": 1.7054689215901604, + "grad_norm": 1.99949049949646, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6822682619094849, + "num_tokens": 401529644.0, + "step": 15530 + }, + { + "epoch": 1.7055787392927741, + "grad_norm": 2.2409462928771973, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7157657146453857, + "num_tokens": 401553693.0, + "step": 15531 + }, + { + "epoch": 1.7056885569953877, + "grad_norm": 2.2882132530212402, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7150810360908508, + "num_tokens": 401577197.0, + "step": 15532 + }, + { + "epoch": 1.7057983746980012, + "grad_norm": 2.0734641551971436, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6957107186317444, + "num_tokens": 401608064.0, + "step": 15533 + }, + { + "epoch": 1.705908192400615, + "grad_norm": 2.577533721923828, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7426164150238037, + "num_tokens": 401628407.0, + "step": 15534 + }, + { + "epoch": 1.7060180101032287, + "grad_norm": 2.1282410621643066, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6900233030319214, + "num_tokens": 401656515.0, + "step": 15535 + }, + { + "epoch": 1.7061278278058423, + "grad_norm": 2.2832634449005127, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7225072383880615, + "num_tokens": 401681436.0, + "step": 15536 + }, + { + "epoch": 1.7062376455084558, + "grad_norm": 2.239224433898926, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7172932624816895, + "num_tokens": 401706904.0, + "step": 15537 + }, + { + "epoch": 1.7063474632110696, + "grad_norm": 2.3304617404937744, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7292006015777588, + "num_tokens": 401730977.0, + "step": 15538 + }, + { + "epoch": 1.7064572809136833, + "grad_norm": 2.0728416442871094, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7313984036445618, + "num_tokens": 401757877.0, + "step": 15539 + }, + { + "epoch": 1.706567098616297, + "grad_norm": 2.1549952030181885, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7263323068618774, + "num_tokens": 401784118.0, + "step": 15540 + }, + { + "epoch": 1.7066769163189106, + "grad_norm": 2.222148895263672, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7538042664527893, + "num_tokens": 401809143.0, + "step": 15541 + }, + { + "epoch": 1.7067867340215241, + "grad_norm": 2.5123510360717773, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7158983945846558, + "num_tokens": 401830948.0, + "step": 15542 + }, + { + "epoch": 1.706896551724138, + "grad_norm": 2.4042625427246094, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.731632649898529, + "num_tokens": 401853041.0, + "step": 15543 + }, + { + "epoch": 1.7070063694267517, + "grad_norm": 2.0946786403656006, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7096689939498901, + "num_tokens": 401883021.0, + "step": 15544 + }, + { + "epoch": 1.7071161871293654, + "grad_norm": 2.4350101947784424, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7155749797821045, + "num_tokens": 401905419.0, + "step": 15545 + }, + { + "epoch": 1.707226004831979, + "grad_norm": 2.0921311378479004, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7040184736251831, + "num_tokens": 401934862.0, + "step": 15546 + }, + { + "epoch": 1.7073358225345925, + "grad_norm": 1.9997001886367798, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6932741403579712, + "num_tokens": 401965412.0, + "step": 15547 + }, + { + "epoch": 1.7074456402372062, + "grad_norm": 2.3168373107910156, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7047329545021057, + "num_tokens": 401990219.0, + "step": 15548 + }, + { + "epoch": 1.70755545793982, + "grad_norm": 2.533609390258789, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7263000011444092, + "num_tokens": 402010596.0, + "step": 15549 + }, + { + "epoch": 1.7076652756424335, + "grad_norm": 2.3622066974639893, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7243548035621643, + "num_tokens": 402035555.0, + "step": 15550 + }, + { + "epoch": 1.707775093345047, + "grad_norm": 2.0937893390655518, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7059201002120972, + "num_tokens": 402064140.0, + "step": 15551 + }, + { + "epoch": 1.7078849110476608, + "grad_norm": 2.446261167526245, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7218326926231384, + "num_tokens": 402085461.0, + "step": 15552 + }, + { + "epoch": 1.7079947287502746, + "grad_norm": 2.106295585632324, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.724595844745636, + "num_tokens": 402113389.0, + "step": 15553 + }, + { + "epoch": 1.7081045464528883, + "grad_norm": 2.4542837142944336, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7209491729736328, + "num_tokens": 402134794.0, + "step": 15554 + }, + { + "epoch": 1.7082143641555019, + "grad_norm": 2.1934146881103516, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.703098475933075, + "num_tokens": 402162204.0, + "step": 15555 + }, + { + "epoch": 1.7083241818581154, + "grad_norm": 2.6455445289611816, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7344076633453369, + "num_tokens": 402180576.0, + "step": 15556 + }, + { + "epoch": 1.7084339995607292, + "grad_norm": 2.323535919189453, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7429611682891846, + "num_tokens": 402202089.0, + "step": 15557 + }, + { + "epoch": 1.708543817263343, + "grad_norm": 2.4863698482513428, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7423783540725708, + "num_tokens": 402221884.0, + "step": 15558 + }, + { + "epoch": 1.7086536349659565, + "grad_norm": 2.4303576946258545, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7176398038864136, + "num_tokens": 402243592.0, + "step": 15559 + }, + { + "epoch": 1.7087634526685702, + "grad_norm": 2.2661800384521484, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7038613557815552, + "num_tokens": 402268787.0, + "step": 15560 + }, + { + "epoch": 1.7088732703711838, + "grad_norm": 2.026916265487671, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7114253044128418, + "num_tokens": 402299378.0, + "step": 15561 + }, + { + "epoch": 1.7089830880737975, + "grad_norm": 2.155120372772217, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7292745113372803, + "num_tokens": 402324054.0, + "step": 15562 + }, + { + "epoch": 1.7090929057764113, + "grad_norm": 2.2578279972076416, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7254893779754639, + "num_tokens": 402346913.0, + "step": 15563 + }, + { + "epoch": 1.7092027234790248, + "grad_norm": 2.2296876907348633, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.70867919921875, + "num_tokens": 402373163.0, + "step": 15564 + }, + { + "epoch": 1.7093125411816383, + "grad_norm": 2.251925230026245, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7418423295021057, + "num_tokens": 402396775.0, + "step": 15565 + }, + { + "epoch": 1.709422358884252, + "grad_norm": 2.1205687522888184, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7215033173561096, + "num_tokens": 402426353.0, + "step": 15566 + }, + { + "epoch": 1.7095321765868658, + "grad_norm": 1.9481948614120483, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.6974360942840576, + "num_tokens": 402457270.0, + "step": 15567 + }, + { + "epoch": 1.7096419942894796, + "grad_norm": 2.207732677459717, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7156600952148438, + "num_tokens": 402483166.0, + "step": 15568 + }, + { + "epoch": 1.7097518119920931, + "grad_norm": 2.119929075241089, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.732597827911377, + "num_tokens": 402510246.0, + "step": 15569 + }, + { + "epoch": 1.7098616296947067, + "grad_norm": 1.9825291633605957, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7280868291854858, + "num_tokens": 402540081.0, + "step": 15570 + }, + { + "epoch": 1.7099714473973204, + "grad_norm": 2.014165163040161, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7032803297042847, + "num_tokens": 402570877.0, + "step": 15571 + }, + { + "epoch": 1.7100812650999342, + "grad_norm": 2.3446919918060303, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7302409410476685, + "num_tokens": 402594434.0, + "step": 15572 + }, + { + "epoch": 1.7101910828025477, + "grad_norm": 2.2611846923828125, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7265802025794983, + "num_tokens": 402618572.0, + "step": 15573 + }, + { + "epoch": 1.7103009005051615, + "grad_norm": 2.611912488937378, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7156051397323608, + "num_tokens": 402639406.0, + "step": 15574 + }, + { + "epoch": 1.710410718207775, + "grad_norm": 2.680936813354492, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7129485607147217, + "num_tokens": 402658305.0, + "step": 15575 + }, + { + "epoch": 1.7105205359103888, + "grad_norm": 2.002978801727295, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.6972720623016357, + "num_tokens": 402690186.0, + "step": 15576 + }, + { + "epoch": 1.7106303536130025, + "grad_norm": 2.0154128074645996, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6874433755874634, + "num_tokens": 402722651.0, + "step": 15577 + }, + { + "epoch": 1.710740171315616, + "grad_norm": 2.3777663707733154, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7394235134124756, + "num_tokens": 402746767.0, + "step": 15578 + }, + { + "epoch": 1.7108499890182296, + "grad_norm": 1.9886304140090942, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7273739576339722, + "num_tokens": 402777349.0, + "step": 15579 + }, + { + "epoch": 1.7109598067208434, + "grad_norm": 2.116990804672241, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7150235176086426, + "num_tokens": 402806854.0, + "step": 15580 + }, + { + "epoch": 1.7110696244234571, + "grad_norm": 2.089219808578491, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.705392599105835, + "num_tokens": 402836077.0, + "step": 15581 + }, + { + "epoch": 1.7111794421260709, + "grad_norm": 2.211606025695801, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7079482078552246, + "num_tokens": 402862507.0, + "step": 15582 + }, + { + "epoch": 1.7112892598286844, + "grad_norm": 1.9138295650482178, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7210739850997925, + "num_tokens": 402895496.0, + "step": 15583 + }, + { + "epoch": 1.711399077531298, + "grad_norm": 2.157519578933716, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7161247730255127, + "num_tokens": 402922000.0, + "step": 15584 + }, + { + "epoch": 1.7115088952339117, + "grad_norm": 2.2142601013183594, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7137605547904968, + "num_tokens": 402946708.0, + "step": 15585 + }, + { + "epoch": 1.7116187129365255, + "grad_norm": 2.1568331718444824, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7145105004310608, + "num_tokens": 402974797.0, + "step": 15586 + }, + { + "epoch": 1.711728530639139, + "grad_norm": 2.243839740753174, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7239788174629211, + "num_tokens": 402999664.0, + "step": 15587 + }, + { + "epoch": 1.7118383483417525, + "grad_norm": 2.0865752696990967, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6960709691047668, + "num_tokens": 403031461.0, + "step": 15588 + }, + { + "epoch": 1.7119481660443663, + "grad_norm": 2.1599373817443848, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7269917726516724, + "num_tokens": 403058704.0, + "step": 15589 + }, + { + "epoch": 1.71205798374698, + "grad_norm": 2.0149850845336914, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7358537912368774, + "num_tokens": 403089446.0, + "step": 15590 + }, + { + "epoch": 1.7121678014495938, + "grad_norm": 2.287956714630127, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6973001956939697, + "num_tokens": 403116757.0, + "step": 15591 + }, + { + "epoch": 1.7122776191522073, + "grad_norm": 2.3135550022125244, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7254020571708679, + "num_tokens": 403138993.0, + "step": 15592 + }, + { + "epoch": 1.7123874368548209, + "grad_norm": 2.366543769836426, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7195674180984497, + "num_tokens": 403162916.0, + "step": 15593 + }, + { + "epoch": 1.7124972545574346, + "grad_norm": 2.0289387702941895, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7251799702644348, + "num_tokens": 403191582.0, + "step": 15594 + }, + { + "epoch": 1.7126070722600484, + "grad_norm": 2.149103879928589, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7060691714286804, + "num_tokens": 403218747.0, + "step": 15595 + }, + { + "epoch": 1.7127168899626621, + "grad_norm": 2.1108598709106445, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7011139392852783, + "num_tokens": 403247716.0, + "step": 15596 + }, + { + "epoch": 1.7128267076652757, + "grad_norm": 1.973168134689331, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7281450629234314, + "num_tokens": 403277059.0, + "step": 15597 + }, + { + "epoch": 1.7129365253678892, + "grad_norm": 2.1444432735443115, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7057775259017944, + "num_tokens": 403304477.0, + "step": 15598 + }, + { + "epoch": 1.713046343070503, + "grad_norm": 2.4314699172973633, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7358523011207581, + "num_tokens": 403327045.0, + "step": 15599 + }, + { + "epoch": 1.7131561607731167, + "grad_norm": 2.796799659729004, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7336211204528809, + "num_tokens": 403346736.0, + "step": 15600 + }, + { + "epoch": 1.7132659784757303, + "grad_norm": 2.2586312294006348, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7213222980499268, + "num_tokens": 403372578.0, + "step": 15601 + }, + { + "epoch": 1.7133757961783438, + "grad_norm": 2.480433940887451, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.713914155960083, + "num_tokens": 403393365.0, + "step": 15602 + }, + { + "epoch": 1.7134856138809575, + "grad_norm": 2.189980983734131, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7138487100601196, + "num_tokens": 403419727.0, + "step": 15603 + }, + { + "epoch": 1.7135954315835713, + "grad_norm": 2.104783773422241, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7186257839202881, + "num_tokens": 403447141.0, + "step": 15604 + }, + { + "epoch": 1.713705249286185, + "grad_norm": 2.3579843044281006, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7115873694419861, + "num_tokens": 403471657.0, + "step": 15605 + }, + { + "epoch": 1.7138150669887986, + "grad_norm": 2.2275545597076416, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7232310175895691, + "num_tokens": 403496854.0, + "step": 15606 + }, + { + "epoch": 1.7139248846914121, + "grad_norm": 2.0620903968811035, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.716982364654541, + "num_tokens": 403524503.0, + "step": 15607 + }, + { + "epoch": 1.7140347023940259, + "grad_norm": 2.088655710220337, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6994579434394836, + "num_tokens": 403554962.0, + "step": 15608 + }, + { + "epoch": 1.7141445200966396, + "grad_norm": 1.9300707578659058, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7122248411178589, + "num_tokens": 403587714.0, + "step": 15609 + }, + { + "epoch": 1.7142543377992534, + "grad_norm": 2.2282755374908447, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7016535401344299, + "num_tokens": 403616592.0, + "step": 15610 + }, + { + "epoch": 1.714364155501867, + "grad_norm": 2.186448097229004, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7254088521003723, + "num_tokens": 403641132.0, + "step": 15611 + }, + { + "epoch": 1.7144739732044805, + "grad_norm": 1.9119329452514648, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.717873752117157, + "num_tokens": 403673395.0, + "step": 15612 + }, + { + "epoch": 1.7145837909070942, + "grad_norm": 2.4269633293151855, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7493449449539185, + "num_tokens": 403695074.0, + "step": 15613 + }, + { + "epoch": 1.714693608609708, + "grad_norm": 2.3030574321746826, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7124073505401611, + "num_tokens": 403722761.0, + "step": 15614 + }, + { + "epoch": 1.7148034263123215, + "grad_norm": 2.565380096435547, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7119921445846558, + "num_tokens": 403743307.0, + "step": 15615 + }, + { + "epoch": 1.714913244014935, + "grad_norm": 2.3635690212249756, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7157784700393677, + "num_tokens": 403767386.0, + "step": 15616 + }, + { + "epoch": 1.7150230617175488, + "grad_norm": 2.09765625, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7261214852333069, + "num_tokens": 403795522.0, + "step": 15617 + }, + { + "epoch": 1.7151328794201626, + "grad_norm": 2.3014743328094482, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7146174907684326, + "num_tokens": 403821048.0, + "step": 15618 + }, + { + "epoch": 1.7152426971227763, + "grad_norm": 1.8968560695648193, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.723351001739502, + "num_tokens": 403854081.0, + "step": 15619 + }, + { + "epoch": 1.7153525148253899, + "grad_norm": 1.911439061164856, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7262452244758606, + "num_tokens": 403887105.0, + "step": 15620 + }, + { + "epoch": 1.7154623325280034, + "grad_norm": 2.317669153213501, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7124320268630981, + "num_tokens": 403911293.0, + "step": 15621 + }, + { + "epoch": 1.7155721502306172, + "grad_norm": 2.404634952545166, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7204582095146179, + "num_tokens": 403932423.0, + "step": 15622 + }, + { + "epoch": 1.715681967933231, + "grad_norm": 1.995007872581482, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7076359987258911, + "num_tokens": 403964248.0, + "step": 15623 + }, + { + "epoch": 1.7157917856358444, + "grad_norm": 2.145318031311035, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7355153560638428, + "num_tokens": 403990120.0, + "step": 15624 + }, + { + "epoch": 1.7159016033384582, + "grad_norm": 2.5798747539520264, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7386325597763062, + "num_tokens": 404009131.0, + "step": 15625 + }, + { + "epoch": 1.7160114210410717, + "grad_norm": 2.1279563903808594, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7118504047393799, + "num_tokens": 404039592.0, + "step": 15626 + }, + { + "epoch": 1.7161212387436855, + "grad_norm": 2.3874380588531494, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7261831760406494, + "num_tokens": 404061611.0, + "step": 15627 + }, + { + "epoch": 1.7162310564462993, + "grad_norm": 2.18172025680542, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7062333822250366, + "num_tokens": 404089891.0, + "step": 15628 + }, + { + "epoch": 1.7163408741489128, + "grad_norm": 2.1174867153167725, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7100992798805237, + "num_tokens": 404118032.0, + "step": 15629 + }, + { + "epoch": 1.7164506918515263, + "grad_norm": 2.660670042037964, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7203391790390015, + "num_tokens": 404139583.0, + "step": 15630 + }, + { + "epoch": 1.71656050955414, + "grad_norm": 2.4655518531799316, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7149228453636169, + "num_tokens": 404162484.0, + "step": 15631 + }, + { + "epoch": 1.7166703272567538, + "grad_norm": 2.308098077774048, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7362499237060547, + "num_tokens": 404185439.0, + "step": 15632 + }, + { + "epoch": 1.7167801449593676, + "grad_norm": 2.272301197052002, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7263578772544861, + "num_tokens": 404209797.0, + "step": 15633 + }, + { + "epoch": 1.7168899626619811, + "grad_norm": 2.387101888656616, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7426809072494507, + "num_tokens": 404230140.0, + "step": 15634 + }, + { + "epoch": 1.7169997803645947, + "grad_norm": 2.4180588722229004, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7098559737205505, + "num_tokens": 404254200.0, + "step": 15635 + }, + { + "epoch": 1.7171095980672084, + "grad_norm": 2.2765870094299316, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7212592959403992, + "num_tokens": 404278519.0, + "step": 15636 + }, + { + "epoch": 1.7172194157698222, + "grad_norm": 2.3011116981506348, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7059619426727295, + "num_tokens": 404303770.0, + "step": 15637 + }, + { + "epoch": 1.7173292334724357, + "grad_norm": 2.481724500656128, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7312202453613281, + "num_tokens": 404325138.0, + "step": 15638 + }, + { + "epoch": 1.7174390511750495, + "grad_norm": 2.3190038204193115, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7043298482894897, + "num_tokens": 404349688.0, + "step": 15639 + }, + { + "epoch": 1.717548868877663, + "grad_norm": 1.8929388523101807, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7110481262207031, + "num_tokens": 404382347.0, + "step": 15640 + }, + { + "epoch": 1.7176586865802768, + "grad_norm": 2.0485565662384033, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7157902717590332, + "num_tokens": 404409840.0, + "step": 15641 + }, + { + "epoch": 1.7177685042828905, + "grad_norm": 1.9998165369033813, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6922776699066162, + "num_tokens": 404441008.0, + "step": 15642 + }, + { + "epoch": 1.717878321985504, + "grad_norm": 2.0810766220092773, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7085080146789551, + "num_tokens": 404468403.0, + "step": 15643 + }, + { + "epoch": 1.7179881396881176, + "grad_norm": 2.087554693222046, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7201318740844727, + "num_tokens": 404497085.0, + "step": 15644 + }, + { + "epoch": 1.7180979573907313, + "grad_norm": 2.0095484256744385, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7033593654632568, + "num_tokens": 404529952.0, + "step": 15645 + }, + { + "epoch": 1.718207775093345, + "grad_norm": 2.020245313644409, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7020641565322876, + "num_tokens": 404559855.0, + "step": 15646 + }, + { + "epoch": 1.7183175927959589, + "grad_norm": 2.16489839553833, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7126728296279907, + "num_tokens": 404585447.0, + "step": 15647 + }, + { + "epoch": 1.7184274104985724, + "grad_norm": 2.0447633266448975, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7126699686050415, + "num_tokens": 404615258.0, + "step": 15648 + }, + { + "epoch": 1.718537228201186, + "grad_norm": 2.248417377471924, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7104274034500122, + "num_tokens": 404640487.0, + "step": 15649 + }, + { + "epoch": 1.7186470459037997, + "grad_norm": 2.1275672912597656, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7039394974708557, + "num_tokens": 404669984.0, + "step": 15650 + }, + { + "epoch": 1.7187568636064134, + "grad_norm": 2.1826906204223633, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7119527459144592, + "num_tokens": 404695795.0, + "step": 15651 + }, + { + "epoch": 1.718866681309027, + "grad_norm": 2.372711420059204, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7331559658050537, + "num_tokens": 404716863.0, + "step": 15652 + }, + { + "epoch": 1.7189764990116405, + "grad_norm": 2.741504430770874, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7331328988075256, + "num_tokens": 404733635.0, + "step": 15653 + }, + { + "epoch": 1.7190863167142543, + "grad_norm": 2.5517940521240234, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7365131378173828, + "num_tokens": 404754534.0, + "step": 15654 + }, + { + "epoch": 1.719196134416868, + "grad_norm": 2.0317575931549072, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.699600338935852, + "num_tokens": 404786352.0, + "step": 15655 + }, + { + "epoch": 1.7193059521194818, + "grad_norm": 2.2318477630615234, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.714701235294342, + "num_tokens": 404812670.0, + "step": 15656 + }, + { + "epoch": 1.7194157698220953, + "grad_norm": 2.3763175010681152, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7285007834434509, + "num_tokens": 404833971.0, + "step": 15657 + }, + { + "epoch": 1.7195255875247089, + "grad_norm": 1.9397279024124146, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7141949534416199, + "num_tokens": 404865009.0, + "step": 15658 + }, + { + "epoch": 1.7196354052273226, + "grad_norm": 2.209860324859619, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6977705955505371, + "num_tokens": 404891573.0, + "step": 15659 + }, + { + "epoch": 1.7197452229299364, + "grad_norm": 2.211819648742676, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.7110339403152466, + "num_tokens": 404918890.0, + "step": 15660 + }, + { + "epoch": 1.7198550406325501, + "grad_norm": 2.172131299972534, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7479938268661499, + "num_tokens": 404943132.0, + "step": 15661 + }, + { + "epoch": 1.7199648583351637, + "grad_norm": 2.2543017864227295, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7099238038063049, + "num_tokens": 404970183.0, + "step": 15662 + }, + { + "epoch": 1.7200746760377772, + "grad_norm": 2.2280900478363037, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7130030393600464, + "num_tokens": 404996090.0, + "step": 15663 + }, + { + "epoch": 1.720184493740391, + "grad_norm": 2.059337615966797, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7214774489402771, + "num_tokens": 405024775.0, + "step": 15664 + }, + { + "epoch": 1.7202943114430047, + "grad_norm": 1.9970463514328003, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7028021216392517, + "num_tokens": 405056691.0, + "step": 15665 + }, + { + "epoch": 1.7204041291456182, + "grad_norm": 2.333357095718384, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7088637948036194, + "num_tokens": 405082271.0, + "step": 15666 + }, + { + "epoch": 1.7205139468482318, + "grad_norm": 2.569216728210449, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7177361249923706, + "num_tokens": 405104542.0, + "step": 15667 + }, + { + "epoch": 1.7206237645508455, + "grad_norm": 2.01216197013855, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7240595817565918, + "num_tokens": 405135485.0, + "step": 15668 + }, + { + "epoch": 1.7207335822534593, + "grad_norm": 2.308793783187866, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7261093854904175, + "num_tokens": 405159707.0, + "step": 15669 + }, + { + "epoch": 1.720843399956073, + "grad_norm": 2.323197603225708, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7191914916038513, + "num_tokens": 405185350.0, + "step": 15670 + }, + { + "epoch": 1.7209532176586866, + "grad_norm": 2.1116549968719482, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7311906814575195, + "num_tokens": 405214522.0, + "step": 15671 + }, + { + "epoch": 1.7210630353613001, + "grad_norm": 2.6290462017059326, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7206859588623047, + "num_tokens": 405236811.0, + "step": 15672 + }, + { + "epoch": 1.7211728530639139, + "grad_norm": 2.0299646854400635, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7451618909835815, + "num_tokens": 405265540.0, + "step": 15673 + }, + { + "epoch": 1.7212826707665276, + "grad_norm": 2.207162618637085, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7084071636199951, + "num_tokens": 405295474.0, + "step": 15674 + }, + { + "epoch": 1.7213924884691414, + "grad_norm": 2.237949848175049, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7234128713607788, + "num_tokens": 405321196.0, + "step": 15675 + }, + { + "epoch": 1.721502306171755, + "grad_norm": 2.0611915588378906, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7002270221710205, + "num_tokens": 405351944.0, + "step": 15676 + }, + { + "epoch": 1.7216121238743685, + "grad_norm": 2.3592517375946045, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7085989713668823, + "num_tokens": 405378336.0, + "step": 15677 + }, + { + "epoch": 1.7217219415769822, + "grad_norm": 2.122591257095337, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7371535301208496, + "num_tokens": 405404115.0, + "step": 15678 + }, + { + "epoch": 1.721831759279596, + "grad_norm": 2.3852698802948, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7130839824676514, + "num_tokens": 405427676.0, + "step": 15679 + }, + { + "epoch": 1.7219415769822095, + "grad_norm": 2.062821388244629, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.6986609697341919, + "num_tokens": 405456316.0, + "step": 15680 + }, + { + "epoch": 1.722051394684823, + "grad_norm": 2.5190396308898926, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7155003547668457, + "num_tokens": 405478827.0, + "step": 15681 + }, + { + "epoch": 1.7221612123874368, + "grad_norm": 2.2051539421081543, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.740128755569458, + "num_tokens": 405503775.0, + "step": 15682 + }, + { + "epoch": 1.7222710300900506, + "grad_norm": 2.2680094242095947, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7087639570236206, + "num_tokens": 405530284.0, + "step": 15683 + }, + { + "epoch": 1.7223808477926643, + "grad_norm": 2.1421258449554443, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7068411707878113, + "num_tokens": 405560487.0, + "step": 15684 + }, + { + "epoch": 1.7224906654952779, + "grad_norm": 2.207338571548462, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7282456159591675, + "num_tokens": 405586029.0, + "step": 15685 + }, + { + "epoch": 1.7226004831978914, + "grad_norm": 2.0552263259887695, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7183421850204468, + "num_tokens": 405614303.0, + "step": 15686 + }, + { + "epoch": 1.7227103009005051, + "grad_norm": 2.3409154415130615, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7202588319778442, + "num_tokens": 405637574.0, + "step": 15687 + }, + { + "epoch": 1.722820118603119, + "grad_norm": 2.2207295894622803, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.6959860324859619, + "num_tokens": 405664647.0, + "step": 15688 + }, + { + "epoch": 1.7229299363057324, + "grad_norm": 1.947136640548706, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7304703593254089, + "num_tokens": 405695245.0, + "step": 15689 + }, + { + "epoch": 1.7230397540083462, + "grad_norm": 2.128239393234253, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.738758385181427, + "num_tokens": 405721043.0, + "step": 15690 + }, + { + "epoch": 1.7231495717109597, + "grad_norm": 2.1655092239379883, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7059152126312256, + "num_tokens": 405748441.0, + "step": 15691 + }, + { + "epoch": 1.7232593894135735, + "grad_norm": 2.4531750679016113, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7151510715484619, + "num_tokens": 405770211.0, + "step": 15692 + }, + { + "epoch": 1.7233692071161872, + "grad_norm": 2.155832529067993, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.735633909702301, + "num_tokens": 405799085.0, + "step": 15693 + }, + { + "epoch": 1.7234790248188008, + "grad_norm": 2.1887617111206055, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7304180860519409, + "num_tokens": 405825810.0, + "step": 15694 + }, + { + "epoch": 1.7235888425214143, + "grad_norm": 2.2055890560150146, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.7011337280273438, + "num_tokens": 405853812.0, + "step": 15695 + }, + { + "epoch": 1.723698660224028, + "grad_norm": 2.3495357036590576, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7298415303230286, + "num_tokens": 405879039.0, + "step": 15696 + }, + { + "epoch": 1.7238084779266418, + "grad_norm": 2.1577765941619873, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7135138511657715, + "num_tokens": 405908301.0, + "step": 15697 + }, + { + "epoch": 1.7239182956292556, + "grad_norm": 2.4855098724365234, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7382300496101379, + "num_tokens": 405928513.0, + "step": 15698 + }, + { + "epoch": 1.7240281133318691, + "grad_norm": 2.048312187194824, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7081887125968933, + "num_tokens": 405958633.0, + "step": 15699 + }, + { + "epoch": 1.7241379310344827, + "grad_norm": 2.2788047790527344, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7316913604736328, + "num_tokens": 405983518.0, + "step": 15700 + }, + { + "epoch": 1.7242477487370964, + "grad_norm": 2.149733781814575, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7273824214935303, + "num_tokens": 406009479.0, + "step": 15701 + }, + { + "epoch": 1.7243575664397102, + "grad_norm": 2.64810848236084, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7208153009414673, + "num_tokens": 406030749.0, + "step": 15702 + }, + { + "epoch": 1.7244673841423237, + "grad_norm": 2.319129228591919, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7183272838592529, + "num_tokens": 406056447.0, + "step": 15703 + }, + { + "epoch": 1.7245772018449375, + "grad_norm": 2.770242214202881, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.754144549369812, + "num_tokens": 406074410.0, + "step": 15704 + }, + { + "epoch": 1.724687019547551, + "grad_norm": 2.5435121059417725, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7430493831634521, + "num_tokens": 406096397.0, + "step": 15705 + }, + { + "epoch": 1.7247968372501647, + "grad_norm": 2.1934149265289307, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7006467580795288, + "num_tokens": 406123869.0, + "step": 15706 + }, + { + "epoch": 1.7249066549527785, + "grad_norm": 2.3044376373291016, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7479874491691589, + "num_tokens": 406147337.0, + "step": 15707 + }, + { + "epoch": 1.725016472655392, + "grad_norm": 2.3662233352661133, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7130177021026611, + "num_tokens": 406171769.0, + "step": 15708 + }, + { + "epoch": 1.7251262903580056, + "grad_norm": 2.259150743484497, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.711820125579834, + "num_tokens": 406196379.0, + "step": 15709 + }, + { + "epoch": 1.7252361080606193, + "grad_norm": 2.100929021835327, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7172933220863342, + "num_tokens": 406226674.0, + "step": 15710 + }, + { + "epoch": 1.725345925763233, + "grad_norm": 2.1712076663970947, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.729974627494812, + "num_tokens": 406251785.0, + "step": 15711 + }, + { + "epoch": 1.7254557434658468, + "grad_norm": 2.438204526901245, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7416102290153503, + "num_tokens": 406272446.0, + "step": 15712 + }, + { + "epoch": 1.7255655611684604, + "grad_norm": 2.3106679916381836, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7278640270233154, + "num_tokens": 406297309.0, + "step": 15713 + }, + { + "epoch": 1.725675378871074, + "grad_norm": 2.3059234619140625, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7168551087379456, + "num_tokens": 406322257.0, + "step": 15714 + }, + { + "epoch": 1.7257851965736877, + "grad_norm": 2.2873375415802, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7173281908035278, + "num_tokens": 406346570.0, + "step": 15715 + }, + { + "epoch": 1.7258950142763014, + "grad_norm": 2.1287004947662354, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7222587466239929, + "num_tokens": 406372128.0, + "step": 15716 + }, + { + "epoch": 1.726004831978915, + "grad_norm": 2.520326614379883, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.742882490158081, + "num_tokens": 406393200.0, + "step": 15717 + }, + { + "epoch": 1.7261146496815285, + "grad_norm": 2.2024290561676025, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6919351816177368, + "num_tokens": 406421258.0, + "step": 15718 + }, + { + "epoch": 1.7262244673841423, + "grad_norm": 1.8982266187667847, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7360650300979614, + "num_tokens": 406454628.0, + "step": 15719 + }, + { + "epoch": 1.726334285086756, + "grad_norm": 2.0219409465789795, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7373833060264587, + "num_tokens": 406484194.0, + "step": 15720 + }, + { + "epoch": 1.7264441027893698, + "grad_norm": 2.3872010707855225, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7186986207962036, + "num_tokens": 406508641.0, + "step": 15721 + }, + { + "epoch": 1.7265539204919833, + "grad_norm": 2.1636643409729004, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6971214413642883, + "num_tokens": 406535886.0, + "step": 15722 + }, + { + "epoch": 1.7266637381945968, + "grad_norm": 2.551543951034546, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7208243608474731, + "num_tokens": 406557711.0, + "step": 15723 + }, + { + "epoch": 1.7267735558972106, + "grad_norm": 2.360867977142334, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7260702848434448, + "num_tokens": 406579582.0, + "step": 15724 + }, + { + "epoch": 1.7268833735998244, + "grad_norm": 2.3182690143585205, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7398562431335449, + "num_tokens": 406603065.0, + "step": 15725 + }, + { + "epoch": 1.7269931913024381, + "grad_norm": 2.2755398750305176, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7128856778144836, + "num_tokens": 406627249.0, + "step": 15726 + }, + { + "epoch": 1.7271030090050516, + "grad_norm": 2.1921966075897217, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7403627038002014, + "num_tokens": 406651523.0, + "step": 15727 + }, + { + "epoch": 1.7272128267076652, + "grad_norm": 2.24725079536438, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6979005336761475, + "num_tokens": 406677929.0, + "step": 15728 + }, + { + "epoch": 1.727322644410279, + "grad_norm": 2.379518985748291, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7187178134918213, + "num_tokens": 406703411.0, + "step": 15729 + }, + { + "epoch": 1.7274324621128927, + "grad_norm": 2.26776123046875, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7163804173469543, + "num_tokens": 406727723.0, + "step": 15730 + }, + { + "epoch": 1.7275422798155062, + "grad_norm": 2.253411054611206, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7034282684326172, + "num_tokens": 406755501.0, + "step": 15731 + }, + { + "epoch": 1.7276520975181198, + "grad_norm": 2.483520984649658, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7333568334579468, + "num_tokens": 406775289.0, + "step": 15732 + }, + { + "epoch": 1.7277619152207335, + "grad_norm": 2.1418967247009277, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7291387319564819, + "num_tokens": 406804107.0, + "step": 15733 + }, + { + "epoch": 1.7278717329233473, + "grad_norm": 2.1941137313842773, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7180120944976807, + "num_tokens": 406832425.0, + "step": 15734 + }, + { + "epoch": 1.727981550625961, + "grad_norm": 2.290783405303955, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7620528936386108, + "num_tokens": 406855526.0, + "step": 15735 + }, + { + "epoch": 1.7280913683285746, + "grad_norm": 2.519462823867798, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7224066257476807, + "num_tokens": 406876907.0, + "step": 15736 + }, + { + "epoch": 1.728201186031188, + "grad_norm": 2.188140869140625, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6993996500968933, + "num_tokens": 406904760.0, + "step": 15737 + }, + { + "epoch": 1.7283110037338019, + "grad_norm": 2.4375972747802734, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7267376780509949, + "num_tokens": 406928549.0, + "step": 15738 + }, + { + "epoch": 1.7284208214364156, + "grad_norm": 2.5793073177337646, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7149097919464111, + "num_tokens": 406949708.0, + "step": 15739 + }, + { + "epoch": 1.7285306391390292, + "grad_norm": 2.075612783432007, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7137771844863892, + "num_tokens": 406977852.0, + "step": 15740 + }, + { + "epoch": 1.728640456841643, + "grad_norm": 2.2132763862609863, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7099641561508179, + "num_tokens": 407004183.0, + "step": 15741 + }, + { + "epoch": 1.7287502745442564, + "grad_norm": 2.189754009246826, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7252386808395386, + "num_tokens": 407029641.0, + "step": 15742 + }, + { + "epoch": 1.7288600922468702, + "grad_norm": 1.9812712669372559, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6947189569473267, + "num_tokens": 407061311.0, + "step": 15743 + }, + { + "epoch": 1.728969909949484, + "grad_norm": 2.4480416774749756, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7346146106719971, + "num_tokens": 407083842.0, + "step": 15744 + }, + { + "epoch": 1.7290797276520975, + "grad_norm": 2.4472262859344482, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7126854658126831, + "num_tokens": 407106403.0, + "step": 15745 + }, + { + "epoch": 1.729189545354711, + "grad_norm": 2.535080671310425, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7363287210464478, + "num_tokens": 407127611.0, + "step": 15746 + }, + { + "epoch": 1.7292993630573248, + "grad_norm": 2.131392002105713, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7080023288726807, + "num_tokens": 407155998.0, + "step": 15747 + }, + { + "epoch": 1.7294091807599385, + "grad_norm": 2.163417339324951, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7563726902008057, + "num_tokens": 407180192.0, + "step": 15748 + }, + { + "epoch": 1.7295189984625523, + "grad_norm": 2.2150049209594727, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7093052268028259, + "num_tokens": 407207098.0, + "step": 15749 + }, + { + "epoch": 1.7296288161651658, + "grad_norm": 2.3604395389556885, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.743060827255249, + "num_tokens": 407229106.0, + "step": 15750 + }, + { + "epoch": 1.7297386338677794, + "grad_norm": 2.413612127304077, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7346233129501343, + "num_tokens": 407251733.0, + "step": 15751 + }, + { + "epoch": 1.7298484515703931, + "grad_norm": 2.237161874771118, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7101751565933228, + "num_tokens": 407276925.0, + "step": 15752 + }, + { + "epoch": 1.7299582692730069, + "grad_norm": 2.256843328475952, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7148797512054443, + "num_tokens": 407301318.0, + "step": 15753 + }, + { + "epoch": 1.7300680869756204, + "grad_norm": 2.508315086364746, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7323791980743408, + "num_tokens": 407323652.0, + "step": 15754 + }, + { + "epoch": 1.7301779046782342, + "grad_norm": 2.023472547531128, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7184664607048035, + "num_tokens": 407353689.0, + "step": 15755 + }, + { + "epoch": 1.7302877223808477, + "grad_norm": 2.313903570175171, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7070099115371704, + "num_tokens": 407378942.0, + "step": 15756 + }, + { + "epoch": 1.7303975400834615, + "grad_norm": 2.2662551403045654, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7221106290817261, + "num_tokens": 407404492.0, + "step": 15757 + }, + { + "epoch": 1.7305073577860752, + "grad_norm": 2.298959970474243, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7219830751419067, + "num_tokens": 407429503.0, + "step": 15758 + }, + { + "epoch": 1.7306171754886888, + "grad_norm": 2.083220958709717, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7216342687606812, + "num_tokens": 407459862.0, + "step": 15759 + }, + { + "epoch": 1.7307269931913023, + "grad_norm": 2.3289847373962402, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7298321723937988, + "num_tokens": 407485344.0, + "step": 15760 + }, + { + "epoch": 1.730836810893916, + "grad_norm": 2.3309953212738037, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7280986905097961, + "num_tokens": 407508639.0, + "step": 15761 + }, + { + "epoch": 1.7309466285965298, + "grad_norm": 2.2249491214752197, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7144913673400879, + "num_tokens": 407535977.0, + "step": 15762 + }, + { + "epoch": 1.7310564462991436, + "grad_norm": 2.401108741760254, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7395973205566406, + "num_tokens": 407557839.0, + "step": 15763 + }, + { + "epoch": 1.731166264001757, + "grad_norm": 2.574714422225952, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7300541400909424, + "num_tokens": 407577992.0, + "step": 15764 + }, + { + "epoch": 1.7312760817043706, + "grad_norm": 2.2338240146636963, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7409233450889587, + "num_tokens": 407602088.0, + "step": 15765 + }, + { + "epoch": 1.7313858994069844, + "grad_norm": 2.5193774700164795, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.73592209815979, + "num_tokens": 407624483.0, + "step": 15766 + }, + { + "epoch": 1.7314957171095982, + "grad_norm": 2.4575514793395996, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.6998345255851746, + "num_tokens": 407648718.0, + "step": 15767 + }, + { + "epoch": 1.7316055348122117, + "grad_norm": 2.4221105575561523, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7228612899780273, + "num_tokens": 407671524.0, + "step": 15768 + }, + { + "epoch": 1.7317153525148252, + "grad_norm": 2.3520050048828125, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7167906761169434, + "num_tokens": 407696736.0, + "step": 15769 + }, + { + "epoch": 1.731825170217439, + "grad_norm": 2.365567207336426, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7359493374824524, + "num_tokens": 407719242.0, + "step": 15770 + }, + { + "epoch": 1.7319349879200527, + "grad_norm": 2.640259027481079, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7312737703323364, + "num_tokens": 407741647.0, + "step": 15771 + }, + { + "epoch": 1.7320448056226665, + "grad_norm": 2.3494174480438232, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7287229895591736, + "num_tokens": 407766431.0, + "step": 15772 + }, + { + "epoch": 1.73215462332528, + "grad_norm": 2.1889843940734863, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7204597592353821, + "num_tokens": 407791725.0, + "step": 15773 + }, + { + "epoch": 1.7322644410278936, + "grad_norm": 2.3826229572296143, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7005892992019653, + "num_tokens": 407814167.0, + "step": 15774 + }, + { + "epoch": 1.7323742587305073, + "grad_norm": 2.0507588386535645, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.6973655819892883, + "num_tokens": 407843181.0, + "step": 15775 + }, + { + "epoch": 1.732484076433121, + "grad_norm": 2.235680103302002, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.69622802734375, + "num_tokens": 407868658.0, + "step": 15776 + }, + { + "epoch": 1.7325938941357348, + "grad_norm": 2.136432647705078, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7350077629089355, + "num_tokens": 407895419.0, + "step": 15777 + }, + { + "epoch": 1.7327037118383484, + "grad_norm": 2.2149922847747803, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7275071144104004, + "num_tokens": 407920037.0, + "step": 15778 + }, + { + "epoch": 1.732813529540962, + "grad_norm": 2.3804354667663574, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7177103161811829, + "num_tokens": 407943819.0, + "step": 15779 + }, + { + "epoch": 1.7329233472435757, + "grad_norm": 1.9880832433700562, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7124190926551819, + "num_tokens": 407974168.0, + "step": 15780 + }, + { + "epoch": 1.7330331649461894, + "grad_norm": 2.2253611087799072, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7302423715591431, + "num_tokens": 408000107.0, + "step": 15781 + }, + { + "epoch": 1.733142982648803, + "grad_norm": 2.1974828243255615, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6845941543579102, + "num_tokens": 408029346.0, + "step": 15782 + }, + { + "epoch": 1.7332528003514165, + "grad_norm": 2.3508660793304443, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7059735059738159, + "num_tokens": 408054076.0, + "step": 15783 + }, + { + "epoch": 1.7333626180540302, + "grad_norm": 1.9424622058868408, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7050198912620544, + "num_tokens": 408087889.0, + "step": 15784 + }, + { + "epoch": 1.733472435756644, + "grad_norm": 2.1189703941345215, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7177542448043823, + "num_tokens": 408115327.0, + "step": 15785 + }, + { + "epoch": 1.7335822534592578, + "grad_norm": 2.486969232559204, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.6983263492584229, + "num_tokens": 408139826.0, + "step": 15786 + }, + { + "epoch": 1.7336920711618713, + "grad_norm": 2.4314074516296387, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7149869203567505, + "num_tokens": 408163102.0, + "step": 15787 + }, + { + "epoch": 1.7338018888644848, + "grad_norm": 2.554763078689575, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7144395112991333, + "num_tokens": 408186929.0, + "step": 15788 + }, + { + "epoch": 1.7339117065670986, + "grad_norm": 2.2801549434661865, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6858620643615723, + "num_tokens": 408213628.0, + "step": 15789 + }, + { + "epoch": 1.7340215242697123, + "grad_norm": 2.12524151802063, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.6977208256721497, + "num_tokens": 408244207.0, + "step": 15790 + }, + { + "epoch": 1.734131341972326, + "grad_norm": 2.0223581790924072, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7300496101379395, + "num_tokens": 408273952.0, + "step": 15791 + }, + { + "epoch": 1.7342411596749396, + "grad_norm": 2.159775495529175, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.72344970703125, + "num_tokens": 408300411.0, + "step": 15792 + }, + { + "epoch": 1.7343509773775532, + "grad_norm": 2.50531268119812, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7366228699684143, + "num_tokens": 408321815.0, + "step": 15793 + }, + { + "epoch": 1.734460795080167, + "grad_norm": 2.492074728012085, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7470223903656006, + "num_tokens": 408343248.0, + "step": 15794 + }, + { + "epoch": 1.7345706127827807, + "grad_norm": 2.316263198852539, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7035059928894043, + "num_tokens": 408368877.0, + "step": 15795 + }, + { + "epoch": 1.7346804304853942, + "grad_norm": 2.289297103881836, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7033712267875671, + "num_tokens": 408394331.0, + "step": 15796 + }, + { + "epoch": 1.7347902481880078, + "grad_norm": 2.1950480937957764, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7023370862007141, + "num_tokens": 408419574.0, + "step": 15797 + }, + { + "epoch": 1.7349000658906215, + "grad_norm": 2.097675085067749, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7215150594711304, + "num_tokens": 408446745.0, + "step": 15798 + }, + { + "epoch": 1.7350098835932353, + "grad_norm": 2.357374429702759, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7139714956283569, + "num_tokens": 408470346.0, + "step": 15799 + }, + { + "epoch": 1.735119701295849, + "grad_norm": 2.149013042449951, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7174907922744751, + "num_tokens": 408497927.0, + "step": 15800 + }, + { + "epoch": 1.7352295189984626, + "grad_norm": 2.2263522148132324, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7078397870063782, + "num_tokens": 408523528.0, + "step": 15801 + }, + { + "epoch": 1.735339336701076, + "grad_norm": 1.8302651643753052, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7109991908073425, + "num_tokens": 408557516.0, + "step": 15802 + }, + { + "epoch": 1.7354491544036899, + "grad_norm": 2.29921555519104, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.709836483001709, + "num_tokens": 408582193.0, + "step": 15803 + }, + { + "epoch": 1.7355589721063036, + "grad_norm": 2.5544183254241943, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7314531207084656, + "num_tokens": 408602622.0, + "step": 15804 + }, + { + "epoch": 1.7356687898089171, + "grad_norm": 2.4985697269439697, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7242077589035034, + "num_tokens": 408624316.0, + "step": 15805 + }, + { + "epoch": 1.735778607511531, + "grad_norm": 2.092804431915283, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7258908748626709, + "num_tokens": 408652235.0, + "step": 15806 + }, + { + "epoch": 1.7358884252141444, + "grad_norm": 2.2602999210357666, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.714159369468689, + "num_tokens": 408677377.0, + "step": 15807 + }, + { + "epoch": 1.7359982429167582, + "grad_norm": 2.226043462753296, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7135199308395386, + "num_tokens": 408703453.0, + "step": 15808 + }, + { + "epoch": 1.736108060619372, + "grad_norm": 2.3318898677825928, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7452284097671509, + "num_tokens": 408727630.0, + "step": 15809 + }, + { + "epoch": 1.7362178783219855, + "grad_norm": 2.348492383956909, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7122780084609985, + "num_tokens": 408750923.0, + "step": 15810 + }, + { + "epoch": 1.736327696024599, + "grad_norm": 2.1982038021087646, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7124805450439453, + "num_tokens": 408778970.0, + "step": 15811 + }, + { + "epoch": 1.7364375137272128, + "grad_norm": 2.2619550228118896, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7450137138366699, + "num_tokens": 408803405.0, + "step": 15812 + }, + { + "epoch": 1.7365473314298265, + "grad_norm": 2.239248752593994, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.6997747421264648, + "num_tokens": 408828951.0, + "step": 15813 + }, + { + "epoch": 1.7366571491324403, + "grad_norm": 1.911409854888916, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6799423098564148, + "num_tokens": 408864782.0, + "step": 15814 + }, + { + "epoch": 1.7367669668350538, + "grad_norm": 2.614166021347046, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7385790348052979, + "num_tokens": 408883987.0, + "step": 15815 + }, + { + "epoch": 1.7368767845376674, + "grad_norm": 2.166194438934326, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7071152925491333, + "num_tokens": 408911114.0, + "step": 15816 + }, + { + "epoch": 1.7369866022402811, + "grad_norm": 2.51246976852417, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7361515164375305, + "num_tokens": 408932732.0, + "step": 15817 + }, + { + "epoch": 1.7370964199428949, + "grad_norm": 2.424257755279541, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7419843673706055, + "num_tokens": 408956139.0, + "step": 15818 + }, + { + "epoch": 1.7372062376455084, + "grad_norm": 2.4700255393981934, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7285988330841064, + "num_tokens": 408979450.0, + "step": 15819 + }, + { + "epoch": 1.7373160553481222, + "grad_norm": 2.371306896209717, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7383731007575989, + "num_tokens": 409002261.0, + "step": 15820 + }, + { + "epoch": 1.7374258730507357, + "grad_norm": 2.2133560180664062, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7120497226715088, + "num_tokens": 409031473.0, + "step": 15821 + }, + { + "epoch": 1.7375356907533495, + "grad_norm": 2.410175085067749, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7216454744338989, + "num_tokens": 409053824.0, + "step": 15822 + }, + { + "epoch": 1.7376455084559632, + "grad_norm": 2.3029391765594482, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.707874059677124, + "num_tokens": 409081507.0, + "step": 15823 + }, + { + "epoch": 1.7377553261585768, + "grad_norm": 2.4468467235565186, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7446048259735107, + "num_tokens": 409104613.0, + "step": 15824 + }, + { + "epoch": 1.7378651438611903, + "grad_norm": 2.192308187484741, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7152726054191589, + "num_tokens": 409131953.0, + "step": 15825 + }, + { + "epoch": 1.737974961563804, + "grad_norm": 2.336782455444336, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7124201059341431, + "num_tokens": 409156621.0, + "step": 15826 + }, + { + "epoch": 1.7380847792664178, + "grad_norm": 2.2805538177490234, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7110339999198914, + "num_tokens": 409183001.0, + "step": 15827 + }, + { + "epoch": 1.7381945969690316, + "grad_norm": 2.432499885559082, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7341305613517761, + "num_tokens": 409203539.0, + "step": 15828 + }, + { + "epoch": 1.738304414671645, + "grad_norm": 2.2044715881347656, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7254377603530884, + "num_tokens": 409230142.0, + "step": 15829 + }, + { + "epoch": 1.7384142323742586, + "grad_norm": 2.2270865440368652, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.701457679271698, + "num_tokens": 409255909.0, + "step": 15830 + }, + { + "epoch": 1.7385240500768724, + "grad_norm": 2.105278968811035, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7357867956161499, + "num_tokens": 409282733.0, + "step": 15831 + }, + { + "epoch": 1.7386338677794861, + "grad_norm": 2.2010700702667236, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7312337160110474, + "num_tokens": 409309821.0, + "step": 15832 + }, + { + "epoch": 1.7387436854820997, + "grad_norm": 2.3019182682037354, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7454438805580139, + "num_tokens": 409334055.0, + "step": 15833 + }, + { + "epoch": 1.7388535031847132, + "grad_norm": 2.792062282562256, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.716529369354248, + "num_tokens": 409354915.0, + "step": 15834 + }, + { + "epoch": 1.738963320887327, + "grad_norm": 2.8111541271209717, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.749744713306427, + "num_tokens": 409372334.0, + "step": 15835 + }, + { + "epoch": 1.7390731385899407, + "grad_norm": 2.0764670372009277, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6943938732147217, + "num_tokens": 409401713.0, + "step": 15836 + }, + { + "epoch": 1.7391829562925545, + "grad_norm": 2.0751748085021973, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.697098970413208, + "num_tokens": 409431433.0, + "step": 15837 + }, + { + "epoch": 1.739292773995168, + "grad_norm": 2.2802350521087646, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7210663557052612, + "num_tokens": 409454487.0, + "step": 15838 + }, + { + "epoch": 1.7394025916977816, + "grad_norm": 2.1698522567749023, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.726492702960968, + "num_tokens": 409479958.0, + "step": 15839 + }, + { + "epoch": 1.7395124094003953, + "grad_norm": 2.05800724029541, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7250919938087463, + "num_tokens": 409509442.0, + "step": 15840 + }, + { + "epoch": 1.739622227103009, + "grad_norm": 2.5208799839019775, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7100529670715332, + "num_tokens": 409530593.0, + "step": 15841 + }, + { + "epoch": 1.7397320448056228, + "grad_norm": 1.941427230834961, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7158872485160828, + "num_tokens": 409563789.0, + "step": 15842 + }, + { + "epoch": 1.7398418625082364, + "grad_norm": 2.048713445663452, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7086446285247803, + "num_tokens": 409592783.0, + "step": 15843 + }, + { + "epoch": 1.73995168021085, + "grad_norm": 2.2704920768737793, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7186212539672852, + "num_tokens": 409617309.0, + "step": 15844 + }, + { + "epoch": 1.7400614979134637, + "grad_norm": 2.2890102863311768, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7120569944381714, + "num_tokens": 409642941.0, + "step": 15845 + }, + { + "epoch": 1.7401713156160774, + "grad_norm": 2.0966644287109375, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7281960248947144, + "num_tokens": 409669861.0, + "step": 15846 + }, + { + "epoch": 1.740281133318691, + "grad_norm": 2.4739062786102295, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7088854312896729, + "num_tokens": 409691814.0, + "step": 15847 + }, + { + "epoch": 1.7403909510213045, + "grad_norm": 2.264115571975708, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7124285697937012, + "num_tokens": 409717653.0, + "step": 15848 + }, + { + "epoch": 1.7405007687239182, + "grad_norm": 2.4418537616729736, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7253508567810059, + "num_tokens": 409738514.0, + "step": 15849 + }, + { + "epoch": 1.740610586426532, + "grad_norm": 2.305419445037842, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7047682404518127, + "num_tokens": 409762917.0, + "step": 15850 + }, + { + "epoch": 1.7407204041291457, + "grad_norm": 2.351778984069824, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7281591892242432, + "num_tokens": 409786764.0, + "step": 15851 + }, + { + "epoch": 1.7408302218317593, + "grad_norm": 2.284956455230713, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7114008069038391, + "num_tokens": 409814012.0, + "step": 15852 + }, + { + "epoch": 1.7409400395343728, + "grad_norm": 2.2495954036712646, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7146894335746765, + "num_tokens": 409841290.0, + "step": 15853 + }, + { + "epoch": 1.7410498572369866, + "grad_norm": 2.2872650623321533, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7148825526237488, + "num_tokens": 409865997.0, + "step": 15854 + }, + { + "epoch": 1.7411596749396003, + "grad_norm": 2.361323356628418, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7075817584991455, + "num_tokens": 409888915.0, + "step": 15855 + }, + { + "epoch": 1.741269492642214, + "grad_norm": 1.875856876373291, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7007119059562683, + "num_tokens": 409922378.0, + "step": 15856 + }, + { + "epoch": 1.7413793103448276, + "grad_norm": 2.2209084033966064, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7547110915184021, + "num_tokens": 409945678.0, + "step": 15857 + }, + { + "epoch": 1.7414891280474412, + "grad_norm": 2.2915446758270264, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7141517996788025, + "num_tokens": 409968835.0, + "step": 15858 + }, + { + "epoch": 1.741598945750055, + "grad_norm": 2.100926637649536, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7124449014663696, + "num_tokens": 409997469.0, + "step": 15859 + }, + { + "epoch": 1.7417087634526687, + "grad_norm": 2.267190456390381, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7059056758880615, + "num_tokens": 410022375.0, + "step": 15860 + }, + { + "epoch": 1.7418185811552822, + "grad_norm": 2.0495810508728027, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6928166151046753, + "num_tokens": 410054020.0, + "step": 15861 + }, + { + "epoch": 1.7419283988578957, + "grad_norm": 2.161606788635254, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7036263346672058, + "num_tokens": 410079572.0, + "step": 15862 + }, + { + "epoch": 1.7420382165605095, + "grad_norm": 2.105419635772705, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7296443581581116, + "num_tokens": 410106514.0, + "step": 15863 + }, + { + "epoch": 1.7421480342631233, + "grad_norm": 2.1940739154815674, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7172524929046631, + "num_tokens": 410132343.0, + "step": 15864 + }, + { + "epoch": 1.742257851965737, + "grad_norm": 2.103766918182373, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7194254398345947, + "num_tokens": 410161866.0, + "step": 15865 + }, + { + "epoch": 1.7423676696683505, + "grad_norm": 2.272383213043213, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.6951209306716919, + "num_tokens": 410187236.0, + "step": 15866 + }, + { + "epoch": 1.742477487370964, + "grad_norm": 2.1264166831970215, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7191685438156128, + "num_tokens": 410216432.0, + "step": 15867 + }, + { + "epoch": 1.7425873050735778, + "grad_norm": 2.1113057136535645, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7268831729888916, + "num_tokens": 410244252.0, + "step": 15868 + }, + { + "epoch": 1.7426971227761916, + "grad_norm": 2.1738221645355225, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7010548114776611, + "num_tokens": 410271135.0, + "step": 15869 + }, + { + "epoch": 1.7428069404788051, + "grad_norm": 2.326772928237915, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7198001742362976, + "num_tokens": 410296705.0, + "step": 15870 + }, + { + "epoch": 1.742916758181419, + "grad_norm": 2.2139902114868164, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6864593625068665, + "num_tokens": 410324785.0, + "step": 15871 + }, + { + "epoch": 1.7430265758840324, + "grad_norm": 2.1872243881225586, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7294253706932068, + "num_tokens": 410350793.0, + "step": 15872 + }, + { + "epoch": 1.7431363935866462, + "grad_norm": 2.1633658409118652, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7221323847770691, + "num_tokens": 410380193.0, + "step": 15873 + }, + { + "epoch": 1.74324621128926, + "grad_norm": 2.392852544784546, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7422956228256226, + "num_tokens": 410403374.0, + "step": 15874 + }, + { + "epoch": 1.7433560289918735, + "grad_norm": 2.059368371963501, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7079618573188782, + "num_tokens": 410431523.0, + "step": 15875 + }, + { + "epoch": 1.743465846694487, + "grad_norm": 2.2810184955596924, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7368678450584412, + "num_tokens": 410455780.0, + "step": 15876 + }, + { + "epoch": 1.7435756643971008, + "grad_norm": 2.1469810009002686, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6987369060516357, + "num_tokens": 410485812.0, + "step": 15877 + }, + { + "epoch": 1.7436854820997145, + "grad_norm": 2.6141574382781982, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7122232913970947, + "num_tokens": 410507164.0, + "step": 15878 + }, + { + "epoch": 1.7437952998023283, + "grad_norm": 2.1189839839935303, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7292866706848145, + "num_tokens": 410534324.0, + "step": 15879 + }, + { + "epoch": 1.7439051175049418, + "grad_norm": 2.1637187004089355, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6858140826225281, + "num_tokens": 410564035.0, + "step": 15880 + }, + { + "epoch": 1.7440149352075554, + "grad_norm": 2.727384328842163, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7282907962799072, + "num_tokens": 410582722.0, + "step": 15881 + }, + { + "epoch": 1.744124752910169, + "grad_norm": 2.3418920040130615, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7343201637268066, + "num_tokens": 410606232.0, + "step": 15882 + }, + { + "epoch": 1.7442345706127829, + "grad_norm": 2.1424386501312256, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7078145742416382, + "num_tokens": 410631951.0, + "step": 15883 + }, + { + "epoch": 1.7443443883153964, + "grad_norm": 2.000359058380127, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7511367201805115, + "num_tokens": 410660421.0, + "step": 15884 + }, + { + "epoch": 1.7444542060180102, + "grad_norm": 2.2584030628204346, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7025808095932007, + "num_tokens": 410688148.0, + "step": 15885 + }, + { + "epoch": 1.7445640237206237, + "grad_norm": 2.0023159980773926, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7037674188613892, + "num_tokens": 410721790.0, + "step": 15886 + }, + { + "epoch": 1.7446738414232374, + "grad_norm": 2.7619476318359375, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7313825488090515, + "num_tokens": 410740257.0, + "step": 15887 + }, + { + "epoch": 1.7447836591258512, + "grad_norm": 2.2262990474700928, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7333145141601562, + "num_tokens": 410763734.0, + "step": 15888 + }, + { + "epoch": 1.7448934768284647, + "grad_norm": 2.502392292022705, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7309558987617493, + "num_tokens": 410785032.0, + "step": 15889 + }, + { + "epoch": 1.7450032945310783, + "grad_norm": 2.2228851318359375, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.6941950917243958, + "num_tokens": 410810970.0, + "step": 15890 + }, + { + "epoch": 1.745113112233692, + "grad_norm": 2.124372959136963, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6920546889305115, + "num_tokens": 410841451.0, + "step": 15891 + }, + { + "epoch": 1.7452229299363058, + "grad_norm": 2.300407648086548, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7214652895927429, + "num_tokens": 410865315.0, + "step": 15892 + }, + { + "epoch": 1.7453327476389195, + "grad_norm": 2.380750894546509, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7102932929992676, + "num_tokens": 410889008.0, + "step": 15893 + }, + { + "epoch": 1.745442565341533, + "grad_norm": 2.4567737579345703, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7025822401046753, + "num_tokens": 410912808.0, + "step": 15894 + }, + { + "epoch": 1.7455523830441466, + "grad_norm": 2.2239880561828613, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7181767225265503, + "num_tokens": 410940907.0, + "step": 15895 + }, + { + "epoch": 1.7456622007467604, + "grad_norm": 2.4686758518218994, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7167810797691345, + "num_tokens": 410964706.0, + "step": 15896 + }, + { + "epoch": 1.7457720184493741, + "grad_norm": 2.2033684253692627, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7088262438774109, + "num_tokens": 410992747.0, + "step": 15897 + }, + { + "epoch": 1.7458818361519877, + "grad_norm": 2.2808098793029785, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.757405161857605, + "num_tokens": 411016194.0, + "step": 15898 + }, + { + "epoch": 1.7459916538546012, + "grad_norm": 2.5164411067962646, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7063699960708618, + "num_tokens": 411037537.0, + "step": 15899 + }, + { + "epoch": 1.746101471557215, + "grad_norm": 2.3309638500213623, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7291857004165649, + "num_tokens": 411061701.0, + "step": 15900 + }, + { + "epoch": 1.7462112892598287, + "grad_norm": 2.3035061359405518, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7251166105270386, + "num_tokens": 411086094.0, + "step": 15901 + }, + { + "epoch": 1.7463211069624425, + "grad_norm": 2.4618349075317383, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7155428528785706, + "num_tokens": 411107452.0, + "step": 15902 + }, + { + "epoch": 1.746430924665056, + "grad_norm": 2.4216408729553223, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7387949228286743, + "num_tokens": 411129754.0, + "step": 15903 + }, + { + "epoch": 1.7465407423676695, + "grad_norm": 2.3286099433898926, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7126320600509644, + "num_tokens": 411155350.0, + "step": 15904 + }, + { + "epoch": 1.7466505600702833, + "grad_norm": 2.046367645263672, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7206077575683594, + "num_tokens": 411183490.0, + "step": 15905 + }, + { + "epoch": 1.746760377772897, + "grad_norm": 2.305985450744629, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7099658250808716, + "num_tokens": 411210915.0, + "step": 15906 + }, + { + "epoch": 1.7468701954755108, + "grad_norm": 2.1956734657287598, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7253642678260803, + "num_tokens": 411237205.0, + "step": 15907 + }, + { + "epoch": 1.7469800131781243, + "grad_norm": 2.2603402137756348, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7072789669036865, + "num_tokens": 411264664.0, + "step": 15908 + }, + { + "epoch": 1.7470898308807379, + "grad_norm": 2.3018274307250977, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7193674445152283, + "num_tokens": 411291043.0, + "step": 15909 + }, + { + "epoch": 1.7471996485833516, + "grad_norm": 2.425706148147583, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7090561389923096, + "num_tokens": 411313227.0, + "step": 15910 + }, + { + "epoch": 1.7473094662859654, + "grad_norm": 2.2965400218963623, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.6972528100013733, + "num_tokens": 411339882.0, + "step": 15911 + }, + { + "epoch": 1.747419283988579, + "grad_norm": 2.2394168376922607, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7179604172706604, + "num_tokens": 411366020.0, + "step": 15912 + }, + { + "epoch": 1.7475291016911925, + "grad_norm": 2.214695453643799, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7278250455856323, + "num_tokens": 411388987.0, + "step": 15913 + }, + { + "epoch": 1.7476389193938062, + "grad_norm": 2.10414719581604, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.734950840473175, + "num_tokens": 411417028.0, + "step": 15914 + }, + { + "epoch": 1.74774873709642, + "grad_norm": 2.4418253898620605, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7105810046195984, + "num_tokens": 411439589.0, + "step": 15915 + }, + { + "epoch": 1.7478585547990337, + "grad_norm": 2.53499698638916, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7005093097686768, + "num_tokens": 411461548.0, + "step": 15916 + }, + { + "epoch": 1.7479683725016473, + "grad_norm": 2.0166964530944824, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7304250001907349, + "num_tokens": 411489361.0, + "step": 15917 + }, + { + "epoch": 1.7480781902042608, + "grad_norm": 2.0024521350860596, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7122697234153748, + "num_tokens": 411519407.0, + "step": 15918 + }, + { + "epoch": 1.7481880079068746, + "grad_norm": 2.0080461502075195, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7241503596305847, + "num_tokens": 411547927.0, + "step": 15919 + }, + { + "epoch": 1.7482978256094883, + "grad_norm": 2.1029245853424072, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7132973670959473, + "num_tokens": 411575453.0, + "step": 15920 + }, + { + "epoch": 1.7484076433121019, + "grad_norm": 2.103893756866455, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.728911280632019, + "num_tokens": 411603169.0, + "step": 15921 + }, + { + "epoch": 1.7485174610147156, + "grad_norm": 1.9637119770050049, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7023413181304932, + "num_tokens": 411635310.0, + "step": 15922 + }, + { + "epoch": 1.7486272787173291, + "grad_norm": 1.9050886631011963, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7357532978057861, + "num_tokens": 411666693.0, + "step": 15923 + }, + { + "epoch": 1.748737096419943, + "grad_norm": 2.192269802093506, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7103846073150635, + "num_tokens": 411694091.0, + "step": 15924 + }, + { + "epoch": 1.7488469141225567, + "grad_norm": 2.6384928226470947, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7217313051223755, + "num_tokens": 411715521.0, + "step": 15925 + }, + { + "epoch": 1.7489567318251702, + "grad_norm": 2.084219455718994, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7228060960769653, + "num_tokens": 411744107.0, + "step": 15926 + }, + { + "epoch": 1.7490665495277837, + "grad_norm": 1.9635077714920044, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7124412059783936, + "num_tokens": 411774228.0, + "step": 15927 + }, + { + "epoch": 1.7491763672303975, + "grad_norm": 2.7499446868896484, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7292144894599915, + "num_tokens": 411791279.0, + "step": 15928 + }, + { + "epoch": 1.7492861849330112, + "grad_norm": 2.377641439437866, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7323731184005737, + "num_tokens": 411813971.0, + "step": 15929 + }, + { + "epoch": 1.749396002635625, + "grad_norm": 2.3734734058380127, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7240571975708008, + "num_tokens": 411836624.0, + "step": 15930 + }, + { + "epoch": 1.7495058203382385, + "grad_norm": 2.266705274581909, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7177073359489441, + "num_tokens": 411862879.0, + "step": 15931 + }, + { + "epoch": 1.749615638040852, + "grad_norm": 2.1253104209899902, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7203500270843506, + "num_tokens": 411890777.0, + "step": 15932 + }, + { + "epoch": 1.7497254557434658, + "grad_norm": 1.8706636428833008, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.706244945526123, + "num_tokens": 411929648.0, + "step": 15933 + }, + { + "epoch": 1.7498352734460796, + "grad_norm": 1.948684811592102, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7408221364021301, + "num_tokens": 411959716.0, + "step": 15934 + }, + { + "epoch": 1.7499450911486931, + "grad_norm": 2.230236530303955, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7220599055290222, + "num_tokens": 411984294.0, + "step": 15935 + }, + { + "epoch": 1.7500549088513069, + "grad_norm": 2.242126941680908, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7165722846984863, + "num_tokens": 412008278.0, + "step": 15936 + }, + { + "epoch": 1.7501647265539204, + "grad_norm": 2.4495890140533447, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.730442225933075, + "num_tokens": 412035117.0, + "step": 15937 + }, + { + "epoch": 1.7502745442565342, + "grad_norm": 2.234224796295166, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7230087518692017, + "num_tokens": 412059205.0, + "step": 15938 + }, + { + "epoch": 1.750384361959148, + "grad_norm": 2.4192092418670654, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7092370390892029, + "num_tokens": 412082773.0, + "step": 15939 + }, + { + "epoch": 1.7504941796617615, + "grad_norm": 2.1244728565216064, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7166017889976501, + "num_tokens": 412111838.0, + "step": 15940 + }, + { + "epoch": 1.750603997364375, + "grad_norm": 2.702866554260254, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7352656126022339, + "num_tokens": 412131623.0, + "step": 15941 + }, + { + "epoch": 1.7507138150669888, + "grad_norm": 2.0654571056365967, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7190014123916626, + "num_tokens": 412161242.0, + "step": 15942 + }, + { + "epoch": 1.7508236327696025, + "grad_norm": 2.4554452896118164, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7367485165596008, + "num_tokens": 412182544.0, + "step": 15943 + }, + { + "epoch": 1.7509334504722163, + "grad_norm": 2.2753686904907227, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7124592661857605, + "num_tokens": 412207490.0, + "step": 15944 + }, + { + "epoch": 1.7510432681748298, + "grad_norm": 2.1282246112823486, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7077354192733765, + "num_tokens": 412235966.0, + "step": 15945 + }, + { + "epoch": 1.7511530858774433, + "grad_norm": 2.448964834213257, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.729726254940033, + "num_tokens": 412259956.0, + "step": 15946 + }, + { + "epoch": 1.751262903580057, + "grad_norm": 2.156085252761841, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.6983140110969543, + "num_tokens": 412289410.0, + "step": 15947 + }, + { + "epoch": 1.7513727212826709, + "grad_norm": 2.1865081787109375, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7170036435127258, + "num_tokens": 412316136.0, + "step": 15948 + }, + { + "epoch": 1.7514825389852844, + "grad_norm": 1.9865790605545044, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7289638519287109, + "num_tokens": 412344879.0, + "step": 15949 + }, + { + "epoch": 1.7515923566878981, + "grad_norm": 2.1274573802948, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7212591767311096, + "num_tokens": 412371632.0, + "step": 15950 + }, + { + "epoch": 1.7517021743905117, + "grad_norm": 1.9635909795761108, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7281609177589417, + "num_tokens": 412400581.0, + "step": 15951 + }, + { + "epoch": 1.7518119920931254, + "grad_norm": 2.2745347023010254, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.699447512626648, + "num_tokens": 412425480.0, + "step": 15952 + }, + { + "epoch": 1.7519218097957392, + "grad_norm": 2.380155324935913, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7235018014907837, + "num_tokens": 412448255.0, + "step": 15953 + }, + { + "epoch": 1.7520316274983527, + "grad_norm": 2.3180112838745117, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.718651533126831, + "num_tokens": 412471255.0, + "step": 15954 + }, + { + "epoch": 1.7521414452009663, + "grad_norm": 2.5156545639038086, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7417057752609253, + "num_tokens": 412491773.0, + "step": 15955 + }, + { + "epoch": 1.75225126290358, + "grad_norm": 2.1094818115234375, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7204630374908447, + "num_tokens": 412521308.0, + "step": 15956 + }, + { + "epoch": 1.7523610806061938, + "grad_norm": 2.209590435028076, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7115851044654846, + "num_tokens": 412547588.0, + "step": 15957 + }, + { + "epoch": 1.7524708983088075, + "grad_norm": 2.140932083129883, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7055507898330688, + "num_tokens": 412576696.0, + "step": 15958 + }, + { + "epoch": 1.752580716011421, + "grad_norm": 2.464177131652832, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7103872895240784, + "num_tokens": 412598512.0, + "step": 15959 + }, + { + "epoch": 1.7526905337140346, + "grad_norm": 2.299691915512085, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.719711184501648, + "num_tokens": 412624152.0, + "step": 15960 + }, + { + "epoch": 1.7528003514166484, + "grad_norm": 2.192451000213623, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7403850555419922, + "num_tokens": 412649925.0, + "step": 15961 + }, + { + "epoch": 1.7529101691192621, + "grad_norm": 1.7863765954971313, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7040088176727295, + "num_tokens": 412689788.0, + "step": 15962 + }, + { + "epoch": 1.7530199868218757, + "grad_norm": 2.4265003204345703, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7410398721694946, + "num_tokens": 412712857.0, + "step": 15963 + }, + { + "epoch": 1.7531298045244892, + "grad_norm": 2.279005765914917, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7076181769371033, + "num_tokens": 412739070.0, + "step": 15964 + }, + { + "epoch": 1.753239622227103, + "grad_norm": 2.065674304962158, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.723603367805481, + "num_tokens": 412768299.0, + "step": 15965 + }, + { + "epoch": 1.7533494399297167, + "grad_norm": 2.4342219829559326, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7158772945404053, + "num_tokens": 412790910.0, + "step": 15966 + }, + { + "epoch": 1.7534592576323305, + "grad_norm": 2.212232828140259, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7250194549560547, + "num_tokens": 412816326.0, + "step": 15967 + }, + { + "epoch": 1.753569075334944, + "grad_norm": 2.2471346855163574, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7366844415664673, + "num_tokens": 412842083.0, + "step": 15968 + }, + { + "epoch": 1.7536788930375575, + "grad_norm": 2.5156049728393555, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7146973609924316, + "num_tokens": 412862592.0, + "step": 15969 + }, + { + "epoch": 1.7537887107401713, + "grad_norm": 2.2467451095581055, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7203460335731506, + "num_tokens": 412886549.0, + "step": 15970 + }, + { + "epoch": 1.753898528442785, + "grad_norm": 2.1644623279571533, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6965674161911011, + "num_tokens": 412912929.0, + "step": 15971 + }, + { + "epoch": 1.7540083461453988, + "grad_norm": 2.4442386627197266, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7240015268325806, + "num_tokens": 412934306.0, + "step": 15972 + }, + { + "epoch": 1.7541181638480123, + "grad_norm": 2.1414783000946045, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7188129425048828, + "num_tokens": 412963595.0, + "step": 15973 + }, + { + "epoch": 1.7542279815506259, + "grad_norm": 1.9414929151535034, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7131150960922241, + "num_tokens": 412999685.0, + "step": 15974 + }, + { + "epoch": 1.7543377992532396, + "grad_norm": 2.1021227836608887, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7120026350021362, + "num_tokens": 413030652.0, + "step": 15975 + }, + { + "epoch": 1.7544476169558534, + "grad_norm": 2.1918396949768066, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7063223719596863, + "num_tokens": 413058995.0, + "step": 15976 + }, + { + "epoch": 1.754557434658467, + "grad_norm": 2.4165241718292236, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7096606492996216, + "num_tokens": 413081066.0, + "step": 15977 + }, + { + "epoch": 1.7546672523610805, + "grad_norm": 2.1857757568359375, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7219448089599609, + "num_tokens": 413109710.0, + "step": 15978 + }, + { + "epoch": 1.7547770700636942, + "grad_norm": 2.265184164047241, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7291465997695923, + "num_tokens": 413133397.0, + "step": 15979 + }, + { + "epoch": 1.754886887766308, + "grad_norm": 1.9830410480499268, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7109485864639282, + "num_tokens": 413165579.0, + "step": 15980 + }, + { + "epoch": 1.7549967054689217, + "grad_norm": 2.3599724769592285, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.734505295753479, + "num_tokens": 413190182.0, + "step": 15981 + }, + { + "epoch": 1.7551065231715353, + "grad_norm": 2.3112916946411133, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7177851796150208, + "num_tokens": 413214874.0, + "step": 15982 + }, + { + "epoch": 1.7552163408741488, + "grad_norm": 2.1021440029144287, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.715412974357605, + "num_tokens": 413244392.0, + "step": 15983 + }, + { + "epoch": 1.7553261585767626, + "grad_norm": 1.9631924629211426, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7172902822494507, + "num_tokens": 413276386.0, + "step": 15984 + }, + { + "epoch": 1.7554359762793763, + "grad_norm": 2.1649653911590576, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7241106629371643, + "num_tokens": 413304406.0, + "step": 15985 + }, + { + "epoch": 1.7555457939819898, + "grad_norm": 2.20076060295105, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.727048397064209, + "num_tokens": 413329957.0, + "step": 15986 + }, + { + "epoch": 1.7556556116846036, + "grad_norm": 2.1584060192108154, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7171421051025391, + "num_tokens": 413355676.0, + "step": 15987 + }, + { + "epoch": 1.7557654293872171, + "grad_norm": 2.3524622917175293, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7245345115661621, + "num_tokens": 413377392.0, + "step": 15988 + }, + { + "epoch": 1.755875247089831, + "grad_norm": 2.367591142654419, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7422794699668884, + "num_tokens": 413398830.0, + "step": 15989 + }, + { + "epoch": 1.7559850647924446, + "grad_norm": 1.991096019744873, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7049468755722046, + "num_tokens": 413429855.0, + "step": 15990 + }, + { + "epoch": 1.7560948824950582, + "grad_norm": 2.152573823928833, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7230538725852966, + "num_tokens": 413456455.0, + "step": 15991 + }, + { + "epoch": 1.7562047001976717, + "grad_norm": 2.3849833011627197, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7186039686203003, + "num_tokens": 413479740.0, + "step": 15992 + }, + { + "epoch": 1.7563145179002855, + "grad_norm": 2.2735342979431152, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7026541233062744, + "num_tokens": 413505148.0, + "step": 15993 + }, + { + "epoch": 1.7564243356028992, + "grad_norm": 2.2262840270996094, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7211012840270996, + "num_tokens": 413530087.0, + "step": 15994 + }, + { + "epoch": 1.756534153305513, + "grad_norm": 2.2897074222564697, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7378775477409363, + "num_tokens": 413556001.0, + "step": 15995 + }, + { + "epoch": 1.7566439710081265, + "grad_norm": 2.331112861633301, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7481825947761536, + "num_tokens": 413578697.0, + "step": 15996 + }, + { + "epoch": 1.75675378871074, + "grad_norm": 2.169569730758667, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7014785408973694, + "num_tokens": 413607179.0, + "step": 15997 + }, + { + "epoch": 1.7568636064133538, + "grad_norm": 2.0239834785461426, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7127636075019836, + "num_tokens": 413635705.0, + "step": 15998 + }, + { + "epoch": 1.7569734241159676, + "grad_norm": 2.3431813716888428, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7452353239059448, + "num_tokens": 413658062.0, + "step": 15999 + }, + { + "epoch": 1.757083241818581, + "grad_norm": 2.313015937805176, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7380049228668213, + "num_tokens": 413679936.0, + "step": 16000 + }, + { + "epoch": 1.7571930595211949, + "grad_norm": 2.4775338172912598, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7168137431144714, + "num_tokens": 413700924.0, + "step": 16001 + }, + { + "epoch": 1.7573028772238084, + "grad_norm": 2.5102388858795166, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7304425239562988, + "num_tokens": 413720728.0, + "step": 16002 + }, + { + "epoch": 1.7574126949264222, + "grad_norm": 2.1871893405914307, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7365276217460632, + "num_tokens": 413747592.0, + "step": 16003 + }, + { + "epoch": 1.757522512629036, + "grad_norm": 1.9542330503463745, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7188615798950195, + "num_tokens": 413779379.0, + "step": 16004 + }, + { + "epoch": 1.7576323303316495, + "grad_norm": 2.4730641841888428, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7122355699539185, + "num_tokens": 413801675.0, + "step": 16005 + }, + { + "epoch": 1.757742148034263, + "grad_norm": 2.178865432739258, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7068929672241211, + "num_tokens": 413828275.0, + "step": 16006 + }, + { + "epoch": 1.7578519657368767, + "grad_norm": 2.0830883979797363, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7245771884918213, + "num_tokens": 413859114.0, + "step": 16007 + }, + { + "epoch": 1.7579617834394905, + "grad_norm": 2.0235488414764404, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7335506677627563, + "num_tokens": 413886825.0, + "step": 16008 + }, + { + "epoch": 1.7580716011421043, + "grad_norm": 1.860032320022583, + "learning_rate": 1e-06, + "loss": 1.0572, + "mean_token_accuracy": 0.6849192380905151, + "num_tokens": 413925084.0, + "step": 16009 + }, + { + "epoch": 1.7581814188447178, + "grad_norm": 2.2884159088134766, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6955708265304565, + "num_tokens": 413952206.0, + "step": 16010 + }, + { + "epoch": 1.7582912365473313, + "grad_norm": 2.350402593612671, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7038883566856384, + "num_tokens": 413975722.0, + "step": 16011 + }, + { + "epoch": 1.758401054249945, + "grad_norm": 2.1607933044433594, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7383756637573242, + "num_tokens": 414000120.0, + "step": 16012 + }, + { + "epoch": 1.7585108719525588, + "grad_norm": 2.398942470550537, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7236791849136353, + "num_tokens": 414022521.0, + "step": 16013 + }, + { + "epoch": 1.7586206896551724, + "grad_norm": 2.280813694000244, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.706270694732666, + "num_tokens": 414047699.0, + "step": 16014 + }, + { + "epoch": 1.758730507357786, + "grad_norm": 2.391775369644165, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7159923911094666, + "num_tokens": 414071095.0, + "step": 16015 + }, + { + "epoch": 1.7588403250603997, + "grad_norm": 2.147617816925049, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7339105606079102, + "num_tokens": 414096434.0, + "step": 16016 + }, + { + "epoch": 1.7589501427630134, + "grad_norm": 2.260923147201538, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7215685248374939, + "num_tokens": 414120608.0, + "step": 16017 + }, + { + "epoch": 1.7590599604656272, + "grad_norm": 2.245204210281372, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7596333026885986, + "num_tokens": 414142862.0, + "step": 16018 + }, + { + "epoch": 1.7591697781682407, + "grad_norm": 2.0062694549560547, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.6997654438018799, + "num_tokens": 414173434.0, + "step": 16019 + }, + { + "epoch": 1.7592795958708543, + "grad_norm": 2.290921211242676, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7343563437461853, + "num_tokens": 414199609.0, + "step": 16020 + }, + { + "epoch": 1.759389413573468, + "grad_norm": 2.339092493057251, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7148544788360596, + "num_tokens": 414225107.0, + "step": 16021 + }, + { + "epoch": 1.7594992312760818, + "grad_norm": 2.080963611602783, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7350956797599792, + "num_tokens": 414251708.0, + "step": 16022 + }, + { + "epoch": 1.7596090489786955, + "grad_norm": 2.099669933319092, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7507591247558594, + "num_tokens": 414279664.0, + "step": 16023 + }, + { + "epoch": 1.759718866681309, + "grad_norm": 2.0554161071777344, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7023912668228149, + "num_tokens": 414309844.0, + "step": 16024 + }, + { + "epoch": 1.7598286843839226, + "grad_norm": 2.198782205581665, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7217307686805725, + "num_tokens": 414333773.0, + "step": 16025 + }, + { + "epoch": 1.7599385020865363, + "grad_norm": 2.203137159347534, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7135035991668701, + "num_tokens": 414360364.0, + "step": 16026 + }, + { + "epoch": 1.76004831978915, + "grad_norm": 2.1574604511260986, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7439351677894592, + "num_tokens": 414387365.0, + "step": 16027 + }, + { + "epoch": 1.7601581374917636, + "grad_norm": 2.1943674087524414, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7301727533340454, + "num_tokens": 414413067.0, + "step": 16028 + }, + { + "epoch": 1.7602679551943772, + "grad_norm": 2.1050524711608887, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7087783217430115, + "num_tokens": 414440864.0, + "step": 16029 + }, + { + "epoch": 1.760377772896991, + "grad_norm": 2.1797373294830322, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7125777006149292, + "num_tokens": 414467877.0, + "step": 16030 + }, + { + "epoch": 1.7604875905996047, + "grad_norm": 2.2334563732147217, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7375028133392334, + "num_tokens": 414492795.0, + "step": 16031 + }, + { + "epoch": 1.7605974083022184, + "grad_norm": 2.195024251937866, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7308056354522705, + "num_tokens": 414519885.0, + "step": 16032 + }, + { + "epoch": 1.760707226004832, + "grad_norm": 1.9955226182937622, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7259527444839478, + "num_tokens": 414549805.0, + "step": 16033 + }, + { + "epoch": 1.7608170437074455, + "grad_norm": 2.1879632472991943, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7296132445335388, + "num_tokens": 414576865.0, + "step": 16034 + }, + { + "epoch": 1.7609268614100593, + "grad_norm": 2.1148383617401123, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7095330953598022, + "num_tokens": 414605167.0, + "step": 16035 + }, + { + "epoch": 1.761036679112673, + "grad_norm": 2.4733285903930664, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7244802117347717, + "num_tokens": 414626986.0, + "step": 16036 + }, + { + "epoch": 1.7611464968152868, + "grad_norm": 2.219546318054199, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7093430757522583, + "num_tokens": 414655905.0, + "step": 16037 + }, + { + "epoch": 1.7612563145179003, + "grad_norm": 2.2010443210601807, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7122467160224915, + "num_tokens": 414682696.0, + "step": 16038 + }, + { + "epoch": 1.7613661322205139, + "grad_norm": 2.1844735145568848, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.723779559135437, + "num_tokens": 414709072.0, + "step": 16039 + }, + { + "epoch": 1.7614759499231276, + "grad_norm": 2.1085891723632812, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7391706109046936, + "num_tokens": 414734163.0, + "step": 16040 + }, + { + "epoch": 1.7615857676257414, + "grad_norm": 2.2715961933135986, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7432683706283569, + "num_tokens": 414757528.0, + "step": 16041 + }, + { + "epoch": 1.761695585328355, + "grad_norm": 2.325348377227783, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7150459289550781, + "num_tokens": 414781026.0, + "step": 16042 + }, + { + "epoch": 1.7618054030309684, + "grad_norm": 1.9888397455215454, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7203048467636108, + "num_tokens": 414811374.0, + "step": 16043 + }, + { + "epoch": 1.7619152207335822, + "grad_norm": 2.357670307159424, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7254137396812439, + "num_tokens": 414834105.0, + "step": 16044 + }, + { + "epoch": 1.762025038436196, + "grad_norm": 2.1190924644470215, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7073025107383728, + "num_tokens": 414863277.0, + "step": 16045 + }, + { + "epoch": 1.7621348561388097, + "grad_norm": 2.3029167652130127, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7311604022979736, + "num_tokens": 414888365.0, + "step": 16046 + }, + { + "epoch": 1.7622446738414232, + "grad_norm": 2.136028289794922, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7308451533317566, + "num_tokens": 414914796.0, + "step": 16047 + }, + { + "epoch": 1.7623544915440368, + "grad_norm": 1.9774895906448364, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7090244293212891, + "num_tokens": 414946541.0, + "step": 16048 + }, + { + "epoch": 1.7624643092466505, + "grad_norm": 2.3827221393585205, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7217944860458374, + "num_tokens": 414970385.0, + "step": 16049 + }, + { + "epoch": 1.7625741269492643, + "grad_norm": 2.123893976211548, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7170741558074951, + "num_tokens": 414997241.0, + "step": 16050 + }, + { + "epoch": 1.7626839446518778, + "grad_norm": 2.2566325664520264, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7206841707229614, + "num_tokens": 415025244.0, + "step": 16051 + }, + { + "epoch": 1.7627937623544916, + "grad_norm": 2.158705234527588, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7453868389129639, + "num_tokens": 415051255.0, + "step": 16052 + }, + { + "epoch": 1.7629035800571051, + "grad_norm": 2.0454561710357666, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7336732149124146, + "num_tokens": 415080247.0, + "step": 16053 + }, + { + "epoch": 1.7630133977597189, + "grad_norm": 2.0340888500213623, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7113599181175232, + "num_tokens": 415109155.0, + "step": 16054 + }, + { + "epoch": 1.7631232154623326, + "grad_norm": 2.2967209815979004, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7242959141731262, + "num_tokens": 415133469.0, + "step": 16055 + }, + { + "epoch": 1.7632330331649462, + "grad_norm": 2.493622303009033, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7094263434410095, + "num_tokens": 415155715.0, + "step": 16056 + }, + { + "epoch": 1.7633428508675597, + "grad_norm": 1.7681492567062378, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7097257375717163, + "num_tokens": 415192246.0, + "step": 16057 + }, + { + "epoch": 1.7634526685701735, + "grad_norm": 2.320863962173462, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7214595079421997, + "num_tokens": 415219216.0, + "step": 16058 + }, + { + "epoch": 1.7635624862727872, + "grad_norm": 2.276895523071289, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.725885272026062, + "num_tokens": 415245103.0, + "step": 16059 + }, + { + "epoch": 1.763672303975401, + "grad_norm": 2.1883602142333984, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7010157704353333, + "num_tokens": 415276220.0, + "step": 16060 + }, + { + "epoch": 1.7637821216780145, + "grad_norm": 2.223395586013794, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7312869429588318, + "num_tokens": 415303037.0, + "step": 16061 + }, + { + "epoch": 1.763891939380628, + "grad_norm": 2.4670019149780273, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.718416154384613, + "num_tokens": 415324617.0, + "step": 16062 + }, + { + "epoch": 1.7640017570832418, + "grad_norm": 2.2827789783477783, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7197506427764893, + "num_tokens": 415349364.0, + "step": 16063 + }, + { + "epoch": 1.7641115747858556, + "grad_norm": 2.0827999114990234, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7082938551902771, + "num_tokens": 415376114.0, + "step": 16064 + }, + { + "epoch": 1.764221392488469, + "grad_norm": 2.1041722297668457, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.683674693107605, + "num_tokens": 415406244.0, + "step": 16065 + }, + { + "epoch": 1.7643312101910829, + "grad_norm": 2.526817798614502, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7035183906555176, + "num_tokens": 415428018.0, + "step": 16066 + }, + { + "epoch": 1.7644410278936964, + "grad_norm": 2.2142536640167236, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7034684419631958, + "num_tokens": 415455075.0, + "step": 16067 + }, + { + "epoch": 1.7645508455963101, + "grad_norm": 2.0119946002960205, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7024881839752197, + "num_tokens": 415485792.0, + "step": 16068 + }, + { + "epoch": 1.764660663298924, + "grad_norm": 2.1094727516174316, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7079527378082275, + "num_tokens": 415514560.0, + "step": 16069 + }, + { + "epoch": 1.7647704810015374, + "grad_norm": 2.323154926300049, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7231377363204956, + "num_tokens": 415537026.0, + "step": 16070 + }, + { + "epoch": 1.764880298704151, + "grad_norm": 2.101167678833008, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6985359191894531, + "num_tokens": 415565677.0, + "step": 16071 + }, + { + "epoch": 1.7649901164067647, + "grad_norm": 2.1712279319763184, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7110922336578369, + "num_tokens": 415594587.0, + "step": 16072 + }, + { + "epoch": 1.7650999341093785, + "grad_norm": 2.110055685043335, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7144585847854614, + "num_tokens": 415622755.0, + "step": 16073 + }, + { + "epoch": 1.7652097518119922, + "grad_norm": 1.9748319387435913, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7219313383102417, + "num_tokens": 415652436.0, + "step": 16074 + }, + { + "epoch": 1.7653195695146058, + "grad_norm": 2.4571290016174316, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7152653932571411, + "num_tokens": 415676475.0, + "step": 16075 + }, + { + "epoch": 1.7654293872172193, + "grad_norm": 2.1935060024261475, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7109456658363342, + "num_tokens": 415703020.0, + "step": 16076 + }, + { + "epoch": 1.765539204919833, + "grad_norm": 2.1883206367492676, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7092359066009521, + "num_tokens": 415729866.0, + "step": 16077 + }, + { + "epoch": 1.7656490226224468, + "grad_norm": 2.4771738052368164, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7266354560852051, + "num_tokens": 415752142.0, + "step": 16078 + }, + { + "epoch": 1.7657588403250604, + "grad_norm": 2.184737205505371, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7117805480957031, + "num_tokens": 415780469.0, + "step": 16079 + }, + { + "epoch": 1.765868658027674, + "grad_norm": 2.098742961883545, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.706930935382843, + "num_tokens": 415809635.0, + "step": 16080 + }, + { + "epoch": 1.7659784757302877, + "grad_norm": 2.409548044204712, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7144002914428711, + "num_tokens": 415832723.0, + "step": 16081 + }, + { + "epoch": 1.7660882934329014, + "grad_norm": 2.299180507659912, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7031093835830688, + "num_tokens": 415857939.0, + "step": 16082 + }, + { + "epoch": 1.7661981111355152, + "grad_norm": 2.345813512802124, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7185405492782593, + "num_tokens": 415881968.0, + "step": 16083 + }, + { + "epoch": 1.7663079288381287, + "grad_norm": 2.1018917560577393, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7142932415008545, + "num_tokens": 415911357.0, + "step": 16084 + }, + { + "epoch": 1.7664177465407422, + "grad_norm": 2.192391872406006, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.732556939125061, + "num_tokens": 415936779.0, + "step": 16085 + }, + { + "epoch": 1.766527564243356, + "grad_norm": 2.12882661819458, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7032871246337891, + "num_tokens": 415964168.0, + "step": 16086 + }, + { + "epoch": 1.7666373819459698, + "grad_norm": 2.2742772102355957, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7286719083786011, + "num_tokens": 415988411.0, + "step": 16087 + }, + { + "epoch": 1.7667471996485835, + "grad_norm": 1.9741486310958862, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.6979965567588806, + "num_tokens": 416018363.0, + "step": 16088 + }, + { + "epoch": 1.766857017351197, + "grad_norm": 2.1436023712158203, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7317684888839722, + "num_tokens": 416046232.0, + "step": 16089 + }, + { + "epoch": 1.7669668350538106, + "grad_norm": 2.1649906635284424, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.73876953125, + "num_tokens": 416073759.0, + "step": 16090 + }, + { + "epoch": 1.7670766527564243, + "grad_norm": 2.239647150039673, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7179522514343262, + "num_tokens": 416098163.0, + "step": 16091 + }, + { + "epoch": 1.767186470459038, + "grad_norm": 2.1090657711029053, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7159916162490845, + "num_tokens": 416126423.0, + "step": 16092 + }, + { + "epoch": 1.7672962881616516, + "grad_norm": 2.5538511276245117, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7390265464782715, + "num_tokens": 416147882.0, + "step": 16093 + }, + { + "epoch": 1.7674061058642652, + "grad_norm": 2.4144814014434814, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7214409708976746, + "num_tokens": 416169273.0, + "step": 16094 + }, + { + "epoch": 1.767515923566879, + "grad_norm": 2.3511648178100586, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7015165090560913, + "num_tokens": 416192765.0, + "step": 16095 + }, + { + "epoch": 1.7676257412694927, + "grad_norm": 2.4012043476104736, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.72674560546875, + "num_tokens": 416216952.0, + "step": 16096 + }, + { + "epoch": 1.7677355589721064, + "grad_norm": 2.117337226867676, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7288463711738586, + "num_tokens": 416243681.0, + "step": 16097 + }, + { + "epoch": 1.76784537667472, + "grad_norm": 2.1899964809417725, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7300714254379272, + "num_tokens": 416270792.0, + "step": 16098 + }, + { + "epoch": 1.7679551943773335, + "grad_norm": 2.171931028366089, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7109130024909973, + "num_tokens": 416298995.0, + "step": 16099 + }, + { + "epoch": 1.7680650120799473, + "grad_norm": 2.1574652194976807, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7266006469726562, + "num_tokens": 416324419.0, + "step": 16100 + }, + { + "epoch": 1.768174829782561, + "grad_norm": 2.430142402648926, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.730221152305603, + "num_tokens": 416347089.0, + "step": 16101 + }, + { + "epoch": 1.7682846474851748, + "grad_norm": 2.275336980819702, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6917284727096558, + "num_tokens": 416372595.0, + "step": 16102 + }, + { + "epoch": 1.7683944651877883, + "grad_norm": 2.4102375507354736, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7214937210083008, + "num_tokens": 416393401.0, + "step": 16103 + }, + { + "epoch": 1.7685042828904018, + "grad_norm": 2.449497938156128, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7264313697814941, + "num_tokens": 416416802.0, + "step": 16104 + }, + { + "epoch": 1.7686141005930156, + "grad_norm": 2.2392477989196777, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7057793140411377, + "num_tokens": 416446841.0, + "step": 16105 + }, + { + "epoch": 1.7687239182956294, + "grad_norm": 2.465528964996338, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7386535406112671, + "num_tokens": 416471615.0, + "step": 16106 + }, + { + "epoch": 1.768833735998243, + "grad_norm": 2.130014419555664, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7076475024223328, + "num_tokens": 416500878.0, + "step": 16107 + }, + { + "epoch": 1.7689435537008564, + "grad_norm": 2.3412516117095947, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7308027744293213, + "num_tokens": 416524480.0, + "step": 16108 + }, + { + "epoch": 1.7690533714034702, + "grad_norm": 2.2209856510162354, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7319706678390503, + "num_tokens": 416549422.0, + "step": 16109 + }, + { + "epoch": 1.769163189106084, + "grad_norm": 2.29148530960083, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7333775758743286, + "num_tokens": 416574267.0, + "step": 16110 + }, + { + "epoch": 1.7692730068086977, + "grad_norm": 2.123528480529785, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7065141201019287, + "num_tokens": 416602495.0, + "step": 16111 + }, + { + "epoch": 1.7693828245113112, + "grad_norm": 2.108299493789673, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.707263171672821, + "num_tokens": 416631999.0, + "step": 16112 + }, + { + "epoch": 1.7694926422139248, + "grad_norm": 2.0307424068450928, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7089623212814331, + "num_tokens": 416664289.0, + "step": 16113 + }, + { + "epoch": 1.7696024599165385, + "grad_norm": 2.451195478439331, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7202180624008179, + "num_tokens": 416686554.0, + "step": 16114 + }, + { + "epoch": 1.7697122776191523, + "grad_norm": 2.090416431427002, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7132954001426697, + "num_tokens": 416715570.0, + "step": 16115 + }, + { + "epoch": 1.7698220953217658, + "grad_norm": 2.1533350944519043, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7092099189758301, + "num_tokens": 416743489.0, + "step": 16116 + }, + { + "epoch": 1.7699319130243796, + "grad_norm": 2.0399129390716553, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7191468477249146, + "num_tokens": 416771487.0, + "step": 16117 + }, + { + "epoch": 1.7700417307269931, + "grad_norm": 2.094644784927368, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7181692123413086, + "num_tokens": 416801416.0, + "step": 16118 + }, + { + "epoch": 1.7701515484296069, + "grad_norm": 2.4556026458740234, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7302929162979126, + "num_tokens": 416823188.0, + "step": 16119 + }, + { + "epoch": 1.7702613661322206, + "grad_norm": 2.227081060409546, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7180933952331543, + "num_tokens": 416849104.0, + "step": 16120 + }, + { + "epoch": 1.7703711838348342, + "grad_norm": 2.220357894897461, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.704666018486023, + "num_tokens": 416876435.0, + "step": 16121 + }, + { + "epoch": 1.7704810015374477, + "grad_norm": 1.9210715293884277, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7166684865951538, + "num_tokens": 416910832.0, + "step": 16122 + }, + { + "epoch": 1.7705908192400615, + "grad_norm": 2.263828992843628, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7070526480674744, + "num_tokens": 416939844.0, + "step": 16123 + }, + { + "epoch": 1.7707006369426752, + "grad_norm": 2.209954023361206, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7335309982299805, + "num_tokens": 416966003.0, + "step": 16124 + }, + { + "epoch": 1.770810454645289, + "grad_norm": 2.5380749702453613, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7252582907676697, + "num_tokens": 416985736.0, + "step": 16125 + }, + { + "epoch": 1.7709202723479025, + "grad_norm": 2.4135427474975586, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7202621698379517, + "num_tokens": 417007408.0, + "step": 16126 + }, + { + "epoch": 1.771030090050516, + "grad_norm": 1.9119305610656738, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7341279983520508, + "num_tokens": 417039428.0, + "step": 16127 + }, + { + "epoch": 1.7711399077531298, + "grad_norm": 2.268860340118408, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.74413001537323, + "num_tokens": 417062279.0, + "step": 16128 + }, + { + "epoch": 1.7712497254557436, + "grad_norm": 2.2795450687408447, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.730868399143219, + "num_tokens": 417086262.0, + "step": 16129 + }, + { + "epoch": 1.771359543158357, + "grad_norm": 2.160351276397705, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7049300670623779, + "num_tokens": 417113676.0, + "step": 16130 + }, + { + "epoch": 1.7714693608609708, + "grad_norm": 2.330336093902588, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7240628004074097, + "num_tokens": 417137542.0, + "step": 16131 + }, + { + "epoch": 1.7715791785635844, + "grad_norm": 2.368196964263916, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7198228240013123, + "num_tokens": 417160678.0, + "step": 16132 + }, + { + "epoch": 1.7716889962661981, + "grad_norm": 2.443115234375, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.711478590965271, + "num_tokens": 417183766.0, + "step": 16133 + }, + { + "epoch": 1.771798813968812, + "grad_norm": 2.3725674152374268, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.699228048324585, + "num_tokens": 417207715.0, + "step": 16134 + }, + { + "epoch": 1.7719086316714254, + "grad_norm": 2.357501983642578, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7275243997573853, + "num_tokens": 417232832.0, + "step": 16135 + }, + { + "epoch": 1.772018449374039, + "grad_norm": 2.03532338142395, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.718765377998352, + "num_tokens": 417262091.0, + "step": 16136 + }, + { + "epoch": 1.7721282670766527, + "grad_norm": 2.0475409030914307, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.71033775806427, + "num_tokens": 417293057.0, + "step": 16137 + }, + { + "epoch": 1.7722380847792665, + "grad_norm": 2.2631783485412598, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7263987064361572, + "num_tokens": 417317893.0, + "step": 16138 + }, + { + "epoch": 1.7723479024818802, + "grad_norm": 2.1247832775115967, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7312777042388916, + "num_tokens": 417345068.0, + "step": 16139 + }, + { + "epoch": 1.7724577201844938, + "grad_norm": 2.102712392807007, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7325437664985657, + "num_tokens": 417372931.0, + "step": 16140 + }, + { + "epoch": 1.7725675378871073, + "grad_norm": 2.3239285945892334, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7160052061080933, + "num_tokens": 417395646.0, + "step": 16141 + }, + { + "epoch": 1.772677355589721, + "grad_norm": 2.117941379547119, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7091571092605591, + "num_tokens": 417426697.0, + "step": 16142 + }, + { + "epoch": 1.7727871732923348, + "grad_norm": 2.331692934036255, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7484170198440552, + "num_tokens": 417448738.0, + "step": 16143 + }, + { + "epoch": 1.7728969909949484, + "grad_norm": 2.3070461750030518, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7302170991897583, + "num_tokens": 417472669.0, + "step": 16144 + }, + { + "epoch": 1.7730068086975619, + "grad_norm": 2.1455583572387695, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.727837085723877, + "num_tokens": 417500389.0, + "step": 16145 + }, + { + "epoch": 1.7731166264001756, + "grad_norm": 2.114812135696411, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7045661807060242, + "num_tokens": 417529568.0, + "step": 16146 + }, + { + "epoch": 1.7732264441027894, + "grad_norm": 2.2126243114471436, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7215099334716797, + "num_tokens": 417554658.0, + "step": 16147 + }, + { + "epoch": 1.7733362618054032, + "grad_norm": 1.9860901832580566, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7135938405990601, + "num_tokens": 417587294.0, + "step": 16148 + }, + { + "epoch": 1.7734460795080167, + "grad_norm": 2.2818169593811035, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7221630811691284, + "num_tokens": 417612958.0, + "step": 16149 + }, + { + "epoch": 1.7735558972106302, + "grad_norm": 2.5010950565338135, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7293449640274048, + "num_tokens": 417634148.0, + "step": 16150 + }, + { + "epoch": 1.773665714913244, + "grad_norm": 2.1744117736816406, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7125812768936157, + "num_tokens": 417659456.0, + "step": 16151 + }, + { + "epoch": 1.7737755326158577, + "grad_norm": 2.1339035034179688, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7042666673660278, + "num_tokens": 417687557.0, + "step": 16152 + }, + { + "epoch": 1.7738853503184715, + "grad_norm": 2.264716625213623, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7133490443229675, + "num_tokens": 417712667.0, + "step": 16153 + }, + { + "epoch": 1.773995168021085, + "grad_norm": 2.3354458808898926, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7408890724182129, + "num_tokens": 417735849.0, + "step": 16154 + }, + { + "epoch": 1.7741049857236986, + "grad_norm": 2.068845510482788, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7023600935935974, + "num_tokens": 417765359.0, + "step": 16155 + }, + { + "epoch": 1.7742148034263123, + "grad_norm": 1.9952963590621948, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7023237347602844, + "num_tokens": 417797089.0, + "step": 16156 + }, + { + "epoch": 1.774324621128926, + "grad_norm": 2.298025369644165, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.724562406539917, + "num_tokens": 417820062.0, + "step": 16157 + }, + { + "epoch": 1.7744344388315396, + "grad_norm": 2.145604372024536, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.6989649534225464, + "num_tokens": 417848752.0, + "step": 16158 + }, + { + "epoch": 1.7745442565341532, + "grad_norm": 1.9634606838226318, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6875627040863037, + "num_tokens": 417881739.0, + "step": 16159 + }, + { + "epoch": 1.774654074236767, + "grad_norm": 2.036923885345459, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.718059778213501, + "num_tokens": 417913173.0, + "step": 16160 + }, + { + "epoch": 1.7747638919393807, + "grad_norm": 2.3041415214538574, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7195987701416016, + "num_tokens": 417939617.0, + "step": 16161 + }, + { + "epoch": 1.7748737096419944, + "grad_norm": 2.2711126804351807, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7259682416915894, + "num_tokens": 417963686.0, + "step": 16162 + }, + { + "epoch": 1.774983527344608, + "grad_norm": 2.134237051010132, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7075787782669067, + "num_tokens": 417992361.0, + "step": 16163 + }, + { + "epoch": 1.7750933450472215, + "grad_norm": 2.210641622543335, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7433702349662781, + "num_tokens": 418017981.0, + "step": 16164 + }, + { + "epoch": 1.7752031627498353, + "grad_norm": 2.200763702392578, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7384748458862305, + "num_tokens": 418043341.0, + "step": 16165 + }, + { + "epoch": 1.775312980452449, + "grad_norm": 2.2327799797058105, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.6978850364685059, + "num_tokens": 418069128.0, + "step": 16166 + }, + { + "epoch": 1.7754227981550625, + "grad_norm": 2.470799684524536, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7264282703399658, + "num_tokens": 418090271.0, + "step": 16167 + }, + { + "epoch": 1.7755326158576763, + "grad_norm": 2.394986629486084, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7355216145515442, + "num_tokens": 418113119.0, + "step": 16168 + }, + { + "epoch": 1.7756424335602898, + "grad_norm": 2.214729070663452, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7239908576011658, + "num_tokens": 418138183.0, + "step": 16169 + }, + { + "epoch": 1.7757522512629036, + "grad_norm": 2.015462636947632, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6980260610580444, + "num_tokens": 418171349.0, + "step": 16170 + }, + { + "epoch": 1.7758620689655173, + "grad_norm": 1.8987071514129639, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7227814197540283, + "num_tokens": 418201351.0, + "step": 16171 + }, + { + "epoch": 1.7759718866681309, + "grad_norm": 2.2996459007263184, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7376238703727722, + "num_tokens": 418226757.0, + "step": 16172 + }, + { + "epoch": 1.7760817043707444, + "grad_norm": 2.3276195526123047, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.724568247795105, + "num_tokens": 418250369.0, + "step": 16173 + }, + { + "epoch": 1.7761915220733582, + "grad_norm": 2.482670783996582, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7517032623291016, + "num_tokens": 418271077.0, + "step": 16174 + }, + { + "epoch": 1.776301339775972, + "grad_norm": 2.387291193008423, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7349343299865723, + "num_tokens": 418292164.0, + "step": 16175 + }, + { + "epoch": 1.7764111574785857, + "grad_norm": 2.4321916103363037, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7283905744552612, + "num_tokens": 418315319.0, + "step": 16176 + }, + { + "epoch": 1.7765209751811992, + "grad_norm": 2.38706636428833, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7039820551872253, + "num_tokens": 418339650.0, + "step": 16177 + }, + { + "epoch": 1.7766307928838128, + "grad_norm": 1.9821211099624634, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.720225989818573, + "num_tokens": 418370376.0, + "step": 16178 + }, + { + "epoch": 1.7767406105864265, + "grad_norm": 2.3076694011688232, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.6996694207191467, + "num_tokens": 418395633.0, + "step": 16179 + }, + { + "epoch": 1.7768504282890403, + "grad_norm": 2.4059178829193115, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7244167923927307, + "num_tokens": 418417308.0, + "step": 16180 + }, + { + "epoch": 1.7769602459916538, + "grad_norm": 2.3025083541870117, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7252942323684692, + "num_tokens": 418441138.0, + "step": 16181 + }, + { + "epoch": 1.7770700636942676, + "grad_norm": 2.051208257675171, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7025883197784424, + "num_tokens": 418470919.0, + "step": 16182 + }, + { + "epoch": 1.777179881396881, + "grad_norm": 2.5335285663604736, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7175229787826538, + "num_tokens": 418493758.0, + "step": 16183 + }, + { + "epoch": 1.7772896990994949, + "grad_norm": 2.303539514541626, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.704288899898529, + "num_tokens": 418520258.0, + "step": 16184 + }, + { + "epoch": 1.7773995168021086, + "grad_norm": 2.276390552520752, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7376059889793396, + "num_tokens": 418544191.0, + "step": 16185 + }, + { + "epoch": 1.7775093345047221, + "grad_norm": 2.5143749713897705, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7621794939041138, + "num_tokens": 418564807.0, + "step": 16186 + }, + { + "epoch": 1.7776191522073357, + "grad_norm": 2.1273880004882812, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7404295206069946, + "num_tokens": 418590708.0, + "step": 16187 + }, + { + "epoch": 1.7777289699099494, + "grad_norm": 2.252628803253174, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7104372978210449, + "num_tokens": 418617068.0, + "step": 16188 + }, + { + "epoch": 1.7778387876125632, + "grad_norm": 2.228273630142212, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7222718000411987, + "num_tokens": 418640811.0, + "step": 16189 + }, + { + "epoch": 1.777948605315177, + "grad_norm": 2.1065545082092285, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7202895879745483, + "num_tokens": 418667841.0, + "step": 16190 + }, + { + "epoch": 1.7780584230177905, + "grad_norm": 2.068307876586914, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7180870771408081, + "num_tokens": 418696645.0, + "step": 16191 + }, + { + "epoch": 1.778168240720404, + "grad_norm": 2.0167903900146484, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7293902039527893, + "num_tokens": 418726463.0, + "step": 16192 + }, + { + "epoch": 1.7782780584230178, + "grad_norm": 2.0790557861328125, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7122315168380737, + "num_tokens": 418755149.0, + "step": 16193 + }, + { + "epoch": 1.7783878761256315, + "grad_norm": 2.3441269397735596, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7279235124588013, + "num_tokens": 418778806.0, + "step": 16194 + }, + { + "epoch": 1.778497693828245, + "grad_norm": 2.4284260272979736, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7380896806716919, + "num_tokens": 418799077.0, + "step": 16195 + }, + { + "epoch": 1.7786075115308586, + "grad_norm": 2.307548999786377, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7345082759857178, + "num_tokens": 418823886.0, + "step": 16196 + }, + { + "epoch": 1.7787173292334724, + "grad_norm": 2.1819658279418945, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7142883539199829, + "num_tokens": 418850668.0, + "step": 16197 + }, + { + "epoch": 1.7788271469360861, + "grad_norm": 2.1695594787597656, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7236385345458984, + "num_tokens": 418876540.0, + "step": 16198 + }, + { + "epoch": 1.7789369646386999, + "grad_norm": 2.700556516647339, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7574395537376404, + "num_tokens": 418894376.0, + "step": 16199 + }, + { + "epoch": 1.7790467823413134, + "grad_norm": 2.2405529022216797, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7059643268585205, + "num_tokens": 418919237.0, + "step": 16200 + }, + { + "epoch": 1.779156600043927, + "grad_norm": 1.9326173067092896, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6868140697479248, + "num_tokens": 418954808.0, + "step": 16201 + }, + { + "epoch": 1.7792664177465407, + "grad_norm": 2.417023181915283, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7137144804000854, + "num_tokens": 418979222.0, + "step": 16202 + }, + { + "epoch": 1.7793762354491545, + "grad_norm": 2.246279239654541, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7334080338478088, + "num_tokens": 419005298.0, + "step": 16203 + }, + { + "epoch": 1.7794860531517682, + "grad_norm": 2.331331968307495, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7131614089012146, + "num_tokens": 419031203.0, + "step": 16204 + }, + { + "epoch": 1.7795958708543818, + "grad_norm": 2.248671054840088, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7385075688362122, + "num_tokens": 419056050.0, + "step": 16205 + }, + { + "epoch": 1.7797056885569953, + "grad_norm": 2.213951349258423, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7130298614501953, + "num_tokens": 419085664.0, + "step": 16206 + }, + { + "epoch": 1.779815506259609, + "grad_norm": 2.5159811973571777, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7196734547615051, + "num_tokens": 419107035.0, + "step": 16207 + }, + { + "epoch": 1.7799253239622228, + "grad_norm": 2.120678663253784, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7188570499420166, + "num_tokens": 419134962.0, + "step": 16208 + }, + { + "epoch": 1.7800351416648363, + "grad_norm": 2.4977002143859863, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.721977949142456, + "num_tokens": 419156469.0, + "step": 16209 + }, + { + "epoch": 1.7801449593674499, + "grad_norm": 2.4053688049316406, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7119677662849426, + "num_tokens": 419180412.0, + "step": 16210 + }, + { + "epoch": 1.7802547770700636, + "grad_norm": 2.157397508621216, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.717465877532959, + "num_tokens": 419209102.0, + "step": 16211 + }, + { + "epoch": 1.7803645947726774, + "grad_norm": 2.514523506164551, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7333223819732666, + "num_tokens": 419230360.0, + "step": 16212 + }, + { + "epoch": 1.7804744124752911, + "grad_norm": 1.9794049263000488, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6782596111297607, + "num_tokens": 419263585.0, + "step": 16213 + }, + { + "epoch": 1.7805842301779047, + "grad_norm": 2.3038556575775146, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7118793725967407, + "num_tokens": 419290289.0, + "step": 16214 + }, + { + "epoch": 1.7806940478805182, + "grad_norm": 2.1716127395629883, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7072902917861938, + "num_tokens": 419315191.0, + "step": 16215 + }, + { + "epoch": 1.780803865583132, + "grad_norm": 2.0149435997009277, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6831380724906921, + "num_tokens": 419345979.0, + "step": 16216 + }, + { + "epoch": 1.7809136832857457, + "grad_norm": 2.3174405097961426, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.745609700679779, + "num_tokens": 419369204.0, + "step": 16217 + }, + { + "epoch": 1.7810235009883595, + "grad_norm": 2.3094568252563477, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7411277890205383, + "num_tokens": 419393122.0, + "step": 16218 + }, + { + "epoch": 1.781133318690973, + "grad_norm": 2.160243034362793, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7122327089309692, + "num_tokens": 419420255.0, + "step": 16219 + }, + { + "epoch": 1.7812431363935866, + "grad_norm": 2.2648375034332275, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7180125713348389, + "num_tokens": 419445421.0, + "step": 16220 + }, + { + "epoch": 1.7813529540962003, + "grad_norm": 2.17401385307312, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6992741227149963, + "num_tokens": 419473130.0, + "step": 16221 + }, + { + "epoch": 1.781462771798814, + "grad_norm": 2.388700485229492, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7147842049598694, + "num_tokens": 419494795.0, + "step": 16222 + }, + { + "epoch": 1.7815725895014276, + "grad_norm": 2.425811529159546, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7094305157661438, + "num_tokens": 419519442.0, + "step": 16223 + }, + { + "epoch": 1.7816824072040411, + "grad_norm": 2.295605421066284, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.711372435092926, + "num_tokens": 419543835.0, + "step": 16224 + }, + { + "epoch": 1.781792224906655, + "grad_norm": 2.5511465072631836, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7203772068023682, + "num_tokens": 419565539.0, + "step": 16225 + }, + { + "epoch": 1.7819020426092687, + "grad_norm": 2.2930219173431396, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7220039963722229, + "num_tokens": 419589785.0, + "step": 16226 + }, + { + "epoch": 1.7820118603118824, + "grad_norm": 2.133502244949341, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7544049620628357, + "num_tokens": 419614859.0, + "step": 16227 + }, + { + "epoch": 1.782121678014496, + "grad_norm": 2.0750787258148193, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7327483892440796, + "num_tokens": 419645316.0, + "step": 16228 + }, + { + "epoch": 1.7822314957171095, + "grad_norm": 2.314687728881836, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7346276044845581, + "num_tokens": 419669390.0, + "step": 16229 + }, + { + "epoch": 1.7823413134197232, + "grad_norm": 2.4567453861236572, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7052946090698242, + "num_tokens": 419692156.0, + "step": 16230 + }, + { + "epoch": 1.782451131122337, + "grad_norm": 2.2111995220184326, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7230119705200195, + "num_tokens": 419718398.0, + "step": 16231 + }, + { + "epoch": 1.7825609488249505, + "grad_norm": 2.1817312240600586, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7240778207778931, + "num_tokens": 419743950.0, + "step": 16232 + }, + { + "epoch": 1.7826707665275643, + "grad_norm": 2.4882049560546875, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7429400682449341, + "num_tokens": 419764640.0, + "step": 16233 + }, + { + "epoch": 1.7827805842301778, + "grad_norm": 2.196349859237671, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7213183641433716, + "num_tokens": 419792119.0, + "step": 16234 + }, + { + "epoch": 1.7828904019327916, + "grad_norm": 2.0772194862365723, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7254589200019836, + "num_tokens": 419820532.0, + "step": 16235 + }, + { + "epoch": 1.7830002196354053, + "grad_norm": 2.2105026245117188, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7439205646514893, + "num_tokens": 419845548.0, + "step": 16236 + }, + { + "epoch": 1.7831100373380189, + "grad_norm": 2.401336193084717, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7226875424385071, + "num_tokens": 419867079.0, + "step": 16237 + }, + { + "epoch": 1.7832198550406324, + "grad_norm": 2.3372139930725098, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7273134589195251, + "num_tokens": 419891368.0, + "step": 16238 + }, + { + "epoch": 1.7833296727432462, + "grad_norm": 2.368095636367798, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.713042140007019, + "num_tokens": 419915401.0, + "step": 16239 + }, + { + "epoch": 1.78343949044586, + "grad_norm": 2.1590867042541504, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7208467125892639, + "num_tokens": 419941102.0, + "step": 16240 + }, + { + "epoch": 1.7835493081484737, + "grad_norm": 2.6984589099884033, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7276614904403687, + "num_tokens": 419960710.0, + "step": 16241 + }, + { + "epoch": 1.7836591258510872, + "grad_norm": 2.5768930912017822, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7331018447875977, + "num_tokens": 419979796.0, + "step": 16242 + }, + { + "epoch": 1.7837689435537007, + "grad_norm": 2.6119399070739746, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7274540662765503, + "num_tokens": 419999905.0, + "step": 16243 + }, + { + "epoch": 1.7838787612563145, + "grad_norm": 2.041299343109131, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7134506702423096, + "num_tokens": 420029631.0, + "step": 16244 + }, + { + "epoch": 1.7839885789589283, + "grad_norm": 2.4424819946289062, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7305625081062317, + "num_tokens": 420051474.0, + "step": 16245 + }, + { + "epoch": 1.7840983966615418, + "grad_norm": 2.536240816116333, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7261269688606262, + "num_tokens": 420071650.0, + "step": 16246 + }, + { + "epoch": 1.7842082143641556, + "grad_norm": 2.517754554748535, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7199218273162842, + "num_tokens": 420094751.0, + "step": 16247 + }, + { + "epoch": 1.784318032066769, + "grad_norm": 2.2968814373016357, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7243776321411133, + "num_tokens": 420121656.0, + "step": 16248 + }, + { + "epoch": 1.7844278497693828, + "grad_norm": 2.4418294429779053, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7347805500030518, + "num_tokens": 420146275.0, + "step": 16249 + }, + { + "epoch": 1.7845376674719966, + "grad_norm": 2.2142155170440674, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7050121426582336, + "num_tokens": 420173068.0, + "step": 16250 + }, + { + "epoch": 1.7846474851746101, + "grad_norm": 2.087796926498413, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7152459621429443, + "num_tokens": 420202313.0, + "step": 16251 + }, + { + "epoch": 1.7847573028772237, + "grad_norm": 2.3628275394439697, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7512120008468628, + "num_tokens": 420225258.0, + "step": 16252 + }, + { + "epoch": 1.7848671205798374, + "grad_norm": 2.103073835372925, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7195491194725037, + "num_tokens": 420254837.0, + "step": 16253 + }, + { + "epoch": 1.7849769382824512, + "grad_norm": 2.0497488975524902, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7029423713684082, + "num_tokens": 420284882.0, + "step": 16254 + }, + { + "epoch": 1.785086755985065, + "grad_norm": 2.2486414909362793, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7165139317512512, + "num_tokens": 420312526.0, + "step": 16255 + }, + { + "epoch": 1.7851965736876785, + "grad_norm": 2.197622776031494, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7115457653999329, + "num_tokens": 420341823.0, + "step": 16256 + }, + { + "epoch": 1.785306391390292, + "grad_norm": 1.9170881509780884, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7236132025718689, + "num_tokens": 420372663.0, + "step": 16257 + }, + { + "epoch": 1.7854162090929058, + "grad_norm": 2.272148609161377, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7140969038009644, + "num_tokens": 420396983.0, + "step": 16258 + }, + { + "epoch": 1.7855260267955195, + "grad_norm": 2.231929063796997, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7225688695907593, + "num_tokens": 420421503.0, + "step": 16259 + }, + { + "epoch": 1.785635844498133, + "grad_norm": 2.167998790740967, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7014074325561523, + "num_tokens": 420451332.0, + "step": 16260 + }, + { + "epoch": 1.7857456622007466, + "grad_norm": 2.107438564300537, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7153968811035156, + "num_tokens": 420481896.0, + "step": 16261 + }, + { + "epoch": 1.7858554799033604, + "grad_norm": 2.238611936569214, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.729963481426239, + "num_tokens": 420507325.0, + "step": 16262 + }, + { + "epoch": 1.7859652976059741, + "grad_norm": 2.181109666824341, + "learning_rate": 1e-06, + "loss": 0.7637, + "mean_token_accuracy": 0.7548481225967407, + "num_tokens": 420532423.0, + "step": 16263 + }, + { + "epoch": 1.7860751153085879, + "grad_norm": 2.1581954956054688, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7193301916122437, + "num_tokens": 420558188.0, + "step": 16264 + }, + { + "epoch": 1.7861849330112014, + "grad_norm": 2.1687357425689697, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7305014133453369, + "num_tokens": 420583440.0, + "step": 16265 + }, + { + "epoch": 1.786294750713815, + "grad_norm": 2.1655664443969727, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7294942736625671, + "num_tokens": 420608649.0, + "step": 16266 + }, + { + "epoch": 1.7864045684164287, + "grad_norm": 2.7218616008758545, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7281140685081482, + "num_tokens": 420627120.0, + "step": 16267 + }, + { + "epoch": 1.7865143861190425, + "grad_norm": 2.063451051712036, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7093121409416199, + "num_tokens": 420656474.0, + "step": 16268 + }, + { + "epoch": 1.7866242038216562, + "grad_norm": 2.1495792865753174, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7317171096801758, + "num_tokens": 420683699.0, + "step": 16269 + }, + { + "epoch": 1.7867340215242697, + "grad_norm": 2.1601502895355225, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7126269340515137, + "num_tokens": 420708766.0, + "step": 16270 + }, + { + "epoch": 1.7868438392268833, + "grad_norm": 1.9412343502044678, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.6985417604446411, + "num_tokens": 420741260.0, + "step": 16271 + }, + { + "epoch": 1.786953656929497, + "grad_norm": 2.166165828704834, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7182596921920776, + "num_tokens": 420768544.0, + "step": 16272 + }, + { + "epoch": 1.7870634746321108, + "grad_norm": 2.3531627655029297, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7459447383880615, + "num_tokens": 420790272.0, + "step": 16273 + }, + { + "epoch": 1.7871732923347243, + "grad_norm": 2.07437801361084, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7100558280944824, + "num_tokens": 420818546.0, + "step": 16274 + }, + { + "epoch": 1.7872831100373379, + "grad_norm": 2.0785391330718994, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.700588047504425, + "num_tokens": 420848155.0, + "step": 16275 + }, + { + "epoch": 1.7873929277399516, + "grad_norm": 2.1945507526397705, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7281318306922913, + "num_tokens": 420873991.0, + "step": 16276 + }, + { + "epoch": 1.7875027454425654, + "grad_norm": 2.269097089767456, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7217194437980652, + "num_tokens": 420899750.0, + "step": 16277 + }, + { + "epoch": 1.7876125631451791, + "grad_norm": 2.302990674972534, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7162150740623474, + "num_tokens": 420926433.0, + "step": 16278 + }, + { + "epoch": 1.7877223808477927, + "grad_norm": 2.0871946811676025, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7328917980194092, + "num_tokens": 420951268.0, + "step": 16279 + }, + { + "epoch": 1.7878321985504062, + "grad_norm": 2.0467607975006104, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7152873277664185, + "num_tokens": 420981931.0, + "step": 16280 + }, + { + "epoch": 1.78794201625302, + "grad_norm": 2.439807891845703, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6918956637382507, + "num_tokens": 421007442.0, + "step": 16281 + }, + { + "epoch": 1.7880518339556337, + "grad_norm": 2.181680917739868, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.710264265537262, + "num_tokens": 421034920.0, + "step": 16282 + }, + { + "epoch": 1.7881616516582475, + "grad_norm": 1.9830985069274902, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7038567662239075, + "num_tokens": 421066890.0, + "step": 16283 + }, + { + "epoch": 1.788271469360861, + "grad_norm": 2.346313714981079, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.6994553804397583, + "num_tokens": 421090461.0, + "step": 16284 + }, + { + "epoch": 1.7883812870634745, + "grad_norm": 2.429029703140259, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7229701280593872, + "num_tokens": 421112929.0, + "step": 16285 + }, + { + "epoch": 1.7884911047660883, + "grad_norm": 2.255573034286499, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7279866933822632, + "num_tokens": 421137933.0, + "step": 16286 + }, + { + "epoch": 1.788600922468702, + "grad_norm": 2.2006943225860596, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7211903929710388, + "num_tokens": 421165253.0, + "step": 16287 + }, + { + "epoch": 1.7887107401713156, + "grad_norm": 2.2883267402648926, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7496772408485413, + "num_tokens": 421187552.0, + "step": 16288 + }, + { + "epoch": 1.7888205578739291, + "grad_norm": 2.2460575103759766, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7286388874053955, + "num_tokens": 421213194.0, + "step": 16289 + }, + { + "epoch": 1.7889303755765429, + "grad_norm": 2.291811943054199, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7182707786560059, + "num_tokens": 421236247.0, + "step": 16290 + }, + { + "epoch": 1.7890401932791566, + "grad_norm": 2.5593209266662598, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7252593040466309, + "num_tokens": 421257100.0, + "step": 16291 + }, + { + "epoch": 1.7891500109817704, + "grad_norm": 2.1727991104125977, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7333027124404907, + "num_tokens": 421280175.0, + "step": 16292 + }, + { + "epoch": 1.789259828684384, + "grad_norm": 2.4447758197784424, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7506077289581299, + "num_tokens": 421300730.0, + "step": 16293 + }, + { + "epoch": 1.7893696463869975, + "grad_norm": 2.325840473175049, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7423233985900879, + "num_tokens": 421323715.0, + "step": 16294 + }, + { + "epoch": 1.7894794640896112, + "grad_norm": 2.4770164489746094, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7112816572189331, + "num_tokens": 421346426.0, + "step": 16295 + }, + { + "epoch": 1.789589281792225, + "grad_norm": 2.294323682785034, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6924217343330383, + "num_tokens": 421371371.0, + "step": 16296 + }, + { + "epoch": 1.7896990994948385, + "grad_norm": 2.241546630859375, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7039511799812317, + "num_tokens": 421396099.0, + "step": 16297 + }, + { + "epoch": 1.7898089171974523, + "grad_norm": 2.180593967437744, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7231218814849854, + "num_tokens": 421424550.0, + "step": 16298 + }, + { + "epoch": 1.7899187349000658, + "grad_norm": 2.3601059913635254, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7396930456161499, + "num_tokens": 421446253.0, + "step": 16299 + }, + { + "epoch": 1.7900285526026796, + "grad_norm": 1.929510235786438, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7127583026885986, + "num_tokens": 421480936.0, + "step": 16300 + }, + { + "epoch": 1.7901383703052933, + "grad_norm": 2.4296841621398926, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7219479084014893, + "num_tokens": 421501546.0, + "step": 16301 + }, + { + "epoch": 1.7902481880079069, + "grad_norm": 2.4510552883148193, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7159919738769531, + "num_tokens": 421523198.0, + "step": 16302 + }, + { + "epoch": 1.7903580057105204, + "grad_norm": 2.199136972427368, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7214396595954895, + "num_tokens": 421549623.0, + "step": 16303 + }, + { + "epoch": 1.7904678234131342, + "grad_norm": 2.232452869415283, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.6974869966506958, + "num_tokens": 421574848.0, + "step": 16304 + }, + { + "epoch": 1.790577641115748, + "grad_norm": 2.4526169300079346, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7222955226898193, + "num_tokens": 421596570.0, + "step": 16305 + }, + { + "epoch": 1.7906874588183617, + "grad_norm": 2.0198147296905518, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7128028273582458, + "num_tokens": 421626890.0, + "step": 16306 + }, + { + "epoch": 1.7907972765209752, + "grad_norm": 2.4165046215057373, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7236751317977905, + "num_tokens": 421652410.0, + "step": 16307 + }, + { + "epoch": 1.7909070942235887, + "grad_norm": 2.2588748931884766, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7101253867149353, + "num_tokens": 421677817.0, + "step": 16308 + }, + { + "epoch": 1.7910169119262025, + "grad_norm": 2.2574527263641357, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7327165603637695, + "num_tokens": 421701895.0, + "step": 16309 + }, + { + "epoch": 1.7911267296288162, + "grad_norm": 2.1638314723968506, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7231718897819519, + "num_tokens": 421727887.0, + "step": 16310 + }, + { + "epoch": 1.7912365473314298, + "grad_norm": 2.183469533920288, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7217286825180054, + "num_tokens": 421754232.0, + "step": 16311 + }, + { + "epoch": 1.7913463650340435, + "grad_norm": 2.606205940246582, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7137759327888489, + "num_tokens": 421774594.0, + "step": 16312 + }, + { + "epoch": 1.791456182736657, + "grad_norm": 2.38832950592041, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7056493759155273, + "num_tokens": 421799392.0, + "step": 16313 + }, + { + "epoch": 1.7915660004392708, + "grad_norm": 2.229625701904297, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.706279993057251, + "num_tokens": 421827295.0, + "step": 16314 + }, + { + "epoch": 1.7916758181418846, + "grad_norm": 2.2684664726257324, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7195224761962891, + "num_tokens": 421853525.0, + "step": 16315 + }, + { + "epoch": 1.7917856358444981, + "grad_norm": 2.237178087234497, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7346757650375366, + "num_tokens": 421880922.0, + "step": 16316 + }, + { + "epoch": 1.7918954535471117, + "grad_norm": 2.3188376426696777, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.73378586769104, + "num_tokens": 421905972.0, + "step": 16317 + }, + { + "epoch": 1.7920052712497254, + "grad_norm": 2.3830082416534424, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7162820100784302, + "num_tokens": 421928724.0, + "step": 16318 + }, + { + "epoch": 1.7921150889523392, + "grad_norm": 2.1019256114959717, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7140069007873535, + "num_tokens": 421956659.0, + "step": 16319 + }, + { + "epoch": 1.792224906654953, + "grad_norm": 2.2986273765563965, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7118384838104248, + "num_tokens": 421979564.0, + "step": 16320 + }, + { + "epoch": 1.7923347243575665, + "grad_norm": 2.3151679039001465, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7557224035263062, + "num_tokens": 422000886.0, + "step": 16321 + }, + { + "epoch": 1.79244454206018, + "grad_norm": 2.1990599632263184, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7078627943992615, + "num_tokens": 422027894.0, + "step": 16322 + }, + { + "epoch": 1.7925543597627938, + "grad_norm": 2.340818166732788, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7180446982383728, + "num_tokens": 422050366.0, + "step": 16323 + }, + { + "epoch": 1.7926641774654075, + "grad_norm": 2.0318171977996826, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.722301185131073, + "num_tokens": 422078367.0, + "step": 16324 + }, + { + "epoch": 1.792773995168021, + "grad_norm": 2.3014230728149414, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7328777313232422, + "num_tokens": 422102979.0, + "step": 16325 + }, + { + "epoch": 1.7928838128706346, + "grad_norm": 2.342432737350464, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7255558371543884, + "num_tokens": 422127190.0, + "step": 16326 + }, + { + "epoch": 1.7929936305732483, + "grad_norm": 2.3126940727233887, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7055395841598511, + "num_tokens": 422151054.0, + "step": 16327 + }, + { + "epoch": 1.793103448275862, + "grad_norm": 2.180407762527466, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7402607202529907, + "num_tokens": 422176163.0, + "step": 16328 + }, + { + "epoch": 1.7932132659784759, + "grad_norm": 2.1935062408447266, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7184520959854126, + "num_tokens": 422200960.0, + "step": 16329 + }, + { + "epoch": 1.7933230836810894, + "grad_norm": 1.9717743396759033, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7299937605857849, + "num_tokens": 422229630.0, + "step": 16330 + }, + { + "epoch": 1.793432901383703, + "grad_norm": 2.0405969619750977, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7111179232597351, + "num_tokens": 422259552.0, + "step": 16331 + }, + { + "epoch": 1.7935427190863167, + "grad_norm": 2.1348893642425537, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7136709690093994, + "num_tokens": 422289751.0, + "step": 16332 + }, + { + "epoch": 1.7936525367889304, + "grad_norm": 2.326096296310425, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7303165197372437, + "num_tokens": 422314633.0, + "step": 16333 + }, + { + "epoch": 1.7937623544915442, + "grad_norm": 2.0978634357452393, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7240643501281738, + "num_tokens": 422341077.0, + "step": 16334 + }, + { + "epoch": 1.7938721721941577, + "grad_norm": 2.248908042907715, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6957828998565674, + "num_tokens": 422369625.0, + "step": 16335 + }, + { + "epoch": 1.7939819898967713, + "grad_norm": 2.3295583724975586, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7240776419639587, + "num_tokens": 422394613.0, + "step": 16336 + }, + { + "epoch": 1.794091807599385, + "grad_norm": 2.3994202613830566, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.724662184715271, + "num_tokens": 422417982.0, + "step": 16337 + }, + { + "epoch": 1.7942016253019988, + "grad_norm": 2.177779197692871, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7345724105834961, + "num_tokens": 422443533.0, + "step": 16338 + }, + { + "epoch": 1.7943114430046123, + "grad_norm": 2.3369390964508057, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7079898118972778, + "num_tokens": 422468024.0, + "step": 16339 + }, + { + "epoch": 1.7944212607072259, + "grad_norm": 2.44771409034729, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7135967016220093, + "num_tokens": 422490614.0, + "step": 16340 + }, + { + "epoch": 1.7945310784098396, + "grad_norm": 2.4906961917877197, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7287142276763916, + "num_tokens": 422512313.0, + "step": 16341 + }, + { + "epoch": 1.7946408961124534, + "grad_norm": 2.119442939758301, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7299970388412476, + "num_tokens": 422537722.0, + "step": 16342 + }, + { + "epoch": 1.7947507138150671, + "grad_norm": 1.9932461977005005, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7243568897247314, + "num_tokens": 422567382.0, + "step": 16343 + }, + { + "epoch": 1.7948605315176807, + "grad_norm": 2.1190237998962402, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7042073011398315, + "num_tokens": 422595383.0, + "step": 16344 + }, + { + "epoch": 1.7949703492202942, + "grad_norm": 2.2250778675079346, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7179081439971924, + "num_tokens": 422623755.0, + "step": 16345 + }, + { + "epoch": 1.795080166922908, + "grad_norm": 2.2300398349761963, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7124322056770325, + "num_tokens": 422649779.0, + "step": 16346 + }, + { + "epoch": 1.7951899846255217, + "grad_norm": 2.39782977104187, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7311749458312988, + "num_tokens": 422671435.0, + "step": 16347 + }, + { + "epoch": 1.7952998023281352, + "grad_norm": 2.16268253326416, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7407106161117554, + "num_tokens": 422696878.0, + "step": 16348 + }, + { + "epoch": 1.795409620030749, + "grad_norm": 2.3191096782684326, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7445876598358154, + "num_tokens": 422720433.0, + "step": 16349 + }, + { + "epoch": 1.7955194377333625, + "grad_norm": 2.447305679321289, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7018193006515503, + "num_tokens": 422743653.0, + "step": 16350 + }, + { + "epoch": 1.7956292554359763, + "grad_norm": 2.197763442993164, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7145994901657104, + "num_tokens": 422771067.0, + "step": 16351 + }, + { + "epoch": 1.79573907313859, + "grad_norm": 2.1750271320343018, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7000452280044556, + "num_tokens": 422798121.0, + "step": 16352 + }, + { + "epoch": 1.7958488908412036, + "grad_norm": 2.4441020488739014, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.715110182762146, + "num_tokens": 422821499.0, + "step": 16353 + }, + { + "epoch": 1.7959587085438171, + "grad_norm": 2.3079116344451904, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7350637912750244, + "num_tokens": 422843980.0, + "step": 16354 + }, + { + "epoch": 1.7960685262464309, + "grad_norm": 2.3336055278778076, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7261425852775574, + "num_tokens": 422866031.0, + "step": 16355 + }, + { + "epoch": 1.7961783439490446, + "grad_norm": 1.9632999897003174, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7208740711212158, + "num_tokens": 422900563.0, + "step": 16356 + }, + { + "epoch": 1.7962881616516584, + "grad_norm": 2.1849822998046875, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.715450644493103, + "num_tokens": 422926656.0, + "step": 16357 + }, + { + "epoch": 1.796397979354272, + "grad_norm": 2.229280710220337, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7229072451591492, + "num_tokens": 422952485.0, + "step": 16358 + }, + { + "epoch": 1.7965077970568855, + "grad_norm": 1.9963716268539429, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6990211009979248, + "num_tokens": 422984735.0, + "step": 16359 + }, + { + "epoch": 1.7966176147594992, + "grad_norm": 2.25541353225708, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.73176109790802, + "num_tokens": 423008382.0, + "step": 16360 + }, + { + "epoch": 1.796727432462113, + "grad_norm": 2.6153805255889893, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7378144264221191, + "num_tokens": 423028419.0, + "step": 16361 + }, + { + "epoch": 1.7968372501647265, + "grad_norm": 2.253905773162842, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7225048542022705, + "num_tokens": 423053152.0, + "step": 16362 + }, + { + "epoch": 1.7969470678673403, + "grad_norm": 2.1537036895751953, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7356184720993042, + "num_tokens": 423079331.0, + "step": 16363 + }, + { + "epoch": 1.7970568855699538, + "grad_norm": 2.427711009979248, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7161074280738831, + "num_tokens": 423102215.0, + "step": 16364 + }, + { + "epoch": 1.7971667032725676, + "grad_norm": 1.9777084589004517, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.749342143535614, + "num_tokens": 423131041.0, + "step": 16365 + }, + { + "epoch": 1.7972765209751813, + "grad_norm": 2.123513698577881, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7298778295516968, + "num_tokens": 423158313.0, + "step": 16366 + }, + { + "epoch": 1.7973863386777948, + "grad_norm": 2.617635726928711, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7230920195579529, + "num_tokens": 423177086.0, + "step": 16367 + }, + { + "epoch": 1.7974961563804084, + "grad_norm": 2.249077320098877, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7222287654876709, + "num_tokens": 423202341.0, + "step": 16368 + }, + { + "epoch": 1.7976059740830221, + "grad_norm": 2.0121591091156006, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7033063769340515, + "num_tokens": 423232975.0, + "step": 16369 + }, + { + "epoch": 1.797715791785636, + "grad_norm": 2.456740379333496, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7148576974868774, + "num_tokens": 423255271.0, + "step": 16370 + }, + { + "epoch": 1.7978256094882497, + "grad_norm": 2.2827329635620117, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7094825506210327, + "num_tokens": 423279840.0, + "step": 16371 + }, + { + "epoch": 1.7979354271908632, + "grad_norm": 2.3873133659362793, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7336086630821228, + "num_tokens": 423301839.0, + "step": 16372 + }, + { + "epoch": 1.7980452448934767, + "grad_norm": 2.321044683456421, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7087365388870239, + "num_tokens": 423325332.0, + "step": 16373 + }, + { + "epoch": 1.7981550625960905, + "grad_norm": 2.0826241970062256, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7314615845680237, + "num_tokens": 423354366.0, + "step": 16374 + }, + { + "epoch": 1.7982648802987042, + "grad_norm": 2.321592330932617, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7069611549377441, + "num_tokens": 423378460.0, + "step": 16375 + }, + { + "epoch": 1.7983746980013178, + "grad_norm": 2.053483009338379, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.707468569278717, + "num_tokens": 423409450.0, + "step": 16376 + }, + { + "epoch": 1.7984845157039313, + "grad_norm": 2.3309128284454346, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7206442356109619, + "num_tokens": 423432078.0, + "step": 16377 + }, + { + "epoch": 1.798594333406545, + "grad_norm": 1.9731944799423218, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6980172395706177, + "num_tokens": 423463737.0, + "step": 16378 + }, + { + "epoch": 1.7987041511091588, + "grad_norm": 2.191711187362671, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7369890213012695, + "num_tokens": 423490202.0, + "step": 16379 + }, + { + "epoch": 1.7988139688117726, + "grad_norm": 2.077624797821045, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.729198157787323, + "num_tokens": 423518171.0, + "step": 16380 + }, + { + "epoch": 1.7989237865143861, + "grad_norm": 2.014979124069214, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7046626806259155, + "num_tokens": 423549214.0, + "step": 16381 + }, + { + "epoch": 1.7990336042169996, + "grad_norm": 1.9850890636444092, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6923723220825195, + "num_tokens": 423583822.0, + "step": 16382 + }, + { + "epoch": 1.7991434219196134, + "grad_norm": 2.0566353797912598, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7188906073570251, + "num_tokens": 423611450.0, + "step": 16383 + }, + { + "epoch": 1.7992532396222272, + "grad_norm": 2.118513822555542, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7271698713302612, + "num_tokens": 423639246.0, + "step": 16384 + }, + { + "epoch": 1.799363057324841, + "grad_norm": 2.32637357711792, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7166358232498169, + "num_tokens": 423663237.0, + "step": 16385 + }, + { + "epoch": 1.7994728750274545, + "grad_norm": 2.8410565853118896, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7189164161682129, + "num_tokens": 423682575.0, + "step": 16386 + }, + { + "epoch": 1.799582692730068, + "grad_norm": 2.1897895336151123, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7122904658317566, + "num_tokens": 423709095.0, + "step": 16387 + }, + { + "epoch": 1.7996925104326817, + "grad_norm": 2.512176036834717, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7445920705795288, + "num_tokens": 423729133.0, + "step": 16388 + }, + { + "epoch": 1.7998023281352955, + "grad_norm": 2.060654878616333, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7364848852157593, + "num_tokens": 423756251.0, + "step": 16389 + }, + { + "epoch": 1.799912145837909, + "grad_norm": 2.2912871837615967, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7058719396591187, + "num_tokens": 423780483.0, + "step": 16390 + }, + { + "epoch": 1.8000219635405226, + "grad_norm": 2.1035234928131104, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7209548950195312, + "num_tokens": 423807886.0, + "step": 16391 + }, + { + "epoch": 1.8001317812431363, + "grad_norm": 2.4088263511657715, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7364429831504822, + "num_tokens": 423830403.0, + "step": 16392 + }, + { + "epoch": 1.80024159894575, + "grad_norm": 2.0009207725524902, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7042229175567627, + "num_tokens": 423862554.0, + "step": 16393 + }, + { + "epoch": 1.8003514166483638, + "grad_norm": 2.2378811836242676, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7311185598373413, + "num_tokens": 423889404.0, + "step": 16394 + }, + { + "epoch": 1.8004612343509774, + "grad_norm": 1.9926773309707642, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7149407863616943, + "num_tokens": 423920984.0, + "step": 16395 + }, + { + "epoch": 1.800571052053591, + "grad_norm": 2.4537174701690674, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.704433262348175, + "num_tokens": 423944673.0, + "step": 16396 + }, + { + "epoch": 1.8006808697562047, + "grad_norm": 2.0751473903656006, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7108317017555237, + "num_tokens": 423971785.0, + "step": 16397 + }, + { + "epoch": 1.8007906874588184, + "grad_norm": 2.5749876499176025, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7373052835464478, + "num_tokens": 423993324.0, + "step": 16398 + }, + { + "epoch": 1.8009005051614322, + "grad_norm": 2.07940673828125, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7209674119949341, + "num_tokens": 424022594.0, + "step": 16399 + }, + { + "epoch": 1.8010103228640457, + "grad_norm": 2.137388229370117, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7130100727081299, + "num_tokens": 424050715.0, + "step": 16400 + }, + { + "epoch": 1.8011201405666593, + "grad_norm": 2.375281572341919, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7105600833892822, + "num_tokens": 424075604.0, + "step": 16401 + }, + { + "epoch": 1.801229958269273, + "grad_norm": 2.7692699432373047, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7531286478042603, + "num_tokens": 424093160.0, + "step": 16402 + }, + { + "epoch": 1.8013397759718868, + "grad_norm": 2.1136677265167236, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.6991596221923828, + "num_tokens": 424121916.0, + "step": 16403 + }, + { + "epoch": 1.8014495936745003, + "grad_norm": 2.1524927616119385, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.727715790271759, + "num_tokens": 424146915.0, + "step": 16404 + }, + { + "epoch": 1.8015594113771138, + "grad_norm": 2.2301225662231445, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7020271420478821, + "num_tokens": 424174791.0, + "step": 16405 + }, + { + "epoch": 1.8016692290797276, + "grad_norm": 1.9277467727661133, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7148118615150452, + "num_tokens": 424206433.0, + "step": 16406 + }, + { + "epoch": 1.8017790467823414, + "grad_norm": 2.353426694869995, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7627019882202148, + "num_tokens": 424229767.0, + "step": 16407 + }, + { + "epoch": 1.801888864484955, + "grad_norm": 2.2170937061309814, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7207292318344116, + "num_tokens": 424258047.0, + "step": 16408 + }, + { + "epoch": 1.8019986821875686, + "grad_norm": 2.322402238845825, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7050848007202148, + "num_tokens": 424281964.0, + "step": 16409 + }, + { + "epoch": 1.8021084998901822, + "grad_norm": 2.0845859050750732, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7181107997894287, + "num_tokens": 424309618.0, + "step": 16410 + }, + { + "epoch": 1.802218317592796, + "grad_norm": 2.2147529125213623, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7261945605278015, + "num_tokens": 424336400.0, + "step": 16411 + }, + { + "epoch": 1.8023281352954097, + "grad_norm": 2.4687089920043945, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7234134674072266, + "num_tokens": 424358957.0, + "step": 16412 + }, + { + "epoch": 1.8024379529980232, + "grad_norm": 1.925437569618225, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7108051180839539, + "num_tokens": 424390698.0, + "step": 16413 + }, + { + "epoch": 1.802547770700637, + "grad_norm": 2.713406801223755, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7337058782577515, + "num_tokens": 424408802.0, + "step": 16414 + }, + { + "epoch": 1.8026575884032505, + "grad_norm": 2.415456771850586, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7210050225257874, + "num_tokens": 424432093.0, + "step": 16415 + }, + { + "epoch": 1.8027674061058643, + "grad_norm": 2.225308418273926, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7213006019592285, + "num_tokens": 424457333.0, + "step": 16416 + }, + { + "epoch": 1.802877223808478, + "grad_norm": 2.4338412284851074, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7273364067077637, + "num_tokens": 424479372.0, + "step": 16417 + }, + { + "epoch": 1.8029870415110916, + "grad_norm": 2.2091152667999268, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.69853675365448, + "num_tokens": 424506414.0, + "step": 16418 + }, + { + "epoch": 1.803096859213705, + "grad_norm": 2.481820583343506, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.700969934463501, + "num_tokens": 424530747.0, + "step": 16419 + }, + { + "epoch": 1.8032066769163189, + "grad_norm": 2.3265504837036133, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7199565172195435, + "num_tokens": 424554944.0, + "step": 16420 + }, + { + "epoch": 1.8033164946189326, + "grad_norm": 2.299978017807007, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7490437626838684, + "num_tokens": 424577934.0, + "step": 16421 + }, + { + "epoch": 1.8034263123215464, + "grad_norm": 2.3197824954986572, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7119289636611938, + "num_tokens": 424600956.0, + "step": 16422 + }, + { + "epoch": 1.80353613002416, + "grad_norm": 2.1868245601654053, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7275309562683105, + "num_tokens": 424625842.0, + "step": 16423 + }, + { + "epoch": 1.8036459477267734, + "grad_norm": 2.2107980251312256, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7260302305221558, + "num_tokens": 424652210.0, + "step": 16424 + }, + { + "epoch": 1.8037557654293872, + "grad_norm": 2.3838822841644287, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.72957444190979, + "num_tokens": 424676026.0, + "step": 16425 + }, + { + "epoch": 1.803865583132001, + "grad_norm": 2.529869556427002, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.711631178855896, + "num_tokens": 424697535.0, + "step": 16426 + }, + { + "epoch": 1.8039754008346145, + "grad_norm": 2.2681493759155273, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7304285168647766, + "num_tokens": 424725156.0, + "step": 16427 + }, + { + "epoch": 1.8040852185372283, + "grad_norm": 2.539736032485962, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.714198887348175, + "num_tokens": 424746333.0, + "step": 16428 + }, + { + "epoch": 1.8041950362398418, + "grad_norm": 2.2969517707824707, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7358014583587646, + "num_tokens": 424771557.0, + "step": 16429 + }, + { + "epoch": 1.8043048539424555, + "grad_norm": 2.4622650146484375, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.713088870048523, + "num_tokens": 424792970.0, + "step": 16430 + }, + { + "epoch": 1.8044146716450693, + "grad_norm": 2.5218288898468018, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7586596012115479, + "num_tokens": 424812348.0, + "step": 16431 + }, + { + "epoch": 1.8045244893476828, + "grad_norm": 2.1163952350616455, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7070289850234985, + "num_tokens": 424840825.0, + "step": 16432 + }, + { + "epoch": 1.8046343070502964, + "grad_norm": 2.3391432762145996, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7404150366783142, + "num_tokens": 424864673.0, + "step": 16433 + }, + { + "epoch": 1.8047441247529101, + "grad_norm": 2.445235252380371, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7297766208648682, + "num_tokens": 424885996.0, + "step": 16434 + }, + { + "epoch": 1.8048539424555239, + "grad_norm": 2.2511825561523438, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7147265672683716, + "num_tokens": 424913553.0, + "step": 16435 + }, + { + "epoch": 1.8049637601581376, + "grad_norm": 2.4580090045928955, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7097809314727783, + "num_tokens": 424936463.0, + "step": 16436 + }, + { + "epoch": 1.8050735778607512, + "grad_norm": 2.1203043460845947, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7038105726242065, + "num_tokens": 424967622.0, + "step": 16437 + }, + { + "epoch": 1.8051833955633647, + "grad_norm": 2.1523783206939697, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7093074917793274, + "num_tokens": 424996346.0, + "step": 16438 + }, + { + "epoch": 1.8052932132659785, + "grad_norm": 2.536245822906494, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7292723655700684, + "num_tokens": 425016697.0, + "step": 16439 + }, + { + "epoch": 1.8054030309685922, + "grad_norm": 2.3787684440612793, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7224360108375549, + "num_tokens": 425040663.0, + "step": 16440 + }, + { + "epoch": 1.8055128486712058, + "grad_norm": 2.2091875076293945, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7350655794143677, + "num_tokens": 425066619.0, + "step": 16441 + }, + { + "epoch": 1.8056226663738193, + "grad_norm": 2.0811667442321777, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7142896056175232, + "num_tokens": 425095651.0, + "step": 16442 + }, + { + "epoch": 1.805732484076433, + "grad_norm": 1.9308441877365112, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7036311626434326, + "num_tokens": 425129103.0, + "step": 16443 + }, + { + "epoch": 1.8058423017790468, + "grad_norm": 2.121335506439209, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7194443345069885, + "num_tokens": 425158032.0, + "step": 16444 + }, + { + "epoch": 1.8059521194816606, + "grad_norm": 2.3969039916992188, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7178477048873901, + "num_tokens": 425179784.0, + "step": 16445 + }, + { + "epoch": 1.806061937184274, + "grad_norm": 2.227484941482544, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7356188893318176, + "num_tokens": 425204646.0, + "step": 16446 + }, + { + "epoch": 1.8061717548868876, + "grad_norm": 2.3855607509613037, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7006329894065857, + "num_tokens": 425230154.0, + "step": 16447 + }, + { + "epoch": 1.8062815725895014, + "grad_norm": 2.2536938190460205, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7307843565940857, + "num_tokens": 425252949.0, + "step": 16448 + }, + { + "epoch": 1.8063913902921152, + "grad_norm": 2.1927554607391357, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7188746333122253, + "num_tokens": 425279039.0, + "step": 16449 + }, + { + "epoch": 1.806501207994729, + "grad_norm": 1.8010306358337402, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7135805487632751, + "num_tokens": 425314120.0, + "step": 16450 + }, + { + "epoch": 1.8066110256973424, + "grad_norm": 2.1576192378997803, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7254639863967896, + "num_tokens": 425341235.0, + "step": 16451 + }, + { + "epoch": 1.806720843399956, + "grad_norm": 2.356884479522705, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.748183012008667, + "num_tokens": 425362936.0, + "step": 16452 + }, + { + "epoch": 1.8068306611025697, + "grad_norm": 2.274963855743408, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7139297723770142, + "num_tokens": 425387980.0, + "step": 16453 + }, + { + "epoch": 1.8069404788051835, + "grad_norm": 2.2120742797851562, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7292640209197998, + "num_tokens": 425415000.0, + "step": 16454 + }, + { + "epoch": 1.807050296507797, + "grad_norm": 2.4066662788391113, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7222324013710022, + "num_tokens": 425438014.0, + "step": 16455 + }, + { + "epoch": 1.8071601142104106, + "grad_norm": 2.102665424346924, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7076787352561951, + "num_tokens": 425465809.0, + "step": 16456 + }, + { + "epoch": 1.8072699319130243, + "grad_norm": 2.122047185897827, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7066613435745239, + "num_tokens": 425494739.0, + "step": 16457 + }, + { + "epoch": 1.807379749615638, + "grad_norm": 2.213303804397583, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7170995473861694, + "num_tokens": 425519222.0, + "step": 16458 + }, + { + "epoch": 1.8074895673182518, + "grad_norm": 2.197190761566162, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7020981311798096, + "num_tokens": 425546860.0, + "step": 16459 + }, + { + "epoch": 1.8075993850208654, + "grad_norm": 2.0637857913970947, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.690869152545929, + "num_tokens": 425578098.0, + "step": 16460 + }, + { + "epoch": 1.807709202723479, + "grad_norm": 2.2886812686920166, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7115050554275513, + "num_tokens": 425603109.0, + "step": 16461 + }, + { + "epoch": 1.8078190204260927, + "grad_norm": 2.479571580886841, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7016357183456421, + "num_tokens": 425624711.0, + "step": 16462 + }, + { + "epoch": 1.8079288381287064, + "grad_norm": 2.1586477756500244, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.715071439743042, + "num_tokens": 425651333.0, + "step": 16463 + }, + { + "epoch": 1.8080386558313202, + "grad_norm": 2.0706331729888916, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7183641195297241, + "num_tokens": 425681369.0, + "step": 16464 + }, + { + "epoch": 1.8081484735339337, + "grad_norm": 2.2716705799102783, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.751430332660675, + "num_tokens": 425705748.0, + "step": 16465 + }, + { + "epoch": 1.8082582912365472, + "grad_norm": 2.184535503387451, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7493047118186951, + "num_tokens": 425730636.0, + "step": 16466 + }, + { + "epoch": 1.808368108939161, + "grad_norm": 2.1460916996002197, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7096825242042542, + "num_tokens": 425758932.0, + "step": 16467 + }, + { + "epoch": 1.8084779266417748, + "grad_norm": 2.5556247234344482, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7265224456787109, + "num_tokens": 425778446.0, + "step": 16468 + }, + { + "epoch": 1.8085877443443883, + "grad_norm": 2.0285046100616455, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7289148569107056, + "num_tokens": 425807780.0, + "step": 16469 + }, + { + "epoch": 1.8086975620470018, + "grad_norm": 2.2208211421966553, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.6987004280090332, + "num_tokens": 425834352.0, + "step": 16470 + }, + { + "epoch": 1.8088073797496156, + "grad_norm": 2.031707286834717, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7084035873413086, + "num_tokens": 425864174.0, + "step": 16471 + }, + { + "epoch": 1.8089171974522293, + "grad_norm": 2.006243944168091, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7262781858444214, + "num_tokens": 425894335.0, + "step": 16472 + }, + { + "epoch": 1.809027015154843, + "grad_norm": 2.119422674179077, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7071380615234375, + "num_tokens": 425925002.0, + "step": 16473 + }, + { + "epoch": 1.8091368328574566, + "grad_norm": 2.2637202739715576, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7115935683250427, + "num_tokens": 425950776.0, + "step": 16474 + }, + { + "epoch": 1.8092466505600702, + "grad_norm": 2.0807127952575684, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7118194103240967, + "num_tokens": 425977702.0, + "step": 16475 + }, + { + "epoch": 1.809356468262684, + "grad_norm": 2.364982843399048, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7425117492675781, + "num_tokens": 425999777.0, + "step": 16476 + }, + { + "epoch": 1.8094662859652977, + "grad_norm": 2.4047305583953857, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7061265707015991, + "num_tokens": 426023601.0, + "step": 16477 + }, + { + "epoch": 1.8095761036679112, + "grad_norm": 2.0413622856140137, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7224279642105103, + "num_tokens": 426052195.0, + "step": 16478 + }, + { + "epoch": 1.809685921370525, + "grad_norm": 2.235565423965454, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7201252579689026, + "num_tokens": 426076543.0, + "step": 16479 + }, + { + "epoch": 1.8097957390731385, + "grad_norm": 2.239644765853882, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7086519002914429, + "num_tokens": 426102646.0, + "step": 16480 + }, + { + "epoch": 1.8099055567757523, + "grad_norm": 2.2053864002227783, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7268412709236145, + "num_tokens": 426128498.0, + "step": 16481 + }, + { + "epoch": 1.810015374478366, + "grad_norm": 2.285144805908203, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7227012515068054, + "num_tokens": 426152242.0, + "step": 16482 + }, + { + "epoch": 1.8101251921809796, + "grad_norm": 2.1704955101013184, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7168518900871277, + "num_tokens": 426179491.0, + "step": 16483 + }, + { + "epoch": 1.810235009883593, + "grad_norm": 2.092226982116699, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6869089603424072, + "num_tokens": 426210144.0, + "step": 16484 + }, + { + "epoch": 1.8103448275862069, + "grad_norm": 2.0463037490844727, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7070940732955933, + "num_tokens": 426240115.0, + "step": 16485 + }, + { + "epoch": 1.8104546452888206, + "grad_norm": 2.268986701965332, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7208466529846191, + "num_tokens": 426264571.0, + "step": 16486 + }, + { + "epoch": 1.8105644629914344, + "grad_norm": 1.9090605974197388, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6969575881958008, + "num_tokens": 426299083.0, + "step": 16487 + }, + { + "epoch": 1.810674280694048, + "grad_norm": 2.1729631423950195, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.717272162437439, + "num_tokens": 426324702.0, + "step": 16488 + }, + { + "epoch": 1.8107840983966614, + "grad_norm": 2.362433671951294, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7338820695877075, + "num_tokens": 426346607.0, + "step": 16489 + }, + { + "epoch": 1.8108939160992752, + "grad_norm": 2.2268848419189453, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7070837616920471, + "num_tokens": 426370961.0, + "step": 16490 + }, + { + "epoch": 1.811003733801889, + "grad_norm": 2.082318067550659, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7221634984016418, + "num_tokens": 426401716.0, + "step": 16491 + }, + { + "epoch": 1.8111135515045025, + "grad_norm": 2.5602362155914307, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7146661281585693, + "num_tokens": 426423232.0, + "step": 16492 + }, + { + "epoch": 1.8112233692071162, + "grad_norm": 2.58673357963562, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7204972505569458, + "num_tokens": 426444908.0, + "step": 16493 + }, + { + "epoch": 1.8113331869097298, + "grad_norm": 2.568734645843506, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7481810450553894, + "num_tokens": 426464873.0, + "step": 16494 + }, + { + "epoch": 1.8114430046123435, + "grad_norm": 2.000540256500244, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7141556739807129, + "num_tokens": 426493448.0, + "step": 16495 + }, + { + "epoch": 1.8115528223149573, + "grad_norm": 2.5064938068389893, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.720941424369812, + "num_tokens": 426514810.0, + "step": 16496 + }, + { + "epoch": 1.8116626400175708, + "grad_norm": 2.4896528720855713, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7381420731544495, + "num_tokens": 426535324.0, + "step": 16497 + }, + { + "epoch": 1.8117724577201844, + "grad_norm": 2.1243131160736084, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7218759059906006, + "num_tokens": 426563419.0, + "step": 16498 + }, + { + "epoch": 1.8118822754227981, + "grad_norm": 2.1848831176757812, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7308944463729858, + "num_tokens": 426589380.0, + "step": 16499 + }, + { + "epoch": 1.8119920931254119, + "grad_norm": 2.4134628772735596, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7261416912078857, + "num_tokens": 426613698.0, + "step": 16500 + }, + { + "epoch": 1.8121019108280256, + "grad_norm": 1.8198457956314087, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6926013231277466, + "num_tokens": 426652593.0, + "step": 16501 + }, + { + "epoch": 1.8122117285306392, + "grad_norm": 2.058881998062134, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7049511671066284, + "num_tokens": 426682565.0, + "step": 16502 + }, + { + "epoch": 1.8123215462332527, + "grad_norm": 1.7947009801864624, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7131409049034119, + "num_tokens": 426718171.0, + "step": 16503 + }, + { + "epoch": 1.8124313639358665, + "grad_norm": 1.8999196290969849, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.730045735836029, + "num_tokens": 426751561.0, + "step": 16504 + }, + { + "epoch": 1.8125411816384802, + "grad_norm": 2.295985460281372, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.702889084815979, + "num_tokens": 426776114.0, + "step": 16505 + }, + { + "epoch": 1.8126509993410937, + "grad_norm": 2.1021687984466553, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7091863751411438, + "num_tokens": 426805410.0, + "step": 16506 + }, + { + "epoch": 1.8127608170437073, + "grad_norm": 2.213632345199585, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7016621232032776, + "num_tokens": 426831192.0, + "step": 16507 + }, + { + "epoch": 1.812870634746321, + "grad_norm": 2.1459426879882812, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7024326324462891, + "num_tokens": 426858715.0, + "step": 16508 + }, + { + "epoch": 1.8129804524489348, + "grad_norm": 2.124192476272583, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.699607253074646, + "num_tokens": 426888925.0, + "step": 16509 + }, + { + "epoch": 1.8130902701515486, + "grad_norm": 2.3516409397125244, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7249406576156616, + "num_tokens": 426911423.0, + "step": 16510 + }, + { + "epoch": 1.813200087854162, + "grad_norm": 2.439864158630371, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7054432034492493, + "num_tokens": 426935761.0, + "step": 16511 + }, + { + "epoch": 1.8133099055567756, + "grad_norm": 2.1988813877105713, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7276624441146851, + "num_tokens": 426963565.0, + "step": 16512 + }, + { + "epoch": 1.8134197232593894, + "grad_norm": 2.208061695098877, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.73626708984375, + "num_tokens": 426988193.0, + "step": 16513 + }, + { + "epoch": 1.8135295409620031, + "grad_norm": 1.86734938621521, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.6880507469177246, + "num_tokens": 427022948.0, + "step": 16514 + }, + { + "epoch": 1.813639358664617, + "grad_norm": 2.07132625579834, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7196749448776245, + "num_tokens": 427052540.0, + "step": 16515 + }, + { + "epoch": 1.8137491763672304, + "grad_norm": 2.318474054336548, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7201840877532959, + "num_tokens": 427075944.0, + "step": 16516 + }, + { + "epoch": 1.813858994069844, + "grad_norm": 1.9007914066314697, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7248095273971558, + "num_tokens": 427108854.0, + "step": 16517 + }, + { + "epoch": 1.8139688117724577, + "grad_norm": 2.41178297996521, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7185163497924805, + "num_tokens": 427131274.0, + "step": 16518 + }, + { + "epoch": 1.8140786294750715, + "grad_norm": 2.346522569656372, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7154775261878967, + "num_tokens": 427155856.0, + "step": 16519 + }, + { + "epoch": 1.814188447177685, + "grad_norm": 2.1275711059570312, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6846674680709839, + "num_tokens": 427185417.0, + "step": 16520 + }, + { + "epoch": 1.8142982648802986, + "grad_norm": 2.165170907974243, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6940634250640869, + "num_tokens": 427212866.0, + "step": 16521 + }, + { + "epoch": 1.8144080825829123, + "grad_norm": 2.155174732208252, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.709274411201477, + "num_tokens": 427238942.0, + "step": 16522 + }, + { + "epoch": 1.814517900285526, + "grad_norm": 2.17626690864563, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7220412492752075, + "num_tokens": 427264887.0, + "step": 16523 + }, + { + "epoch": 1.8146277179881398, + "grad_norm": 2.0577454566955566, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7343422174453735, + "num_tokens": 427292572.0, + "step": 16524 + }, + { + "epoch": 1.8147375356907534, + "grad_norm": 2.3475236892700195, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7087335586547852, + "num_tokens": 427316787.0, + "step": 16525 + }, + { + "epoch": 1.814847353393367, + "grad_norm": 2.1159780025482178, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6915497779846191, + "num_tokens": 427348035.0, + "step": 16526 + }, + { + "epoch": 1.8149571710959806, + "grad_norm": 2.2813286781311035, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7212738394737244, + "num_tokens": 427373501.0, + "step": 16527 + }, + { + "epoch": 1.8150669887985944, + "grad_norm": 2.5711498260498047, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.734430730342865, + "num_tokens": 427392578.0, + "step": 16528 + }, + { + "epoch": 1.815176806501208, + "grad_norm": 2.127004861831665, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7217617630958557, + "num_tokens": 427420076.0, + "step": 16529 + }, + { + "epoch": 1.8152866242038217, + "grad_norm": 2.2604522705078125, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7066408395767212, + "num_tokens": 427447000.0, + "step": 16530 + }, + { + "epoch": 1.8153964419064352, + "grad_norm": 2.523719310760498, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7148796319961548, + "num_tokens": 427467724.0, + "step": 16531 + }, + { + "epoch": 1.815506259609049, + "grad_norm": 2.0304605960845947, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7299193143844604, + "num_tokens": 427498695.0, + "step": 16532 + }, + { + "epoch": 1.8156160773116627, + "grad_norm": 2.262518882751465, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7214060425758362, + "num_tokens": 427524312.0, + "step": 16533 + }, + { + "epoch": 1.8157258950142763, + "grad_norm": 2.2503092288970947, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6928733587265015, + "num_tokens": 427552420.0, + "step": 16534 + }, + { + "epoch": 1.8158357127168898, + "grad_norm": 2.27337646484375, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7245276570320129, + "num_tokens": 427577265.0, + "step": 16535 + }, + { + "epoch": 1.8159455304195036, + "grad_norm": 2.0019569396972656, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7294973731040955, + "num_tokens": 427607241.0, + "step": 16536 + }, + { + "epoch": 1.8160553481221173, + "grad_norm": 2.633074998855591, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7271848917007446, + "num_tokens": 427626819.0, + "step": 16537 + }, + { + "epoch": 1.816165165824731, + "grad_norm": 2.033841609954834, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7289494872093201, + "num_tokens": 427656171.0, + "step": 16538 + }, + { + "epoch": 1.8162749835273446, + "grad_norm": 2.1287925243377686, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.6993891000747681, + "num_tokens": 427683797.0, + "step": 16539 + }, + { + "epoch": 1.8163848012299582, + "grad_norm": 2.0744664669036865, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7404299974441528, + "num_tokens": 427710480.0, + "step": 16540 + }, + { + "epoch": 1.816494618932572, + "grad_norm": 2.1386334896087646, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7233757972717285, + "num_tokens": 427736371.0, + "step": 16541 + }, + { + "epoch": 1.8166044366351857, + "grad_norm": 2.393559455871582, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7076759338378906, + "num_tokens": 427758803.0, + "step": 16542 + }, + { + "epoch": 1.8167142543377992, + "grad_norm": 2.5174551010131836, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7322686910629272, + "num_tokens": 427779095.0, + "step": 16543 + }, + { + "epoch": 1.816824072040413, + "grad_norm": 2.5523200035095215, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7075115442276001, + "num_tokens": 427799114.0, + "step": 16544 + }, + { + "epoch": 1.8169338897430265, + "grad_norm": 2.1679553985595703, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6962307095527649, + "num_tokens": 427826069.0, + "step": 16545 + }, + { + "epoch": 1.8170437074456403, + "grad_norm": 2.3301901817321777, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7219321131706238, + "num_tokens": 427849250.0, + "step": 16546 + }, + { + "epoch": 1.817153525148254, + "grad_norm": 2.079918622970581, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.706453800201416, + "num_tokens": 427880158.0, + "step": 16547 + }, + { + "epoch": 1.8172633428508675, + "grad_norm": 1.9306175708770752, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7234376072883606, + "num_tokens": 427911055.0, + "step": 16548 + }, + { + "epoch": 1.817373160553481, + "grad_norm": 1.9996492862701416, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.700476884841919, + "num_tokens": 427941798.0, + "step": 16549 + }, + { + "epoch": 1.8174829782560948, + "grad_norm": 2.291424036026001, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7297770977020264, + "num_tokens": 427965018.0, + "step": 16550 + }, + { + "epoch": 1.8175927959587086, + "grad_norm": 2.439685583114624, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7203645706176758, + "num_tokens": 427986831.0, + "step": 16551 + }, + { + "epoch": 1.8177026136613224, + "grad_norm": 2.271312713623047, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7416587471961975, + "num_tokens": 428011835.0, + "step": 16552 + }, + { + "epoch": 1.8178124313639359, + "grad_norm": 2.7321579456329346, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7284828424453735, + "num_tokens": 428030219.0, + "step": 16553 + }, + { + "epoch": 1.8179222490665494, + "grad_norm": 2.544611930847168, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7247447371482849, + "num_tokens": 428050693.0, + "step": 16554 + }, + { + "epoch": 1.8180320667691632, + "grad_norm": 2.244854688644409, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7265774011611938, + "num_tokens": 428075890.0, + "step": 16555 + }, + { + "epoch": 1.818141884471777, + "grad_norm": 2.1187875270843506, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7099450826644897, + "num_tokens": 428104492.0, + "step": 16556 + }, + { + "epoch": 1.8182517021743905, + "grad_norm": 2.5049681663513184, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7242227792739868, + "num_tokens": 428125142.0, + "step": 16557 + }, + { + "epoch": 1.818361519877004, + "grad_norm": 2.2840042114257812, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.716467559337616, + "num_tokens": 428149052.0, + "step": 16558 + }, + { + "epoch": 1.8184713375796178, + "grad_norm": 2.087834358215332, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7483252882957458, + "num_tokens": 428175138.0, + "step": 16559 + }, + { + "epoch": 1.8185811552822315, + "grad_norm": 2.2023205757141113, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7210972309112549, + "num_tokens": 428201616.0, + "step": 16560 + }, + { + "epoch": 1.8186909729848453, + "grad_norm": 2.133300304412842, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7279508113861084, + "num_tokens": 428229204.0, + "step": 16561 + }, + { + "epoch": 1.8188007906874588, + "grad_norm": 2.307417631149292, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7182324528694153, + "num_tokens": 428252398.0, + "step": 16562 + }, + { + "epoch": 1.8189106083900723, + "grad_norm": 2.1193959712982178, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7350753545761108, + "num_tokens": 428280091.0, + "step": 16563 + }, + { + "epoch": 1.819020426092686, + "grad_norm": 2.547050714492798, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7436679601669312, + "num_tokens": 428303463.0, + "step": 16564 + }, + { + "epoch": 1.8191302437952999, + "grad_norm": 2.3701043128967285, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.721586287021637, + "num_tokens": 428326495.0, + "step": 16565 + }, + { + "epoch": 1.8192400614979136, + "grad_norm": 2.048030138015747, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.6997706890106201, + "num_tokens": 428356987.0, + "step": 16566 + }, + { + "epoch": 1.8193498792005272, + "grad_norm": 2.3741281032562256, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7118220329284668, + "num_tokens": 428382039.0, + "step": 16567 + }, + { + "epoch": 1.8194596969031407, + "grad_norm": 2.1562628746032715, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7236146926879883, + "num_tokens": 428407816.0, + "step": 16568 + }, + { + "epoch": 1.8195695146057544, + "grad_norm": 2.1396994590759277, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7185625433921814, + "num_tokens": 428434878.0, + "step": 16569 + }, + { + "epoch": 1.8196793323083682, + "grad_norm": 2.140775442123413, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7136304378509521, + "num_tokens": 428461640.0, + "step": 16570 + }, + { + "epoch": 1.8197891500109817, + "grad_norm": 2.2679879665374756, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7164283394813538, + "num_tokens": 428486922.0, + "step": 16571 + }, + { + "epoch": 1.8198989677135953, + "grad_norm": 2.4436750411987305, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7509687542915344, + "num_tokens": 428509178.0, + "step": 16572 + }, + { + "epoch": 1.820008785416209, + "grad_norm": 2.789712429046631, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7330677509307861, + "num_tokens": 428526688.0, + "step": 16573 + }, + { + "epoch": 1.8201186031188228, + "grad_norm": 2.2601935863494873, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7072263956069946, + "num_tokens": 428552595.0, + "step": 16574 + }, + { + "epoch": 1.8202284208214365, + "grad_norm": 2.16641902923584, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7310718297958374, + "num_tokens": 428580590.0, + "step": 16575 + }, + { + "epoch": 1.82033823852405, + "grad_norm": 1.971818447113037, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7288760542869568, + "num_tokens": 428610668.0, + "step": 16576 + }, + { + "epoch": 1.8204480562266636, + "grad_norm": 2.506964683532715, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7311698794364929, + "num_tokens": 428630137.0, + "step": 16577 + }, + { + "epoch": 1.8205578739292774, + "grad_norm": 2.2206850051879883, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.735346257686615, + "num_tokens": 428656268.0, + "step": 16578 + }, + { + "epoch": 1.8206676916318911, + "grad_norm": 2.1214118003845215, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7280739545822144, + "num_tokens": 428684183.0, + "step": 16579 + }, + { + "epoch": 1.8207775093345049, + "grad_norm": 2.27131724357605, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.738471269607544, + "num_tokens": 428708305.0, + "step": 16580 + }, + { + "epoch": 1.8208873270371184, + "grad_norm": 2.2697360515594482, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7291221022605896, + "num_tokens": 428732776.0, + "step": 16581 + }, + { + "epoch": 1.820997144739732, + "grad_norm": 1.971171259880066, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7113820314407349, + "num_tokens": 428764808.0, + "step": 16582 + }, + { + "epoch": 1.8211069624423457, + "grad_norm": 2.551046133041382, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7160300016403198, + "num_tokens": 428784878.0, + "step": 16583 + }, + { + "epoch": 1.8212167801449595, + "grad_norm": 2.008596181869507, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7382674217224121, + "num_tokens": 428815897.0, + "step": 16584 + }, + { + "epoch": 1.821326597847573, + "grad_norm": 2.465730667114258, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7338926792144775, + "num_tokens": 428837386.0, + "step": 16585 + }, + { + "epoch": 1.8214364155501865, + "grad_norm": 2.0381929874420166, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7293059825897217, + "num_tokens": 428863800.0, + "step": 16586 + }, + { + "epoch": 1.8215462332528003, + "grad_norm": 2.467528820037842, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7499266266822815, + "num_tokens": 428884082.0, + "step": 16587 + }, + { + "epoch": 1.821656050955414, + "grad_norm": 2.5207886695861816, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7431721687316895, + "num_tokens": 428904607.0, + "step": 16588 + }, + { + "epoch": 1.8217658686580278, + "grad_norm": 2.201418161392212, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7173681259155273, + "num_tokens": 428930390.0, + "step": 16589 + }, + { + "epoch": 1.8218756863606413, + "grad_norm": 2.3489482402801514, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7211042642593384, + "num_tokens": 428955621.0, + "step": 16590 + }, + { + "epoch": 1.8219855040632549, + "grad_norm": 2.0662357807159424, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7382863759994507, + "num_tokens": 428986531.0, + "step": 16591 + }, + { + "epoch": 1.8220953217658686, + "grad_norm": 2.1647167205810547, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7354561686515808, + "num_tokens": 429011547.0, + "step": 16592 + }, + { + "epoch": 1.8222051394684824, + "grad_norm": 2.242762804031372, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7181488871574402, + "num_tokens": 429037503.0, + "step": 16593 + }, + { + "epoch": 1.822314957171096, + "grad_norm": 2.2967536449432373, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7110762596130371, + "num_tokens": 429063438.0, + "step": 16594 + }, + { + "epoch": 1.8224247748737097, + "grad_norm": 2.1916778087615967, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.726719319820404, + "num_tokens": 429089960.0, + "step": 16595 + }, + { + "epoch": 1.8225345925763232, + "grad_norm": 2.377441883087158, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7211235761642456, + "num_tokens": 429112940.0, + "step": 16596 + }, + { + "epoch": 1.822644410278937, + "grad_norm": 2.1382734775543213, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7415986061096191, + "num_tokens": 429139904.0, + "step": 16597 + }, + { + "epoch": 1.8227542279815507, + "grad_norm": 1.965072751045227, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.712212085723877, + "num_tokens": 429172666.0, + "step": 16598 + }, + { + "epoch": 1.8228640456841643, + "grad_norm": 2.3964288234710693, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7263495326042175, + "num_tokens": 429195114.0, + "step": 16599 + }, + { + "epoch": 1.8229738633867778, + "grad_norm": 2.1085004806518555, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7021282911300659, + "num_tokens": 429222275.0, + "step": 16600 + }, + { + "epoch": 1.8230836810893916, + "grad_norm": 2.1934280395507812, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7037276029586792, + "num_tokens": 429248468.0, + "step": 16601 + }, + { + "epoch": 1.8231934987920053, + "grad_norm": 2.0833449363708496, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7124001979827881, + "num_tokens": 429276517.0, + "step": 16602 + }, + { + "epoch": 1.823303316494619, + "grad_norm": 2.030622959136963, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7095558643341064, + "num_tokens": 429305965.0, + "step": 16603 + }, + { + "epoch": 1.8234131341972326, + "grad_norm": 2.296823024749756, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7129817008972168, + "num_tokens": 429329134.0, + "step": 16604 + }, + { + "epoch": 1.8235229518998461, + "grad_norm": 2.112946033477783, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7111250162124634, + "num_tokens": 429358836.0, + "step": 16605 + }, + { + "epoch": 1.82363276960246, + "grad_norm": 2.527461528778076, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7381908893585205, + "num_tokens": 429380611.0, + "step": 16606 + }, + { + "epoch": 1.8237425873050737, + "grad_norm": 2.201589584350586, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7036325931549072, + "num_tokens": 429407180.0, + "step": 16607 + }, + { + "epoch": 1.8238524050076872, + "grad_norm": 2.0918657779693604, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.696540355682373, + "num_tokens": 429435290.0, + "step": 16608 + }, + { + "epoch": 1.823962222710301, + "grad_norm": 2.3333840370178223, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7221404314041138, + "num_tokens": 429460877.0, + "step": 16609 + }, + { + "epoch": 1.8240720404129145, + "grad_norm": 2.049591302871704, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7256457209587097, + "num_tokens": 429490257.0, + "step": 16610 + }, + { + "epoch": 1.8241818581155282, + "grad_norm": 2.1135497093200684, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7008626461029053, + "num_tokens": 429518656.0, + "step": 16611 + }, + { + "epoch": 1.824291675818142, + "grad_norm": 1.9539722204208374, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6944885849952698, + "num_tokens": 429551292.0, + "step": 16612 + }, + { + "epoch": 1.8244014935207555, + "grad_norm": 1.9845038652420044, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6997768878936768, + "num_tokens": 429581395.0, + "step": 16613 + }, + { + "epoch": 1.824511311223369, + "grad_norm": 2.062574863433838, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7248589396476746, + "num_tokens": 429610545.0, + "step": 16614 + }, + { + "epoch": 1.8246211289259828, + "grad_norm": 2.044811487197876, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7047567963600159, + "num_tokens": 429640094.0, + "step": 16615 + }, + { + "epoch": 1.8247309466285966, + "grad_norm": 2.3312461376190186, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.6949363350868225, + "num_tokens": 429664718.0, + "step": 16616 + }, + { + "epoch": 1.8248407643312103, + "grad_norm": 2.0207695960998535, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7247512340545654, + "num_tokens": 429693462.0, + "step": 16617 + }, + { + "epoch": 1.8249505820338239, + "grad_norm": 2.1314046382904053, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6945478320121765, + "num_tokens": 429722076.0, + "step": 16618 + }, + { + "epoch": 1.8250603997364374, + "grad_norm": 2.04488468170166, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7046912312507629, + "num_tokens": 429751375.0, + "step": 16619 + }, + { + "epoch": 1.8251702174390512, + "grad_norm": 2.398529291152954, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7219186425209045, + "num_tokens": 429773201.0, + "step": 16620 + }, + { + "epoch": 1.825280035141665, + "grad_norm": 2.0744946002960205, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7428469657897949, + "num_tokens": 429801196.0, + "step": 16621 + }, + { + "epoch": 1.8253898528442785, + "grad_norm": 2.3473427295684814, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7245831489562988, + "num_tokens": 429823930.0, + "step": 16622 + }, + { + "epoch": 1.825499670546892, + "grad_norm": 2.151416301727295, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.71527099609375, + "num_tokens": 429852022.0, + "step": 16623 + }, + { + "epoch": 1.8256094882495058, + "grad_norm": 2.0600974559783936, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.6996821165084839, + "num_tokens": 429881626.0, + "step": 16624 + }, + { + "epoch": 1.8257193059521195, + "grad_norm": 2.0659122467041016, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7186083793640137, + "num_tokens": 429910784.0, + "step": 16625 + }, + { + "epoch": 1.8258291236547333, + "grad_norm": 2.0254786014556885, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7221134901046753, + "num_tokens": 429940358.0, + "step": 16626 + }, + { + "epoch": 1.8259389413573468, + "grad_norm": 2.0812582969665527, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.6978870034217834, + "num_tokens": 429969579.0, + "step": 16627 + }, + { + "epoch": 1.8260487590599603, + "grad_norm": 2.2609570026397705, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7065083384513855, + "num_tokens": 429996881.0, + "step": 16628 + }, + { + "epoch": 1.826158576762574, + "grad_norm": 2.241360902786255, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7263699769973755, + "num_tokens": 430022276.0, + "step": 16629 + }, + { + "epoch": 1.8262683944651878, + "grad_norm": 2.237835168838501, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7249477505683899, + "num_tokens": 430047514.0, + "step": 16630 + }, + { + "epoch": 1.8263782121678016, + "grad_norm": 2.3782124519348145, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7207564115524292, + "num_tokens": 430068455.0, + "step": 16631 + }, + { + "epoch": 1.8264880298704151, + "grad_norm": 2.3877131938934326, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.71831876039505, + "num_tokens": 430091890.0, + "step": 16632 + }, + { + "epoch": 1.8265978475730287, + "grad_norm": 2.00504994392395, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6928269267082214, + "num_tokens": 430122910.0, + "step": 16633 + }, + { + "epoch": 1.8267076652756424, + "grad_norm": 2.2941577434539795, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7099766731262207, + "num_tokens": 430148544.0, + "step": 16634 + }, + { + "epoch": 1.8268174829782562, + "grad_norm": 2.121708393096924, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.696678876876831, + "num_tokens": 430177519.0, + "step": 16635 + }, + { + "epoch": 1.8269273006808697, + "grad_norm": 2.334012269973755, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7174640893936157, + "num_tokens": 430202491.0, + "step": 16636 + }, + { + "epoch": 1.8270371183834833, + "grad_norm": 2.253232955932617, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7266045808792114, + "num_tokens": 430228359.0, + "step": 16637 + }, + { + "epoch": 1.827146936086097, + "grad_norm": 2.657503366470337, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7238866090774536, + "num_tokens": 430246665.0, + "step": 16638 + }, + { + "epoch": 1.8272567537887108, + "grad_norm": 2.252260684967041, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7265132665634155, + "num_tokens": 430271257.0, + "step": 16639 + }, + { + "epoch": 1.8273665714913245, + "grad_norm": 2.173760414123535, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7458215951919556, + "num_tokens": 430295702.0, + "step": 16640 + }, + { + "epoch": 1.827476389193938, + "grad_norm": 2.3204681873321533, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7268906831741333, + "num_tokens": 430319395.0, + "step": 16641 + }, + { + "epoch": 1.8275862068965516, + "grad_norm": 2.2824485301971436, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7154425382614136, + "num_tokens": 430344876.0, + "step": 16642 + }, + { + "epoch": 1.8276960245991654, + "grad_norm": 2.1339566707611084, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7380674481391907, + "num_tokens": 430369863.0, + "step": 16643 + }, + { + "epoch": 1.8278058423017791, + "grad_norm": 2.375598192214966, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7132800221443176, + "num_tokens": 430394954.0, + "step": 16644 + }, + { + "epoch": 1.8279156600043929, + "grad_norm": 1.9570674896240234, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.6985839605331421, + "num_tokens": 430426202.0, + "step": 16645 + }, + { + "epoch": 1.8280254777070064, + "grad_norm": 2.2400577068328857, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7082582712173462, + "num_tokens": 430452728.0, + "step": 16646 + }, + { + "epoch": 1.82813529540962, + "grad_norm": 2.3120195865631104, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7216784954071045, + "num_tokens": 430476884.0, + "step": 16647 + }, + { + "epoch": 1.8282451131122337, + "grad_norm": 2.2147765159606934, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.723013699054718, + "num_tokens": 430501811.0, + "step": 16648 + }, + { + "epoch": 1.8283549308148475, + "grad_norm": 2.1572506427764893, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.734440803527832, + "num_tokens": 430525781.0, + "step": 16649 + }, + { + "epoch": 1.828464748517461, + "grad_norm": 1.9969189167022705, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7273134589195251, + "num_tokens": 430556236.0, + "step": 16650 + }, + { + "epoch": 1.8285745662200745, + "grad_norm": 2.6790027618408203, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7328813672065735, + "num_tokens": 430575557.0, + "step": 16651 + }, + { + "epoch": 1.8286843839226883, + "grad_norm": 2.02987003326416, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7390758991241455, + "num_tokens": 430602998.0, + "step": 16652 + }, + { + "epoch": 1.828794201625302, + "grad_norm": 2.300358533859253, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7229150533676147, + "num_tokens": 430626485.0, + "step": 16653 + }, + { + "epoch": 1.8289040193279158, + "grad_norm": 2.1570863723754883, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7224974036216736, + "num_tokens": 430651993.0, + "step": 16654 + }, + { + "epoch": 1.8290138370305293, + "grad_norm": 2.1302318572998047, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7104777097702026, + "num_tokens": 430677563.0, + "step": 16655 + }, + { + "epoch": 1.8291236547331429, + "grad_norm": 2.083440065383911, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6985704898834229, + "num_tokens": 430705592.0, + "step": 16656 + }, + { + "epoch": 1.8292334724357566, + "grad_norm": 2.7798898220062256, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7213151454925537, + "num_tokens": 430725170.0, + "step": 16657 + }, + { + "epoch": 1.8293432901383704, + "grad_norm": 2.173525333404541, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7068815231323242, + "num_tokens": 430751860.0, + "step": 16658 + }, + { + "epoch": 1.829453107840984, + "grad_norm": 2.136051654815674, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7074798345565796, + "num_tokens": 430781725.0, + "step": 16659 + }, + { + "epoch": 1.8295629255435977, + "grad_norm": 2.1442904472351074, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7396957874298096, + "num_tokens": 430807699.0, + "step": 16660 + }, + { + "epoch": 1.8296727432462112, + "grad_norm": 2.269808769226074, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7283599972724915, + "num_tokens": 430832221.0, + "step": 16661 + }, + { + "epoch": 1.829782560948825, + "grad_norm": 2.2848823070526123, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7216598391532898, + "num_tokens": 430856288.0, + "step": 16662 + }, + { + "epoch": 1.8298923786514387, + "grad_norm": 2.3744428157806396, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.717482328414917, + "num_tokens": 430878692.0, + "step": 16663 + }, + { + "epoch": 1.8300021963540523, + "grad_norm": 2.2342376708984375, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7340015769004822, + "num_tokens": 430902390.0, + "step": 16664 + }, + { + "epoch": 1.8301120140566658, + "grad_norm": 2.2096455097198486, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7163025140762329, + "num_tokens": 430927656.0, + "step": 16665 + }, + { + "epoch": 1.8302218317592795, + "grad_norm": 2.157156229019165, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7227200269699097, + "num_tokens": 430955307.0, + "step": 16666 + }, + { + "epoch": 1.8303316494618933, + "grad_norm": 2.0871880054473877, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7176950573921204, + "num_tokens": 430984328.0, + "step": 16667 + }, + { + "epoch": 1.830441467164507, + "grad_norm": 1.858437180519104, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7064897418022156, + "num_tokens": 431019994.0, + "step": 16668 + }, + { + "epoch": 1.8305512848671206, + "grad_norm": 2.352552652359009, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7366361618041992, + "num_tokens": 431042177.0, + "step": 16669 + }, + { + "epoch": 1.8306611025697341, + "grad_norm": 2.1202175617218018, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7168506979942322, + "num_tokens": 431068939.0, + "step": 16670 + }, + { + "epoch": 1.830770920272348, + "grad_norm": 2.4436280727386475, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7071062326431274, + "num_tokens": 431089876.0, + "step": 16671 + }, + { + "epoch": 1.8308807379749616, + "grad_norm": 2.268125057220459, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.6977260112762451, + "num_tokens": 431114712.0, + "step": 16672 + }, + { + "epoch": 1.8309905556775752, + "grad_norm": 2.405569076538086, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.734117329120636, + "num_tokens": 431136945.0, + "step": 16673 + }, + { + "epoch": 1.831100373380189, + "grad_norm": 2.451748847961426, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7205733060836792, + "num_tokens": 431159197.0, + "step": 16674 + }, + { + "epoch": 1.8312101910828025, + "grad_norm": 2.319526433944702, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7102648615837097, + "num_tokens": 431183436.0, + "step": 16675 + }, + { + "epoch": 1.8313200087854162, + "grad_norm": 2.2262353897094727, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7168006896972656, + "num_tokens": 431208000.0, + "step": 16676 + }, + { + "epoch": 1.83142982648803, + "grad_norm": 2.4647867679595947, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7162668704986572, + "num_tokens": 431229066.0, + "step": 16677 + }, + { + "epoch": 1.8315396441906435, + "grad_norm": 2.318782329559326, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7320374250411987, + "num_tokens": 431253228.0, + "step": 16678 + }, + { + "epoch": 1.831649461893257, + "grad_norm": 2.2902424335479736, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.715785562992096, + "num_tokens": 431277606.0, + "step": 16679 + }, + { + "epoch": 1.8317592795958708, + "grad_norm": 2.0441553592681885, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.6976625919342041, + "num_tokens": 431306796.0, + "step": 16680 + }, + { + "epoch": 1.8318690972984846, + "grad_norm": 2.1857845783233643, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7138246297836304, + "num_tokens": 431332376.0, + "step": 16681 + }, + { + "epoch": 1.8319789150010983, + "grad_norm": 1.9610236883163452, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7230642437934875, + "num_tokens": 431361858.0, + "step": 16682 + }, + { + "epoch": 1.8320887327037119, + "grad_norm": 2.3090426921844482, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7214925289154053, + "num_tokens": 431385909.0, + "step": 16683 + }, + { + "epoch": 1.8321985504063254, + "grad_norm": 2.195322275161743, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.6995261311531067, + "num_tokens": 431411587.0, + "step": 16684 + }, + { + "epoch": 1.8323083681089392, + "grad_norm": 2.2642080783843994, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7182564735412598, + "num_tokens": 431438484.0, + "step": 16685 + }, + { + "epoch": 1.832418185811553, + "grad_norm": 2.3087871074676514, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7025800943374634, + "num_tokens": 431463481.0, + "step": 16686 + }, + { + "epoch": 1.8325280035141664, + "grad_norm": 2.1107351779937744, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7207306623458862, + "num_tokens": 431491588.0, + "step": 16687 + }, + { + "epoch": 1.83263782121678, + "grad_norm": 2.2436769008636475, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.746036171913147, + "num_tokens": 431517339.0, + "step": 16688 + }, + { + "epoch": 1.8327476389193937, + "grad_norm": 2.154855251312256, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7496795058250427, + "num_tokens": 431543952.0, + "step": 16689 + }, + { + "epoch": 1.8328574566220075, + "grad_norm": 2.4440114498138428, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.714612603187561, + "num_tokens": 431566628.0, + "step": 16690 + }, + { + "epoch": 1.8329672743246213, + "grad_norm": 2.205618143081665, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7104355692863464, + "num_tokens": 431595692.0, + "step": 16691 + }, + { + "epoch": 1.8330770920272348, + "grad_norm": 2.48323392868042, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7141217589378357, + "num_tokens": 431617405.0, + "step": 16692 + }, + { + "epoch": 1.8331869097298483, + "grad_norm": 2.5370378494262695, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7249317765235901, + "num_tokens": 431639808.0, + "step": 16693 + }, + { + "epoch": 1.833296727432462, + "grad_norm": 2.2272696495056152, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7262475490570068, + "num_tokens": 431665541.0, + "step": 16694 + }, + { + "epoch": 1.8334065451350758, + "grad_norm": 2.4348037242889404, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7456352114677429, + "num_tokens": 431686833.0, + "step": 16695 + }, + { + "epoch": 1.8335163628376896, + "grad_norm": 2.5163092613220215, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7153241634368896, + "num_tokens": 431707512.0, + "step": 16696 + }, + { + "epoch": 1.8336261805403031, + "grad_norm": 2.2140772342681885, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.721839189529419, + "num_tokens": 431734053.0, + "step": 16697 + }, + { + "epoch": 1.8337359982429167, + "grad_norm": 2.1890437602996826, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7108548283576965, + "num_tokens": 431762374.0, + "step": 16698 + }, + { + "epoch": 1.8338458159455304, + "grad_norm": 2.186720609664917, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7103770971298218, + "num_tokens": 431791432.0, + "step": 16699 + }, + { + "epoch": 1.8339556336481442, + "grad_norm": 2.2078542709350586, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7194938063621521, + "num_tokens": 431815212.0, + "step": 16700 + }, + { + "epoch": 1.8340654513507577, + "grad_norm": 2.185211181640625, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7405262589454651, + "num_tokens": 431840246.0, + "step": 16701 + }, + { + "epoch": 1.8341752690533712, + "grad_norm": 2.2295711040496826, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7134406566619873, + "num_tokens": 431865239.0, + "step": 16702 + }, + { + "epoch": 1.834285086755985, + "grad_norm": 2.2749083042144775, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7148614525794983, + "num_tokens": 431889686.0, + "step": 16703 + }, + { + "epoch": 1.8343949044585988, + "grad_norm": 2.0335075855255127, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7130590677261353, + "num_tokens": 431916534.0, + "step": 16704 + }, + { + "epoch": 1.8345047221612125, + "grad_norm": 2.3782217502593994, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7155420184135437, + "num_tokens": 431939591.0, + "step": 16705 + }, + { + "epoch": 1.834614539863826, + "grad_norm": 2.230478525161743, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7374222874641418, + "num_tokens": 431962665.0, + "step": 16706 + }, + { + "epoch": 1.8347243575664396, + "grad_norm": 2.2822346687316895, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7258961200714111, + "num_tokens": 431986591.0, + "step": 16707 + }, + { + "epoch": 1.8348341752690533, + "grad_norm": 2.2190520763397217, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7325854301452637, + "num_tokens": 432012179.0, + "step": 16708 + }, + { + "epoch": 1.834943992971667, + "grad_norm": 2.365400552749634, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7233357429504395, + "num_tokens": 432035559.0, + "step": 16709 + }, + { + "epoch": 1.8350538106742806, + "grad_norm": 2.3644344806671143, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.734413206577301, + "num_tokens": 432058218.0, + "step": 16710 + }, + { + "epoch": 1.8351636283768944, + "grad_norm": 2.162261486053467, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7292863130569458, + "num_tokens": 432087561.0, + "step": 16711 + }, + { + "epoch": 1.835273446079508, + "grad_norm": 2.075221538543701, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7332041263580322, + "num_tokens": 432115675.0, + "step": 16712 + }, + { + "epoch": 1.8353832637821217, + "grad_norm": 2.1574347019195557, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7049413919448853, + "num_tokens": 432143933.0, + "step": 16713 + }, + { + "epoch": 1.8354930814847354, + "grad_norm": 1.982373833656311, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7056832909584045, + "num_tokens": 432174272.0, + "step": 16714 + }, + { + "epoch": 1.835602899187349, + "grad_norm": 2.438434362411499, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7055885791778564, + "num_tokens": 432196508.0, + "step": 16715 + }, + { + "epoch": 1.8357127168899625, + "grad_norm": 2.490598440170288, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7149651050567627, + "num_tokens": 432219663.0, + "step": 16716 + }, + { + "epoch": 1.8358225345925763, + "grad_norm": 2.0326881408691406, + "learning_rate": 1e-06, + "loss": 1.0686, + "mean_token_accuracy": 0.6817591786384583, + "num_tokens": 432251645.0, + "step": 16717 + }, + { + "epoch": 1.83593235229519, + "grad_norm": 2.022427797317505, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7179131507873535, + "num_tokens": 432279998.0, + "step": 16718 + }, + { + "epoch": 1.8360421699978038, + "grad_norm": 2.1469197273254395, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.709595263004303, + "num_tokens": 432309369.0, + "step": 16719 + }, + { + "epoch": 1.8361519877004173, + "grad_norm": 2.1359355449676514, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7396444082260132, + "num_tokens": 432335263.0, + "step": 16720 + }, + { + "epoch": 1.8362618054030309, + "grad_norm": 2.312584161758423, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7142722010612488, + "num_tokens": 432362386.0, + "step": 16721 + }, + { + "epoch": 1.8363716231056446, + "grad_norm": 2.2934765815734863, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6936334371566772, + "num_tokens": 432388428.0, + "step": 16722 + }, + { + "epoch": 1.8364814408082584, + "grad_norm": 2.1261003017425537, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7064051628112793, + "num_tokens": 432415030.0, + "step": 16723 + }, + { + "epoch": 1.836591258510872, + "grad_norm": 2.229947805404663, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7210499048233032, + "num_tokens": 432438623.0, + "step": 16724 + }, + { + "epoch": 1.8367010762134857, + "grad_norm": 2.553518533706665, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7366694211959839, + "num_tokens": 432458752.0, + "step": 16725 + }, + { + "epoch": 1.8368108939160992, + "grad_norm": 2.2832469940185547, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7135212421417236, + "num_tokens": 432482944.0, + "step": 16726 + }, + { + "epoch": 1.836920711618713, + "grad_norm": 2.1023988723754883, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7174547910690308, + "num_tokens": 432510600.0, + "step": 16727 + }, + { + "epoch": 1.8370305293213267, + "grad_norm": 2.4127392768859863, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7441619634628296, + "num_tokens": 432532220.0, + "step": 16728 + }, + { + "epoch": 1.8371403470239402, + "grad_norm": 2.344301223754883, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7287974953651428, + "num_tokens": 432556467.0, + "step": 16729 + }, + { + "epoch": 1.8372501647265538, + "grad_norm": 1.9907537698745728, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7072030305862427, + "num_tokens": 432588323.0, + "step": 16730 + }, + { + "epoch": 1.8373599824291675, + "grad_norm": 2.3947668075561523, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7062705159187317, + "num_tokens": 432610998.0, + "step": 16731 + }, + { + "epoch": 1.8374698001317813, + "grad_norm": 2.2105531692504883, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7091994285583496, + "num_tokens": 432639635.0, + "step": 16732 + }, + { + "epoch": 1.837579617834395, + "grad_norm": 2.2481417655944824, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7390474081039429, + "num_tokens": 432665412.0, + "step": 16733 + }, + { + "epoch": 1.8376894355370086, + "grad_norm": 2.644152879714966, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.731860876083374, + "num_tokens": 432684610.0, + "step": 16734 + }, + { + "epoch": 1.8377992532396221, + "grad_norm": 2.4365334510803223, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7168312072753906, + "num_tokens": 432707798.0, + "step": 16735 + }, + { + "epoch": 1.8379090709422359, + "grad_norm": 2.367863416671753, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7304041981697083, + "num_tokens": 432731428.0, + "step": 16736 + }, + { + "epoch": 1.8380188886448496, + "grad_norm": 2.4205708503723145, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7139910459518433, + "num_tokens": 432755040.0, + "step": 16737 + }, + { + "epoch": 1.8381287063474632, + "grad_norm": 2.5260889530181885, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7466297149658203, + "num_tokens": 432775161.0, + "step": 16738 + }, + { + "epoch": 1.838238524050077, + "grad_norm": 2.43778395652771, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7268604040145874, + "num_tokens": 432797686.0, + "step": 16739 + }, + { + "epoch": 1.8383483417526905, + "grad_norm": 1.9349521398544312, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7033929824829102, + "num_tokens": 432831722.0, + "step": 16740 + }, + { + "epoch": 1.8384581594553042, + "grad_norm": 2.204676628112793, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7224957942962646, + "num_tokens": 432857199.0, + "step": 16741 + }, + { + "epoch": 1.838567977157918, + "grad_norm": 2.287090301513672, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7075140476226807, + "num_tokens": 432884640.0, + "step": 16742 + }, + { + "epoch": 1.8386777948605315, + "grad_norm": 2.5508484840393066, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7181421518325806, + "num_tokens": 432906955.0, + "step": 16743 + }, + { + "epoch": 1.838787612563145, + "grad_norm": 2.026571273803711, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7235708832740784, + "num_tokens": 432936136.0, + "step": 16744 + }, + { + "epoch": 1.8388974302657588, + "grad_norm": 2.123323678970337, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7157847285270691, + "num_tokens": 432962710.0, + "step": 16745 + }, + { + "epoch": 1.8390072479683726, + "grad_norm": 2.3850648403167725, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7243114113807678, + "num_tokens": 432985909.0, + "step": 16746 + }, + { + "epoch": 1.8391170656709863, + "grad_norm": 2.433487892150879, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7314807176589966, + "num_tokens": 433005395.0, + "step": 16747 + }, + { + "epoch": 1.8392268833735999, + "grad_norm": 2.1748387813568115, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7367171049118042, + "num_tokens": 433031676.0, + "step": 16748 + }, + { + "epoch": 1.8393367010762134, + "grad_norm": 2.1945242881774902, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7201108336448669, + "num_tokens": 433058830.0, + "step": 16749 + }, + { + "epoch": 1.8394465187788271, + "grad_norm": 2.3112118244171143, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7302029132843018, + "num_tokens": 433082843.0, + "step": 16750 + }, + { + "epoch": 1.839556336481441, + "grad_norm": 2.5268545150756836, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7292686700820923, + "num_tokens": 433102919.0, + "step": 16751 + }, + { + "epoch": 1.8396661541840544, + "grad_norm": 2.7844181060791016, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7392006516456604, + "num_tokens": 433120615.0, + "step": 16752 + }, + { + "epoch": 1.839775971886668, + "grad_norm": 2.432065010070801, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7180951237678528, + "num_tokens": 433144967.0, + "step": 16753 + }, + { + "epoch": 1.8398857895892817, + "grad_norm": 2.464296340942383, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7173798680305481, + "num_tokens": 433167433.0, + "step": 16754 + }, + { + "epoch": 1.8399956072918955, + "grad_norm": 2.1644015312194824, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7160420417785645, + "num_tokens": 433194566.0, + "step": 16755 + }, + { + "epoch": 1.8401054249945092, + "grad_norm": 2.054502010345459, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.72747403383255, + "num_tokens": 433222826.0, + "step": 16756 + }, + { + "epoch": 1.8402152426971228, + "grad_norm": 2.272691011428833, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7212520837783813, + "num_tokens": 433246856.0, + "step": 16757 + }, + { + "epoch": 1.8403250603997363, + "grad_norm": 2.2309415340423584, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.73216712474823, + "num_tokens": 433270364.0, + "step": 16758 + }, + { + "epoch": 1.84043487810235, + "grad_norm": 2.256159782409668, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7030110359191895, + "num_tokens": 433298091.0, + "step": 16759 + }, + { + "epoch": 1.8405446958049638, + "grad_norm": 2.0918304920196533, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7444525957107544, + "num_tokens": 433326190.0, + "step": 16760 + }, + { + "epoch": 1.8406545135075776, + "grad_norm": 1.9196075201034546, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7223430871963501, + "num_tokens": 433356604.0, + "step": 16761 + }, + { + "epoch": 1.8407643312101911, + "grad_norm": 2.1744062900543213, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.711205244064331, + "num_tokens": 433383447.0, + "step": 16762 + }, + { + "epoch": 1.8408741489128047, + "grad_norm": 2.176143169403076, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6956647634506226, + "num_tokens": 433411016.0, + "step": 16763 + }, + { + "epoch": 1.8409839666154184, + "grad_norm": 2.2012691497802734, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.710728645324707, + "num_tokens": 433437278.0, + "step": 16764 + }, + { + "epoch": 1.8410937843180322, + "grad_norm": 2.4178173542022705, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.733224630355835, + "num_tokens": 433458780.0, + "step": 16765 + }, + { + "epoch": 1.8412036020206457, + "grad_norm": 2.1596851348876953, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.705899178981781, + "num_tokens": 433483918.0, + "step": 16766 + }, + { + "epoch": 1.8413134197232592, + "grad_norm": 2.425497055053711, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.724372386932373, + "num_tokens": 433505207.0, + "step": 16767 + }, + { + "epoch": 1.841423237425873, + "grad_norm": 2.4145095348358154, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7465522289276123, + "num_tokens": 433527139.0, + "step": 16768 + }, + { + "epoch": 1.8415330551284868, + "grad_norm": 2.281081438064575, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7162187695503235, + "num_tokens": 433552161.0, + "step": 16769 + }, + { + "epoch": 1.8416428728311005, + "grad_norm": 2.2964835166931152, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6937717795372009, + "num_tokens": 433578670.0, + "step": 16770 + }, + { + "epoch": 1.841752690533714, + "grad_norm": 2.4154958724975586, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7153809666633606, + "num_tokens": 433599905.0, + "step": 16771 + }, + { + "epoch": 1.8418625082363276, + "grad_norm": 2.1778404712677, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7049260139465332, + "num_tokens": 433628044.0, + "step": 16772 + }, + { + "epoch": 1.8419723259389413, + "grad_norm": 2.1788089275360107, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7188296914100647, + "num_tokens": 433654532.0, + "step": 16773 + }, + { + "epoch": 1.842082143641555, + "grad_norm": 1.9526163339614868, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7254464030265808, + "num_tokens": 433685377.0, + "step": 16774 + }, + { + "epoch": 1.8421919613441686, + "grad_norm": 2.2664740085601807, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7179001569747925, + "num_tokens": 433708832.0, + "step": 16775 + }, + { + "epoch": 1.8423017790467824, + "grad_norm": 2.1446220874786377, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7480013370513916, + "num_tokens": 433733598.0, + "step": 16776 + }, + { + "epoch": 1.842411596749396, + "grad_norm": 2.2491369247436523, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7405701279640198, + "num_tokens": 433757291.0, + "step": 16777 + }, + { + "epoch": 1.8425214144520097, + "grad_norm": 2.227306842803955, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.73114013671875, + "num_tokens": 433781578.0, + "step": 16778 + }, + { + "epoch": 1.8426312321546234, + "grad_norm": 2.206787109375, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7058510780334473, + "num_tokens": 433807409.0, + "step": 16779 + }, + { + "epoch": 1.842741049857237, + "grad_norm": 1.985365629196167, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7203454971313477, + "num_tokens": 433836391.0, + "step": 16780 + }, + { + "epoch": 1.8428508675598505, + "grad_norm": 1.9598846435546875, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6952719688415527, + "num_tokens": 433869430.0, + "step": 16781 + }, + { + "epoch": 1.8429606852624643, + "grad_norm": 1.9850902557373047, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7203489542007446, + "num_tokens": 433899256.0, + "step": 16782 + }, + { + "epoch": 1.843070502965078, + "grad_norm": 2.4470901489257812, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7155265808105469, + "num_tokens": 433922331.0, + "step": 16783 + }, + { + "epoch": 1.8431803206676918, + "grad_norm": 1.9761042594909668, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7178952693939209, + "num_tokens": 433952389.0, + "step": 16784 + }, + { + "epoch": 1.8432901383703053, + "grad_norm": 2.5395333766937256, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7324095964431763, + "num_tokens": 433972878.0, + "step": 16785 + }, + { + "epoch": 1.8433999560729188, + "grad_norm": 2.3603720664978027, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.729453444480896, + "num_tokens": 433996169.0, + "step": 16786 + }, + { + "epoch": 1.8435097737755326, + "grad_norm": 2.102583408355713, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7425114512443542, + "num_tokens": 434023402.0, + "step": 16787 + }, + { + "epoch": 1.8436195914781464, + "grad_norm": 2.229494094848633, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7250781059265137, + "num_tokens": 434048443.0, + "step": 16788 + }, + { + "epoch": 1.84372940918076, + "grad_norm": 2.3010196685791016, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7336565256118774, + "num_tokens": 434074861.0, + "step": 16789 + }, + { + "epoch": 1.8438392268833736, + "grad_norm": 2.137507915496826, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.708802342414856, + "num_tokens": 434105980.0, + "step": 16790 + }, + { + "epoch": 1.8439490445859872, + "grad_norm": 2.339733600616455, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7195898294448853, + "num_tokens": 434129951.0, + "step": 16791 + }, + { + "epoch": 1.844058862288601, + "grad_norm": 2.481307029724121, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7399607300758362, + "num_tokens": 434151178.0, + "step": 16792 + }, + { + "epoch": 1.8441686799912147, + "grad_norm": 2.1335787773132324, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7324903011322021, + "num_tokens": 434179016.0, + "step": 16793 + }, + { + "epoch": 1.8442784976938282, + "grad_norm": 2.6594924926757812, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7140353918075562, + "num_tokens": 434198588.0, + "step": 16794 + }, + { + "epoch": 1.8443883153964418, + "grad_norm": 2.3549203872680664, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7394662499427795, + "num_tokens": 434220242.0, + "step": 16795 + }, + { + "epoch": 1.8444981330990555, + "grad_norm": 2.151956796646118, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7145386338233948, + "num_tokens": 434246939.0, + "step": 16796 + }, + { + "epoch": 1.8446079508016693, + "grad_norm": 2.080315113067627, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6962114572525024, + "num_tokens": 434275946.0, + "step": 16797 + }, + { + "epoch": 1.844717768504283, + "grad_norm": 2.4143991470336914, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7116893529891968, + "num_tokens": 434299740.0, + "step": 16798 + }, + { + "epoch": 1.8448275862068966, + "grad_norm": 2.0114715099334717, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7002491354942322, + "num_tokens": 434330720.0, + "step": 16799 + }, + { + "epoch": 1.84493740390951, + "grad_norm": 2.1053643226623535, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.717846155166626, + "num_tokens": 434359474.0, + "step": 16800 + }, + { + "epoch": 1.8450472216121239, + "grad_norm": 2.2199442386627197, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7308112382888794, + "num_tokens": 434383772.0, + "step": 16801 + }, + { + "epoch": 1.8451570393147376, + "grad_norm": 2.0526206493377686, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7211198806762695, + "num_tokens": 434412512.0, + "step": 16802 + }, + { + "epoch": 1.8452668570173512, + "grad_norm": 2.285949230194092, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7201783657073975, + "num_tokens": 434435345.0, + "step": 16803 + }, + { + "epoch": 1.8453766747199647, + "grad_norm": 2.303044080734253, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7144376635551453, + "num_tokens": 434459575.0, + "step": 16804 + }, + { + "epoch": 1.8454864924225785, + "grad_norm": 2.192790985107422, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7108519077301025, + "num_tokens": 434487834.0, + "step": 16805 + }, + { + "epoch": 1.8455963101251922, + "grad_norm": 2.59725022315979, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7231836915016174, + "num_tokens": 434508043.0, + "step": 16806 + }, + { + "epoch": 1.845706127827806, + "grad_norm": 2.254270315170288, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7362889051437378, + "num_tokens": 434531909.0, + "step": 16807 + }, + { + "epoch": 1.8458159455304195, + "grad_norm": 2.2868926525115967, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7201262712478638, + "num_tokens": 434556812.0, + "step": 16808 + }, + { + "epoch": 1.845925763233033, + "grad_norm": 2.10847806930542, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.729557454586029, + "num_tokens": 434585807.0, + "step": 16809 + }, + { + "epoch": 1.8460355809356468, + "grad_norm": 1.966902494430542, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7128334641456604, + "num_tokens": 434619430.0, + "step": 16810 + }, + { + "epoch": 1.8461453986382605, + "grad_norm": 2.347632884979248, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7130830883979797, + "num_tokens": 434644649.0, + "step": 16811 + }, + { + "epoch": 1.8462552163408743, + "grad_norm": 1.9791803359985352, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7266513109207153, + "num_tokens": 434675536.0, + "step": 16812 + }, + { + "epoch": 1.8463650340434878, + "grad_norm": 2.1822690963745117, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7158351540565491, + "num_tokens": 434702111.0, + "step": 16813 + }, + { + "epoch": 1.8464748517461014, + "grad_norm": 2.1418890953063965, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7409111857414246, + "num_tokens": 434727742.0, + "step": 16814 + }, + { + "epoch": 1.8465846694487151, + "grad_norm": 2.179424524307251, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7234916090965271, + "num_tokens": 434753236.0, + "step": 16815 + }, + { + "epoch": 1.846694487151329, + "grad_norm": 2.146770477294922, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7106218338012695, + "num_tokens": 434780497.0, + "step": 16816 + }, + { + "epoch": 1.8468043048539424, + "grad_norm": 2.4730312824249268, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7418303489685059, + "num_tokens": 434801305.0, + "step": 16817 + }, + { + "epoch": 1.846914122556556, + "grad_norm": 2.3265933990478516, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7367572784423828, + "num_tokens": 434825040.0, + "step": 16818 + }, + { + "epoch": 1.8470239402591697, + "grad_norm": 2.4070658683776855, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7123368978500366, + "num_tokens": 434848252.0, + "step": 16819 + }, + { + "epoch": 1.8471337579617835, + "grad_norm": 2.2730212211608887, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7140690684318542, + "num_tokens": 434874096.0, + "step": 16820 + }, + { + "epoch": 1.8472435756643972, + "grad_norm": 2.4538519382476807, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7425855994224548, + "num_tokens": 434894708.0, + "step": 16821 + }, + { + "epoch": 1.8473533933670108, + "grad_norm": 2.0519046783447266, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7150695323944092, + "num_tokens": 434923205.0, + "step": 16822 + }, + { + "epoch": 1.8474632110696243, + "grad_norm": 2.556586265563965, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7408787608146667, + "num_tokens": 434944794.0, + "step": 16823 + }, + { + "epoch": 1.847573028772238, + "grad_norm": 2.226353645324707, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7150527238845825, + "num_tokens": 434969538.0, + "step": 16824 + }, + { + "epoch": 1.8476828464748518, + "grad_norm": 2.3861846923828125, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.718152642250061, + "num_tokens": 434991412.0, + "step": 16825 + }, + { + "epoch": 1.8477926641774656, + "grad_norm": 2.130504608154297, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6994298100471497, + "num_tokens": 435021016.0, + "step": 16826 + }, + { + "epoch": 1.847902481880079, + "grad_norm": 2.3739030361175537, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7093774676322937, + "num_tokens": 435044767.0, + "step": 16827 + }, + { + "epoch": 1.8480122995826926, + "grad_norm": 2.2528226375579834, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7264089584350586, + "num_tokens": 435068460.0, + "step": 16828 + }, + { + "epoch": 1.8481221172853064, + "grad_norm": 2.2285993099212646, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7231293320655823, + "num_tokens": 435092176.0, + "step": 16829 + }, + { + "epoch": 1.8482319349879202, + "grad_norm": 2.0608808994293213, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7095995545387268, + "num_tokens": 435122503.0, + "step": 16830 + }, + { + "epoch": 1.8483417526905337, + "grad_norm": 2.311501979827881, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7222713232040405, + "num_tokens": 435147829.0, + "step": 16831 + }, + { + "epoch": 1.8484515703931472, + "grad_norm": 2.3269078731536865, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7073871493339539, + "num_tokens": 435170486.0, + "step": 16832 + }, + { + "epoch": 1.848561388095761, + "grad_norm": 2.16441011428833, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.7164469957351685, + "num_tokens": 435198505.0, + "step": 16833 + }, + { + "epoch": 1.8486712057983747, + "grad_norm": 2.554919481277466, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7176709175109863, + "num_tokens": 435220991.0, + "step": 16834 + }, + { + "epoch": 1.8487810235009885, + "grad_norm": 2.3344361782073975, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7145626544952393, + "num_tokens": 435247749.0, + "step": 16835 + }, + { + "epoch": 1.848890841203602, + "grad_norm": 2.684659004211426, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7164008617401123, + "num_tokens": 435266982.0, + "step": 16836 + }, + { + "epoch": 1.8490006589062156, + "grad_norm": 2.597047805786133, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7412082552909851, + "num_tokens": 435286124.0, + "step": 16837 + }, + { + "epoch": 1.8491104766088293, + "grad_norm": 2.3193166255950928, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7439119815826416, + "num_tokens": 435308497.0, + "step": 16838 + }, + { + "epoch": 1.849220294311443, + "grad_norm": 2.1191842555999756, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6844556331634521, + "num_tokens": 435337722.0, + "step": 16839 + }, + { + "epoch": 1.8493301120140566, + "grad_norm": 2.4321227073669434, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7380249500274658, + "num_tokens": 435358826.0, + "step": 16840 + }, + { + "epoch": 1.8494399297166704, + "grad_norm": 2.4046471118927, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7061707377433777, + "num_tokens": 435383016.0, + "step": 16841 + }, + { + "epoch": 1.849549747419284, + "grad_norm": 2.188776969909668, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.7026826739311218, + "num_tokens": 435411040.0, + "step": 16842 + }, + { + "epoch": 1.8496595651218977, + "grad_norm": 2.231658458709717, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.732277512550354, + "num_tokens": 435437714.0, + "step": 16843 + }, + { + "epoch": 1.8497693828245114, + "grad_norm": 2.2127127647399902, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7224904298782349, + "num_tokens": 435463892.0, + "step": 16844 + }, + { + "epoch": 1.849879200527125, + "grad_norm": 2.0290796756744385, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7031494379043579, + "num_tokens": 435494700.0, + "step": 16845 + }, + { + "epoch": 1.8499890182297385, + "grad_norm": 2.6571476459503174, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7309175729751587, + "num_tokens": 435514755.0, + "step": 16846 + }, + { + "epoch": 1.8500988359323522, + "grad_norm": 2.5951662063598633, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7479904890060425, + "num_tokens": 435534376.0, + "step": 16847 + }, + { + "epoch": 1.850208653634966, + "grad_norm": 2.445307731628418, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7166981101036072, + "num_tokens": 435557910.0, + "step": 16848 + }, + { + "epoch": 1.8503184713375798, + "grad_norm": 1.825677752494812, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.704023540019989, + "num_tokens": 435593600.0, + "step": 16849 + }, + { + "epoch": 1.8504282890401933, + "grad_norm": 2.436795949935913, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7007193565368652, + "num_tokens": 435615993.0, + "step": 16850 + }, + { + "epoch": 1.8505381067428068, + "grad_norm": 2.2238266468048096, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7170199155807495, + "num_tokens": 435640678.0, + "step": 16851 + }, + { + "epoch": 1.8506479244454206, + "grad_norm": 2.264580011367798, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.708818793296814, + "num_tokens": 435666368.0, + "step": 16852 + }, + { + "epoch": 1.8507577421480343, + "grad_norm": 2.4676666259765625, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7300310134887695, + "num_tokens": 435686611.0, + "step": 16853 + }, + { + "epoch": 1.8508675598506479, + "grad_norm": 2.1107451915740967, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7170460224151611, + "num_tokens": 435715453.0, + "step": 16854 + }, + { + "epoch": 1.8509773775532616, + "grad_norm": 2.4158835411071777, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7021744251251221, + "num_tokens": 435739985.0, + "step": 16855 + }, + { + "epoch": 1.8510871952558752, + "grad_norm": 2.1451523303985596, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7124717235565186, + "num_tokens": 435767221.0, + "step": 16856 + }, + { + "epoch": 1.851197012958489, + "grad_norm": 2.3643250465393066, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7455204129219055, + "num_tokens": 435790554.0, + "step": 16857 + }, + { + "epoch": 1.8513068306611027, + "grad_norm": 2.1614484786987305, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.727523922920227, + "num_tokens": 435817734.0, + "step": 16858 + }, + { + "epoch": 1.8514166483637162, + "grad_norm": 2.1310322284698486, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7049798965454102, + "num_tokens": 435845409.0, + "step": 16859 + }, + { + "epoch": 1.8515264660663298, + "grad_norm": 2.471583604812622, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7249372005462646, + "num_tokens": 435867146.0, + "step": 16860 + }, + { + "epoch": 1.8516362837689435, + "grad_norm": 2.2587695121765137, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7150977849960327, + "num_tokens": 435896313.0, + "step": 16861 + }, + { + "epoch": 1.8517461014715573, + "grad_norm": 2.6019325256347656, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7137993574142456, + "num_tokens": 435916873.0, + "step": 16862 + }, + { + "epoch": 1.851855919174171, + "grad_norm": 2.2361979484558105, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7344062924385071, + "num_tokens": 435940352.0, + "step": 16863 + }, + { + "epoch": 1.8519657368767846, + "grad_norm": 2.2695555686950684, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7309380769729614, + "num_tokens": 435965409.0, + "step": 16864 + }, + { + "epoch": 1.852075554579398, + "grad_norm": 2.338731527328491, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7365492582321167, + "num_tokens": 435989123.0, + "step": 16865 + }, + { + "epoch": 1.8521853722820119, + "grad_norm": 2.371605396270752, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7385955452919006, + "num_tokens": 436010896.0, + "step": 16866 + }, + { + "epoch": 1.8522951899846256, + "grad_norm": 2.2179007530212402, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.709740400314331, + "num_tokens": 436037438.0, + "step": 16867 + }, + { + "epoch": 1.8524050076872391, + "grad_norm": 2.5827996730804443, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7118457555770874, + "num_tokens": 436056645.0, + "step": 16868 + }, + { + "epoch": 1.8525148253898527, + "grad_norm": 2.1958038806915283, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7021200656890869, + "num_tokens": 436084261.0, + "step": 16869 + }, + { + "epoch": 1.8526246430924664, + "grad_norm": 2.0747828483581543, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7097319960594177, + "num_tokens": 436115170.0, + "step": 16870 + }, + { + "epoch": 1.8527344607950802, + "grad_norm": 2.4242451190948486, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7233171463012695, + "num_tokens": 436138017.0, + "step": 16871 + }, + { + "epoch": 1.852844278497694, + "grad_norm": 2.277536153793335, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7086991667747498, + "num_tokens": 436164826.0, + "step": 16872 + }, + { + "epoch": 1.8529540962003075, + "grad_norm": 2.4681336879730225, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7370694875717163, + "num_tokens": 436186950.0, + "step": 16873 + }, + { + "epoch": 1.853063913902921, + "grad_norm": 2.059325695037842, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.700087308883667, + "num_tokens": 436218554.0, + "step": 16874 + }, + { + "epoch": 1.8531737316055348, + "grad_norm": 2.216187000274658, + "learning_rate": 1e-06, + "loss": 1.0471, + "mean_token_accuracy": 0.6882269978523254, + "num_tokens": 436245732.0, + "step": 16875 + }, + { + "epoch": 1.8532835493081485, + "grad_norm": 2.0265021324157715, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7303386926651001, + "num_tokens": 436275907.0, + "step": 16876 + }, + { + "epoch": 1.8533933670107623, + "grad_norm": 2.3554723262786865, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7517789602279663, + "num_tokens": 436298559.0, + "step": 16877 + }, + { + "epoch": 1.8535031847133758, + "grad_norm": 2.073834180831909, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7091556191444397, + "num_tokens": 436328711.0, + "step": 16878 + }, + { + "epoch": 1.8536130024159894, + "grad_norm": 2.3697214126586914, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7241072654724121, + "num_tokens": 436351002.0, + "step": 16879 + }, + { + "epoch": 1.8537228201186031, + "grad_norm": 2.221237897872925, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.709794282913208, + "num_tokens": 436376844.0, + "step": 16880 + }, + { + "epoch": 1.8538326378212169, + "grad_norm": 2.2327775955200195, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.6973333358764648, + "num_tokens": 436403455.0, + "step": 16881 + }, + { + "epoch": 1.8539424555238304, + "grad_norm": 2.3029706478118896, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.718116044998169, + "num_tokens": 436427524.0, + "step": 16882 + }, + { + "epoch": 1.854052273226444, + "grad_norm": 2.134084701538086, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7292547225952148, + "num_tokens": 436454649.0, + "step": 16883 + }, + { + "epoch": 1.8541620909290577, + "grad_norm": 2.3496363162994385, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7358102202415466, + "num_tokens": 436477195.0, + "step": 16884 + }, + { + "epoch": 1.8542719086316715, + "grad_norm": 2.050844192504883, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7193213105201721, + "num_tokens": 436505904.0, + "step": 16885 + }, + { + "epoch": 1.8543817263342852, + "grad_norm": 2.29712176322937, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7193706631660461, + "num_tokens": 436531585.0, + "step": 16886 + }, + { + "epoch": 1.8544915440368988, + "grad_norm": 2.025233268737793, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7023119330406189, + "num_tokens": 436563798.0, + "step": 16887 + }, + { + "epoch": 1.8546013617395123, + "grad_norm": 2.0832583904266357, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7184745669364929, + "num_tokens": 436592580.0, + "step": 16888 + }, + { + "epoch": 1.854711179442126, + "grad_norm": 2.2616894245147705, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7335827946662903, + "num_tokens": 436615814.0, + "step": 16889 + }, + { + "epoch": 1.8548209971447398, + "grad_norm": 2.150878429412842, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7253657579421997, + "num_tokens": 436642617.0, + "step": 16890 + }, + { + "epoch": 1.8549308148473536, + "grad_norm": 2.344627618789673, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7110117077827454, + "num_tokens": 436666971.0, + "step": 16891 + }, + { + "epoch": 1.855040632549967, + "grad_norm": 2.1162467002868652, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7078588008880615, + "num_tokens": 436695065.0, + "step": 16892 + }, + { + "epoch": 1.8551504502525806, + "grad_norm": 1.9990302324295044, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7134863138198853, + "num_tokens": 436724658.0, + "step": 16893 + }, + { + "epoch": 1.8552602679551944, + "grad_norm": 2.199735164642334, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7189353704452515, + "num_tokens": 436750097.0, + "step": 16894 + }, + { + "epoch": 1.8553700856578081, + "grad_norm": 2.209273338317871, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7326751947402954, + "num_tokens": 436774023.0, + "step": 16895 + }, + { + "epoch": 1.8554799033604217, + "grad_norm": 2.156916379928589, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7181159257888794, + "num_tokens": 436800885.0, + "step": 16896 + }, + { + "epoch": 1.8555897210630352, + "grad_norm": 2.5743658542633057, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7128798365592957, + "num_tokens": 436824212.0, + "step": 16897 + }, + { + "epoch": 1.855699538765649, + "grad_norm": 2.6453447341918945, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7416449785232544, + "num_tokens": 436843050.0, + "step": 16898 + }, + { + "epoch": 1.8558093564682627, + "grad_norm": 1.9339377880096436, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7014369964599609, + "num_tokens": 436880926.0, + "step": 16899 + }, + { + "epoch": 1.8559191741708765, + "grad_norm": 2.1738154888153076, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.716607928276062, + "num_tokens": 436908418.0, + "step": 16900 + }, + { + "epoch": 1.85602899187349, + "grad_norm": 2.2584102153778076, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7408483028411865, + "num_tokens": 436933281.0, + "step": 16901 + }, + { + "epoch": 1.8561388095761036, + "grad_norm": 2.096608877182007, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7187499403953552, + "num_tokens": 436959755.0, + "step": 16902 + }, + { + "epoch": 1.8562486272787173, + "grad_norm": 2.239501714706421, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7509368658065796, + "num_tokens": 436983183.0, + "step": 16903 + }, + { + "epoch": 1.856358444981331, + "grad_norm": 2.6297171115875244, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7623805403709412, + "num_tokens": 437000635.0, + "step": 16904 + }, + { + "epoch": 1.8564682626839446, + "grad_norm": 2.166705369949341, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7127285003662109, + "num_tokens": 437025503.0, + "step": 16905 + }, + { + "epoch": 1.8565780803865584, + "grad_norm": 2.218172311782837, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.722761869430542, + "num_tokens": 437051630.0, + "step": 16906 + }, + { + "epoch": 1.856687898089172, + "grad_norm": 2.354095220565796, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7416812181472778, + "num_tokens": 437076770.0, + "step": 16907 + }, + { + "epoch": 1.8567977157917857, + "grad_norm": 1.9946528673171997, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7270876169204712, + "num_tokens": 437108400.0, + "step": 16908 + }, + { + "epoch": 1.8569075334943994, + "grad_norm": 2.024962902069092, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7093047499656677, + "num_tokens": 437140555.0, + "step": 16909 + }, + { + "epoch": 1.857017351197013, + "grad_norm": 2.46478009223938, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7248328924179077, + "num_tokens": 437162231.0, + "step": 16910 + }, + { + "epoch": 1.8571271688996265, + "grad_norm": 2.147390842437744, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7127806544303894, + "num_tokens": 437188757.0, + "step": 16911 + }, + { + "epoch": 1.8572369866022402, + "grad_norm": 2.0857367515563965, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7454479932785034, + "num_tokens": 437215715.0, + "step": 16912 + }, + { + "epoch": 1.857346804304854, + "grad_norm": 2.2543203830718994, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.722368597984314, + "num_tokens": 437240293.0, + "step": 16913 + }, + { + "epoch": 1.8574566220074678, + "grad_norm": 2.091426134109497, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7268868088722229, + "num_tokens": 437271806.0, + "step": 16914 + }, + { + "epoch": 1.8575664397100813, + "grad_norm": 2.277886390686035, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.736322283744812, + "num_tokens": 437295395.0, + "step": 16915 + }, + { + "epoch": 1.8576762574126948, + "grad_norm": 2.541585683822632, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7186395525932312, + "num_tokens": 437316129.0, + "step": 16916 + }, + { + "epoch": 1.8577860751153086, + "grad_norm": 2.2005414962768555, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7028318047523499, + "num_tokens": 437342790.0, + "step": 16917 + }, + { + "epoch": 1.8578958928179223, + "grad_norm": 2.111647129058838, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.70013427734375, + "num_tokens": 437372755.0, + "step": 16918 + }, + { + "epoch": 1.8580057105205359, + "grad_norm": 1.9127365350723267, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7174980640411377, + "num_tokens": 437406550.0, + "step": 16919 + }, + { + "epoch": 1.8581155282231496, + "grad_norm": 2.3308167457580566, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7457320690155029, + "num_tokens": 437428676.0, + "step": 16920 + }, + { + "epoch": 1.8582253459257632, + "grad_norm": 2.2248494625091553, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7040042281150818, + "num_tokens": 437455018.0, + "step": 16921 + }, + { + "epoch": 1.858335163628377, + "grad_norm": 2.3693931102752686, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7055197954177856, + "num_tokens": 437479847.0, + "step": 16922 + }, + { + "epoch": 1.8584449813309907, + "grad_norm": 2.217799186706543, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7077584266662598, + "num_tokens": 437507113.0, + "step": 16923 + }, + { + "epoch": 1.8585547990336042, + "grad_norm": 2.4495108127593994, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7064228057861328, + "num_tokens": 437529316.0, + "step": 16924 + }, + { + "epoch": 1.8586646167362177, + "grad_norm": 2.0334081649780273, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6918035745620728, + "num_tokens": 437560522.0, + "step": 16925 + }, + { + "epoch": 1.8587744344388315, + "grad_norm": 2.0288889408111572, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.698768138885498, + "num_tokens": 437590821.0, + "step": 16926 + }, + { + "epoch": 1.8588842521414453, + "grad_norm": 2.2268049716949463, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.7020539045333862, + "num_tokens": 437619416.0, + "step": 16927 + }, + { + "epoch": 1.858994069844059, + "grad_norm": 2.3692195415496826, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7047147750854492, + "num_tokens": 437643616.0, + "step": 16928 + }, + { + "epoch": 1.8591038875466726, + "grad_norm": 2.1168742179870605, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7302197217941284, + "num_tokens": 437670853.0, + "step": 16929 + }, + { + "epoch": 1.859213705249286, + "grad_norm": 2.4172613620758057, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7269044518470764, + "num_tokens": 437692695.0, + "step": 16930 + }, + { + "epoch": 1.8593235229518998, + "grad_norm": 2.1379599571228027, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7262179851531982, + "num_tokens": 437719371.0, + "step": 16931 + }, + { + "epoch": 1.8594333406545136, + "grad_norm": 2.2177374362945557, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.733942985534668, + "num_tokens": 437743999.0, + "step": 16932 + }, + { + "epoch": 1.8595431583571271, + "grad_norm": 2.3487420082092285, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7113450169563293, + "num_tokens": 437766725.0, + "step": 16933 + }, + { + "epoch": 1.8596529760597407, + "grad_norm": 2.299302816390991, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.747571587562561, + "num_tokens": 437790488.0, + "step": 16934 + }, + { + "epoch": 1.8597627937623544, + "grad_norm": 2.304074287414551, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7142577171325684, + "num_tokens": 437814868.0, + "step": 16935 + }, + { + "epoch": 1.8598726114649682, + "grad_norm": 2.4826173782348633, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7286191582679749, + "num_tokens": 437834285.0, + "step": 16936 + }, + { + "epoch": 1.859982429167582, + "grad_norm": 2.0397050380706787, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.6976876258850098, + "num_tokens": 437862652.0, + "step": 16937 + }, + { + "epoch": 1.8600922468701955, + "grad_norm": 2.219278335571289, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7177765965461731, + "num_tokens": 437886820.0, + "step": 16938 + }, + { + "epoch": 1.860202064572809, + "grad_norm": 2.0097460746765137, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7289405465126038, + "num_tokens": 437919256.0, + "step": 16939 + }, + { + "epoch": 1.8603118822754228, + "grad_norm": 2.167025327682495, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7021873593330383, + "num_tokens": 437947549.0, + "step": 16940 + }, + { + "epoch": 1.8604216999780365, + "grad_norm": 2.2466559410095215, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.707220733165741, + "num_tokens": 437972843.0, + "step": 16941 + }, + { + "epoch": 1.8605315176806503, + "grad_norm": 2.2787539958953857, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7299794554710388, + "num_tokens": 437998600.0, + "step": 16942 + }, + { + "epoch": 1.8606413353832638, + "grad_norm": 2.009105920791626, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7206099033355713, + "num_tokens": 438028102.0, + "step": 16943 + }, + { + "epoch": 1.8607511530858774, + "grad_norm": 2.3508260250091553, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7048932909965515, + "num_tokens": 438052822.0, + "step": 16944 + }, + { + "epoch": 1.860860970788491, + "grad_norm": 2.081047773361206, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7118147611618042, + "num_tokens": 438082844.0, + "step": 16945 + }, + { + "epoch": 1.8609707884911049, + "grad_norm": 2.2395074367523193, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7195066809654236, + "num_tokens": 438109890.0, + "step": 16946 + }, + { + "epoch": 1.8610806061937184, + "grad_norm": 1.914347767829895, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7174314260482788, + "num_tokens": 438140895.0, + "step": 16947 + }, + { + "epoch": 1.861190423896332, + "grad_norm": 2.2712419033050537, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7341262102127075, + "num_tokens": 438165806.0, + "step": 16948 + }, + { + "epoch": 1.8613002415989457, + "grad_norm": 2.3530590534210205, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7326838970184326, + "num_tokens": 438190317.0, + "step": 16949 + }, + { + "epoch": 1.8614100593015594, + "grad_norm": 2.1497714519500732, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7178146839141846, + "num_tokens": 438215540.0, + "step": 16950 + }, + { + "epoch": 1.8615198770041732, + "grad_norm": 2.276232957839966, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7330737113952637, + "num_tokens": 438239349.0, + "step": 16951 + }, + { + "epoch": 1.8616296947067867, + "grad_norm": 2.0193490982055664, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7022560834884644, + "num_tokens": 438272407.0, + "step": 16952 + }, + { + "epoch": 1.8617395124094003, + "grad_norm": 2.5242621898651123, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7543468475341797, + "num_tokens": 438292060.0, + "step": 16953 + }, + { + "epoch": 1.861849330112014, + "grad_norm": 2.0025362968444824, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7051029205322266, + "num_tokens": 438323163.0, + "step": 16954 + }, + { + "epoch": 1.8619591478146278, + "grad_norm": 2.210526466369629, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6990877389907837, + "num_tokens": 438349393.0, + "step": 16955 + }, + { + "epoch": 1.8620689655172413, + "grad_norm": 2.241661310195923, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7220618724822998, + "num_tokens": 438374889.0, + "step": 16956 + }, + { + "epoch": 1.862178783219855, + "grad_norm": 2.098200559616089, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7068067789077759, + "num_tokens": 438404116.0, + "step": 16957 + }, + { + "epoch": 1.8622886009224686, + "grad_norm": 2.261457681655884, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7192516922950745, + "num_tokens": 438428559.0, + "step": 16958 + }, + { + "epoch": 1.8623984186250824, + "grad_norm": 2.4056801795959473, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7073333263397217, + "num_tokens": 438452106.0, + "step": 16959 + }, + { + "epoch": 1.8625082363276961, + "grad_norm": 2.2822582721710205, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7154427766799927, + "num_tokens": 438476888.0, + "step": 16960 + }, + { + "epoch": 1.8626180540303097, + "grad_norm": 1.9570820331573486, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.6964194774627686, + "num_tokens": 438509799.0, + "step": 16961 + }, + { + "epoch": 1.8627278717329232, + "grad_norm": 2.2712488174438477, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7227808833122253, + "num_tokens": 438533819.0, + "step": 16962 + }, + { + "epoch": 1.862837689435537, + "grad_norm": 2.1766247749328613, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7089377045631409, + "num_tokens": 438559396.0, + "step": 16963 + }, + { + "epoch": 1.8629475071381507, + "grad_norm": 2.278172731399536, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.752535879611969, + "num_tokens": 438581493.0, + "step": 16964 + }, + { + "epoch": 1.8630573248407645, + "grad_norm": 2.3816890716552734, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7407535314559937, + "num_tokens": 438603969.0, + "step": 16965 + }, + { + "epoch": 1.863167142543378, + "grad_norm": 2.160163640975952, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7344825267791748, + "num_tokens": 438629259.0, + "step": 16966 + }, + { + "epoch": 1.8632769602459915, + "grad_norm": 2.02878475189209, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6870511770248413, + "num_tokens": 438659759.0, + "step": 16967 + }, + { + "epoch": 1.8633867779486053, + "grad_norm": 1.8999292850494385, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.694612443447113, + "num_tokens": 438693993.0, + "step": 16968 + }, + { + "epoch": 1.863496595651219, + "grad_norm": 2.344256639480591, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7325184345245361, + "num_tokens": 438717877.0, + "step": 16969 + }, + { + "epoch": 1.8636064133538326, + "grad_norm": 2.275463819503784, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7203797698020935, + "num_tokens": 438741555.0, + "step": 16970 + }, + { + "epoch": 1.8637162310564463, + "grad_norm": 2.1137595176696777, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7090029120445251, + "num_tokens": 438772483.0, + "step": 16971 + }, + { + "epoch": 1.8638260487590599, + "grad_norm": 2.277331829071045, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7364288568496704, + "num_tokens": 438796946.0, + "step": 16972 + }, + { + "epoch": 1.8639358664616736, + "grad_norm": 1.96968674659729, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7064568996429443, + "num_tokens": 438828630.0, + "step": 16973 + }, + { + "epoch": 1.8640456841642874, + "grad_norm": 2.349141836166382, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7248650193214417, + "num_tokens": 438854638.0, + "step": 16974 + }, + { + "epoch": 1.864155501866901, + "grad_norm": 2.196805477142334, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7342265844345093, + "num_tokens": 438880277.0, + "step": 16975 + }, + { + "epoch": 1.8642653195695145, + "grad_norm": 1.977681040763855, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7074726819992065, + "num_tokens": 438913785.0, + "step": 16976 + }, + { + "epoch": 1.8643751372721282, + "grad_norm": 2.5189566612243652, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7285877466201782, + "num_tokens": 438935297.0, + "step": 16977 + }, + { + "epoch": 1.864484954974742, + "grad_norm": 2.1712276935577393, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7266359925270081, + "num_tokens": 438960935.0, + "step": 16978 + }, + { + "epoch": 1.8645947726773557, + "grad_norm": 2.22029447555542, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7257840633392334, + "num_tokens": 438985536.0, + "step": 16979 + }, + { + "epoch": 1.8647045903799693, + "grad_norm": 2.192976713180542, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.6980859041213989, + "num_tokens": 439014239.0, + "step": 16980 + }, + { + "epoch": 1.8648144080825828, + "grad_norm": 2.26406192779541, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7202916145324707, + "num_tokens": 439039532.0, + "step": 16981 + }, + { + "epoch": 1.8649242257851966, + "grad_norm": 2.0073022842407227, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7212517261505127, + "num_tokens": 439070929.0, + "step": 16982 + }, + { + "epoch": 1.8650340434878103, + "grad_norm": 2.3338773250579834, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7242014408111572, + "num_tokens": 439093825.0, + "step": 16983 + }, + { + "epoch": 1.8651438611904239, + "grad_norm": 2.387420415878296, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7122723460197449, + "num_tokens": 439117753.0, + "step": 16984 + }, + { + "epoch": 1.8652536788930374, + "grad_norm": 2.122637987136841, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7511636018753052, + "num_tokens": 439144930.0, + "step": 16985 + }, + { + "epoch": 1.8653634965956511, + "grad_norm": 2.6044952869415283, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7220019102096558, + "num_tokens": 439165820.0, + "step": 16986 + }, + { + "epoch": 1.865473314298265, + "grad_norm": 2.3904449939727783, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7426772117614746, + "num_tokens": 439188198.0, + "step": 16987 + }, + { + "epoch": 1.8655831320008787, + "grad_norm": 2.0755181312561035, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7326592206954956, + "num_tokens": 439215580.0, + "step": 16988 + }, + { + "epoch": 1.8656929497034922, + "grad_norm": 2.2495172023773193, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7294447422027588, + "num_tokens": 439238300.0, + "step": 16989 + }, + { + "epoch": 1.8658027674061057, + "grad_norm": 2.241213798522949, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7094243764877319, + "num_tokens": 439263492.0, + "step": 16990 + }, + { + "epoch": 1.8659125851087195, + "grad_norm": 2.2908356189727783, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7158089876174927, + "num_tokens": 439290625.0, + "step": 16991 + }, + { + "epoch": 1.8660224028113332, + "grad_norm": 2.370412826538086, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6945548057556152, + "num_tokens": 439316988.0, + "step": 16992 + }, + { + "epoch": 1.866132220513947, + "grad_norm": 2.429887294769287, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7291918992996216, + "num_tokens": 439338441.0, + "step": 16993 + }, + { + "epoch": 1.8662420382165605, + "grad_norm": 2.61198091506958, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7315719723701477, + "num_tokens": 439359117.0, + "step": 16994 + }, + { + "epoch": 1.866351855919174, + "grad_norm": 2.319438934326172, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.711207926273346, + "num_tokens": 439383259.0, + "step": 16995 + }, + { + "epoch": 1.8664616736217878, + "grad_norm": 2.2237603664398193, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7159655094146729, + "num_tokens": 439407786.0, + "step": 16996 + }, + { + "epoch": 1.8665714913244016, + "grad_norm": 2.0224623680114746, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7126191854476929, + "num_tokens": 439437130.0, + "step": 16997 + }, + { + "epoch": 1.8666813090270151, + "grad_norm": 2.0717804431915283, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7349668145179749, + "num_tokens": 439465442.0, + "step": 16998 + }, + { + "epoch": 1.8667911267296287, + "grad_norm": 2.1268808841705322, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7157250642776489, + "num_tokens": 439494648.0, + "step": 16999 + }, + { + "epoch": 1.8669009444322424, + "grad_norm": 2.1962223052978516, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7252922058105469, + "num_tokens": 439520459.0, + "step": 17000 + }, + { + "epoch": 1.8670107621348562, + "grad_norm": 2.281187057495117, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7090580463409424, + "num_tokens": 439546501.0, + "step": 17001 + }, + { + "epoch": 1.86712057983747, + "grad_norm": 2.019387722015381, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7240411043167114, + "num_tokens": 439576103.0, + "step": 17002 + }, + { + "epoch": 1.8672303975400835, + "grad_norm": 2.6025354862213135, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7155494689941406, + "num_tokens": 439597541.0, + "step": 17003 + }, + { + "epoch": 1.867340215242697, + "grad_norm": 2.1011595726013184, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7274344563484192, + "num_tokens": 439627409.0, + "step": 17004 + }, + { + "epoch": 1.8674500329453108, + "grad_norm": 2.374498128890991, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7161783576011658, + "num_tokens": 439649990.0, + "step": 17005 + }, + { + "epoch": 1.8675598506479245, + "grad_norm": 2.284292221069336, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7281942367553711, + "num_tokens": 439676100.0, + "step": 17006 + }, + { + "epoch": 1.8676696683505383, + "grad_norm": 2.2727866172790527, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7073558568954468, + "num_tokens": 439701954.0, + "step": 17007 + }, + { + "epoch": 1.8677794860531518, + "grad_norm": 2.1580889225006104, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7231030464172363, + "num_tokens": 439729123.0, + "step": 17008 + }, + { + "epoch": 1.8678893037557653, + "grad_norm": 2.249577760696411, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7015011310577393, + "num_tokens": 439755615.0, + "step": 17009 + }, + { + "epoch": 1.867999121458379, + "grad_norm": 2.04190993309021, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7090294361114502, + "num_tokens": 439786232.0, + "step": 17010 + }, + { + "epoch": 1.8681089391609929, + "grad_norm": 2.2842156887054443, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7253985404968262, + "num_tokens": 439810890.0, + "step": 17011 + }, + { + "epoch": 1.8682187568636064, + "grad_norm": 2.4239094257354736, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7166200280189514, + "num_tokens": 439833482.0, + "step": 17012 + }, + { + "epoch": 1.86832857456622, + "grad_norm": 2.1234562397003174, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7285469770431519, + "num_tokens": 439859092.0, + "step": 17013 + }, + { + "epoch": 1.8684383922688337, + "grad_norm": 2.0925397872924805, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7238560914993286, + "num_tokens": 439886906.0, + "step": 17014 + }, + { + "epoch": 1.8685482099714474, + "grad_norm": 2.2331645488739014, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7059720158576965, + "num_tokens": 439912722.0, + "step": 17015 + }, + { + "epoch": 1.8686580276740612, + "grad_norm": 2.407700300216675, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7331275939941406, + "num_tokens": 439936368.0, + "step": 17016 + }, + { + "epoch": 1.8687678453766747, + "grad_norm": 2.3564586639404297, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7255584001541138, + "num_tokens": 439961316.0, + "step": 17017 + }, + { + "epoch": 1.8688776630792883, + "grad_norm": 2.2051265239715576, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6988705396652222, + "num_tokens": 439989532.0, + "step": 17018 + }, + { + "epoch": 1.868987480781902, + "grad_norm": 2.2136309146881104, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.716035008430481, + "num_tokens": 440016373.0, + "step": 17019 + }, + { + "epoch": 1.8690972984845158, + "grad_norm": 2.154707908630371, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7404053211212158, + "num_tokens": 440041652.0, + "step": 17020 + }, + { + "epoch": 1.8692071161871293, + "grad_norm": 2.2668256759643555, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7148367762565613, + "num_tokens": 440065463.0, + "step": 17021 + }, + { + "epoch": 1.869316933889743, + "grad_norm": 2.1543309688568115, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.704146146774292, + "num_tokens": 440092608.0, + "step": 17022 + }, + { + "epoch": 1.8694267515923566, + "grad_norm": 2.006415367126465, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7180370092391968, + "num_tokens": 440122046.0, + "step": 17023 + }, + { + "epoch": 1.8695365692949704, + "grad_norm": 1.9642815589904785, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7305564284324646, + "num_tokens": 440152237.0, + "step": 17024 + }, + { + "epoch": 1.8696463869975841, + "grad_norm": 2.112710475921631, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7067071199417114, + "num_tokens": 440179011.0, + "step": 17025 + }, + { + "epoch": 1.8697562047001977, + "grad_norm": 2.2519216537475586, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7305555939674377, + "num_tokens": 440202428.0, + "step": 17026 + }, + { + "epoch": 1.8698660224028112, + "grad_norm": 2.3014256954193115, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7021865248680115, + "num_tokens": 440228411.0, + "step": 17027 + }, + { + "epoch": 1.869975840105425, + "grad_norm": 2.20330548286438, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.726421058177948, + "num_tokens": 440253868.0, + "step": 17028 + }, + { + "epoch": 1.8700856578080387, + "grad_norm": 2.2183127403259277, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7285683155059814, + "num_tokens": 440276775.0, + "step": 17029 + }, + { + "epoch": 1.8701954755106525, + "grad_norm": 2.082531213760376, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.711585283279419, + "num_tokens": 440304245.0, + "step": 17030 + }, + { + "epoch": 1.870305293213266, + "grad_norm": 2.335247278213501, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7256222367286682, + "num_tokens": 440328836.0, + "step": 17031 + }, + { + "epoch": 1.8704151109158795, + "grad_norm": 2.4197275638580322, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7244569063186646, + "num_tokens": 440350971.0, + "step": 17032 + }, + { + "epoch": 1.8705249286184933, + "grad_norm": 2.1606812477111816, + "learning_rate": 1e-06, + "loss": 1.0383, + "mean_token_accuracy": 0.6801868677139282, + "num_tokens": 440377766.0, + "step": 17033 + }, + { + "epoch": 1.870634746321107, + "grad_norm": 1.9963020086288452, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7392866015434265, + "num_tokens": 440407060.0, + "step": 17034 + }, + { + "epoch": 1.8707445640237206, + "grad_norm": 2.079679489135742, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7040193676948547, + "num_tokens": 440433321.0, + "step": 17035 + }, + { + "epoch": 1.8708543817263343, + "grad_norm": 2.1058874130249023, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7266153693199158, + "num_tokens": 440461191.0, + "step": 17036 + }, + { + "epoch": 1.8709641994289479, + "grad_norm": 2.25582218170166, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6980556845664978, + "num_tokens": 440485817.0, + "step": 17037 + }, + { + "epoch": 1.8710740171315616, + "grad_norm": 2.082547426223755, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7180829644203186, + "num_tokens": 440512880.0, + "step": 17038 + }, + { + "epoch": 1.8711838348341754, + "grad_norm": 2.0165367126464844, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7264814376831055, + "num_tokens": 440543790.0, + "step": 17039 + }, + { + "epoch": 1.871293652536789, + "grad_norm": 2.2908692359924316, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7126232385635376, + "num_tokens": 440568853.0, + "step": 17040 + }, + { + "epoch": 1.8714034702394025, + "grad_norm": 2.002903938293457, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7054021954536438, + "num_tokens": 440599282.0, + "step": 17041 + }, + { + "epoch": 1.8715132879420162, + "grad_norm": 2.0025768280029297, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7496194839477539, + "num_tokens": 440629140.0, + "step": 17042 + }, + { + "epoch": 1.87162310564463, + "grad_norm": 2.2613468170166016, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7206473350524902, + "num_tokens": 440653455.0, + "step": 17043 + }, + { + "epoch": 1.8717329233472437, + "grad_norm": 2.1816062927246094, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.724769115447998, + "num_tokens": 440680585.0, + "step": 17044 + }, + { + "epoch": 1.8718427410498573, + "grad_norm": 2.4902682304382324, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.726598858833313, + "num_tokens": 440703091.0, + "step": 17045 + }, + { + "epoch": 1.8719525587524708, + "grad_norm": 2.2129247188568115, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7227422595024109, + "num_tokens": 440729516.0, + "step": 17046 + }, + { + "epoch": 1.8720623764550846, + "grad_norm": 1.961178183555603, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7279337048530579, + "num_tokens": 440764333.0, + "step": 17047 + }, + { + "epoch": 1.8721721941576983, + "grad_norm": 2.017038106918335, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7072640657424927, + "num_tokens": 440794112.0, + "step": 17048 + }, + { + "epoch": 1.8722820118603118, + "grad_norm": 2.345635414123535, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7381199598312378, + "num_tokens": 440816068.0, + "step": 17049 + }, + { + "epoch": 1.8723918295629254, + "grad_norm": 2.386084794998169, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7301315069198608, + "num_tokens": 440839606.0, + "step": 17050 + }, + { + "epoch": 1.8725016472655391, + "grad_norm": 2.1391351222991943, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7082157135009766, + "num_tokens": 440867155.0, + "step": 17051 + }, + { + "epoch": 1.872611464968153, + "grad_norm": 2.1989734172821045, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7154281139373779, + "num_tokens": 440892677.0, + "step": 17052 + }, + { + "epoch": 1.8727212826707667, + "grad_norm": 2.163003444671631, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.711211085319519, + "num_tokens": 440920143.0, + "step": 17053 + }, + { + "epoch": 1.8728311003733802, + "grad_norm": 2.396080732345581, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7681758999824524, + "num_tokens": 440940554.0, + "step": 17054 + }, + { + "epoch": 1.8729409180759937, + "grad_norm": 2.027358293533325, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7346588373184204, + "num_tokens": 440971347.0, + "step": 17055 + }, + { + "epoch": 1.8730507357786075, + "grad_norm": 2.221855401992798, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7150883674621582, + "num_tokens": 440997428.0, + "step": 17056 + }, + { + "epoch": 1.8731605534812212, + "grad_norm": 2.2433817386627197, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7318795323371887, + "num_tokens": 441020541.0, + "step": 17057 + }, + { + "epoch": 1.873270371183835, + "grad_norm": 2.189485549926758, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7606000900268555, + "num_tokens": 441044714.0, + "step": 17058 + }, + { + "epoch": 1.8733801888864485, + "grad_norm": 2.2447760105133057, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7329670190811157, + "num_tokens": 441068532.0, + "step": 17059 + }, + { + "epoch": 1.873490006589062, + "grad_norm": 2.480306386947632, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7486000061035156, + "num_tokens": 441091202.0, + "step": 17060 + }, + { + "epoch": 1.8735998242916758, + "grad_norm": 2.529222011566162, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7165319919586182, + "num_tokens": 441112498.0, + "step": 17061 + }, + { + "epoch": 1.8737096419942896, + "grad_norm": 2.1089155673980713, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7126781344413757, + "num_tokens": 441142109.0, + "step": 17062 + }, + { + "epoch": 1.8738194596969031, + "grad_norm": 2.7110369205474854, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.700005292892456, + "num_tokens": 441164003.0, + "step": 17063 + }, + { + "epoch": 1.8739292773995166, + "grad_norm": 2.1902871131896973, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.696171760559082, + "num_tokens": 441189905.0, + "step": 17064 + }, + { + "epoch": 1.8740390951021304, + "grad_norm": 2.2234387397766113, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.723690390586853, + "num_tokens": 441215805.0, + "step": 17065 + }, + { + "epoch": 1.8741489128047442, + "grad_norm": 2.099902629852295, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7091149687767029, + "num_tokens": 441246601.0, + "step": 17066 + }, + { + "epoch": 1.874258730507358, + "grad_norm": 2.052159070968628, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7313410043716431, + "num_tokens": 441274797.0, + "step": 17067 + }, + { + "epoch": 1.8743685482099715, + "grad_norm": 2.26375150680542, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7051870226860046, + "num_tokens": 441301370.0, + "step": 17068 + }, + { + "epoch": 1.874478365912585, + "grad_norm": 2.462700128555298, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7258487939834595, + "num_tokens": 441323660.0, + "step": 17069 + }, + { + "epoch": 1.8745881836151987, + "grad_norm": 2.0852248668670654, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7133265137672424, + "num_tokens": 441352209.0, + "step": 17070 + }, + { + "epoch": 1.8746980013178125, + "grad_norm": 1.942598581314087, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7182430624961853, + "num_tokens": 441383721.0, + "step": 17071 + }, + { + "epoch": 1.8748078190204263, + "grad_norm": 2.1919147968292236, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7115254998207092, + "num_tokens": 441409299.0, + "step": 17072 + }, + { + "epoch": 1.8749176367230398, + "grad_norm": 2.2464723587036133, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7193213701248169, + "num_tokens": 441434302.0, + "step": 17073 + }, + { + "epoch": 1.8750274544256533, + "grad_norm": 2.127978563308716, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7204844951629639, + "num_tokens": 441461110.0, + "step": 17074 + }, + { + "epoch": 1.875137272128267, + "grad_norm": 1.9104952812194824, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7173978090286255, + "num_tokens": 441493487.0, + "step": 17075 + }, + { + "epoch": 1.8752470898308808, + "grad_norm": 2.180657386779785, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7456372976303101, + "num_tokens": 441518779.0, + "step": 17076 + }, + { + "epoch": 1.8753569075334944, + "grad_norm": 2.282294511795044, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7419949769973755, + "num_tokens": 441542646.0, + "step": 17077 + }, + { + "epoch": 1.875466725236108, + "grad_norm": 1.8931878805160522, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7247294783592224, + "num_tokens": 441573331.0, + "step": 17078 + }, + { + "epoch": 1.8755765429387217, + "grad_norm": 2.0722849369049072, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7321228981018066, + "num_tokens": 441602259.0, + "step": 17079 + }, + { + "epoch": 1.8756863606413354, + "grad_norm": 2.0064096450805664, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6968038082122803, + "num_tokens": 441634513.0, + "step": 17080 + }, + { + "epoch": 1.8757961783439492, + "grad_norm": 2.3912060260772705, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7255694270133972, + "num_tokens": 441657125.0, + "step": 17081 + }, + { + "epoch": 1.8759059960465627, + "grad_norm": 2.1508021354675293, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.712361216545105, + "num_tokens": 441683874.0, + "step": 17082 + }, + { + "epoch": 1.8760158137491763, + "grad_norm": 2.210014581680298, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.6990885734558105, + "num_tokens": 441710732.0, + "step": 17083 + }, + { + "epoch": 1.87612563145179, + "grad_norm": 2.310124158859253, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7325401306152344, + "num_tokens": 441736131.0, + "step": 17084 + }, + { + "epoch": 1.8762354491544038, + "grad_norm": 2.42842435836792, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7157084345817566, + "num_tokens": 441758858.0, + "step": 17085 + }, + { + "epoch": 1.8763452668570173, + "grad_norm": 2.315366744995117, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7300729751586914, + "num_tokens": 441782241.0, + "step": 17086 + }, + { + "epoch": 1.876455084559631, + "grad_norm": 2.2527058124542236, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7313591837882996, + "num_tokens": 441805904.0, + "step": 17087 + }, + { + "epoch": 1.8765649022622446, + "grad_norm": 2.551832675933838, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7295515537261963, + "num_tokens": 441826314.0, + "step": 17088 + }, + { + "epoch": 1.8766747199648584, + "grad_norm": 2.282440423965454, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7194865942001343, + "num_tokens": 441852105.0, + "step": 17089 + }, + { + "epoch": 1.876784537667472, + "grad_norm": 2.0120670795440674, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7115550637245178, + "num_tokens": 441882518.0, + "step": 17090 + }, + { + "epoch": 1.8768943553700856, + "grad_norm": 2.5474720001220703, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.707331120967865, + "num_tokens": 441905581.0, + "step": 17091 + }, + { + "epoch": 1.8770041730726992, + "grad_norm": 2.7990996837615967, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7510743737220764, + "num_tokens": 441921785.0, + "step": 17092 + }, + { + "epoch": 1.877113990775313, + "grad_norm": 2.297525644302368, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7357379198074341, + "num_tokens": 441944956.0, + "step": 17093 + }, + { + "epoch": 1.8772238084779267, + "grad_norm": 2.4212141036987305, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.723966121673584, + "num_tokens": 441967212.0, + "step": 17094 + }, + { + "epoch": 1.8773336261805404, + "grad_norm": 2.0201351642608643, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7163159251213074, + "num_tokens": 441997960.0, + "step": 17095 + }, + { + "epoch": 1.877443443883154, + "grad_norm": 2.263949155807495, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7155415415763855, + "num_tokens": 442022187.0, + "step": 17096 + }, + { + "epoch": 1.8775532615857675, + "grad_norm": 2.1560137271881104, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.6995608806610107, + "num_tokens": 442050708.0, + "step": 17097 + }, + { + "epoch": 1.8776630792883813, + "grad_norm": 2.224172592163086, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7458227872848511, + "num_tokens": 442075492.0, + "step": 17098 + }, + { + "epoch": 1.877772896990995, + "grad_norm": 2.154397964477539, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7027547359466553, + "num_tokens": 442102548.0, + "step": 17099 + }, + { + "epoch": 1.8778827146936086, + "grad_norm": 2.1521494388580322, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.710162878036499, + "num_tokens": 442130805.0, + "step": 17100 + }, + { + "epoch": 1.8779925323962223, + "grad_norm": 2.4495849609375, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.711675226688385, + "num_tokens": 442154279.0, + "step": 17101 + }, + { + "epoch": 1.8781023500988359, + "grad_norm": 2.1656694412231445, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7141727209091187, + "num_tokens": 442182025.0, + "step": 17102 + }, + { + "epoch": 1.8782121678014496, + "grad_norm": 2.364508867263794, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7325058579444885, + "num_tokens": 442205503.0, + "step": 17103 + }, + { + "epoch": 1.8783219855040634, + "grad_norm": 2.3218910694122314, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7343572378158569, + "num_tokens": 442228249.0, + "step": 17104 + }, + { + "epoch": 1.878431803206677, + "grad_norm": 2.4152636528015137, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7202723026275635, + "num_tokens": 442249551.0, + "step": 17105 + }, + { + "epoch": 1.8785416209092904, + "grad_norm": 2.3567423820495605, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7329573631286621, + "num_tokens": 442271194.0, + "step": 17106 + }, + { + "epoch": 1.8786514386119042, + "grad_norm": 2.103943109512329, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7226606607437134, + "num_tokens": 442300042.0, + "step": 17107 + }, + { + "epoch": 1.878761256314518, + "grad_norm": 2.2468364238739014, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7065691947937012, + "num_tokens": 442328394.0, + "step": 17108 + }, + { + "epoch": 1.8788710740171317, + "grad_norm": 2.011626958847046, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.700446605682373, + "num_tokens": 442360783.0, + "step": 17109 + }, + { + "epoch": 1.8789808917197452, + "grad_norm": 2.3067564964294434, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7188385128974915, + "num_tokens": 442385450.0, + "step": 17110 + }, + { + "epoch": 1.8790907094223588, + "grad_norm": 2.261631488800049, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7133034467697144, + "num_tokens": 442410895.0, + "step": 17111 + }, + { + "epoch": 1.8792005271249725, + "grad_norm": 2.293743848800659, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7389310002326965, + "num_tokens": 442433978.0, + "step": 17112 + }, + { + "epoch": 1.8793103448275863, + "grad_norm": 2.2813000679016113, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7171081304550171, + "num_tokens": 442459487.0, + "step": 17113 + }, + { + "epoch": 1.8794201625301998, + "grad_norm": 2.194662094116211, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7163358926773071, + "num_tokens": 442488748.0, + "step": 17114 + }, + { + "epoch": 1.8795299802328134, + "grad_norm": 2.14337158203125, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7396522760391235, + "num_tokens": 442513277.0, + "step": 17115 + }, + { + "epoch": 1.8796397979354271, + "grad_norm": 2.3535163402557373, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7431483268737793, + "num_tokens": 442535248.0, + "step": 17116 + }, + { + "epoch": 1.8797496156380409, + "grad_norm": 2.0229063034057617, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7192592620849609, + "num_tokens": 442565238.0, + "step": 17117 + }, + { + "epoch": 1.8798594333406546, + "grad_norm": 2.150944232940674, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.731997013092041, + "num_tokens": 442592048.0, + "step": 17118 + }, + { + "epoch": 1.8799692510432682, + "grad_norm": 2.192225217819214, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7070716619491577, + "num_tokens": 442619496.0, + "step": 17119 + }, + { + "epoch": 1.8800790687458817, + "grad_norm": 2.3068675994873047, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7168205976486206, + "num_tokens": 442644476.0, + "step": 17120 + }, + { + "epoch": 1.8801888864484955, + "grad_norm": 2.259427547454834, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7068068385124207, + "num_tokens": 442670967.0, + "step": 17121 + }, + { + "epoch": 1.8802987041511092, + "grad_norm": 2.319673538208008, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7205381393432617, + "num_tokens": 442694055.0, + "step": 17122 + }, + { + "epoch": 1.880408521853723, + "grad_norm": 2.2860500812530518, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6981269121170044, + "num_tokens": 442719608.0, + "step": 17123 + }, + { + "epoch": 1.8805183395563365, + "grad_norm": 2.212418556213379, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7191596031188965, + "num_tokens": 442745522.0, + "step": 17124 + }, + { + "epoch": 1.88062815725895, + "grad_norm": 2.0042097568511963, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7451581358909607, + "num_tokens": 442774483.0, + "step": 17125 + }, + { + "epoch": 1.8807379749615638, + "grad_norm": 2.3264071941375732, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7230054140090942, + "num_tokens": 442799053.0, + "step": 17126 + }, + { + "epoch": 1.8808477926641776, + "grad_norm": 1.953932523727417, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7007887363433838, + "num_tokens": 442830422.0, + "step": 17127 + }, + { + "epoch": 1.880957610366791, + "grad_norm": 2.1285955905914307, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7077798843383789, + "num_tokens": 442858114.0, + "step": 17128 + }, + { + "epoch": 1.8810674280694046, + "grad_norm": 2.206672430038452, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.698340892791748, + "num_tokens": 442885656.0, + "step": 17129 + }, + { + "epoch": 1.8811772457720184, + "grad_norm": 1.9106887578964233, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7353742718696594, + "num_tokens": 442915149.0, + "step": 17130 + }, + { + "epoch": 1.8812870634746321, + "grad_norm": 2.211275100708008, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.720358669757843, + "num_tokens": 442940307.0, + "step": 17131 + }, + { + "epoch": 1.881396881177246, + "grad_norm": 1.9426509141921997, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7156614661216736, + "num_tokens": 442971327.0, + "step": 17132 + }, + { + "epoch": 1.8815066988798594, + "grad_norm": 2.4801063537597656, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7266755104064941, + "num_tokens": 442993456.0, + "step": 17133 + }, + { + "epoch": 1.881616516582473, + "grad_norm": 2.0628128051757812, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7123522162437439, + "num_tokens": 443022173.0, + "step": 17134 + }, + { + "epoch": 1.8817263342850867, + "grad_norm": 2.1842052936553955, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7307441234588623, + "num_tokens": 443048031.0, + "step": 17135 + }, + { + "epoch": 1.8818361519877005, + "grad_norm": 2.599499464035034, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7244629859924316, + "num_tokens": 443068998.0, + "step": 17136 + }, + { + "epoch": 1.881945969690314, + "grad_norm": 2.4237492084503174, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7046416401863098, + "num_tokens": 443090374.0, + "step": 17137 + }, + { + "epoch": 1.8820557873929278, + "grad_norm": 2.207818031311035, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7291253209114075, + "num_tokens": 443116477.0, + "step": 17138 + }, + { + "epoch": 1.8821656050955413, + "grad_norm": 2.2407751083374023, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7447729110717773, + "num_tokens": 443142580.0, + "step": 17139 + }, + { + "epoch": 1.882275422798155, + "grad_norm": 2.5265400409698486, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.717668890953064, + "num_tokens": 443165982.0, + "step": 17140 + }, + { + "epoch": 1.8823852405007688, + "grad_norm": 1.9713934659957886, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7292642593383789, + "num_tokens": 443197490.0, + "step": 17141 + }, + { + "epoch": 1.8824950582033824, + "grad_norm": 2.517961263656616, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7515721321105957, + "num_tokens": 443216578.0, + "step": 17142 + }, + { + "epoch": 1.882604875905996, + "grad_norm": 2.050507068634033, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7362532019615173, + "num_tokens": 443246303.0, + "step": 17143 + }, + { + "epoch": 1.8827146936086097, + "grad_norm": 2.236409902572632, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7166711688041687, + "num_tokens": 443272511.0, + "step": 17144 + }, + { + "epoch": 1.8828245113112234, + "grad_norm": 2.225597620010376, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7131053805351257, + "num_tokens": 443297898.0, + "step": 17145 + }, + { + "epoch": 1.8829343290138372, + "grad_norm": 2.3317103385925293, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7286467552185059, + "num_tokens": 443320756.0, + "step": 17146 + }, + { + "epoch": 1.8830441467164507, + "grad_norm": 2.3054869174957275, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7157937288284302, + "num_tokens": 443344590.0, + "step": 17147 + }, + { + "epoch": 1.8831539644190642, + "grad_norm": 2.163010358810425, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7065784931182861, + "num_tokens": 443372794.0, + "step": 17148 + }, + { + "epoch": 1.883263782121678, + "grad_norm": 2.2298738956451416, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7020618915557861, + "num_tokens": 443401270.0, + "step": 17149 + }, + { + "epoch": 1.8833735998242918, + "grad_norm": 1.9670684337615967, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7122107744216919, + "num_tokens": 443432763.0, + "step": 17150 + }, + { + "epoch": 1.8834834175269053, + "grad_norm": 2.065330982208252, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7023380994796753, + "num_tokens": 443462146.0, + "step": 17151 + }, + { + "epoch": 1.883593235229519, + "grad_norm": 2.1078808307647705, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7410955429077148, + "num_tokens": 443489265.0, + "step": 17152 + }, + { + "epoch": 1.8837030529321326, + "grad_norm": 2.0695583820343018, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7101521492004395, + "num_tokens": 443521043.0, + "step": 17153 + }, + { + "epoch": 1.8838128706347463, + "grad_norm": 2.153315544128418, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7214174866676331, + "num_tokens": 443548868.0, + "step": 17154 + }, + { + "epoch": 1.88392268833736, + "grad_norm": 2.3513994216918945, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7573180794715881, + "num_tokens": 443572380.0, + "step": 17155 + }, + { + "epoch": 1.8840325060399736, + "grad_norm": 2.340447187423706, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7246741056442261, + "num_tokens": 443595791.0, + "step": 17156 + }, + { + "epoch": 1.8841423237425872, + "grad_norm": 2.0932486057281494, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7120018005371094, + "num_tokens": 443623415.0, + "step": 17157 + }, + { + "epoch": 1.884252141445201, + "grad_norm": 2.0895304679870605, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7137110829353333, + "num_tokens": 443653345.0, + "step": 17158 + }, + { + "epoch": 1.8843619591478147, + "grad_norm": 2.1933629512786865, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7306094169616699, + "num_tokens": 443681577.0, + "step": 17159 + }, + { + "epoch": 1.8844717768504284, + "grad_norm": 2.2133171558380127, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7250685691833496, + "num_tokens": 443707466.0, + "step": 17160 + }, + { + "epoch": 1.884581594553042, + "grad_norm": 2.326287031173706, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7159541845321655, + "num_tokens": 443730996.0, + "step": 17161 + }, + { + "epoch": 1.8846914122556555, + "grad_norm": 2.402956008911133, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7096828818321228, + "num_tokens": 443754314.0, + "step": 17162 + }, + { + "epoch": 1.8848012299582693, + "grad_norm": 2.5185210704803467, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.729429304599762, + "num_tokens": 443775221.0, + "step": 17163 + }, + { + "epoch": 1.884911047660883, + "grad_norm": 2.32672381401062, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.6989413499832153, + "num_tokens": 443799455.0, + "step": 17164 + }, + { + "epoch": 1.8850208653634966, + "grad_norm": 2.3022079467773438, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7082246541976929, + "num_tokens": 443823852.0, + "step": 17165 + }, + { + "epoch": 1.88513068306611, + "grad_norm": 2.1631038188934326, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.698264479637146, + "num_tokens": 443850175.0, + "step": 17166 + }, + { + "epoch": 1.8852405007687238, + "grad_norm": 2.3636960983276367, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7339774966239929, + "num_tokens": 443874583.0, + "step": 17167 + }, + { + "epoch": 1.8853503184713376, + "grad_norm": 2.5643765926361084, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7387784719467163, + "num_tokens": 443894730.0, + "step": 17168 + }, + { + "epoch": 1.8854601361739514, + "grad_norm": 2.5739641189575195, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7457238435745239, + "num_tokens": 443914815.0, + "step": 17169 + }, + { + "epoch": 1.885569953876565, + "grad_norm": 1.9915318489074707, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6935207843780518, + "num_tokens": 443946817.0, + "step": 17170 + }, + { + "epoch": 1.8856797715791784, + "grad_norm": 2.2768139839172363, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7107672095298767, + "num_tokens": 443972576.0, + "step": 17171 + }, + { + "epoch": 1.8857895892817922, + "grad_norm": 2.1755478382110596, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7299512624740601, + "num_tokens": 443997637.0, + "step": 17172 + }, + { + "epoch": 1.885899406984406, + "grad_norm": 2.6351845264434814, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.728927731513977, + "num_tokens": 444016602.0, + "step": 17173 + }, + { + "epoch": 1.8860092246870197, + "grad_norm": 2.1634035110473633, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7279007434844971, + "num_tokens": 444042596.0, + "step": 17174 + }, + { + "epoch": 1.8861190423896332, + "grad_norm": 1.9674406051635742, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7401471734046936, + "num_tokens": 444071027.0, + "step": 17175 + }, + { + "epoch": 1.8862288600922468, + "grad_norm": 2.2254042625427246, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7074413299560547, + "num_tokens": 444097511.0, + "step": 17176 + }, + { + "epoch": 1.8863386777948605, + "grad_norm": 2.203036069869995, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7111716270446777, + "num_tokens": 444123105.0, + "step": 17177 + }, + { + "epoch": 1.8864484954974743, + "grad_norm": 2.023407459259033, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6986115574836731, + "num_tokens": 444152880.0, + "step": 17178 + }, + { + "epoch": 1.8865583132000878, + "grad_norm": 2.2356600761413574, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7123507261276245, + "num_tokens": 444177159.0, + "step": 17179 + }, + { + "epoch": 1.8866681309027014, + "grad_norm": 2.0103328227996826, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7304377555847168, + "num_tokens": 444204765.0, + "step": 17180 + }, + { + "epoch": 1.8867779486053151, + "grad_norm": 1.7868657112121582, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7067182064056396, + "num_tokens": 444240019.0, + "step": 17181 + }, + { + "epoch": 1.8868877663079289, + "grad_norm": 2.107013463973999, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.727180004119873, + "num_tokens": 444265462.0, + "step": 17182 + }, + { + "epoch": 1.8869975840105426, + "grad_norm": 2.2072086334228516, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7464401721954346, + "num_tokens": 444290496.0, + "step": 17183 + }, + { + "epoch": 1.8871074017131562, + "grad_norm": 2.1211678981781006, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7351466417312622, + "num_tokens": 444316974.0, + "step": 17184 + }, + { + "epoch": 1.8872172194157697, + "grad_norm": 2.093319892883301, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7181426882743835, + "num_tokens": 444347182.0, + "step": 17185 + }, + { + "epoch": 1.8873270371183835, + "grad_norm": 2.2682723999023438, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7087410092353821, + "num_tokens": 444373363.0, + "step": 17186 + }, + { + "epoch": 1.8874368548209972, + "grad_norm": 2.547746181488037, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7091357707977295, + "num_tokens": 444395418.0, + "step": 17187 + }, + { + "epoch": 1.887546672523611, + "grad_norm": 2.023376703262329, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7435751557350159, + "num_tokens": 444423375.0, + "step": 17188 + }, + { + "epoch": 1.8876564902262245, + "grad_norm": 2.394167184829712, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7284590005874634, + "num_tokens": 444444920.0, + "step": 17189 + }, + { + "epoch": 1.887766307928838, + "grad_norm": 2.139106035232544, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7090500593185425, + "num_tokens": 444474529.0, + "step": 17190 + }, + { + "epoch": 1.8878761256314518, + "grad_norm": 2.25355863571167, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7085590362548828, + "num_tokens": 444500093.0, + "step": 17191 + }, + { + "epoch": 1.8879859433340656, + "grad_norm": 2.206144332885742, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.706115186214447, + "num_tokens": 444527583.0, + "step": 17192 + }, + { + "epoch": 1.888095761036679, + "grad_norm": 2.095292806625366, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7463536262512207, + "num_tokens": 444554468.0, + "step": 17193 + }, + { + "epoch": 1.8882055787392926, + "grad_norm": 2.4527201652526855, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7115116119384766, + "num_tokens": 444576911.0, + "step": 17194 + }, + { + "epoch": 1.8883153964419064, + "grad_norm": 2.177919387817383, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.710392951965332, + "num_tokens": 444606474.0, + "step": 17195 + }, + { + "epoch": 1.8884252141445201, + "grad_norm": 2.1499381065368652, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.6977782845497131, + "num_tokens": 444634777.0, + "step": 17196 + }, + { + "epoch": 1.888535031847134, + "grad_norm": 2.2170450687408447, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7190183401107788, + "num_tokens": 444659687.0, + "step": 17197 + }, + { + "epoch": 1.8886448495497474, + "grad_norm": 2.319296360015869, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7353146076202393, + "num_tokens": 444684014.0, + "step": 17198 + }, + { + "epoch": 1.888754667252361, + "grad_norm": 2.3968656063079834, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7060245275497437, + "num_tokens": 444706289.0, + "step": 17199 + }, + { + "epoch": 1.8888644849549747, + "grad_norm": 2.4859778881073, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.720076322555542, + "num_tokens": 444727952.0, + "step": 17200 + }, + { + "epoch": 1.8889743026575885, + "grad_norm": 2.0611133575439453, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7222867608070374, + "num_tokens": 444757089.0, + "step": 17201 + }, + { + "epoch": 1.889084120360202, + "grad_norm": 1.9767204523086548, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.702790379524231, + "num_tokens": 444787948.0, + "step": 17202 + }, + { + "epoch": 1.8891939380628158, + "grad_norm": 2.006538152694702, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.6932421922683716, + "num_tokens": 444821121.0, + "step": 17203 + }, + { + "epoch": 1.8893037557654293, + "grad_norm": 2.440877676010132, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7236744165420532, + "num_tokens": 444843954.0, + "step": 17204 + }, + { + "epoch": 1.889413573468043, + "grad_norm": 2.4165289402008057, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7045992016792297, + "num_tokens": 444865200.0, + "step": 17205 + }, + { + "epoch": 1.8895233911706568, + "grad_norm": 2.3356106281280518, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7263287901878357, + "num_tokens": 444888083.0, + "step": 17206 + }, + { + "epoch": 1.8896332088732704, + "grad_norm": 2.285144567489624, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7371195554733276, + "num_tokens": 444912017.0, + "step": 17207 + }, + { + "epoch": 1.8897430265758839, + "grad_norm": 2.406111717224121, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7357926964759827, + "num_tokens": 444934303.0, + "step": 17208 + }, + { + "epoch": 1.8898528442784976, + "grad_norm": 2.4839589595794678, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7225465178489685, + "num_tokens": 444956792.0, + "step": 17209 + }, + { + "epoch": 1.8899626619811114, + "grad_norm": 2.401303768157959, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7242069244384766, + "num_tokens": 444978512.0, + "step": 17210 + }, + { + "epoch": 1.8900724796837252, + "grad_norm": 2.1787917613983154, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7164653539657593, + "num_tokens": 445007847.0, + "step": 17211 + }, + { + "epoch": 1.8901822973863387, + "grad_norm": 2.1763336658477783, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7161573171615601, + "num_tokens": 445034565.0, + "step": 17212 + }, + { + "epoch": 1.8902921150889522, + "grad_norm": 1.9852086305618286, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.70975661277771, + "num_tokens": 445066671.0, + "step": 17213 + }, + { + "epoch": 1.890401932791566, + "grad_norm": 1.974450945854187, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7054085731506348, + "num_tokens": 445100563.0, + "step": 17214 + }, + { + "epoch": 1.8905117504941797, + "grad_norm": 2.0481719970703125, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7304192781448364, + "num_tokens": 445129719.0, + "step": 17215 + }, + { + "epoch": 1.8906215681967933, + "grad_norm": 2.4545366764068604, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7125369310379028, + "num_tokens": 445152273.0, + "step": 17216 + }, + { + "epoch": 1.890731385899407, + "grad_norm": 2.1981194019317627, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7329657077789307, + "num_tokens": 445176937.0, + "step": 17217 + }, + { + "epoch": 1.8908412036020206, + "grad_norm": 1.9802011251449585, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.6999946236610413, + "num_tokens": 445207189.0, + "step": 17218 + }, + { + "epoch": 1.8909510213046343, + "grad_norm": 2.4215023517608643, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7194986343383789, + "num_tokens": 445229436.0, + "step": 17219 + }, + { + "epoch": 1.891060839007248, + "grad_norm": 2.2220139503479004, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7210842967033386, + "num_tokens": 445257816.0, + "step": 17220 + }, + { + "epoch": 1.8911706567098616, + "grad_norm": 2.1938095092773438, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7123371362686157, + "num_tokens": 445282658.0, + "step": 17221 + }, + { + "epoch": 1.8912804744124752, + "grad_norm": 2.1004879474639893, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7111753225326538, + "num_tokens": 445309648.0, + "step": 17222 + }, + { + "epoch": 1.891390292115089, + "grad_norm": 2.2391762733459473, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7211377620697021, + "num_tokens": 445336103.0, + "step": 17223 + }, + { + "epoch": 1.8915001098177027, + "grad_norm": 1.984817385673523, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7196491956710815, + "num_tokens": 445366873.0, + "step": 17224 + }, + { + "epoch": 1.8916099275203164, + "grad_norm": 2.2208688259124756, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7033231854438782, + "num_tokens": 445394047.0, + "step": 17225 + }, + { + "epoch": 1.89171974522293, + "grad_norm": 2.35351824760437, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7157394289970398, + "num_tokens": 445418118.0, + "step": 17226 + }, + { + "epoch": 1.8918295629255435, + "grad_norm": 2.093502998352051, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7169396281242371, + "num_tokens": 445446647.0, + "step": 17227 + }, + { + "epoch": 1.8919393806281573, + "grad_norm": 2.268420696258545, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7275571823120117, + "num_tokens": 445470464.0, + "step": 17228 + }, + { + "epoch": 1.892049198330771, + "grad_norm": 2.4107913970947266, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7377595901489258, + "num_tokens": 445492287.0, + "step": 17229 + }, + { + "epoch": 1.8921590160333845, + "grad_norm": 2.3118185997009277, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7149352431297302, + "num_tokens": 445516898.0, + "step": 17230 + }, + { + "epoch": 1.892268833735998, + "grad_norm": 2.293886423110962, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.710302472114563, + "num_tokens": 445541226.0, + "step": 17231 + }, + { + "epoch": 1.8923786514386118, + "grad_norm": 2.341815233230591, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7308875322341919, + "num_tokens": 445564130.0, + "step": 17232 + }, + { + "epoch": 1.8924884691412256, + "grad_norm": 2.1682960987091064, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7286867499351501, + "num_tokens": 445589156.0, + "step": 17233 + }, + { + "epoch": 1.8925982868438394, + "grad_norm": 2.28092622756958, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7208375334739685, + "num_tokens": 445613269.0, + "step": 17234 + }, + { + "epoch": 1.8927081045464529, + "grad_norm": 2.168440103530884, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.707480251789093, + "num_tokens": 445642926.0, + "step": 17235 + }, + { + "epoch": 1.8928179222490664, + "grad_norm": 2.4544107913970947, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7505584359169006, + "num_tokens": 445663012.0, + "step": 17236 + }, + { + "epoch": 1.8929277399516802, + "grad_norm": 2.131990909576416, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7267509698867798, + "num_tokens": 445690161.0, + "step": 17237 + }, + { + "epoch": 1.893037557654294, + "grad_norm": 2.2115235328674316, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7087301015853882, + "num_tokens": 445716987.0, + "step": 17238 + }, + { + "epoch": 1.8931473753569077, + "grad_norm": 2.273376703262329, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7446264624595642, + "num_tokens": 445741219.0, + "step": 17239 + }, + { + "epoch": 1.8932571930595212, + "grad_norm": 2.33957576751709, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7080694437026978, + "num_tokens": 445766150.0, + "step": 17240 + }, + { + "epoch": 1.8933670107621348, + "grad_norm": 2.081845760345459, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7087562084197998, + "num_tokens": 445793915.0, + "step": 17241 + }, + { + "epoch": 1.8934768284647485, + "grad_norm": 2.1749231815338135, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7110491991043091, + "num_tokens": 445819999.0, + "step": 17242 + }, + { + "epoch": 1.8935866461673623, + "grad_norm": 2.135279655456543, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7121226191520691, + "num_tokens": 445849330.0, + "step": 17243 + }, + { + "epoch": 1.8936964638699758, + "grad_norm": 2.158219337463379, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.706019401550293, + "num_tokens": 445876150.0, + "step": 17244 + }, + { + "epoch": 1.8938062815725893, + "grad_norm": 2.061284303665161, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7211770415306091, + "num_tokens": 445906024.0, + "step": 17245 + }, + { + "epoch": 1.893916099275203, + "grad_norm": 2.0709211826324463, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7025633454322815, + "num_tokens": 445934276.0, + "step": 17246 + }, + { + "epoch": 1.8940259169778169, + "grad_norm": 2.1482760906219482, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7074959874153137, + "num_tokens": 445961624.0, + "step": 17247 + }, + { + "epoch": 1.8941357346804306, + "grad_norm": 2.1231493949890137, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7226855754852295, + "num_tokens": 445991363.0, + "step": 17248 + }, + { + "epoch": 1.8942455523830442, + "grad_norm": 2.4224867820739746, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.710659921169281, + "num_tokens": 446014534.0, + "step": 17249 + }, + { + "epoch": 1.8943553700856577, + "grad_norm": 2.4986519813537598, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7385139465332031, + "num_tokens": 446036606.0, + "step": 17250 + }, + { + "epoch": 1.8944651877882714, + "grad_norm": 2.5793206691741943, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.725176215171814, + "num_tokens": 446056241.0, + "step": 17251 + }, + { + "epoch": 1.8945750054908852, + "grad_norm": 2.8861773014068604, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7287338972091675, + "num_tokens": 446073179.0, + "step": 17252 + }, + { + "epoch": 1.894684823193499, + "grad_norm": 2.2321879863739014, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7040448784828186, + "num_tokens": 446102627.0, + "step": 17253 + }, + { + "epoch": 1.8947946408961125, + "grad_norm": 2.18329119682312, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7283111810684204, + "num_tokens": 446129432.0, + "step": 17254 + }, + { + "epoch": 1.894904458598726, + "grad_norm": 2.356879472732544, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7311956882476807, + "num_tokens": 446151081.0, + "step": 17255 + }, + { + "epoch": 1.8950142763013398, + "grad_norm": 2.384204864501953, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7139342427253723, + "num_tokens": 446176354.0, + "step": 17256 + }, + { + "epoch": 1.8951240940039535, + "grad_norm": 2.2213873863220215, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7197306156158447, + "num_tokens": 446202852.0, + "step": 17257 + }, + { + "epoch": 1.895233911706567, + "grad_norm": 2.4579200744628906, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6931710243225098, + "num_tokens": 446229433.0, + "step": 17258 + }, + { + "epoch": 1.8953437294091806, + "grad_norm": 2.4222588539123535, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.733405590057373, + "num_tokens": 446250044.0, + "step": 17259 + }, + { + "epoch": 1.8954535471117944, + "grad_norm": 2.3162219524383545, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7120134830474854, + "num_tokens": 446274982.0, + "step": 17260 + }, + { + "epoch": 1.8955633648144081, + "grad_norm": 2.4691643714904785, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7276889681816101, + "num_tokens": 446296475.0, + "step": 17261 + }, + { + "epoch": 1.8956731825170219, + "grad_norm": 1.958840250968933, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7122964262962341, + "num_tokens": 446329801.0, + "step": 17262 + }, + { + "epoch": 1.8957830002196354, + "grad_norm": 2.2261340618133545, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7051915526390076, + "num_tokens": 446354878.0, + "step": 17263 + }, + { + "epoch": 1.895892817922249, + "grad_norm": 1.9841194152832031, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7205137610435486, + "num_tokens": 446388505.0, + "step": 17264 + }, + { + "epoch": 1.8960026356248627, + "grad_norm": 2.460338592529297, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7110326290130615, + "num_tokens": 446409303.0, + "step": 17265 + }, + { + "epoch": 1.8961124533274765, + "grad_norm": 2.1162354946136475, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.713363528251648, + "num_tokens": 446437388.0, + "step": 17266 + }, + { + "epoch": 1.89622227103009, + "grad_norm": 2.0704774856567383, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7038212418556213, + "num_tokens": 446467633.0, + "step": 17267 + }, + { + "epoch": 1.8963320887327038, + "grad_norm": 2.212968349456787, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7133147716522217, + "num_tokens": 446495235.0, + "step": 17268 + }, + { + "epoch": 1.8964419064353173, + "grad_norm": 2.1441683769226074, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.70741868019104, + "num_tokens": 446526089.0, + "step": 17269 + }, + { + "epoch": 1.896551724137931, + "grad_norm": 2.59769344329834, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7254543304443359, + "num_tokens": 446546020.0, + "step": 17270 + }, + { + "epoch": 1.8966615418405448, + "grad_norm": 2.06846284866333, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7240279912948608, + "num_tokens": 446573748.0, + "step": 17271 + }, + { + "epoch": 1.8967713595431583, + "grad_norm": 2.2389159202575684, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7288724780082703, + "num_tokens": 446600106.0, + "step": 17272 + }, + { + "epoch": 1.8968811772457719, + "grad_norm": 2.607191562652588, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7346644401550293, + "num_tokens": 446621102.0, + "step": 17273 + }, + { + "epoch": 1.8969909949483856, + "grad_norm": 2.169569253921509, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6911879777908325, + "num_tokens": 446647799.0, + "step": 17274 + }, + { + "epoch": 1.8971008126509994, + "grad_norm": 2.1662633419036865, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7178394198417664, + "num_tokens": 446673387.0, + "step": 17275 + }, + { + "epoch": 1.8972106303536131, + "grad_norm": 2.264033794403076, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7206531167030334, + "num_tokens": 446698085.0, + "step": 17276 + }, + { + "epoch": 1.8973204480562267, + "grad_norm": 2.2330212593078613, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.689509391784668, + "num_tokens": 446724016.0, + "step": 17277 + }, + { + "epoch": 1.8974302657588402, + "grad_norm": 2.2147133350372314, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7301857471466064, + "num_tokens": 446749350.0, + "step": 17278 + }, + { + "epoch": 1.897540083461454, + "grad_norm": 2.0435757637023926, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7268607020378113, + "num_tokens": 446779779.0, + "step": 17279 + }, + { + "epoch": 1.8976499011640677, + "grad_norm": 2.1895737648010254, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7285727262496948, + "num_tokens": 446805939.0, + "step": 17280 + }, + { + "epoch": 1.8977597188666813, + "grad_norm": 2.156015634536743, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7132093906402588, + "num_tokens": 446837087.0, + "step": 17281 + }, + { + "epoch": 1.897869536569295, + "grad_norm": 2.2256381511688232, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7256419658660889, + "num_tokens": 446862282.0, + "step": 17282 + }, + { + "epoch": 1.8979793542719086, + "grad_norm": 1.994559645652771, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7317855954170227, + "num_tokens": 446892472.0, + "step": 17283 + }, + { + "epoch": 1.8980891719745223, + "grad_norm": 2.2277672290802, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7351477146148682, + "num_tokens": 446917216.0, + "step": 17284 + }, + { + "epoch": 1.898198989677136, + "grad_norm": 2.1002941131591797, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7242897748947144, + "num_tokens": 446942707.0, + "step": 17285 + }, + { + "epoch": 1.8983088073797496, + "grad_norm": 2.1265676021575928, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7172478437423706, + "num_tokens": 446970127.0, + "step": 17286 + }, + { + "epoch": 1.8984186250823631, + "grad_norm": 2.6172432899475098, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.716649055480957, + "num_tokens": 446990487.0, + "step": 17287 + }, + { + "epoch": 1.898528442784977, + "grad_norm": 2.1853649616241455, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7314637899398804, + "num_tokens": 447016971.0, + "step": 17288 + }, + { + "epoch": 1.8986382604875907, + "grad_norm": 1.9788670539855957, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6877965331077576, + "num_tokens": 447049599.0, + "step": 17289 + }, + { + "epoch": 1.8987480781902044, + "grad_norm": 2.123168706893921, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7206094861030579, + "num_tokens": 447077909.0, + "step": 17290 + }, + { + "epoch": 1.898857895892818, + "grad_norm": 2.2352442741394043, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7462947368621826, + "num_tokens": 447102460.0, + "step": 17291 + }, + { + "epoch": 1.8989677135954315, + "grad_norm": 2.2262773513793945, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7252726554870605, + "num_tokens": 447127190.0, + "step": 17292 + }, + { + "epoch": 1.8990775312980452, + "grad_norm": 2.2422895431518555, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7366548776626587, + "num_tokens": 447152851.0, + "step": 17293 + }, + { + "epoch": 1.899187349000659, + "grad_norm": 2.072401523590088, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7120875120162964, + "num_tokens": 447181968.0, + "step": 17294 + }, + { + "epoch": 1.8992971667032725, + "grad_norm": 2.363825798034668, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7294365167617798, + "num_tokens": 447204277.0, + "step": 17295 + }, + { + "epoch": 1.899406984405886, + "grad_norm": 2.3958706855773926, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7323154211044312, + "num_tokens": 447226752.0, + "step": 17296 + }, + { + "epoch": 1.8995168021084998, + "grad_norm": 2.384974479675293, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7497067451477051, + "num_tokens": 447248056.0, + "step": 17297 + }, + { + "epoch": 1.8996266198111136, + "grad_norm": 2.2797648906707764, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7113667726516724, + "num_tokens": 447273500.0, + "step": 17298 + }, + { + "epoch": 1.8997364375137273, + "grad_norm": 2.142242431640625, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.718326210975647, + "num_tokens": 447302646.0, + "step": 17299 + }, + { + "epoch": 1.8998462552163409, + "grad_norm": 2.211507558822632, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7206088304519653, + "num_tokens": 447328778.0, + "step": 17300 + }, + { + "epoch": 1.8999560729189544, + "grad_norm": 2.2935540676116943, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.725957989692688, + "num_tokens": 447354432.0, + "step": 17301 + }, + { + "epoch": 1.9000658906215682, + "grad_norm": 2.148080825805664, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7290579080581665, + "num_tokens": 447381128.0, + "step": 17302 + }, + { + "epoch": 1.900175708324182, + "grad_norm": 2.4271132946014404, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7185367941856384, + "num_tokens": 447406027.0, + "step": 17303 + }, + { + "epoch": 1.9002855260267957, + "grad_norm": 2.2446935176849365, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7503941059112549, + "num_tokens": 447430713.0, + "step": 17304 + }, + { + "epoch": 1.9003953437294092, + "grad_norm": 2.3744451999664307, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7287220358848572, + "num_tokens": 447455069.0, + "step": 17305 + }, + { + "epoch": 1.9005051614320227, + "grad_norm": 2.173218250274658, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7131326198577881, + "num_tokens": 447483421.0, + "step": 17306 + }, + { + "epoch": 1.9006149791346365, + "grad_norm": 2.047150135040283, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7093358039855957, + "num_tokens": 447511348.0, + "step": 17307 + }, + { + "epoch": 1.9007247968372503, + "grad_norm": 2.4052011966705322, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7346876859664917, + "num_tokens": 447533441.0, + "step": 17308 + }, + { + "epoch": 1.9008346145398638, + "grad_norm": 2.1813700199127197, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.723800003528595, + "num_tokens": 447558387.0, + "step": 17309 + }, + { + "epoch": 1.9009444322424773, + "grad_norm": 1.925612449645996, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7038509845733643, + "num_tokens": 447592242.0, + "step": 17310 + }, + { + "epoch": 1.901054249945091, + "grad_norm": 2.3688266277313232, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7052590847015381, + "num_tokens": 447615537.0, + "step": 17311 + }, + { + "epoch": 1.9011640676477048, + "grad_norm": 2.063685894012451, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7255934476852417, + "num_tokens": 447645715.0, + "step": 17312 + }, + { + "epoch": 1.9012738853503186, + "grad_norm": 2.0967180728912354, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7318373918533325, + "num_tokens": 447674361.0, + "step": 17313 + }, + { + "epoch": 1.9013837030529321, + "grad_norm": 2.1451025009155273, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7111419439315796, + "num_tokens": 447700851.0, + "step": 17314 + }, + { + "epoch": 1.9014935207555457, + "grad_norm": 2.0559306144714355, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7194375991821289, + "num_tokens": 447729638.0, + "step": 17315 + }, + { + "epoch": 1.9016033384581594, + "grad_norm": 1.9826021194458008, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7054346799850464, + "num_tokens": 447760890.0, + "step": 17316 + }, + { + "epoch": 1.9017131561607732, + "grad_norm": 2.061476945877075, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7447738647460938, + "num_tokens": 447785658.0, + "step": 17317 + }, + { + "epoch": 1.9018229738633867, + "grad_norm": 2.274059534072876, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7085310220718384, + "num_tokens": 447811159.0, + "step": 17318 + }, + { + "epoch": 1.9019327915660005, + "grad_norm": 2.118701934814453, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7314608693122864, + "num_tokens": 447838044.0, + "step": 17319 + }, + { + "epoch": 1.902042609268614, + "grad_norm": 2.268540620803833, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7312124967575073, + "num_tokens": 447865087.0, + "step": 17320 + }, + { + "epoch": 1.9021524269712278, + "grad_norm": 2.191845655441284, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7142958641052246, + "num_tokens": 447890684.0, + "step": 17321 + }, + { + "epoch": 1.9022622446738415, + "grad_norm": 2.2830333709716797, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7259253859519958, + "num_tokens": 447915223.0, + "step": 17322 + }, + { + "epoch": 1.902372062376455, + "grad_norm": 2.1510298252105713, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7095251083374023, + "num_tokens": 447943562.0, + "step": 17323 + }, + { + "epoch": 1.9024818800790686, + "grad_norm": 2.4210762977600098, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7277559041976929, + "num_tokens": 447967291.0, + "step": 17324 + }, + { + "epoch": 1.9025916977816824, + "grad_norm": 2.219947099685669, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7105647921562195, + "num_tokens": 447993169.0, + "step": 17325 + }, + { + "epoch": 1.9027015154842961, + "grad_norm": 2.0734646320343018, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7082562446594238, + "num_tokens": 448021760.0, + "step": 17326 + }, + { + "epoch": 1.9028113331869099, + "grad_norm": 1.9633028507232666, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.6981832981109619, + "num_tokens": 448054045.0, + "step": 17327 + }, + { + "epoch": 1.9029211508895234, + "grad_norm": 2.0175600051879883, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7220975160598755, + "num_tokens": 448083280.0, + "step": 17328 + }, + { + "epoch": 1.903030968592137, + "grad_norm": 2.1561737060546875, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7205675840377808, + "num_tokens": 448110152.0, + "step": 17329 + }, + { + "epoch": 1.9031407862947507, + "grad_norm": 2.0218420028686523, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6984981894493103, + "num_tokens": 448141465.0, + "step": 17330 + }, + { + "epoch": 1.9032506039973645, + "grad_norm": 2.1635022163391113, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.73838210105896, + "num_tokens": 448165574.0, + "step": 17331 + }, + { + "epoch": 1.903360421699978, + "grad_norm": 2.060753345489502, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7016230225563049, + "num_tokens": 448193919.0, + "step": 17332 + }, + { + "epoch": 1.9034702394025917, + "grad_norm": 2.1360416412353516, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7300967574119568, + "num_tokens": 448219763.0, + "step": 17333 + }, + { + "epoch": 1.9035800571052053, + "grad_norm": 2.0289266109466553, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.69411301612854, + "num_tokens": 448251785.0, + "step": 17334 + }, + { + "epoch": 1.903689874807819, + "grad_norm": 2.1888697147369385, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7223449349403381, + "num_tokens": 448278044.0, + "step": 17335 + }, + { + "epoch": 1.9037996925104328, + "grad_norm": 2.189121961593628, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7300189733505249, + "num_tokens": 448303866.0, + "step": 17336 + }, + { + "epoch": 1.9039095102130463, + "grad_norm": 1.9896550178527832, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7092885971069336, + "num_tokens": 448334191.0, + "step": 17337 + }, + { + "epoch": 1.9040193279156599, + "grad_norm": 2.2766125202178955, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7152989506721497, + "num_tokens": 448361400.0, + "step": 17338 + }, + { + "epoch": 1.9041291456182736, + "grad_norm": 2.638709545135498, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7116416692733765, + "num_tokens": 448381714.0, + "step": 17339 + }, + { + "epoch": 1.9042389633208874, + "grad_norm": 2.2056500911712646, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7151999473571777, + "num_tokens": 448407440.0, + "step": 17340 + }, + { + "epoch": 1.9043487810235011, + "grad_norm": 2.471172332763672, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7246243953704834, + "num_tokens": 448429954.0, + "step": 17341 + }, + { + "epoch": 1.9044585987261147, + "grad_norm": 2.2736995220184326, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7374876737594604, + "num_tokens": 448455485.0, + "step": 17342 + }, + { + "epoch": 1.9045684164287282, + "grad_norm": 2.3885865211486816, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7443387508392334, + "num_tokens": 448480074.0, + "step": 17343 + }, + { + "epoch": 1.904678234131342, + "grad_norm": 2.1187074184417725, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7389171123504639, + "num_tokens": 448507473.0, + "step": 17344 + }, + { + "epoch": 1.9047880518339557, + "grad_norm": 2.4290661811828613, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7189823985099792, + "num_tokens": 448529085.0, + "step": 17345 + }, + { + "epoch": 1.9048978695365693, + "grad_norm": 2.361887216567993, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7205357551574707, + "num_tokens": 448553224.0, + "step": 17346 + }, + { + "epoch": 1.9050076872391828, + "grad_norm": 2.3553519248962402, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.70411217212677, + "num_tokens": 448577920.0, + "step": 17347 + }, + { + "epoch": 1.9051175049417965, + "grad_norm": 2.1656012535095215, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7265235185623169, + "num_tokens": 448604254.0, + "step": 17348 + }, + { + "epoch": 1.9052273226444103, + "grad_norm": 2.4050681591033936, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7205495238304138, + "num_tokens": 448626666.0, + "step": 17349 + }, + { + "epoch": 1.905337140347024, + "grad_norm": 1.8742045164108276, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7096636295318604, + "num_tokens": 448661643.0, + "step": 17350 + }, + { + "epoch": 1.9054469580496376, + "grad_norm": 2.1982879638671875, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7147519588470459, + "num_tokens": 448688672.0, + "step": 17351 + }, + { + "epoch": 1.9055567757522511, + "grad_norm": 2.8677616119384766, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7251814603805542, + "num_tokens": 448707192.0, + "step": 17352 + }, + { + "epoch": 1.9056665934548649, + "grad_norm": 2.3129327297210693, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7026395797729492, + "num_tokens": 448733459.0, + "step": 17353 + }, + { + "epoch": 1.9057764111574786, + "grad_norm": 2.142033815383911, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7320284247398376, + "num_tokens": 448758398.0, + "step": 17354 + }, + { + "epoch": 1.9058862288600924, + "grad_norm": 2.452726364135742, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7188330888748169, + "num_tokens": 448779754.0, + "step": 17355 + }, + { + "epoch": 1.905996046562706, + "grad_norm": 2.39969801902771, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7104219198226929, + "num_tokens": 448804077.0, + "step": 17356 + }, + { + "epoch": 1.9061058642653195, + "grad_norm": 2.5911917686462402, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.727257251739502, + "num_tokens": 448825247.0, + "step": 17357 + }, + { + "epoch": 1.9062156819679332, + "grad_norm": 2.1675665378570557, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7264000177383423, + "num_tokens": 448850877.0, + "step": 17358 + }, + { + "epoch": 1.906325499670547, + "grad_norm": 2.3426356315612793, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7097970247268677, + "num_tokens": 448873759.0, + "step": 17359 + }, + { + "epoch": 1.9064353173731605, + "grad_norm": 2.0833466053009033, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7023653984069824, + "num_tokens": 448902244.0, + "step": 17360 + }, + { + "epoch": 1.906545135075774, + "grad_norm": 1.8850363492965698, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7220478653907776, + "num_tokens": 448937166.0, + "step": 17361 + }, + { + "epoch": 1.9066549527783878, + "grad_norm": 2.294694662094116, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7196491956710815, + "num_tokens": 448961982.0, + "step": 17362 + }, + { + "epoch": 1.9067647704810016, + "grad_norm": 2.1165573596954346, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7080932855606079, + "num_tokens": 448993537.0, + "step": 17363 + }, + { + "epoch": 1.9068745881836153, + "grad_norm": 2.190164089202881, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7236175537109375, + "num_tokens": 449020416.0, + "step": 17364 + }, + { + "epoch": 1.9069844058862289, + "grad_norm": 2.185495138168335, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7381270527839661, + "num_tokens": 449046315.0, + "step": 17365 + }, + { + "epoch": 1.9070942235888424, + "grad_norm": 2.276343822479248, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7058851718902588, + "num_tokens": 449072172.0, + "step": 17366 + }, + { + "epoch": 1.9072040412914562, + "grad_norm": 2.4230740070343018, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7287805080413818, + "num_tokens": 449094238.0, + "step": 17367 + }, + { + "epoch": 1.90731385899407, + "grad_norm": 2.319685220718384, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.719973623752594, + "num_tokens": 449118465.0, + "step": 17368 + }, + { + "epoch": 1.9074236766966837, + "grad_norm": 2.406346559524536, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7082687020301819, + "num_tokens": 449140523.0, + "step": 17369 + }, + { + "epoch": 1.9075334943992972, + "grad_norm": 2.3292605876922607, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7103695869445801, + "num_tokens": 449167458.0, + "step": 17370 + }, + { + "epoch": 1.9076433121019107, + "grad_norm": 2.558854341506958, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7316104769706726, + "num_tokens": 449187811.0, + "step": 17371 + }, + { + "epoch": 1.9077531298045245, + "grad_norm": 2.223780870437622, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7397134900093079, + "num_tokens": 449212999.0, + "step": 17372 + }, + { + "epoch": 1.9078629475071383, + "grad_norm": 2.249845027923584, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7008066177368164, + "num_tokens": 449238024.0, + "step": 17373 + }, + { + "epoch": 1.9079727652097518, + "grad_norm": 2.0457825660705566, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7186484336853027, + "num_tokens": 449266862.0, + "step": 17374 + }, + { + "epoch": 1.9080825829123653, + "grad_norm": 2.0410571098327637, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7351837158203125, + "num_tokens": 449296136.0, + "step": 17375 + }, + { + "epoch": 1.908192400614979, + "grad_norm": 2.2181222438812256, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7062668204307556, + "num_tokens": 449322698.0, + "step": 17376 + }, + { + "epoch": 1.9083022183175928, + "grad_norm": 2.1229889392852783, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.6989525556564331, + "num_tokens": 449351229.0, + "step": 17377 + }, + { + "epoch": 1.9084120360202066, + "grad_norm": 2.330153465270996, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7344605922698975, + "num_tokens": 449374457.0, + "step": 17378 + }, + { + "epoch": 1.9085218537228201, + "grad_norm": 2.150834798812866, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7179412841796875, + "num_tokens": 449401738.0, + "step": 17379 + }, + { + "epoch": 1.9086316714254337, + "grad_norm": 2.122573137283325, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7155981063842773, + "num_tokens": 449428375.0, + "step": 17380 + }, + { + "epoch": 1.9087414891280474, + "grad_norm": 2.062260389328003, + "learning_rate": 1e-06, + "loss": 1.0576, + "mean_token_accuracy": 0.6848207712173462, + "num_tokens": 449460367.0, + "step": 17381 + }, + { + "epoch": 1.9088513068306612, + "grad_norm": 2.4106383323669434, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7366237640380859, + "num_tokens": 449481939.0, + "step": 17382 + }, + { + "epoch": 1.9089611245332747, + "grad_norm": 2.1863503456115723, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7162681818008423, + "num_tokens": 449507779.0, + "step": 17383 + }, + { + "epoch": 1.9090709422358885, + "grad_norm": 2.244488477706909, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.727520227432251, + "num_tokens": 449536049.0, + "step": 17384 + }, + { + "epoch": 1.909180759938502, + "grad_norm": 2.5491011142730713, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7372865676879883, + "num_tokens": 449556604.0, + "step": 17385 + }, + { + "epoch": 1.9092905776411158, + "grad_norm": 2.0750572681427, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7246237993240356, + "num_tokens": 449583690.0, + "step": 17386 + }, + { + "epoch": 1.9094003953437295, + "grad_norm": 2.192517042160034, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7353091835975647, + "num_tokens": 449609306.0, + "step": 17387 + }, + { + "epoch": 1.909510213046343, + "grad_norm": 2.0983493328094482, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7248268127441406, + "num_tokens": 449637102.0, + "step": 17388 + }, + { + "epoch": 1.9096200307489566, + "grad_norm": 2.288731098175049, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7346929311752319, + "num_tokens": 449659902.0, + "step": 17389 + }, + { + "epoch": 1.9097298484515703, + "grad_norm": 2.2351059913635254, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7247766256332397, + "num_tokens": 449685570.0, + "step": 17390 + }, + { + "epoch": 1.909839666154184, + "grad_norm": 2.204967498779297, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7283709049224854, + "num_tokens": 449709286.0, + "step": 17391 + }, + { + "epoch": 1.9099494838567979, + "grad_norm": 2.200387477874756, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7151718735694885, + "num_tokens": 449735615.0, + "step": 17392 + }, + { + "epoch": 1.9100593015594114, + "grad_norm": 2.132976531982422, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7142307758331299, + "num_tokens": 449762331.0, + "step": 17393 + }, + { + "epoch": 1.910169119262025, + "grad_norm": 1.9844826459884644, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7136993408203125, + "num_tokens": 449795113.0, + "step": 17394 + }, + { + "epoch": 1.9102789369646387, + "grad_norm": 2.2825369834899902, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7302430868148804, + "num_tokens": 449821302.0, + "step": 17395 + }, + { + "epoch": 1.9103887546672524, + "grad_norm": 2.2028615474700928, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7140424847602844, + "num_tokens": 449847719.0, + "step": 17396 + }, + { + "epoch": 1.910498572369866, + "grad_norm": 2.1897079944610596, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7014427781105042, + "num_tokens": 449874132.0, + "step": 17397 + }, + { + "epoch": 1.9106083900724797, + "grad_norm": 2.6558754444122314, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7312891483306885, + "num_tokens": 449892049.0, + "step": 17398 + }, + { + "epoch": 1.9107182077750933, + "grad_norm": 2.123675584793091, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7174186706542969, + "num_tokens": 449918827.0, + "step": 17399 + }, + { + "epoch": 1.910828025477707, + "grad_norm": 2.5299766063690186, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7147663235664368, + "num_tokens": 449940306.0, + "step": 17400 + }, + { + "epoch": 1.9109378431803208, + "grad_norm": 2.0688822269439697, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7065550088882446, + "num_tokens": 449968701.0, + "step": 17401 + }, + { + "epoch": 1.9110476608829343, + "grad_norm": 2.3004038333892822, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.6997077465057373, + "num_tokens": 449994837.0, + "step": 17402 + }, + { + "epoch": 1.9111574785855479, + "grad_norm": 2.2348837852478027, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7077292203903198, + "num_tokens": 450021966.0, + "step": 17403 + }, + { + "epoch": 1.9112672962881616, + "grad_norm": 2.122487783432007, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.706474781036377, + "num_tokens": 450050687.0, + "step": 17404 + }, + { + "epoch": 1.9113771139907754, + "grad_norm": 2.240964651107788, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7224105596542358, + "num_tokens": 450076238.0, + "step": 17405 + }, + { + "epoch": 1.9114869316933891, + "grad_norm": 2.491520881652832, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7210843563079834, + "num_tokens": 450096935.0, + "step": 17406 + }, + { + "epoch": 1.9115967493960027, + "grad_norm": 2.352322578430176, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7212738990783691, + "num_tokens": 450120989.0, + "step": 17407 + }, + { + "epoch": 1.9117065670986162, + "grad_norm": 2.225743055343628, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7259466052055359, + "num_tokens": 450148570.0, + "step": 17408 + }, + { + "epoch": 1.91181638480123, + "grad_norm": 1.848976969718933, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.758751392364502, + "num_tokens": 450180517.0, + "step": 17409 + }, + { + "epoch": 1.9119262025038437, + "grad_norm": 2.37566876411438, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7347838282585144, + "num_tokens": 450204352.0, + "step": 17410 + }, + { + "epoch": 1.9120360202064572, + "grad_norm": 2.542711019515991, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7201778292655945, + "num_tokens": 450223829.0, + "step": 17411 + }, + { + "epoch": 1.9121458379090708, + "grad_norm": 2.303662061691284, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7138263583183289, + "num_tokens": 450249659.0, + "step": 17412 + }, + { + "epoch": 1.9122556556116845, + "grad_norm": 2.3711366653442383, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7409694790840149, + "num_tokens": 450271616.0, + "step": 17413 + }, + { + "epoch": 1.9123654733142983, + "grad_norm": 2.348165512084961, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.741196870803833, + "num_tokens": 450293969.0, + "step": 17414 + }, + { + "epoch": 1.912475291016912, + "grad_norm": 2.4276909828186035, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7285038232803345, + "num_tokens": 450317278.0, + "step": 17415 + }, + { + "epoch": 1.9125851087195256, + "grad_norm": 2.3470911979675293, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7097508907318115, + "num_tokens": 450343184.0, + "step": 17416 + }, + { + "epoch": 1.9126949264221391, + "grad_norm": 2.6852657794952393, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7482138872146606, + "num_tokens": 450360937.0, + "step": 17417 + }, + { + "epoch": 1.9128047441247529, + "grad_norm": 2.5019822120666504, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7197549939155579, + "num_tokens": 450384911.0, + "step": 17418 + }, + { + "epoch": 1.9129145618273666, + "grad_norm": 2.807067394256592, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7472873330116272, + "num_tokens": 450403271.0, + "step": 17419 + }, + { + "epoch": 1.9130243795299804, + "grad_norm": 2.366739273071289, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7422713041305542, + "num_tokens": 450425048.0, + "step": 17420 + }, + { + "epoch": 1.913134197232594, + "grad_norm": 2.1683712005615234, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7571635246276855, + "num_tokens": 450449798.0, + "step": 17421 + }, + { + "epoch": 1.9132440149352075, + "grad_norm": 2.454324245452881, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7259392738342285, + "num_tokens": 450471716.0, + "step": 17422 + }, + { + "epoch": 1.9133538326378212, + "grad_norm": 2.3214643001556396, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6940670013427734, + "num_tokens": 450499535.0, + "step": 17423 + }, + { + "epoch": 1.913463650340435, + "grad_norm": 2.246310234069824, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7236292362213135, + "num_tokens": 450523739.0, + "step": 17424 + }, + { + "epoch": 1.9135734680430485, + "grad_norm": 2.125232219696045, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6797724962234497, + "num_tokens": 450556393.0, + "step": 17425 + }, + { + "epoch": 1.913683285745662, + "grad_norm": 2.416086435317993, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7299081087112427, + "num_tokens": 450577971.0, + "step": 17426 + }, + { + "epoch": 1.9137931034482758, + "grad_norm": 2.007504940032959, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.711513102054596, + "num_tokens": 450609518.0, + "step": 17427 + }, + { + "epoch": 1.9139029211508896, + "grad_norm": 2.5630033016204834, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7364250421524048, + "num_tokens": 450629031.0, + "step": 17428 + }, + { + "epoch": 1.9140127388535033, + "grad_norm": 2.1243441104888916, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7256096601486206, + "num_tokens": 450656033.0, + "step": 17429 + }, + { + "epoch": 1.9141225565561168, + "grad_norm": 2.244205951690674, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.712999701499939, + "num_tokens": 450681630.0, + "step": 17430 + }, + { + "epoch": 1.9142323742587304, + "grad_norm": 2.469109535217285, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7507545948028564, + "num_tokens": 450701588.0, + "step": 17431 + }, + { + "epoch": 1.9143421919613441, + "grad_norm": 2.3356001377105713, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.731452226638794, + "num_tokens": 450724736.0, + "step": 17432 + }, + { + "epoch": 1.914452009663958, + "grad_norm": 2.3243298530578613, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7265722155570984, + "num_tokens": 450748266.0, + "step": 17433 + }, + { + "epoch": 1.9145618273665717, + "grad_norm": 2.0752460956573486, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7250484228134155, + "num_tokens": 450777544.0, + "step": 17434 + }, + { + "epoch": 1.9146716450691852, + "grad_norm": 2.2587969303131104, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.72613525390625, + "num_tokens": 450802881.0, + "step": 17435 + }, + { + "epoch": 1.9147814627717987, + "grad_norm": 2.1406240463256836, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7083186507225037, + "num_tokens": 450832711.0, + "step": 17436 + }, + { + "epoch": 1.9148912804744125, + "grad_norm": 2.089233160018921, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7103756666183472, + "num_tokens": 450862156.0, + "step": 17437 + }, + { + "epoch": 1.9150010981770262, + "grad_norm": 2.1647775173187256, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7028473019599915, + "num_tokens": 450887857.0, + "step": 17438 + }, + { + "epoch": 1.9151109158796398, + "grad_norm": 2.105926513671875, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7168487310409546, + "num_tokens": 450916891.0, + "step": 17439 + }, + { + "epoch": 1.9152207335822533, + "grad_norm": 2.196495294570923, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7021936178207397, + "num_tokens": 450943925.0, + "step": 17440 + }, + { + "epoch": 1.915330551284867, + "grad_norm": 2.409799337387085, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7492766380310059, + "num_tokens": 450964413.0, + "step": 17441 + }, + { + "epoch": 1.9154403689874808, + "grad_norm": 2.2699978351593018, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7213699817657471, + "num_tokens": 450989912.0, + "step": 17442 + }, + { + "epoch": 1.9155501866900946, + "grad_norm": 2.3527345657348633, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7148926854133606, + "num_tokens": 451014695.0, + "step": 17443 + }, + { + "epoch": 1.9156600043927081, + "grad_norm": 2.501472234725952, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7077620029449463, + "num_tokens": 451035290.0, + "step": 17444 + }, + { + "epoch": 1.9157698220953217, + "grad_norm": 2.5280675888061523, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.729245662689209, + "num_tokens": 451057017.0, + "step": 17445 + }, + { + "epoch": 1.9158796397979354, + "grad_norm": 2.282323122024536, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7308081388473511, + "num_tokens": 451081191.0, + "step": 17446 + }, + { + "epoch": 1.9159894575005492, + "grad_norm": 2.1825389862060547, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7290080785751343, + "num_tokens": 451107356.0, + "step": 17447 + }, + { + "epoch": 1.9160992752031627, + "grad_norm": 1.925093412399292, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7081854343414307, + "num_tokens": 451142762.0, + "step": 17448 + }, + { + "epoch": 1.9162090929057765, + "grad_norm": 2.1396515369415283, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7254847288131714, + "num_tokens": 451170603.0, + "step": 17449 + }, + { + "epoch": 1.91631891060839, + "grad_norm": 2.1377832889556885, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7073051929473877, + "num_tokens": 451200087.0, + "step": 17450 + }, + { + "epoch": 1.9164287283110037, + "grad_norm": 2.334137439727783, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7100178599357605, + "num_tokens": 451223252.0, + "step": 17451 + }, + { + "epoch": 1.9165385460136175, + "grad_norm": 2.3763043880462646, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7139555215835571, + "num_tokens": 451245659.0, + "step": 17452 + }, + { + "epoch": 1.916648363716231, + "grad_norm": 2.3837978839874268, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7367405891418457, + "num_tokens": 451267037.0, + "step": 17453 + }, + { + "epoch": 1.9167581814188446, + "grad_norm": 2.54054594039917, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7265626192092896, + "num_tokens": 451288055.0, + "step": 17454 + }, + { + "epoch": 1.9168679991214583, + "grad_norm": 2.582531213760376, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7083480954170227, + "num_tokens": 451309635.0, + "step": 17455 + }, + { + "epoch": 1.916977816824072, + "grad_norm": 2.0519607067108154, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7204185724258423, + "num_tokens": 451341564.0, + "step": 17456 + }, + { + "epoch": 1.9170876345266858, + "grad_norm": 2.101057291030884, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6950279474258423, + "num_tokens": 451369371.0, + "step": 17457 + }, + { + "epoch": 1.9171974522292994, + "grad_norm": 2.258429527282715, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7282407283782959, + "num_tokens": 451393862.0, + "step": 17458 + }, + { + "epoch": 1.917307269931913, + "grad_norm": 2.208038330078125, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7163589000701904, + "num_tokens": 451421284.0, + "step": 17459 + }, + { + "epoch": 1.9174170876345267, + "grad_norm": 2.3814592361450195, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.709994912147522, + "num_tokens": 451445744.0, + "step": 17460 + }, + { + "epoch": 1.9175269053371404, + "grad_norm": 2.2736763954162598, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7257762551307678, + "num_tokens": 451469195.0, + "step": 17461 + }, + { + "epoch": 1.917636723039754, + "grad_norm": 2.365898609161377, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7207137942314148, + "num_tokens": 451492045.0, + "step": 17462 + }, + { + "epoch": 1.9177465407423677, + "grad_norm": 2.46755313873291, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7204346656799316, + "num_tokens": 451514767.0, + "step": 17463 + }, + { + "epoch": 1.9178563584449813, + "grad_norm": 2.189262866973877, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7247305512428284, + "num_tokens": 451541390.0, + "step": 17464 + }, + { + "epoch": 1.917966176147595, + "grad_norm": 2.250483512878418, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7252829074859619, + "num_tokens": 451565160.0, + "step": 17465 + }, + { + "epoch": 1.9180759938502088, + "grad_norm": 2.4401562213897705, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7088719606399536, + "num_tokens": 451587544.0, + "step": 17466 + }, + { + "epoch": 1.9181858115528223, + "grad_norm": 2.312713384628296, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7453891038894653, + "num_tokens": 451612067.0, + "step": 17467 + }, + { + "epoch": 1.9182956292554358, + "grad_norm": 2.1515512466430664, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7355731725692749, + "num_tokens": 451638093.0, + "step": 17468 + }, + { + "epoch": 1.9184054469580496, + "grad_norm": 2.134749412536621, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7224855422973633, + "num_tokens": 451665022.0, + "step": 17469 + }, + { + "epoch": 1.9185152646606634, + "grad_norm": 2.223219156265259, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7092068195343018, + "num_tokens": 451692246.0, + "step": 17470 + }, + { + "epoch": 1.9186250823632771, + "grad_norm": 2.336442232131958, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7465714812278748, + "num_tokens": 451715934.0, + "step": 17471 + }, + { + "epoch": 1.9187349000658906, + "grad_norm": 1.9433549642562866, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6942788362503052, + "num_tokens": 451751060.0, + "step": 17472 + }, + { + "epoch": 1.9188447177685042, + "grad_norm": 1.9652961492538452, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7111595273017883, + "num_tokens": 451782865.0, + "step": 17473 + }, + { + "epoch": 1.918954535471118, + "grad_norm": 2.05123233795166, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7265563607215881, + "num_tokens": 451812176.0, + "step": 17474 + }, + { + "epoch": 1.9190643531737317, + "grad_norm": 2.2354018688201904, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7407188415527344, + "num_tokens": 451835676.0, + "step": 17475 + }, + { + "epoch": 1.9191741708763452, + "grad_norm": 2.05961012840271, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7158565521240234, + "num_tokens": 451864951.0, + "step": 17476 + }, + { + "epoch": 1.9192839885789588, + "grad_norm": 2.2798664569854736, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7194480895996094, + "num_tokens": 451888418.0, + "step": 17477 + }, + { + "epoch": 1.9193938062815725, + "grad_norm": 2.0669355392456055, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7028013467788696, + "num_tokens": 451917245.0, + "step": 17478 + }, + { + "epoch": 1.9195036239841863, + "grad_norm": 2.2699825763702393, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7516298294067383, + "num_tokens": 451940714.0, + "step": 17479 + }, + { + "epoch": 1.9196134416868, + "grad_norm": 2.103024959564209, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7111363410949707, + "num_tokens": 451971952.0, + "step": 17480 + }, + { + "epoch": 1.9197232593894136, + "grad_norm": 2.1244983673095703, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7032262086868286, + "num_tokens": 452002024.0, + "step": 17481 + }, + { + "epoch": 1.919833077092027, + "grad_norm": 2.193464756011963, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7208784222602844, + "num_tokens": 452029407.0, + "step": 17482 + }, + { + "epoch": 1.9199428947946409, + "grad_norm": 2.086059331893921, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7147266864776611, + "num_tokens": 452056850.0, + "step": 17483 + }, + { + "epoch": 1.9200527124972546, + "grad_norm": 2.125562906265259, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7164808511734009, + "num_tokens": 452084598.0, + "step": 17484 + }, + { + "epoch": 1.9201625301998684, + "grad_norm": 2.1111981868743896, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7292110919952393, + "num_tokens": 452112922.0, + "step": 17485 + }, + { + "epoch": 1.920272347902482, + "grad_norm": 2.040076494216919, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7096351385116577, + "num_tokens": 452142665.0, + "step": 17486 + }, + { + "epoch": 1.9203821656050954, + "grad_norm": 2.321214199066162, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7227077484130859, + "num_tokens": 452165606.0, + "step": 17487 + }, + { + "epoch": 1.9204919833077092, + "grad_norm": 2.552947998046875, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7469372749328613, + "num_tokens": 452184198.0, + "step": 17488 + }, + { + "epoch": 1.920601801010323, + "grad_norm": 2.4598913192749023, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7195839881896973, + "num_tokens": 452206770.0, + "step": 17489 + }, + { + "epoch": 1.9207116187129365, + "grad_norm": 2.2307746410369873, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7157223224639893, + "num_tokens": 452233802.0, + "step": 17490 + }, + { + "epoch": 1.92082143641555, + "grad_norm": 2.3883187770843506, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7232592105865479, + "num_tokens": 452257788.0, + "step": 17491 + }, + { + "epoch": 1.9209312541181638, + "grad_norm": 2.0658092498779297, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.714490532875061, + "num_tokens": 452285869.0, + "step": 17492 + }, + { + "epoch": 1.9210410718207775, + "grad_norm": 1.9648385047912598, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.726758599281311, + "num_tokens": 452317624.0, + "step": 17493 + }, + { + "epoch": 1.9211508895233913, + "grad_norm": 2.2784478664398193, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7243733406066895, + "num_tokens": 452341393.0, + "step": 17494 + }, + { + "epoch": 1.9212607072260048, + "grad_norm": 2.3172593116760254, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7211063504219055, + "num_tokens": 452366505.0, + "step": 17495 + }, + { + "epoch": 1.9213705249286184, + "grad_norm": 2.465036392211914, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7442154884338379, + "num_tokens": 452387653.0, + "step": 17496 + }, + { + "epoch": 1.9214803426312321, + "grad_norm": 2.341484308242798, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7391079664230347, + "num_tokens": 452410836.0, + "step": 17497 + }, + { + "epoch": 1.9215901603338459, + "grad_norm": 1.9797420501708984, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7146233320236206, + "num_tokens": 452441639.0, + "step": 17498 + }, + { + "epoch": 1.9216999780364594, + "grad_norm": 2.49131178855896, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.718408465385437, + "num_tokens": 452462803.0, + "step": 17499 + }, + { + "epoch": 1.9218097957390732, + "grad_norm": 2.313720703125, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7359394431114197, + "num_tokens": 452485567.0, + "step": 17500 + }, + { + "epoch": 1.9219196134416867, + "grad_norm": 2.2642199993133545, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7064287662506104, + "num_tokens": 452512177.0, + "step": 17501 + }, + { + "epoch": 1.9220294311443005, + "grad_norm": 2.056802272796631, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7260240912437439, + "num_tokens": 452540671.0, + "step": 17502 + }, + { + "epoch": 1.9221392488469142, + "grad_norm": 2.020998001098633, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7201974391937256, + "num_tokens": 452572954.0, + "step": 17503 + }, + { + "epoch": 1.9222490665495278, + "grad_norm": 2.459172487258911, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7167328596115112, + "num_tokens": 452595664.0, + "step": 17504 + }, + { + "epoch": 1.9223588842521413, + "grad_norm": 2.4800307750701904, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.73016357421875, + "num_tokens": 452616085.0, + "step": 17505 + }, + { + "epoch": 1.922468701954755, + "grad_norm": 2.4215762615203857, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7041453123092651, + "num_tokens": 452641245.0, + "step": 17506 + }, + { + "epoch": 1.9225785196573688, + "grad_norm": 2.277182102203369, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7118823528289795, + "num_tokens": 452666302.0, + "step": 17507 + }, + { + "epoch": 1.9226883373599826, + "grad_norm": 2.284240961074829, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7268389463424683, + "num_tokens": 452689351.0, + "step": 17508 + }, + { + "epoch": 1.922798155062596, + "grad_norm": 2.6012890338897705, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7300071716308594, + "num_tokens": 452709981.0, + "step": 17509 + }, + { + "epoch": 1.9229079727652096, + "grad_norm": 2.016165256500244, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7168080806732178, + "num_tokens": 452739967.0, + "step": 17510 + }, + { + "epoch": 1.9230177904678234, + "grad_norm": 2.240593671798706, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7252436876296997, + "num_tokens": 452766530.0, + "step": 17511 + }, + { + "epoch": 1.9231276081704372, + "grad_norm": 2.1220245361328125, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6973296403884888, + "num_tokens": 452795231.0, + "step": 17512 + }, + { + "epoch": 1.9232374258730507, + "grad_norm": 2.1211254596710205, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7171075344085693, + "num_tokens": 452823578.0, + "step": 17513 + }, + { + "epoch": 1.9233472435756644, + "grad_norm": 2.0087456703186035, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7105309963226318, + "num_tokens": 452855213.0, + "step": 17514 + }, + { + "epoch": 1.923457061278278, + "grad_norm": 2.216247320175171, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6917546987533569, + "num_tokens": 452885761.0, + "step": 17515 + }, + { + "epoch": 1.9235668789808917, + "grad_norm": 2.273064136505127, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7343419790267944, + "num_tokens": 452909656.0, + "step": 17516 + }, + { + "epoch": 1.9236766966835055, + "grad_norm": 1.9394121170043945, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7224434614181519, + "num_tokens": 452945783.0, + "step": 17517 + }, + { + "epoch": 1.923786514386119, + "grad_norm": 2.3486125469207764, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7268143892288208, + "num_tokens": 452967176.0, + "step": 17518 + }, + { + "epoch": 1.9238963320887326, + "grad_norm": 2.354957103729248, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7014680504798889, + "num_tokens": 452991130.0, + "step": 17519 + }, + { + "epoch": 1.9240061497913463, + "grad_norm": 2.1534366607666016, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7327800989151001, + "num_tokens": 453017318.0, + "step": 17520 + }, + { + "epoch": 1.92411596749396, + "grad_norm": 2.349802255630493, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7045562267303467, + "num_tokens": 453042097.0, + "step": 17521 + }, + { + "epoch": 1.9242257851965738, + "grad_norm": 2.121633529663086, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.710412859916687, + "num_tokens": 453070639.0, + "step": 17522 + }, + { + "epoch": 1.9243356028991874, + "grad_norm": 2.343604326248169, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7431668639183044, + "num_tokens": 453092593.0, + "step": 17523 + }, + { + "epoch": 1.924445420601801, + "grad_norm": 1.9793682098388672, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7437893152236938, + "num_tokens": 453121695.0, + "step": 17524 + }, + { + "epoch": 1.9245552383044147, + "grad_norm": 2.0697247982025146, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7080832123756409, + "num_tokens": 453150979.0, + "step": 17525 + }, + { + "epoch": 1.9246650560070284, + "grad_norm": 1.911435842514038, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7051881551742554, + "num_tokens": 453184250.0, + "step": 17526 + }, + { + "epoch": 1.924774873709642, + "grad_norm": 2.009688377380371, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6901472210884094, + "num_tokens": 453215976.0, + "step": 17527 + }, + { + "epoch": 1.9248846914122557, + "grad_norm": 2.3277246952056885, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7505539655685425, + "num_tokens": 453239722.0, + "step": 17528 + }, + { + "epoch": 1.9249945091148692, + "grad_norm": 2.1380813121795654, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7042555809020996, + "num_tokens": 453267887.0, + "step": 17529 + }, + { + "epoch": 1.925104326817483, + "grad_norm": 2.0699429512023926, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7097062468528748, + "num_tokens": 453297401.0, + "step": 17530 + }, + { + "epoch": 1.9252141445200968, + "grad_norm": 1.9474248886108398, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7229402661323547, + "num_tokens": 453328276.0, + "step": 17531 + }, + { + "epoch": 1.9253239622227103, + "grad_norm": 2.0142099857330322, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7029174566268921, + "num_tokens": 453361354.0, + "step": 17532 + }, + { + "epoch": 1.9254337799253238, + "grad_norm": 2.2576770782470703, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7240450382232666, + "num_tokens": 453385604.0, + "step": 17533 + }, + { + "epoch": 1.9255435976279376, + "grad_norm": 3.426253318786621, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7293702363967896, + "num_tokens": 453411241.0, + "step": 17534 + }, + { + "epoch": 1.9256534153305513, + "grad_norm": 2.3928022384643555, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7527113556861877, + "num_tokens": 453432239.0, + "step": 17535 + }, + { + "epoch": 1.925763233033165, + "grad_norm": 2.0530850887298584, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7176432609558105, + "num_tokens": 453461818.0, + "step": 17536 + }, + { + "epoch": 1.9258730507357786, + "grad_norm": 2.631348133087158, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7305785417556763, + "num_tokens": 453482038.0, + "step": 17537 + }, + { + "epoch": 1.9259828684383922, + "grad_norm": 2.124793767929077, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7294108867645264, + "num_tokens": 453508784.0, + "step": 17538 + }, + { + "epoch": 1.926092686141006, + "grad_norm": 2.02669095993042, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.695880651473999, + "num_tokens": 453538687.0, + "step": 17539 + }, + { + "epoch": 1.9262025038436197, + "grad_norm": 2.2034428119659424, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.6997049450874329, + "num_tokens": 453564911.0, + "step": 17540 + }, + { + "epoch": 1.9263123215462332, + "grad_norm": 1.944844126701355, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6959265470504761, + "num_tokens": 453599779.0, + "step": 17541 + }, + { + "epoch": 1.9264221392488468, + "grad_norm": 2.244907855987549, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7115249633789062, + "num_tokens": 453624549.0, + "step": 17542 + }, + { + "epoch": 1.9265319569514605, + "grad_norm": 2.323208808898926, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7208227515220642, + "num_tokens": 453648337.0, + "step": 17543 + }, + { + "epoch": 1.9266417746540743, + "grad_norm": 2.166351079940796, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7533069252967834, + "num_tokens": 453675954.0, + "step": 17544 + }, + { + "epoch": 1.926751592356688, + "grad_norm": 2.3600246906280518, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.6989980936050415, + "num_tokens": 453702509.0, + "step": 17545 + }, + { + "epoch": 1.9268614100593016, + "grad_norm": 2.0647051334381104, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.738456666469574, + "num_tokens": 453730629.0, + "step": 17546 + }, + { + "epoch": 1.926971227761915, + "grad_norm": 2.4444546699523926, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7291476130485535, + "num_tokens": 453752699.0, + "step": 17547 + }, + { + "epoch": 1.9270810454645289, + "grad_norm": 2.315575122833252, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6941546201705933, + "num_tokens": 453779843.0, + "step": 17548 + }, + { + "epoch": 1.9271908631671426, + "grad_norm": 2.1594254970550537, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7263867855072021, + "num_tokens": 453805125.0, + "step": 17549 + }, + { + "epoch": 1.9273006808697564, + "grad_norm": 2.041231155395508, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7067588567733765, + "num_tokens": 453834983.0, + "step": 17550 + }, + { + "epoch": 1.92741049857237, + "grad_norm": 2.1869051456451416, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7319937944412231, + "num_tokens": 453861241.0, + "step": 17551 + }, + { + "epoch": 1.9275203162749834, + "grad_norm": 2.277254581451416, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7145644426345825, + "num_tokens": 453887612.0, + "step": 17552 + }, + { + "epoch": 1.9276301339775972, + "grad_norm": 2.4567723274230957, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7442050576210022, + "num_tokens": 453909501.0, + "step": 17553 + }, + { + "epoch": 1.927739951680211, + "grad_norm": 2.59186053276062, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7338497638702393, + "num_tokens": 453930561.0, + "step": 17554 + }, + { + "epoch": 1.9278497693828245, + "grad_norm": 2.0880849361419678, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.6936607360839844, + "num_tokens": 453960082.0, + "step": 17555 + }, + { + "epoch": 1.927959587085438, + "grad_norm": 1.8376089334487915, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7172834277153015, + "num_tokens": 453994622.0, + "step": 17556 + }, + { + "epoch": 1.9280694047880518, + "grad_norm": 2.2087643146514893, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.737766683101654, + "num_tokens": 454019594.0, + "step": 17557 + }, + { + "epoch": 1.9281792224906655, + "grad_norm": 2.373793363571167, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7495522499084473, + "num_tokens": 454042410.0, + "step": 17558 + }, + { + "epoch": 1.9282890401932793, + "grad_norm": 1.9484400749206543, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7301967144012451, + "num_tokens": 454074885.0, + "step": 17559 + }, + { + "epoch": 1.9283988578958928, + "grad_norm": 2.300513982772827, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7045453190803528, + "num_tokens": 454099407.0, + "step": 17560 + }, + { + "epoch": 1.9285086755985064, + "grad_norm": 2.1791133880615234, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7224191427230835, + "num_tokens": 454125813.0, + "step": 17561 + }, + { + "epoch": 1.9286184933011201, + "grad_norm": 2.0900819301605225, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7466921806335449, + "num_tokens": 454152954.0, + "step": 17562 + }, + { + "epoch": 1.9287283110037339, + "grad_norm": 2.2132210731506348, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7271406650543213, + "num_tokens": 454179351.0, + "step": 17563 + }, + { + "epoch": 1.9288381287063474, + "grad_norm": 2.167005777359009, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7154631018638611, + "num_tokens": 454205510.0, + "step": 17564 + }, + { + "epoch": 1.9289479464089612, + "grad_norm": 2.3622872829437256, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7334126234054565, + "num_tokens": 454228764.0, + "step": 17565 + }, + { + "epoch": 1.9290577641115747, + "grad_norm": 2.538982391357422, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7358686327934265, + "num_tokens": 454248196.0, + "step": 17566 + }, + { + "epoch": 1.9291675818141885, + "grad_norm": 2.360460042953491, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.734804093837738, + "num_tokens": 454269712.0, + "step": 17567 + }, + { + "epoch": 1.9292773995168022, + "grad_norm": 2.138115406036377, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7005757689476013, + "num_tokens": 454297607.0, + "step": 17568 + }, + { + "epoch": 1.9293872172194158, + "grad_norm": 2.412205934524536, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7222515940666199, + "num_tokens": 454321168.0, + "step": 17569 + }, + { + "epoch": 1.9294970349220293, + "grad_norm": 1.9655648469924927, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7192136645317078, + "num_tokens": 454352864.0, + "step": 17570 + }, + { + "epoch": 1.929606852624643, + "grad_norm": 2.221025228500366, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7337484359741211, + "num_tokens": 454377003.0, + "step": 17571 + }, + { + "epoch": 1.9297166703272568, + "grad_norm": 2.3030364513397217, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7097820043563843, + "num_tokens": 454403443.0, + "step": 17572 + }, + { + "epoch": 1.9298264880298706, + "grad_norm": 2.1750473976135254, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7261328101158142, + "num_tokens": 454428688.0, + "step": 17573 + }, + { + "epoch": 1.929936305732484, + "grad_norm": 2.4227113723754883, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7488559484481812, + "num_tokens": 454448632.0, + "step": 17574 + }, + { + "epoch": 1.9300461234350976, + "grad_norm": 2.0569088459014893, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7292965054512024, + "num_tokens": 454478079.0, + "step": 17575 + }, + { + "epoch": 1.9301559411377114, + "grad_norm": 2.2256603240966797, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7300925850868225, + "num_tokens": 454503560.0, + "step": 17576 + }, + { + "epoch": 1.9302657588403251, + "grad_norm": 2.1559176445007324, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.706598162651062, + "num_tokens": 454532357.0, + "step": 17577 + }, + { + "epoch": 1.9303755765429387, + "grad_norm": 2.0623104572296143, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7131619453430176, + "num_tokens": 454561608.0, + "step": 17578 + }, + { + "epoch": 1.9304853942455524, + "grad_norm": 2.1616411209106445, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7151998281478882, + "num_tokens": 454590535.0, + "step": 17579 + }, + { + "epoch": 1.930595211948166, + "grad_norm": 2.3665032386779785, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7412195205688477, + "num_tokens": 454613021.0, + "step": 17580 + }, + { + "epoch": 1.9307050296507797, + "grad_norm": 2.289177656173706, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7036945819854736, + "num_tokens": 454638183.0, + "step": 17581 + }, + { + "epoch": 1.9308148473533935, + "grad_norm": 2.1982626914978027, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7260434627532959, + "num_tokens": 454664254.0, + "step": 17582 + }, + { + "epoch": 1.930924665056007, + "grad_norm": 1.857326626777649, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6973872184753418, + "num_tokens": 454702713.0, + "step": 17583 + }, + { + "epoch": 1.9310344827586206, + "grad_norm": 2.116654872894287, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7146598696708679, + "num_tokens": 454729770.0, + "step": 17584 + }, + { + "epoch": 1.9311443004612343, + "grad_norm": 2.148574113845825, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7229433655738831, + "num_tokens": 454754170.0, + "step": 17585 + }, + { + "epoch": 1.931254118163848, + "grad_norm": 2.273083209991455, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7145777344703674, + "num_tokens": 454779132.0, + "step": 17586 + }, + { + "epoch": 1.9313639358664618, + "grad_norm": 2.2720940113067627, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7088661193847656, + "num_tokens": 454803807.0, + "step": 17587 + }, + { + "epoch": 1.9314737535690754, + "grad_norm": 2.224095344543457, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7029596567153931, + "num_tokens": 454831878.0, + "step": 17588 + }, + { + "epoch": 1.931583571271689, + "grad_norm": 2.1905176639556885, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7129281163215637, + "num_tokens": 454859770.0, + "step": 17589 + }, + { + "epoch": 1.9316933889743026, + "grad_norm": 1.638731837272644, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7077262997627258, + "num_tokens": 454905089.0, + "step": 17590 + }, + { + "epoch": 1.9318032066769164, + "grad_norm": 2.109973907470703, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7188329696655273, + "num_tokens": 454934153.0, + "step": 17591 + }, + { + "epoch": 1.93191302437953, + "grad_norm": 2.018695592880249, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6892281174659729, + "num_tokens": 454965708.0, + "step": 17592 + }, + { + "epoch": 1.9320228420821435, + "grad_norm": 2.1401259899139404, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.701611340045929, + "num_tokens": 454994316.0, + "step": 17593 + }, + { + "epoch": 1.9321326597847572, + "grad_norm": 2.399287462234497, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7348198890686035, + "num_tokens": 455016511.0, + "step": 17594 + }, + { + "epoch": 1.932242477487371, + "grad_norm": 2.4666919708251953, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7531956434249878, + "num_tokens": 455036822.0, + "step": 17595 + }, + { + "epoch": 1.9323522951899847, + "grad_norm": 2.1198112964630127, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.723487913608551, + "num_tokens": 455065624.0, + "step": 17596 + }, + { + "epoch": 1.9324621128925983, + "grad_norm": 2.2739205360412598, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7174346446990967, + "num_tokens": 455092259.0, + "step": 17597 + }, + { + "epoch": 1.9325719305952118, + "grad_norm": 2.513033628463745, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7220741510391235, + "num_tokens": 455114492.0, + "step": 17598 + }, + { + "epoch": 1.9326817482978256, + "grad_norm": 2.4106311798095703, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7382572293281555, + "num_tokens": 455135634.0, + "step": 17599 + }, + { + "epoch": 1.9327915660004393, + "grad_norm": 2.064488649368286, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7178784608840942, + "num_tokens": 455167182.0, + "step": 17600 + }, + { + "epoch": 1.932901383703053, + "grad_norm": 2.282886266708374, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7197971343994141, + "num_tokens": 455191465.0, + "step": 17601 + }, + { + "epoch": 1.9330112014056666, + "grad_norm": 1.9457048177719116, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7207546830177307, + "num_tokens": 455223966.0, + "step": 17602 + }, + { + "epoch": 1.9331210191082802, + "grad_norm": 2.0910162925720215, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7192672491073608, + "num_tokens": 455250006.0, + "step": 17603 + }, + { + "epoch": 1.933230836810894, + "grad_norm": 2.3632733821868896, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7141472101211548, + "num_tokens": 455273183.0, + "step": 17604 + }, + { + "epoch": 1.9333406545135077, + "grad_norm": 2.235088348388672, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7052476406097412, + "num_tokens": 455302408.0, + "step": 17605 + }, + { + "epoch": 1.9334504722161212, + "grad_norm": 2.1828842163085938, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.714898943901062, + "num_tokens": 455330213.0, + "step": 17606 + }, + { + "epoch": 1.9335602899187347, + "grad_norm": 2.431771993637085, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.6994935274124146, + "num_tokens": 455354541.0, + "step": 17607 + }, + { + "epoch": 1.9336701076213485, + "grad_norm": 2.1389307975769043, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7092083692550659, + "num_tokens": 455382122.0, + "step": 17608 + }, + { + "epoch": 1.9337799253239623, + "grad_norm": 2.29870343208313, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7124735713005066, + "num_tokens": 455408109.0, + "step": 17609 + }, + { + "epoch": 1.933889743026576, + "grad_norm": 2.4958534240722656, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.732171893119812, + "num_tokens": 455429011.0, + "step": 17610 + }, + { + "epoch": 1.9339995607291895, + "grad_norm": 2.2952523231506348, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7234317064285278, + "num_tokens": 455453465.0, + "step": 17611 + }, + { + "epoch": 1.934109378431803, + "grad_norm": 1.9850212335586548, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7133949995040894, + "num_tokens": 455484298.0, + "step": 17612 + }, + { + "epoch": 1.9342191961344168, + "grad_norm": 2.1887636184692383, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.706842303276062, + "num_tokens": 455510085.0, + "step": 17613 + }, + { + "epoch": 1.9343290138370306, + "grad_norm": 2.2727956771850586, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7050389051437378, + "num_tokens": 455534673.0, + "step": 17614 + }, + { + "epoch": 1.9344388315396444, + "grad_norm": 2.2767295837402344, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7154125571250916, + "num_tokens": 455559873.0, + "step": 17615 + }, + { + "epoch": 1.934548649242258, + "grad_norm": 2.0684659481048584, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7165191769599915, + "num_tokens": 455588007.0, + "step": 17616 + }, + { + "epoch": 1.9346584669448714, + "grad_norm": 2.2254581451416016, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7400544881820679, + "num_tokens": 455612967.0, + "step": 17617 + }, + { + "epoch": 1.9347682846474852, + "grad_norm": 2.169732093811035, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.715290904045105, + "num_tokens": 455641419.0, + "step": 17618 + }, + { + "epoch": 1.934878102350099, + "grad_norm": 2.3877172470092773, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7263607978820801, + "num_tokens": 455663530.0, + "step": 17619 + }, + { + "epoch": 1.9349879200527125, + "grad_norm": 2.136859893798828, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7220140695571899, + "num_tokens": 455690929.0, + "step": 17620 + }, + { + "epoch": 1.935097737755326, + "grad_norm": 2.2309298515319824, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7226679921150208, + "num_tokens": 455716597.0, + "step": 17621 + }, + { + "epoch": 1.9352075554579398, + "grad_norm": 2.1243059635162354, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7062549591064453, + "num_tokens": 455745358.0, + "step": 17622 + }, + { + "epoch": 1.9353173731605535, + "grad_norm": 2.0482687950134277, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.704119086265564, + "num_tokens": 455773826.0, + "step": 17623 + }, + { + "epoch": 1.9354271908631673, + "grad_norm": 2.0556368827819824, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.701478898525238, + "num_tokens": 455802065.0, + "step": 17624 + }, + { + "epoch": 1.9355370085657808, + "grad_norm": 2.007890224456787, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7168316841125488, + "num_tokens": 455831596.0, + "step": 17625 + }, + { + "epoch": 1.9356468262683943, + "grad_norm": 2.0224714279174805, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6988268494606018, + "num_tokens": 455860969.0, + "step": 17626 + }, + { + "epoch": 1.935756643971008, + "grad_norm": 2.3568637371063232, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7209842205047607, + "num_tokens": 455884957.0, + "step": 17627 + }, + { + "epoch": 1.9358664616736219, + "grad_norm": 2.1480982303619385, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.6991990804672241, + "num_tokens": 455912229.0, + "step": 17628 + }, + { + "epoch": 1.9359762793762354, + "grad_norm": 2.159472942352295, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6995754837989807, + "num_tokens": 455939141.0, + "step": 17629 + }, + { + "epoch": 1.9360860970788492, + "grad_norm": 2.0021417140960693, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7070620656013489, + "num_tokens": 455970042.0, + "step": 17630 + }, + { + "epoch": 1.9361959147814627, + "grad_norm": 2.4613282680511475, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7384223937988281, + "num_tokens": 455991574.0, + "step": 17631 + }, + { + "epoch": 1.9363057324840764, + "grad_norm": 2.353843927383423, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7275550961494446, + "num_tokens": 456014313.0, + "step": 17632 + }, + { + "epoch": 1.9364155501866902, + "grad_norm": 2.2500722408294678, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7342263460159302, + "num_tokens": 456038844.0, + "step": 17633 + }, + { + "epoch": 1.9365253678893037, + "grad_norm": 2.107337236404419, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7098575830459595, + "num_tokens": 456065470.0, + "step": 17634 + }, + { + "epoch": 1.9366351855919173, + "grad_norm": 2.0720505714416504, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7267213463783264, + "num_tokens": 456094501.0, + "step": 17635 + }, + { + "epoch": 1.936745003294531, + "grad_norm": 2.203735828399658, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7186565399169922, + "num_tokens": 456121203.0, + "step": 17636 + }, + { + "epoch": 1.9368548209971448, + "grad_norm": 2.1650969982147217, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7005449533462524, + "num_tokens": 456147979.0, + "step": 17637 + }, + { + "epoch": 1.9369646386997585, + "grad_norm": 2.305218458175659, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7482564449310303, + "num_tokens": 456169219.0, + "step": 17638 + }, + { + "epoch": 1.937074456402372, + "grad_norm": 2.113020658493042, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7100672125816345, + "num_tokens": 456197959.0, + "step": 17639 + }, + { + "epoch": 1.9371842741049856, + "grad_norm": 2.171626329421997, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7177016735076904, + "num_tokens": 456225101.0, + "step": 17640 + }, + { + "epoch": 1.9372940918075994, + "grad_norm": 2.260547399520874, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7042279243469238, + "num_tokens": 456251271.0, + "step": 17641 + }, + { + "epoch": 1.9374039095102131, + "grad_norm": 2.18290376663208, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7023757696151733, + "num_tokens": 456279573.0, + "step": 17642 + }, + { + "epoch": 1.9375137272128267, + "grad_norm": 2.2651524543762207, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7214863300323486, + "num_tokens": 456304106.0, + "step": 17643 + }, + { + "epoch": 1.9376235449154404, + "grad_norm": 2.453096628189087, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7146234512329102, + "num_tokens": 456326000.0, + "step": 17644 + }, + { + "epoch": 1.937733362618054, + "grad_norm": 2.06177020072937, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7032318711280823, + "num_tokens": 456357308.0, + "step": 17645 + }, + { + "epoch": 1.9378431803206677, + "grad_norm": 2.3215107917785645, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7152230739593506, + "num_tokens": 456379764.0, + "step": 17646 + }, + { + "epoch": 1.9379529980232815, + "grad_norm": 2.2019898891448975, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7327172756195068, + "num_tokens": 456404692.0, + "step": 17647 + }, + { + "epoch": 1.938062815725895, + "grad_norm": 2.252683401107788, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7052390575408936, + "num_tokens": 456429190.0, + "step": 17648 + }, + { + "epoch": 1.9381726334285085, + "grad_norm": 2.0135672092437744, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7178059220314026, + "num_tokens": 456459855.0, + "step": 17649 + }, + { + "epoch": 1.9382824511311223, + "grad_norm": 1.9584016799926758, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7150527238845825, + "num_tokens": 456489508.0, + "step": 17650 + }, + { + "epoch": 1.938392268833736, + "grad_norm": 2.240431785583496, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7055266499519348, + "num_tokens": 456514346.0, + "step": 17651 + }, + { + "epoch": 1.9385020865363498, + "grad_norm": 1.9189070463180542, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.694168210029602, + "num_tokens": 456547450.0, + "step": 17652 + }, + { + "epoch": 1.9386119042389633, + "grad_norm": 2.671191692352295, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7462962865829468, + "num_tokens": 456565621.0, + "step": 17653 + }, + { + "epoch": 1.9387217219415769, + "grad_norm": 2.528646230697632, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7128139734268188, + "num_tokens": 456585911.0, + "step": 17654 + }, + { + "epoch": 1.9388315396441906, + "grad_norm": 2.1276581287384033, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.7020905017852783, + "num_tokens": 456614068.0, + "step": 17655 + }, + { + "epoch": 1.9389413573468044, + "grad_norm": 1.901899814605713, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7140920162200928, + "num_tokens": 456646528.0, + "step": 17656 + }, + { + "epoch": 1.939051175049418, + "grad_norm": 2.244192123413086, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7027390599250793, + "num_tokens": 456672194.0, + "step": 17657 + }, + { + "epoch": 1.9391609927520315, + "grad_norm": 2.4753947257995605, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7205745577812195, + "num_tokens": 456692698.0, + "step": 17658 + }, + { + "epoch": 1.9392708104546452, + "grad_norm": 2.294390916824341, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7165993452072144, + "num_tokens": 456715664.0, + "step": 17659 + }, + { + "epoch": 1.939380628157259, + "grad_norm": 2.1180617809295654, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7226839065551758, + "num_tokens": 456741854.0, + "step": 17660 + }, + { + "epoch": 1.9394904458598727, + "grad_norm": 2.1031153202056885, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7115080952644348, + "num_tokens": 456768364.0, + "step": 17661 + }, + { + "epoch": 1.9396002635624863, + "grad_norm": 2.0549635887145996, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6951403617858887, + "num_tokens": 456799532.0, + "step": 17662 + }, + { + "epoch": 1.9397100812650998, + "grad_norm": 2.2779250144958496, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7414216995239258, + "num_tokens": 456823868.0, + "step": 17663 + }, + { + "epoch": 1.9398198989677136, + "grad_norm": 2.7132341861724854, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7499886751174927, + "num_tokens": 456843968.0, + "step": 17664 + }, + { + "epoch": 1.9399297166703273, + "grad_norm": 2.699348211288452, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7334173917770386, + "num_tokens": 456862883.0, + "step": 17665 + }, + { + "epoch": 1.940039534372941, + "grad_norm": 2.3589189052581787, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7157666087150574, + "num_tokens": 456886445.0, + "step": 17666 + }, + { + "epoch": 1.9401493520755546, + "grad_norm": 2.4708499908447266, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7228169441223145, + "num_tokens": 456908085.0, + "step": 17667 + }, + { + "epoch": 1.9402591697781681, + "grad_norm": 2.1563894748687744, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7229707837104797, + "num_tokens": 456937381.0, + "step": 17668 + }, + { + "epoch": 1.940368987480782, + "grad_norm": 2.292797803878784, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7314720749855042, + "num_tokens": 456961903.0, + "step": 17669 + }, + { + "epoch": 1.9404788051833957, + "grad_norm": 2.2980103492736816, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6960182189941406, + "num_tokens": 456986959.0, + "step": 17670 + }, + { + "epoch": 1.9405886228860092, + "grad_norm": 2.5213539600372314, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7142918109893799, + "num_tokens": 457006516.0, + "step": 17671 + }, + { + "epoch": 1.9406984405886227, + "grad_norm": 1.8822314739227295, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.731647253036499, + "num_tokens": 457038982.0, + "step": 17672 + }, + { + "epoch": 1.9408082582912365, + "grad_norm": 1.9294198751449585, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7305885553359985, + "num_tokens": 457070470.0, + "step": 17673 + }, + { + "epoch": 1.9409180759938502, + "grad_norm": 2.5456321239471436, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7410088181495667, + "num_tokens": 457090603.0, + "step": 17674 + }, + { + "epoch": 1.941027893696464, + "grad_norm": 2.3102352619171143, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.721136748790741, + "num_tokens": 457116720.0, + "step": 17675 + }, + { + "epoch": 1.9411377113990775, + "grad_norm": 2.4373557567596436, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7397100925445557, + "num_tokens": 457138482.0, + "step": 17676 + }, + { + "epoch": 1.941247529101691, + "grad_norm": 2.2143990993499756, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.73031085729599, + "num_tokens": 457162724.0, + "step": 17677 + }, + { + "epoch": 1.9413573468043048, + "grad_norm": 2.2200212478637695, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7326425313949585, + "num_tokens": 457187631.0, + "step": 17678 + }, + { + "epoch": 1.9414671645069186, + "grad_norm": 2.100433349609375, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7166085243225098, + "num_tokens": 457217336.0, + "step": 17679 + }, + { + "epoch": 1.9415769822095323, + "grad_norm": 2.4793453216552734, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7373905181884766, + "num_tokens": 457239708.0, + "step": 17680 + }, + { + "epoch": 1.9416867999121459, + "grad_norm": 2.125037908554077, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6949429512023926, + "num_tokens": 457273370.0, + "step": 17681 + }, + { + "epoch": 1.9417966176147594, + "grad_norm": 1.9926294088363647, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.721827507019043, + "num_tokens": 457303652.0, + "step": 17682 + }, + { + "epoch": 1.9419064353173732, + "grad_norm": 2.285440683364868, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7173944711685181, + "num_tokens": 457329030.0, + "step": 17683 + }, + { + "epoch": 1.942016253019987, + "grad_norm": 2.7727160453796387, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7368001937866211, + "num_tokens": 457346283.0, + "step": 17684 + }, + { + "epoch": 1.9421260707226005, + "grad_norm": 2.72318696975708, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7314196825027466, + "num_tokens": 457366083.0, + "step": 17685 + }, + { + "epoch": 1.942235888425214, + "grad_norm": 2.0988686084747314, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7018487453460693, + "num_tokens": 457397897.0, + "step": 17686 + }, + { + "epoch": 1.9423457061278278, + "grad_norm": 2.1994869709014893, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7181651592254639, + "num_tokens": 457422345.0, + "step": 17687 + }, + { + "epoch": 1.9424555238304415, + "grad_norm": 2.146139144897461, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.727401614189148, + "num_tokens": 457448229.0, + "step": 17688 + }, + { + "epoch": 1.9425653415330553, + "grad_norm": 2.0345261096954346, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7131026387214661, + "num_tokens": 457477440.0, + "step": 17689 + }, + { + "epoch": 1.9426751592356688, + "grad_norm": 2.086872100830078, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.706709086894989, + "num_tokens": 457505166.0, + "step": 17690 + }, + { + "epoch": 1.9427849769382823, + "grad_norm": 2.035403251647949, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.700142502784729, + "num_tokens": 457536226.0, + "step": 17691 + }, + { + "epoch": 1.942894794640896, + "grad_norm": 2.0252251625061035, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7197480201721191, + "num_tokens": 457565023.0, + "step": 17692 + }, + { + "epoch": 1.9430046123435099, + "grad_norm": 2.123046398162842, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7262902855873108, + "num_tokens": 457592425.0, + "step": 17693 + }, + { + "epoch": 1.9431144300461234, + "grad_norm": 2.194530725479126, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7093064785003662, + "num_tokens": 457621475.0, + "step": 17694 + }, + { + "epoch": 1.9432242477487371, + "grad_norm": 2.038097858428955, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7067509293556213, + "num_tokens": 457652399.0, + "step": 17695 + }, + { + "epoch": 1.9433340654513507, + "grad_norm": 2.2147908210754395, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.682631254196167, + "num_tokens": 457680403.0, + "step": 17696 + }, + { + "epoch": 1.9434438831539644, + "grad_norm": 2.2209150791168213, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7143330574035645, + "num_tokens": 457706330.0, + "step": 17697 + }, + { + "epoch": 1.9435537008565782, + "grad_norm": 2.349268674850464, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6979990005493164, + "num_tokens": 457728851.0, + "step": 17698 + }, + { + "epoch": 1.9436635185591917, + "grad_norm": 2.5367441177368164, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7262993454933167, + "num_tokens": 457749633.0, + "step": 17699 + }, + { + "epoch": 1.9437733362618053, + "grad_norm": 2.262834072113037, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7159770727157593, + "num_tokens": 457774304.0, + "step": 17700 + }, + { + "epoch": 1.943883153964419, + "grad_norm": 2.1107451915740967, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7429239153862, + "num_tokens": 457801445.0, + "step": 17701 + }, + { + "epoch": 1.9439929716670328, + "grad_norm": 2.228241443634033, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7299782037734985, + "num_tokens": 457825155.0, + "step": 17702 + }, + { + "epoch": 1.9441027893696465, + "grad_norm": 2.147989511489868, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7182676792144775, + "num_tokens": 457852221.0, + "step": 17703 + }, + { + "epoch": 1.94421260707226, + "grad_norm": 2.179959297180176, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7141587734222412, + "num_tokens": 457879946.0, + "step": 17704 + }, + { + "epoch": 1.9443224247748736, + "grad_norm": 2.0724220275878906, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7170377373695374, + "num_tokens": 457910546.0, + "step": 17705 + }, + { + "epoch": 1.9444322424774874, + "grad_norm": 2.3848328590393066, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7307351231575012, + "num_tokens": 457933478.0, + "step": 17706 + }, + { + "epoch": 1.9445420601801011, + "grad_norm": 2.0370872020721436, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7154510021209717, + "num_tokens": 457964229.0, + "step": 17707 + }, + { + "epoch": 1.9446518778827147, + "grad_norm": 2.2939181327819824, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7085033059120178, + "num_tokens": 457988781.0, + "step": 17708 + }, + { + "epoch": 1.9447616955853284, + "grad_norm": 2.1471517086029053, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7070237398147583, + "num_tokens": 458016238.0, + "step": 17709 + }, + { + "epoch": 1.944871513287942, + "grad_norm": 2.0244436264038086, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6959443092346191, + "num_tokens": 458047596.0, + "step": 17710 + }, + { + "epoch": 1.9449813309905557, + "grad_norm": 2.0411593914031982, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6962155103683472, + "num_tokens": 458077736.0, + "step": 17711 + }, + { + "epoch": 1.9450911486931695, + "grad_norm": 2.3840956687927246, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.71634840965271, + "num_tokens": 458102156.0, + "step": 17712 + }, + { + "epoch": 1.945200966395783, + "grad_norm": 2.1268441677093506, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7229259610176086, + "num_tokens": 458132768.0, + "step": 17713 + }, + { + "epoch": 1.9453107840983965, + "grad_norm": 2.1772844791412354, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7065223455429077, + "num_tokens": 458159677.0, + "step": 17714 + }, + { + "epoch": 1.9454206018010103, + "grad_norm": 2.283755302429199, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7179502248764038, + "num_tokens": 458184722.0, + "step": 17715 + }, + { + "epoch": 1.945530419503624, + "grad_norm": 1.909035325050354, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7196479439735413, + "num_tokens": 458217156.0, + "step": 17716 + }, + { + "epoch": 1.9456402372062378, + "grad_norm": 2.443389415740967, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7199077010154724, + "num_tokens": 458238832.0, + "step": 17717 + }, + { + "epoch": 1.9457500549088513, + "grad_norm": 2.3225324153900146, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7089014053344727, + "num_tokens": 458262370.0, + "step": 17718 + }, + { + "epoch": 1.9458598726114649, + "grad_norm": 2.605238437652588, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7404751777648926, + "num_tokens": 458281145.0, + "step": 17719 + }, + { + "epoch": 1.9459696903140786, + "grad_norm": 2.1457080841064453, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7213093042373657, + "num_tokens": 458308876.0, + "step": 17720 + }, + { + "epoch": 1.9460795080166924, + "grad_norm": 2.0058882236480713, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7031328678131104, + "num_tokens": 458338453.0, + "step": 17721 + }, + { + "epoch": 1.946189325719306, + "grad_norm": 2.109940767288208, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7072433233261108, + "num_tokens": 458364645.0, + "step": 17722 + }, + { + "epoch": 1.9462991434219195, + "grad_norm": 2.3231899738311768, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7081409692764282, + "num_tokens": 458389238.0, + "step": 17723 + }, + { + "epoch": 1.9464089611245332, + "grad_norm": 2.2769455909729004, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7187035083770752, + "num_tokens": 458412718.0, + "step": 17724 + }, + { + "epoch": 1.946518778827147, + "grad_norm": 2.174483060836792, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7369269132614136, + "num_tokens": 458440883.0, + "step": 17725 + }, + { + "epoch": 1.9466285965297607, + "grad_norm": 1.9931777715682983, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.71168452501297, + "num_tokens": 458471234.0, + "step": 17726 + }, + { + "epoch": 1.9467384142323743, + "grad_norm": 2.5923025608062744, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.712093710899353, + "num_tokens": 458491313.0, + "step": 17727 + }, + { + "epoch": 1.9468482319349878, + "grad_norm": 1.9736329317092896, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7300193905830383, + "num_tokens": 458521269.0, + "step": 17728 + }, + { + "epoch": 1.9469580496376016, + "grad_norm": 2.2267885208129883, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7064406871795654, + "num_tokens": 458549404.0, + "step": 17729 + }, + { + "epoch": 1.9470678673402153, + "grad_norm": 2.281398057937622, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7034302949905396, + "num_tokens": 458576406.0, + "step": 17730 + }, + { + "epoch": 1.947177685042829, + "grad_norm": 2.332674741744995, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7228797674179077, + "num_tokens": 458601780.0, + "step": 17731 + }, + { + "epoch": 1.9472875027454426, + "grad_norm": 2.1911277770996094, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7371847629547119, + "num_tokens": 458626965.0, + "step": 17732 + }, + { + "epoch": 1.9473973204480561, + "grad_norm": 2.2160885334014893, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7341464161872864, + "num_tokens": 458652567.0, + "step": 17733 + }, + { + "epoch": 1.94750713815067, + "grad_norm": 2.37223219871521, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7106729745864868, + "num_tokens": 458678063.0, + "step": 17734 + }, + { + "epoch": 1.9476169558532836, + "grad_norm": 2.2029664516448975, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7296144962310791, + "num_tokens": 458703002.0, + "step": 17735 + }, + { + "epoch": 1.9477267735558972, + "grad_norm": 2.3042941093444824, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7266951203346252, + "num_tokens": 458726273.0, + "step": 17736 + }, + { + "epoch": 1.9478365912585107, + "grad_norm": 2.3436379432678223, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7210988998413086, + "num_tokens": 458748947.0, + "step": 17737 + }, + { + "epoch": 1.9479464089611245, + "grad_norm": 2.33454966545105, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7204195261001587, + "num_tokens": 458773159.0, + "step": 17738 + }, + { + "epoch": 1.9480562266637382, + "grad_norm": 2.0892436504364014, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7128013968467712, + "num_tokens": 458802385.0, + "step": 17739 + }, + { + "epoch": 1.948166044366352, + "grad_norm": 2.192828893661499, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7218558192253113, + "num_tokens": 458828374.0, + "step": 17740 + }, + { + "epoch": 1.9482758620689655, + "grad_norm": 2.274435043334961, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7063838243484497, + "num_tokens": 458852194.0, + "step": 17741 + }, + { + "epoch": 1.948385679771579, + "grad_norm": 2.161975383758545, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7379101514816284, + "num_tokens": 458879585.0, + "step": 17742 + }, + { + "epoch": 1.9484954974741928, + "grad_norm": 2.0316121578216553, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7210647463798523, + "num_tokens": 458910244.0, + "step": 17743 + }, + { + "epoch": 1.9486053151768066, + "grad_norm": 2.106898784637451, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7176342010498047, + "num_tokens": 458937185.0, + "step": 17744 + }, + { + "epoch": 1.94871513287942, + "grad_norm": 2.2088310718536377, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7384604215621948, + "num_tokens": 458963727.0, + "step": 17745 + }, + { + "epoch": 1.9488249505820339, + "grad_norm": 1.9409912824630737, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7048022747039795, + "num_tokens": 458996251.0, + "step": 17746 + }, + { + "epoch": 1.9489347682846474, + "grad_norm": 2.0108208656311035, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7143905162811279, + "num_tokens": 459027137.0, + "step": 17747 + }, + { + "epoch": 1.9490445859872612, + "grad_norm": 2.302950143814087, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7352890968322754, + "num_tokens": 459050086.0, + "step": 17748 + }, + { + "epoch": 1.949154403689875, + "grad_norm": 2.263791799545288, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7360836267471313, + "num_tokens": 459073889.0, + "step": 17749 + }, + { + "epoch": 1.9492642213924884, + "grad_norm": 2.2109968662261963, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7123980522155762, + "num_tokens": 459100458.0, + "step": 17750 + }, + { + "epoch": 1.949374039095102, + "grad_norm": 2.575453042984009, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7352614402770996, + "num_tokens": 459119890.0, + "step": 17751 + }, + { + "epoch": 1.9494838567977157, + "grad_norm": 2.2086966037750244, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7066483497619629, + "num_tokens": 459147450.0, + "step": 17752 + }, + { + "epoch": 1.9495936745003295, + "grad_norm": 2.1352531909942627, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7504262924194336, + "num_tokens": 459172636.0, + "step": 17753 + }, + { + "epoch": 1.9497034922029433, + "grad_norm": 2.204728364944458, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7079355120658875, + "num_tokens": 459200885.0, + "step": 17754 + }, + { + "epoch": 1.9498133099055568, + "grad_norm": 2.021061897277832, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7285854816436768, + "num_tokens": 459229408.0, + "step": 17755 + }, + { + "epoch": 1.9499231276081703, + "grad_norm": 2.4153034687042236, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7416186332702637, + "num_tokens": 459254265.0, + "step": 17756 + }, + { + "epoch": 1.950032945310784, + "grad_norm": 2.2186357975006104, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7108157873153687, + "num_tokens": 459280332.0, + "step": 17757 + }, + { + "epoch": 1.9501427630133978, + "grad_norm": 2.1934762001037598, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7257488369941711, + "num_tokens": 459305595.0, + "step": 17758 + }, + { + "epoch": 1.9502525807160114, + "grad_norm": 2.5889651775360107, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7450463771820068, + "num_tokens": 459325299.0, + "step": 17759 + }, + { + "epoch": 1.9503623984186251, + "grad_norm": 2.24507474899292, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7184392809867859, + "num_tokens": 459352836.0, + "step": 17760 + }, + { + "epoch": 1.9504722161212387, + "grad_norm": 2.2591238021850586, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.705870509147644, + "num_tokens": 459379227.0, + "step": 17761 + }, + { + "epoch": 1.9505820338238524, + "grad_norm": 2.0320518016815186, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6975772380828857, + "num_tokens": 459411692.0, + "step": 17762 + }, + { + "epoch": 1.9506918515264662, + "grad_norm": 1.8778035640716553, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.726474940776825, + "num_tokens": 459445915.0, + "step": 17763 + }, + { + "epoch": 1.9508016692290797, + "grad_norm": 2.3392422199249268, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7038555145263672, + "num_tokens": 459471396.0, + "step": 17764 + }, + { + "epoch": 1.9509114869316933, + "grad_norm": 2.4941210746765137, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7231453061103821, + "num_tokens": 459493047.0, + "step": 17765 + }, + { + "epoch": 1.951021304634307, + "grad_norm": 2.4504685401916504, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7287617325782776, + "num_tokens": 459515684.0, + "step": 17766 + }, + { + "epoch": 1.9511311223369208, + "grad_norm": 2.291320323944092, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7286868095397949, + "num_tokens": 459540503.0, + "step": 17767 + }, + { + "epoch": 1.9512409400395345, + "grad_norm": 2.084695816040039, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7144932746887207, + "num_tokens": 459569352.0, + "step": 17768 + }, + { + "epoch": 1.951350757742148, + "grad_norm": 2.536186456680298, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7459691762924194, + "num_tokens": 459591727.0, + "step": 17769 + }, + { + "epoch": 1.9514605754447616, + "grad_norm": 2.4941155910491943, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7343938946723938, + "num_tokens": 459613008.0, + "step": 17770 + }, + { + "epoch": 1.9515703931473753, + "grad_norm": 2.107172966003418, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7240207195281982, + "num_tokens": 459640519.0, + "step": 17771 + }, + { + "epoch": 1.951680210849989, + "grad_norm": 2.1849112510681152, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7175218462944031, + "num_tokens": 459668946.0, + "step": 17772 + }, + { + "epoch": 1.9517900285526026, + "grad_norm": 1.8815079927444458, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7234096527099609, + "num_tokens": 459701833.0, + "step": 17773 + }, + { + "epoch": 1.9518998462552162, + "grad_norm": 2.1254067420959473, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7291892170906067, + "num_tokens": 459729092.0, + "step": 17774 + }, + { + "epoch": 1.95200966395783, + "grad_norm": 2.316538095474243, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7085785865783691, + "num_tokens": 459755210.0, + "step": 17775 + }, + { + "epoch": 1.9521194816604437, + "grad_norm": 2.037518262863159, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6970654129981995, + "num_tokens": 459787978.0, + "step": 17776 + }, + { + "epoch": 1.9522292993630574, + "grad_norm": 2.133347511291504, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.732170820236206, + "num_tokens": 459814198.0, + "step": 17777 + }, + { + "epoch": 1.952339117065671, + "grad_norm": 1.993656039237976, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7480747699737549, + "num_tokens": 459843631.0, + "step": 17778 + }, + { + "epoch": 1.9524489347682845, + "grad_norm": 2.3932886123657227, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7302210330963135, + "num_tokens": 459865903.0, + "step": 17779 + }, + { + "epoch": 1.9525587524708983, + "grad_norm": 2.4418089389801025, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7018827199935913, + "num_tokens": 459890911.0, + "step": 17780 + }, + { + "epoch": 1.952668570173512, + "grad_norm": 2.183695077896118, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7327996492385864, + "num_tokens": 459916634.0, + "step": 17781 + }, + { + "epoch": 1.9527783878761258, + "grad_norm": 2.0739314556121826, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7067065834999084, + "num_tokens": 459945596.0, + "step": 17782 + }, + { + "epoch": 1.9528882055787393, + "grad_norm": 2.187594413757324, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6879065036773682, + "num_tokens": 459971509.0, + "step": 17783 + }, + { + "epoch": 1.9529980232813529, + "grad_norm": 2.2502918243408203, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7147401571273804, + "num_tokens": 459995866.0, + "step": 17784 + }, + { + "epoch": 1.9531078409839666, + "grad_norm": 2.4173266887664795, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7196774482727051, + "num_tokens": 460019422.0, + "step": 17785 + }, + { + "epoch": 1.9532176586865804, + "grad_norm": 2.0643396377563477, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7111110091209412, + "num_tokens": 460051120.0, + "step": 17786 + }, + { + "epoch": 1.953327476389194, + "grad_norm": 2.2289326190948486, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7292441725730896, + "num_tokens": 460075998.0, + "step": 17787 + }, + { + "epoch": 1.9534372940918074, + "grad_norm": 2.2891736030578613, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7401121854782104, + "num_tokens": 460102115.0, + "step": 17788 + }, + { + "epoch": 1.9535471117944212, + "grad_norm": 2.364959478378296, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7187451720237732, + "num_tokens": 460124918.0, + "step": 17789 + }, + { + "epoch": 1.953656929497035, + "grad_norm": 2.133833646774292, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7149856686592102, + "num_tokens": 460151773.0, + "step": 17790 + }, + { + "epoch": 1.9537667471996487, + "grad_norm": 2.058505058288574, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7137162685394287, + "num_tokens": 460180962.0, + "step": 17791 + }, + { + "epoch": 1.9538765649022622, + "grad_norm": 2.321948289871216, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6951586604118347, + "num_tokens": 460204462.0, + "step": 17792 + }, + { + "epoch": 1.9539863826048758, + "grad_norm": 2.1632726192474365, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7045948505401611, + "num_tokens": 460231154.0, + "step": 17793 + }, + { + "epoch": 1.9540962003074895, + "grad_norm": 2.4818038940429688, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7362980842590332, + "num_tokens": 460251325.0, + "step": 17794 + }, + { + "epoch": 1.9542060180101033, + "grad_norm": 2.3556737899780273, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7128115892410278, + "num_tokens": 460275225.0, + "step": 17795 + }, + { + "epoch": 1.954315835712717, + "grad_norm": 2.222468852996826, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7339364290237427, + "num_tokens": 460300141.0, + "step": 17796 + }, + { + "epoch": 1.9544256534153306, + "grad_norm": 2.435396194458008, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7318664789199829, + "num_tokens": 460321850.0, + "step": 17797 + }, + { + "epoch": 1.9545354711179441, + "grad_norm": 1.9190915822982788, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7382285594940186, + "num_tokens": 460353093.0, + "step": 17798 + }, + { + "epoch": 1.9546452888205579, + "grad_norm": 2.2245469093322754, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6866883039474487, + "num_tokens": 460380598.0, + "step": 17799 + }, + { + "epoch": 1.9547551065231716, + "grad_norm": 2.1892988681793213, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7131233215332031, + "num_tokens": 460405826.0, + "step": 17800 + }, + { + "epoch": 1.9548649242257852, + "grad_norm": 2.287276268005371, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7227910757064819, + "num_tokens": 460429248.0, + "step": 17801 + }, + { + "epoch": 1.9549747419283987, + "grad_norm": 2.215717315673828, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7413212060928345, + "num_tokens": 460453280.0, + "step": 17802 + }, + { + "epoch": 1.9550845596310125, + "grad_norm": 2.3203585147857666, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7254450917243958, + "num_tokens": 460476265.0, + "step": 17803 + }, + { + "epoch": 1.9551943773336262, + "grad_norm": 1.9493615627288818, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7152276039123535, + "num_tokens": 460506821.0, + "step": 17804 + }, + { + "epoch": 1.95530419503624, + "grad_norm": 2.05171799659729, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7051205635070801, + "num_tokens": 460535198.0, + "step": 17805 + }, + { + "epoch": 1.9554140127388535, + "grad_norm": 2.3335721492767334, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7074348330497742, + "num_tokens": 460559697.0, + "step": 17806 + }, + { + "epoch": 1.955523830441467, + "grad_norm": 2.146548271179199, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7067323923110962, + "num_tokens": 460587873.0, + "step": 17807 + }, + { + "epoch": 1.9556336481440808, + "grad_norm": 2.210827350616455, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6987267732620239, + "num_tokens": 460613492.0, + "step": 17808 + }, + { + "epoch": 1.9557434658466946, + "grad_norm": 2.474519968032837, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7376132607460022, + "num_tokens": 460633636.0, + "step": 17809 + }, + { + "epoch": 1.955853283549308, + "grad_norm": 2.117144823074341, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7334731817245483, + "num_tokens": 460660312.0, + "step": 17810 + }, + { + "epoch": 1.9559631012519219, + "grad_norm": 2.6721014976501465, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7221480011940002, + "num_tokens": 460678619.0, + "step": 17811 + }, + { + "epoch": 1.9560729189545354, + "grad_norm": 2.480729579925537, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7344570159912109, + "num_tokens": 460699511.0, + "step": 17812 + }, + { + "epoch": 1.9561827366571491, + "grad_norm": 2.1753995418548584, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7210930585861206, + "num_tokens": 460726563.0, + "step": 17813 + }, + { + "epoch": 1.956292554359763, + "grad_norm": 2.0243477821350098, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7349052429199219, + "num_tokens": 460755421.0, + "step": 17814 + }, + { + "epoch": 1.9564023720623764, + "grad_norm": 2.144059658050537, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.721985936164856, + "num_tokens": 460781504.0, + "step": 17815 + }, + { + "epoch": 1.95651218976499, + "grad_norm": 2.1797516345977783, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7251889109611511, + "num_tokens": 460808805.0, + "step": 17816 + }, + { + "epoch": 1.9566220074676037, + "grad_norm": 2.083575963973999, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7116860151290894, + "num_tokens": 460836943.0, + "step": 17817 + }, + { + "epoch": 1.9567318251702175, + "grad_norm": 2.2190029621124268, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7086197137832642, + "num_tokens": 460863149.0, + "step": 17818 + }, + { + "epoch": 1.9568416428728312, + "grad_norm": 2.1182563304901123, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7332686185836792, + "num_tokens": 460891102.0, + "step": 17819 + }, + { + "epoch": 1.9569514605754448, + "grad_norm": 2.1330208778381348, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.70774245262146, + "num_tokens": 460917830.0, + "step": 17820 + }, + { + "epoch": 1.9570612782780583, + "grad_norm": 2.1014328002929688, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.713001549243927, + "num_tokens": 460943706.0, + "step": 17821 + }, + { + "epoch": 1.957171095980672, + "grad_norm": 2.064354419708252, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7360918521881104, + "num_tokens": 460970926.0, + "step": 17822 + }, + { + "epoch": 1.9572809136832858, + "grad_norm": 2.1836018562316895, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7297326326370239, + "num_tokens": 460995967.0, + "step": 17823 + }, + { + "epoch": 1.9573907313858994, + "grad_norm": 2.0037126541137695, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7125816345214844, + "num_tokens": 461026643.0, + "step": 17824 + }, + { + "epoch": 1.9575005490885131, + "grad_norm": 2.2388598918914795, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7098495960235596, + "num_tokens": 461052393.0, + "step": 17825 + }, + { + "epoch": 1.9576103667911267, + "grad_norm": 1.9704569578170776, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7129508852958679, + "num_tokens": 461083688.0, + "step": 17826 + }, + { + "epoch": 1.9577201844937404, + "grad_norm": 2.3449432849884033, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7166174650192261, + "num_tokens": 461106965.0, + "step": 17827 + }, + { + "epoch": 1.9578300021963542, + "grad_norm": 2.183265209197998, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6910771727561951, + "num_tokens": 461134676.0, + "step": 17828 + }, + { + "epoch": 1.9579398198989677, + "grad_norm": 2.174151659011841, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6987636089324951, + "num_tokens": 461163256.0, + "step": 17829 + }, + { + "epoch": 1.9580496376015812, + "grad_norm": 2.2659130096435547, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7375473380088806, + "num_tokens": 461186258.0, + "step": 17830 + }, + { + "epoch": 1.958159455304195, + "grad_norm": 1.9985781908035278, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.705642580986023, + "num_tokens": 461218074.0, + "step": 17831 + }, + { + "epoch": 1.9582692730068088, + "grad_norm": 2.142765760421753, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7285491824150085, + "num_tokens": 461243234.0, + "step": 17832 + }, + { + "epoch": 1.9583790907094225, + "grad_norm": 2.035511016845703, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7116605639457703, + "num_tokens": 461272954.0, + "step": 17833 + }, + { + "epoch": 1.958488908412036, + "grad_norm": 2.311338186264038, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7414731383323669, + "num_tokens": 461295956.0, + "step": 17834 + }, + { + "epoch": 1.9585987261146496, + "grad_norm": 2.106546401977539, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7204602360725403, + "num_tokens": 461323378.0, + "step": 17835 + }, + { + "epoch": 1.9587085438172633, + "grad_norm": 2.144925117492676, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7127256393432617, + "num_tokens": 461349861.0, + "step": 17836 + }, + { + "epoch": 1.958818361519877, + "grad_norm": 2.5451669692993164, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7324902415275574, + "num_tokens": 461370899.0, + "step": 17837 + }, + { + "epoch": 1.9589281792224906, + "grad_norm": 2.262019157409668, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7346209287643433, + "num_tokens": 461397994.0, + "step": 17838 + }, + { + "epoch": 1.9590379969251042, + "grad_norm": 2.4144816398620605, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7226382493972778, + "num_tokens": 461421247.0, + "step": 17839 + }, + { + "epoch": 1.959147814627718, + "grad_norm": 2.37316632270813, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.715053915977478, + "num_tokens": 461443922.0, + "step": 17840 + }, + { + "epoch": 1.9592576323303317, + "grad_norm": 2.4240565299987793, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7224789261817932, + "num_tokens": 461466502.0, + "step": 17841 + }, + { + "epoch": 1.9593674500329454, + "grad_norm": 1.9629462957382202, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7012312412261963, + "num_tokens": 461502352.0, + "step": 17842 + }, + { + "epoch": 1.959477267735559, + "grad_norm": 2.219780445098877, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.727900505065918, + "num_tokens": 461526845.0, + "step": 17843 + }, + { + "epoch": 1.9595870854381725, + "grad_norm": 2.1258583068847656, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7396063804626465, + "num_tokens": 461552144.0, + "step": 17844 + }, + { + "epoch": 1.9596969031407863, + "grad_norm": 2.032327890396118, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7248027920722961, + "num_tokens": 461581473.0, + "step": 17845 + }, + { + "epoch": 1.9598067208434, + "grad_norm": 2.1666455268859863, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7305042743682861, + "num_tokens": 461607343.0, + "step": 17846 + }, + { + "epoch": 1.9599165385460138, + "grad_norm": 2.0978212356567383, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7154167890548706, + "num_tokens": 461634676.0, + "step": 17847 + }, + { + "epoch": 1.9600263562486273, + "grad_norm": 1.9674196243286133, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.708293080329895, + "num_tokens": 461667066.0, + "step": 17848 + }, + { + "epoch": 1.9601361739512408, + "grad_norm": 2.519320011138916, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7718709111213684, + "num_tokens": 461687263.0, + "step": 17849 + }, + { + "epoch": 1.9602459916538546, + "grad_norm": 2.125831127166748, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7394863367080688, + "num_tokens": 461712768.0, + "step": 17850 + }, + { + "epoch": 1.9603558093564684, + "grad_norm": 2.2433905601501465, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.6998845934867859, + "num_tokens": 461738675.0, + "step": 17851 + }, + { + "epoch": 1.960465627059082, + "grad_norm": 2.2411048412323, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7177528738975525, + "num_tokens": 461764686.0, + "step": 17852 + }, + { + "epoch": 1.9605754447616954, + "grad_norm": 2.1123666763305664, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7219274044036865, + "num_tokens": 461792000.0, + "step": 17853 + }, + { + "epoch": 1.9606852624643092, + "grad_norm": 2.4016711711883545, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7180080413818359, + "num_tokens": 461815714.0, + "step": 17854 + }, + { + "epoch": 1.960795080166923, + "grad_norm": 2.2338173389434814, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7157076001167297, + "num_tokens": 461840276.0, + "step": 17855 + }, + { + "epoch": 1.9609048978695367, + "grad_norm": 1.9811259508132935, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6961681246757507, + "num_tokens": 461871940.0, + "step": 17856 + }, + { + "epoch": 1.9610147155721502, + "grad_norm": 2.092406988143921, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7373872995376587, + "num_tokens": 461900224.0, + "step": 17857 + }, + { + "epoch": 1.9611245332747638, + "grad_norm": 2.214587926864624, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7207850813865662, + "num_tokens": 461925930.0, + "step": 17858 + }, + { + "epoch": 1.9612343509773775, + "grad_norm": 2.278686285018921, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7238807678222656, + "num_tokens": 461951480.0, + "step": 17859 + }, + { + "epoch": 1.9613441686799913, + "grad_norm": 2.229663372039795, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.719232439994812, + "num_tokens": 461975796.0, + "step": 17860 + }, + { + "epoch": 1.961453986382605, + "grad_norm": 2.509561061859131, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7185275554656982, + "num_tokens": 461996344.0, + "step": 17861 + }, + { + "epoch": 1.9615638040852186, + "grad_norm": 2.1449108123779297, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.732284665107727, + "num_tokens": 462023553.0, + "step": 17862 + }, + { + "epoch": 1.9616736217878321, + "grad_norm": 2.086888074874878, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7048985362052917, + "num_tokens": 462049915.0, + "step": 17863 + }, + { + "epoch": 1.9617834394904459, + "grad_norm": 2.120511054992676, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7047088146209717, + "num_tokens": 462079966.0, + "step": 17864 + }, + { + "epoch": 1.9618932571930596, + "grad_norm": 2.405211925506592, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7157363295555115, + "num_tokens": 462103348.0, + "step": 17865 + }, + { + "epoch": 1.9620030748956732, + "grad_norm": 2.197854518890381, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7209017276763916, + "num_tokens": 462131788.0, + "step": 17866 + }, + { + "epoch": 1.9621128925982867, + "grad_norm": 2.4902799129486084, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7199798226356506, + "num_tokens": 462154304.0, + "step": 17867 + }, + { + "epoch": 1.9622227103009005, + "grad_norm": 2.165755271911621, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7334253787994385, + "num_tokens": 462180987.0, + "step": 17868 + }, + { + "epoch": 1.9623325280035142, + "grad_norm": 2.0494608879089355, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7178495526313782, + "num_tokens": 462210465.0, + "step": 17869 + }, + { + "epoch": 1.962442345706128, + "grad_norm": 2.208439350128174, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7159236669540405, + "num_tokens": 462236827.0, + "step": 17870 + }, + { + "epoch": 1.9625521634087415, + "grad_norm": 2.0882019996643066, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7237499952316284, + "num_tokens": 462264662.0, + "step": 17871 + }, + { + "epoch": 1.962661981111355, + "grad_norm": 2.309009075164795, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7267380952835083, + "num_tokens": 462288633.0, + "step": 17872 + }, + { + "epoch": 1.9627717988139688, + "grad_norm": 2.135274648666382, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7094308137893677, + "num_tokens": 462315207.0, + "step": 17873 + }, + { + "epoch": 1.9628816165165826, + "grad_norm": 2.2152507305145264, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7078168988227844, + "num_tokens": 462343361.0, + "step": 17874 + }, + { + "epoch": 1.962991434219196, + "grad_norm": 2.4339826107025146, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7153892517089844, + "num_tokens": 462365597.0, + "step": 17875 + }, + { + "epoch": 1.9631012519218098, + "grad_norm": 2.2477335929870605, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7035595178604126, + "num_tokens": 462391846.0, + "step": 17876 + }, + { + "epoch": 1.9632110696244234, + "grad_norm": 2.110262393951416, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.717465877532959, + "num_tokens": 462418944.0, + "step": 17877 + }, + { + "epoch": 1.9633208873270371, + "grad_norm": 2.25803804397583, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7132724523544312, + "num_tokens": 462446632.0, + "step": 17878 + }, + { + "epoch": 1.963430705029651, + "grad_norm": 2.323960542678833, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7319461107254028, + "num_tokens": 462471488.0, + "step": 17879 + }, + { + "epoch": 1.9635405227322644, + "grad_norm": 2.1277360916137695, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6936169862747192, + "num_tokens": 462499777.0, + "step": 17880 + }, + { + "epoch": 1.963650340434878, + "grad_norm": 2.5857088565826416, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7177330255508423, + "num_tokens": 462521390.0, + "step": 17881 + }, + { + "epoch": 1.9637601581374917, + "grad_norm": 2.1948397159576416, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7255121469497681, + "num_tokens": 462548402.0, + "step": 17882 + }, + { + "epoch": 1.9638699758401055, + "grad_norm": 2.052069664001465, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7234154343605042, + "num_tokens": 462579022.0, + "step": 17883 + }, + { + "epoch": 1.9639797935427192, + "grad_norm": 2.492170572280884, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.739119291305542, + "num_tokens": 462600350.0, + "step": 17884 + }, + { + "epoch": 1.9640896112453328, + "grad_norm": 2.090250253677368, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7161139249801636, + "num_tokens": 462628836.0, + "step": 17885 + }, + { + "epoch": 1.9641994289479463, + "grad_norm": 2.5823256969451904, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7289657592773438, + "num_tokens": 462648404.0, + "step": 17886 + }, + { + "epoch": 1.96430924665056, + "grad_norm": 2.030050754547119, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7328877449035645, + "num_tokens": 462675662.0, + "step": 17887 + }, + { + "epoch": 1.9644190643531738, + "grad_norm": 2.143463611602783, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7093313932418823, + "num_tokens": 462702373.0, + "step": 17888 + }, + { + "epoch": 1.9645288820557874, + "grad_norm": 2.432312488555908, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7273381948471069, + "num_tokens": 462724181.0, + "step": 17889 + }, + { + "epoch": 1.964638699758401, + "grad_norm": 2.138875961303711, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7127394676208496, + "num_tokens": 462751807.0, + "step": 17890 + }, + { + "epoch": 1.9647485174610146, + "grad_norm": 2.0102272033691406, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.717627763748169, + "num_tokens": 462783331.0, + "step": 17891 + }, + { + "epoch": 1.9648583351636284, + "grad_norm": 2.203613519668579, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7340422868728638, + "num_tokens": 462812266.0, + "step": 17892 + }, + { + "epoch": 1.9649681528662422, + "grad_norm": 2.219480276107788, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7245098352432251, + "num_tokens": 462841811.0, + "step": 17893 + }, + { + "epoch": 1.9650779705688557, + "grad_norm": 2.024350166320801, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7211065888404846, + "num_tokens": 462870788.0, + "step": 17894 + }, + { + "epoch": 1.9651877882714692, + "grad_norm": 2.1744203567504883, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6976222395896912, + "num_tokens": 462899120.0, + "step": 17895 + }, + { + "epoch": 1.965297605974083, + "grad_norm": 1.9381444454193115, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7315666079521179, + "num_tokens": 462928270.0, + "step": 17896 + }, + { + "epoch": 1.9654074236766967, + "grad_norm": 1.9331063032150269, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.7022149562835693, + "num_tokens": 462960055.0, + "step": 17897 + }, + { + "epoch": 1.9655172413793105, + "grad_norm": 2.099416494369507, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7226319313049316, + "num_tokens": 462989831.0, + "step": 17898 + }, + { + "epoch": 1.965627059081924, + "grad_norm": 2.535623550415039, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7011237144470215, + "num_tokens": 463011219.0, + "step": 17899 + }, + { + "epoch": 1.9657368767845376, + "grad_norm": 1.9032281637191772, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6960353851318359, + "num_tokens": 463047470.0, + "step": 17900 + }, + { + "epoch": 1.9658466944871513, + "grad_norm": 2.572544813156128, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7250465154647827, + "num_tokens": 463068746.0, + "step": 17901 + }, + { + "epoch": 1.965956512189765, + "grad_norm": 2.2431721687316895, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.727910041809082, + "num_tokens": 463095132.0, + "step": 17902 + }, + { + "epoch": 1.9660663298923786, + "grad_norm": 1.9249788522720337, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7331019639968872, + "num_tokens": 463124970.0, + "step": 17903 + }, + { + "epoch": 1.9661761475949922, + "grad_norm": 2.115152597427368, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7168059349060059, + "num_tokens": 463154144.0, + "step": 17904 + }, + { + "epoch": 1.966285965297606, + "grad_norm": 2.4025444984436035, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7171317338943481, + "num_tokens": 463179158.0, + "step": 17905 + }, + { + "epoch": 1.9663957830002197, + "grad_norm": 2.030355215072632, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7519051432609558, + "num_tokens": 463206314.0, + "step": 17906 + }, + { + "epoch": 1.9665056007028334, + "grad_norm": 2.5814905166625977, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7200311422348022, + "num_tokens": 463228266.0, + "step": 17907 + }, + { + "epoch": 1.966615418405447, + "grad_norm": 2.178952217102051, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7059273719787598, + "num_tokens": 463255744.0, + "step": 17908 + }, + { + "epoch": 1.9667252361080605, + "grad_norm": 1.9962819814682007, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7122012376785278, + "num_tokens": 463287159.0, + "step": 17909 + }, + { + "epoch": 1.9668350538106742, + "grad_norm": 2.111515522003174, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7474125027656555, + "num_tokens": 463313671.0, + "step": 17910 + }, + { + "epoch": 1.966944871513288, + "grad_norm": 2.2863643169403076, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7150174975395203, + "num_tokens": 463338766.0, + "step": 17911 + }, + { + "epoch": 1.9670546892159018, + "grad_norm": 2.108663558959961, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7263177633285522, + "num_tokens": 463365966.0, + "step": 17912 + }, + { + "epoch": 1.9671645069185153, + "grad_norm": 2.2500431537628174, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.716643214225769, + "num_tokens": 463389649.0, + "step": 17913 + }, + { + "epoch": 1.9672743246211288, + "grad_norm": 2.106595277786255, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7145419120788574, + "num_tokens": 463418898.0, + "step": 17914 + }, + { + "epoch": 1.9673841423237426, + "grad_norm": 2.414149522781372, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7126479744911194, + "num_tokens": 463442296.0, + "step": 17915 + }, + { + "epoch": 1.9674939600263563, + "grad_norm": 2.0850963592529297, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7158743143081665, + "num_tokens": 463470572.0, + "step": 17916 + }, + { + "epoch": 1.9676037777289699, + "grad_norm": 2.0438392162323, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7127823233604431, + "num_tokens": 463501542.0, + "step": 17917 + }, + { + "epoch": 1.9677135954315834, + "grad_norm": 2.2821805477142334, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7011624574661255, + "num_tokens": 463528562.0, + "step": 17918 + }, + { + "epoch": 1.9678234131341972, + "grad_norm": 2.1247200965881348, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7223504781723022, + "num_tokens": 463557046.0, + "step": 17919 + }, + { + "epoch": 1.967933230836811, + "grad_norm": 2.256814956665039, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7259087562561035, + "num_tokens": 463581800.0, + "step": 17920 + }, + { + "epoch": 1.9680430485394247, + "grad_norm": 2.187361240386963, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7237414121627808, + "num_tokens": 463607472.0, + "step": 17921 + }, + { + "epoch": 1.9681528662420382, + "grad_norm": 2.1104767322540283, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7000352144241333, + "num_tokens": 463635685.0, + "step": 17922 + }, + { + "epoch": 1.9682626839446518, + "grad_norm": 2.0378096103668213, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6959023475646973, + "num_tokens": 463667783.0, + "step": 17923 + }, + { + "epoch": 1.9683725016472655, + "grad_norm": 2.0251989364624023, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.718139111995697, + "num_tokens": 463699725.0, + "step": 17924 + }, + { + "epoch": 1.9684823193498793, + "grad_norm": 2.381098985671997, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7352466583251953, + "num_tokens": 463721407.0, + "step": 17925 + }, + { + "epoch": 1.9685921370524928, + "grad_norm": 2.2152657508850098, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.7093828916549683, + "num_tokens": 463749097.0, + "step": 17926 + }, + { + "epoch": 1.9687019547551066, + "grad_norm": 2.1095821857452393, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7258527874946594, + "num_tokens": 463777892.0, + "step": 17927 + }, + { + "epoch": 1.96881177245772, + "grad_norm": 3.199571132659912, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7231816053390503, + "num_tokens": 463802494.0, + "step": 17928 + }, + { + "epoch": 1.9689215901603339, + "grad_norm": 2.185001850128174, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7133206129074097, + "num_tokens": 463829345.0, + "step": 17929 + }, + { + "epoch": 1.9690314078629476, + "grad_norm": 2.330692768096924, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7154103517532349, + "num_tokens": 463852750.0, + "step": 17930 + }, + { + "epoch": 1.9691412255655611, + "grad_norm": 2.255382776260376, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7378858923912048, + "num_tokens": 463876341.0, + "step": 17931 + }, + { + "epoch": 1.9692510432681747, + "grad_norm": 2.174497127532959, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6964017152786255, + "num_tokens": 463903977.0, + "step": 17932 + }, + { + "epoch": 1.9693608609707884, + "grad_norm": 1.9731699228286743, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7046810388565063, + "num_tokens": 463935305.0, + "step": 17933 + }, + { + "epoch": 1.9694706786734022, + "grad_norm": 2.0934972763061523, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.724212646484375, + "num_tokens": 463964995.0, + "step": 17934 + }, + { + "epoch": 1.969580496376016, + "grad_norm": 2.131538152694702, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7173489928245544, + "num_tokens": 463995158.0, + "step": 17935 + }, + { + "epoch": 1.9696903140786295, + "grad_norm": 2.1655654907226562, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7193522453308105, + "num_tokens": 464022617.0, + "step": 17936 + }, + { + "epoch": 1.969800131781243, + "grad_norm": 2.4627134799957275, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.727695107460022, + "num_tokens": 464044568.0, + "step": 17937 + }, + { + "epoch": 1.9699099494838568, + "grad_norm": 2.3887109756469727, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7329883575439453, + "num_tokens": 464066711.0, + "step": 17938 + }, + { + "epoch": 1.9700197671864705, + "grad_norm": 2.007065534591675, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7142012119293213, + "num_tokens": 464096915.0, + "step": 17939 + }, + { + "epoch": 1.970129584889084, + "grad_norm": 2.142180919647217, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7165201902389526, + "num_tokens": 464123541.0, + "step": 17940 + }, + { + "epoch": 1.9702394025916978, + "grad_norm": 2.1007800102233887, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7115744352340698, + "num_tokens": 464151300.0, + "step": 17941 + }, + { + "epoch": 1.9703492202943114, + "grad_norm": 2.1460957527160645, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7132664322853088, + "num_tokens": 464178356.0, + "step": 17942 + }, + { + "epoch": 1.9704590379969251, + "grad_norm": 2.241441249847412, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6944103240966797, + "num_tokens": 464206079.0, + "step": 17943 + }, + { + "epoch": 1.9705688556995389, + "grad_norm": 1.9353256225585938, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6901668310165405, + "num_tokens": 464238899.0, + "step": 17944 + }, + { + "epoch": 1.9706786734021524, + "grad_norm": 2.0556907653808594, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.708977222442627, + "num_tokens": 464269439.0, + "step": 17945 + }, + { + "epoch": 1.970788491104766, + "grad_norm": 2.33553409576416, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.726940929889679, + "num_tokens": 464292768.0, + "step": 17946 + }, + { + "epoch": 1.9708983088073797, + "grad_norm": 2.0513274669647217, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7176005840301514, + "num_tokens": 464322796.0, + "step": 17947 + }, + { + "epoch": 1.9710081265099935, + "grad_norm": 1.9932305812835693, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7156692743301392, + "num_tokens": 464353679.0, + "step": 17948 + }, + { + "epoch": 1.9711179442126072, + "grad_norm": 2.6906843185424805, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7279196381568909, + "num_tokens": 464372732.0, + "step": 17949 + }, + { + "epoch": 1.9712277619152208, + "grad_norm": 1.9009031057357788, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7313039302825928, + "num_tokens": 464405373.0, + "step": 17950 + }, + { + "epoch": 1.9713375796178343, + "grad_norm": 2.3488144874572754, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7110255360603333, + "num_tokens": 464430443.0, + "step": 17951 + }, + { + "epoch": 1.971447397320448, + "grad_norm": 2.143425941467285, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7033129930496216, + "num_tokens": 464457515.0, + "step": 17952 + }, + { + "epoch": 1.9715572150230618, + "grad_norm": 2.647979736328125, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7165980339050293, + "num_tokens": 464477028.0, + "step": 17953 + }, + { + "epoch": 1.9716670327256753, + "grad_norm": 2.158148765563965, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7167705297470093, + "num_tokens": 464502140.0, + "step": 17954 + }, + { + "epoch": 1.9717768504282889, + "grad_norm": 2.046393871307373, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7266772389411926, + "num_tokens": 464530992.0, + "step": 17955 + }, + { + "epoch": 1.9718866681309026, + "grad_norm": 2.0020856857299805, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6921308040618896, + "num_tokens": 464565271.0, + "step": 17956 + }, + { + "epoch": 1.9719964858335164, + "grad_norm": 2.1755688190460205, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7455799579620361, + "num_tokens": 464592186.0, + "step": 17957 + }, + { + "epoch": 1.9721063035361301, + "grad_norm": 2.238443374633789, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7294585704803467, + "num_tokens": 464616573.0, + "step": 17958 + }, + { + "epoch": 1.9722161212387437, + "grad_norm": 2.35180926322937, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7025025486946106, + "num_tokens": 464641020.0, + "step": 17959 + }, + { + "epoch": 1.9723259389413572, + "grad_norm": 2.0469391345977783, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7378782033920288, + "num_tokens": 464671457.0, + "step": 17960 + }, + { + "epoch": 1.972435756643971, + "grad_norm": 1.9728111028671265, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7335910797119141, + "num_tokens": 464701479.0, + "step": 17961 + }, + { + "epoch": 1.9725455743465847, + "grad_norm": 2.0896193981170654, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7095612287521362, + "num_tokens": 464730943.0, + "step": 17962 + }, + { + "epoch": 1.9726553920491985, + "grad_norm": 2.4709112644195557, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7117135524749756, + "num_tokens": 464753566.0, + "step": 17963 + }, + { + "epoch": 1.972765209751812, + "grad_norm": 2.278184652328491, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6979490518569946, + "num_tokens": 464781410.0, + "step": 17964 + }, + { + "epoch": 1.9728750274544256, + "grad_norm": 2.3458099365234375, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7262696027755737, + "num_tokens": 464806484.0, + "step": 17965 + }, + { + "epoch": 1.9729848451570393, + "grad_norm": 2.1781249046325684, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7130343317985535, + "num_tokens": 464833348.0, + "step": 17966 + }, + { + "epoch": 1.973094662859653, + "grad_norm": 2.410680055618286, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7110175490379333, + "num_tokens": 464856986.0, + "step": 17967 + }, + { + "epoch": 1.9732044805622666, + "grad_norm": 1.9728777408599854, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7438764572143555, + "num_tokens": 464887986.0, + "step": 17968 + }, + { + "epoch": 1.9733142982648801, + "grad_norm": 2.565565824508667, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7362260222434998, + "num_tokens": 464907957.0, + "step": 17969 + }, + { + "epoch": 1.973424115967494, + "grad_norm": 2.09452748298645, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7363215684890747, + "num_tokens": 464933222.0, + "step": 17970 + }, + { + "epoch": 1.9735339336701077, + "grad_norm": 2.1903879642486572, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7177824378013611, + "num_tokens": 464959829.0, + "step": 17971 + }, + { + "epoch": 1.9736437513727214, + "grad_norm": 2.323697566986084, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7248519659042358, + "num_tokens": 464983039.0, + "step": 17972 + }, + { + "epoch": 1.973753569075335, + "grad_norm": 2.025787591934204, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7003628015518188, + "num_tokens": 465012388.0, + "step": 17973 + }, + { + "epoch": 1.9738633867779485, + "grad_norm": 2.042306900024414, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.705451250076294, + "num_tokens": 465042863.0, + "step": 17974 + }, + { + "epoch": 1.9739732044805622, + "grad_norm": 2.2547338008880615, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7285997867584229, + "num_tokens": 465066140.0, + "step": 17975 + }, + { + "epoch": 1.974083022183176, + "grad_norm": 2.2139732837677, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7301636934280396, + "num_tokens": 465091795.0, + "step": 17976 + }, + { + "epoch": 1.9741928398857898, + "grad_norm": 2.421485424041748, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7318637371063232, + "num_tokens": 465115843.0, + "step": 17977 + }, + { + "epoch": 1.9743026575884033, + "grad_norm": 2.0006532669067383, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7191557884216309, + "num_tokens": 465143631.0, + "step": 17978 + }, + { + "epoch": 1.9744124752910168, + "grad_norm": 2.165722608566284, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6985576152801514, + "num_tokens": 465170006.0, + "step": 17979 + }, + { + "epoch": 1.9745222929936306, + "grad_norm": 2.1137850284576416, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7197834253311157, + "num_tokens": 465196759.0, + "step": 17980 + }, + { + "epoch": 1.9746321106962443, + "grad_norm": 2.079357624053955, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7468479871749878, + "num_tokens": 465223186.0, + "step": 17981 + }, + { + "epoch": 1.9747419283988579, + "grad_norm": 2.4222307205200195, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.716172456741333, + "num_tokens": 465244970.0, + "step": 17982 + }, + { + "epoch": 1.9748517461014714, + "grad_norm": 2.304187774658203, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7296317219734192, + "num_tokens": 465267666.0, + "step": 17983 + }, + { + "epoch": 1.9749615638040852, + "grad_norm": 1.9315754175186157, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6944819688796997, + "num_tokens": 465300502.0, + "step": 17984 + }, + { + "epoch": 1.975071381506699, + "grad_norm": 2.1145739555358887, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.689390242099762, + "num_tokens": 465330538.0, + "step": 17985 + }, + { + "epoch": 1.9751811992093127, + "grad_norm": 2.1664743423461914, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.721648097038269, + "num_tokens": 465356579.0, + "step": 17986 + }, + { + "epoch": 1.9752910169119262, + "grad_norm": 2.5522937774658203, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7350434064865112, + "num_tokens": 465375455.0, + "step": 17987 + }, + { + "epoch": 1.9754008346145397, + "grad_norm": 2.0715391635894775, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.689134955406189, + "num_tokens": 465406089.0, + "step": 17988 + }, + { + "epoch": 1.9755106523171535, + "grad_norm": 2.397580146789551, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7419155836105347, + "num_tokens": 465428073.0, + "step": 17989 + }, + { + "epoch": 1.9756204700197673, + "grad_norm": 2.122342824935913, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7147152423858643, + "num_tokens": 465454586.0, + "step": 17990 + }, + { + "epoch": 1.9757302877223808, + "grad_norm": 2.231544256210327, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7190462350845337, + "num_tokens": 465479020.0, + "step": 17991 + }, + { + "epoch": 1.9758401054249946, + "grad_norm": 2.465794086456299, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7302337884902954, + "num_tokens": 465499620.0, + "step": 17992 + }, + { + "epoch": 1.975949923127608, + "grad_norm": 2.325087070465088, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7304152250289917, + "num_tokens": 465523489.0, + "step": 17993 + }, + { + "epoch": 1.9760597408302218, + "grad_norm": 2.2613120079040527, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7282127141952515, + "num_tokens": 465550394.0, + "step": 17994 + }, + { + "epoch": 1.9761695585328356, + "grad_norm": 2.380000114440918, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7361921072006226, + "num_tokens": 465574074.0, + "step": 17995 + }, + { + "epoch": 1.9762793762354491, + "grad_norm": 1.797910451889038, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7378149032592773, + "num_tokens": 465610552.0, + "step": 17996 + }, + { + "epoch": 1.9763891939380627, + "grad_norm": 2.312467098236084, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7081217765808105, + "num_tokens": 465638687.0, + "step": 17997 + }, + { + "epoch": 1.9764990116406764, + "grad_norm": 2.6772267818450928, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7214646339416504, + "num_tokens": 465659469.0, + "step": 17998 + }, + { + "epoch": 1.9766088293432902, + "grad_norm": 2.0214455127716064, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7156853675842285, + "num_tokens": 465690690.0, + "step": 17999 + }, + { + "epoch": 1.976718647045904, + "grad_norm": 2.4108200073242188, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7345397472381592, + "num_tokens": 465713521.0, + "step": 18000 + }, + { + "epoch": 1.9768284647485175, + "grad_norm": 2.085792303085327, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7098087072372437, + "num_tokens": 465741454.0, + "step": 18001 + }, + { + "epoch": 1.976938282451131, + "grad_norm": 2.208704710006714, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7212396860122681, + "num_tokens": 465767905.0, + "step": 18002 + }, + { + "epoch": 1.9770481001537448, + "grad_norm": 2.183043956756592, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6943840980529785, + "num_tokens": 465795347.0, + "step": 18003 + }, + { + "epoch": 1.9771579178563585, + "grad_norm": 2.8206264972686768, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7281920909881592, + "num_tokens": 465812978.0, + "step": 18004 + }, + { + "epoch": 1.977267735558972, + "grad_norm": 2.2547249794006348, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7110023498535156, + "num_tokens": 465838833.0, + "step": 18005 + }, + { + "epoch": 1.9773775532615858, + "grad_norm": 2.1728994846343994, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7405635118484497, + "num_tokens": 465865017.0, + "step": 18006 + }, + { + "epoch": 1.9774873709641994, + "grad_norm": 2.291215181350708, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7117813229560852, + "num_tokens": 465887636.0, + "step": 18007 + }, + { + "epoch": 1.977597188666813, + "grad_norm": 1.9997758865356445, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7005338072776794, + "num_tokens": 465917893.0, + "step": 18008 + }, + { + "epoch": 1.9777070063694269, + "grad_norm": 2.049686908721924, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7269171476364136, + "num_tokens": 465946956.0, + "step": 18009 + }, + { + "epoch": 1.9778168240720404, + "grad_norm": 2.316973924636841, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7259370684623718, + "num_tokens": 465971921.0, + "step": 18010 + }, + { + "epoch": 1.977926641774654, + "grad_norm": 2.186582088470459, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.718770444393158, + "num_tokens": 465998156.0, + "step": 18011 + }, + { + "epoch": 1.9780364594772677, + "grad_norm": 2.000467538833618, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.714991569519043, + "num_tokens": 466029346.0, + "step": 18012 + }, + { + "epoch": 1.9781462771798815, + "grad_norm": 1.998225450515747, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7304432392120361, + "num_tokens": 466059698.0, + "step": 18013 + }, + { + "epoch": 1.9782560948824952, + "grad_norm": 2.3389971256256104, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7008033394813538, + "num_tokens": 466084530.0, + "step": 18014 + }, + { + "epoch": 1.9783659125851087, + "grad_norm": 1.9164965152740479, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6939849853515625, + "num_tokens": 466118536.0, + "step": 18015 + }, + { + "epoch": 1.9784757302877223, + "grad_norm": 2.278907537460327, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7320001721382141, + "num_tokens": 466142907.0, + "step": 18016 + }, + { + "epoch": 1.978585547990336, + "grad_norm": 1.9694089889526367, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7242182493209839, + "num_tokens": 466175850.0, + "step": 18017 + }, + { + "epoch": 1.9786953656929498, + "grad_norm": 2.3740696907043457, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.699378490447998, + "num_tokens": 466201036.0, + "step": 18018 + }, + { + "epoch": 1.9788051833955633, + "grad_norm": 2.400695562362671, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7317720651626587, + "num_tokens": 466221700.0, + "step": 18019 + }, + { + "epoch": 1.9789150010981769, + "grad_norm": 2.6221137046813965, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7233211994171143, + "num_tokens": 466241325.0, + "step": 18020 + }, + { + "epoch": 1.9790248188007906, + "grad_norm": 1.9872331619262695, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7142162322998047, + "num_tokens": 466272510.0, + "step": 18021 + }, + { + "epoch": 1.9791346365034044, + "grad_norm": 2.205064535140991, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7177656888961792, + "num_tokens": 466297912.0, + "step": 18022 + }, + { + "epoch": 1.9792444542060181, + "grad_norm": 2.1949880123138428, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7143793106079102, + "num_tokens": 466324169.0, + "step": 18023 + }, + { + "epoch": 1.9793542719086317, + "grad_norm": 2.0883724689483643, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7117363214492798, + "num_tokens": 466351938.0, + "step": 18024 + }, + { + "epoch": 1.9794640896112452, + "grad_norm": 2.1232898235321045, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7128950357437134, + "num_tokens": 466381266.0, + "step": 18025 + }, + { + "epoch": 1.979573907313859, + "grad_norm": 2.091137170791626, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7312471866607666, + "num_tokens": 466408555.0, + "step": 18026 + }, + { + "epoch": 1.9796837250164727, + "grad_norm": 2.490973949432373, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7203403115272522, + "num_tokens": 466429737.0, + "step": 18027 + }, + { + "epoch": 1.9797935427190865, + "grad_norm": 2.2233731746673584, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7233174443244934, + "num_tokens": 466454995.0, + "step": 18028 + }, + { + "epoch": 1.9799033604217, + "grad_norm": 2.0262985229492188, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7075334787368774, + "num_tokens": 466485326.0, + "step": 18029 + }, + { + "epoch": 1.9800131781243135, + "grad_norm": 1.958042860031128, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7011622190475464, + "num_tokens": 466517105.0, + "step": 18030 + }, + { + "epoch": 1.9801229958269273, + "grad_norm": 2.2914061546325684, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7322900295257568, + "num_tokens": 466542822.0, + "step": 18031 + }, + { + "epoch": 1.980232813529541, + "grad_norm": 2.405571460723877, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7254049777984619, + "num_tokens": 466566406.0, + "step": 18032 + }, + { + "epoch": 1.9803426312321546, + "grad_norm": 2.141181707382202, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7226719856262207, + "num_tokens": 466594203.0, + "step": 18033 + }, + { + "epoch": 1.9804524489347681, + "grad_norm": 2.0369088649749756, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7152400612831116, + "num_tokens": 466624685.0, + "step": 18034 + }, + { + "epoch": 1.9805622666373819, + "grad_norm": 2.085783004760742, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7207373380661011, + "num_tokens": 466652801.0, + "step": 18035 + }, + { + "epoch": 1.9806720843399956, + "grad_norm": 2.218390941619873, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7192710638046265, + "num_tokens": 466678910.0, + "step": 18036 + }, + { + "epoch": 1.9807819020426094, + "grad_norm": 2.1485438346862793, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7243642807006836, + "num_tokens": 466707742.0, + "step": 18037 + }, + { + "epoch": 1.980891719745223, + "grad_norm": 2.3798909187316895, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7101939916610718, + "num_tokens": 466730741.0, + "step": 18038 + }, + { + "epoch": 1.9810015374478365, + "grad_norm": 2.1220579147338867, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7111641764640808, + "num_tokens": 466757928.0, + "step": 18039 + }, + { + "epoch": 1.9811113551504502, + "grad_norm": 2.296684503555298, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7161740064620972, + "num_tokens": 466781909.0, + "step": 18040 + }, + { + "epoch": 1.981221172853064, + "grad_norm": 2.336155414581299, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7095416784286499, + "num_tokens": 466807173.0, + "step": 18041 + }, + { + "epoch": 1.9813309905556777, + "grad_norm": 2.3249542713165283, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.730625331401825, + "num_tokens": 466829869.0, + "step": 18042 + }, + { + "epoch": 1.9814408082582913, + "grad_norm": 2.0436935424804688, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7268912196159363, + "num_tokens": 466857696.0, + "step": 18043 + }, + { + "epoch": 1.9815506259609048, + "grad_norm": 2.406280755996704, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7335036396980286, + "num_tokens": 466879323.0, + "step": 18044 + }, + { + "epoch": 1.9816604436635186, + "grad_norm": 2.2660231590270996, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7192791700363159, + "num_tokens": 466904558.0, + "step": 18045 + }, + { + "epoch": 1.9817702613661323, + "grad_norm": 2.4986448287963867, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7221503257751465, + "num_tokens": 466924408.0, + "step": 18046 + }, + { + "epoch": 1.9818800790687459, + "grad_norm": 2.32271146774292, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7466673254966736, + "num_tokens": 466945415.0, + "step": 18047 + }, + { + "epoch": 1.9819898967713594, + "grad_norm": 2.1320903301239014, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7241604328155518, + "num_tokens": 466971936.0, + "step": 18048 + }, + { + "epoch": 1.9820997144739732, + "grad_norm": 2.1857426166534424, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.712210476398468, + "num_tokens": 466999690.0, + "step": 18049 + }, + { + "epoch": 1.982209532176587, + "grad_norm": 2.573208808898926, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7439859509468079, + "num_tokens": 467019527.0, + "step": 18050 + }, + { + "epoch": 1.9823193498792007, + "grad_norm": 2.0457944869995117, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7294968962669373, + "num_tokens": 467047899.0, + "step": 18051 + }, + { + "epoch": 1.9824291675818142, + "grad_norm": 2.5009231567382812, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7324029207229614, + "num_tokens": 467069050.0, + "step": 18052 + }, + { + "epoch": 1.9825389852844277, + "grad_norm": 2.2976276874542236, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7419068813323975, + "num_tokens": 467091624.0, + "step": 18053 + }, + { + "epoch": 1.9826488029870415, + "grad_norm": 2.196281671524048, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.697248637676239, + "num_tokens": 467119004.0, + "step": 18054 + }, + { + "epoch": 1.9827586206896552, + "grad_norm": 2.0869333744049072, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6854488849639893, + "num_tokens": 467150163.0, + "step": 18055 + }, + { + "epoch": 1.9828684383922688, + "grad_norm": 2.02713680267334, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7428895831108093, + "num_tokens": 467180353.0, + "step": 18056 + }, + { + "epoch": 1.9829782560948825, + "grad_norm": 2.379965305328369, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7570786476135254, + "num_tokens": 467200651.0, + "step": 18057 + }, + { + "epoch": 1.983088073797496, + "grad_norm": 2.2161600589752197, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7392452359199524, + "num_tokens": 467225799.0, + "step": 18058 + }, + { + "epoch": 1.9831978915001098, + "grad_norm": 2.8363962173461914, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7280161380767822, + "num_tokens": 467243075.0, + "step": 18059 + }, + { + "epoch": 1.9833077092027236, + "grad_norm": 2.0767924785614014, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.698296308517456, + "num_tokens": 467272737.0, + "step": 18060 + }, + { + "epoch": 1.9834175269053371, + "grad_norm": 1.959243655204773, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7321960926055908, + "num_tokens": 467301245.0, + "step": 18061 + }, + { + "epoch": 1.9835273446079507, + "grad_norm": 1.9810935258865356, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7053809762001038, + "num_tokens": 467331176.0, + "step": 18062 + }, + { + "epoch": 1.9836371623105644, + "grad_norm": 2.4781062602996826, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7480830550193787, + "num_tokens": 467350422.0, + "step": 18063 + }, + { + "epoch": 1.9837469800131782, + "grad_norm": 2.1311357021331787, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7320375442504883, + "num_tokens": 467377058.0, + "step": 18064 + }, + { + "epoch": 1.983856797715792, + "grad_norm": 2.1409695148468018, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7042946815490723, + "num_tokens": 467405378.0, + "step": 18065 + }, + { + "epoch": 1.9839666154184055, + "grad_norm": 2.1746737957000732, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7180653214454651, + "num_tokens": 467433938.0, + "step": 18066 + }, + { + "epoch": 1.984076433121019, + "grad_norm": 2.4451208114624023, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7006096243858337, + "num_tokens": 467456400.0, + "step": 18067 + }, + { + "epoch": 1.9841862508236328, + "grad_norm": 2.0627381801605225, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7140493988990784, + "num_tokens": 467485415.0, + "step": 18068 + }, + { + "epoch": 1.9842960685262465, + "grad_norm": 2.0210514068603516, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6900218725204468, + "num_tokens": 467516613.0, + "step": 18069 + }, + { + "epoch": 1.98440588622886, + "grad_norm": 2.159592628479004, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7236756086349487, + "num_tokens": 467541554.0, + "step": 18070 + }, + { + "epoch": 1.9845157039314738, + "grad_norm": 2.4697234630584717, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7371264696121216, + "num_tokens": 467562551.0, + "step": 18071 + }, + { + "epoch": 1.9846255216340873, + "grad_norm": 2.2788307666778564, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.714976966381073, + "num_tokens": 467587875.0, + "step": 18072 + }, + { + "epoch": 1.984735339336701, + "grad_norm": 2.414853572845459, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7381976842880249, + "num_tokens": 467609473.0, + "step": 18073 + }, + { + "epoch": 1.9848451570393149, + "grad_norm": 2.061340093612671, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7316209673881531, + "num_tokens": 467638318.0, + "step": 18074 + }, + { + "epoch": 1.9849549747419284, + "grad_norm": 2.0486114025115967, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6933335065841675, + "num_tokens": 467669113.0, + "step": 18075 + }, + { + "epoch": 1.985064792444542, + "grad_norm": 2.279872417449951, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7028051614761353, + "num_tokens": 467695282.0, + "step": 18076 + }, + { + "epoch": 1.9851746101471557, + "grad_norm": 2.3864355087280273, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7234201431274414, + "num_tokens": 467718225.0, + "step": 18077 + }, + { + "epoch": 1.9852844278497694, + "grad_norm": 2.504497766494751, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.6985131502151489, + "num_tokens": 467741915.0, + "step": 18078 + }, + { + "epoch": 1.9853942455523832, + "grad_norm": 2.4363505840301514, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.730530858039856, + "num_tokens": 467762792.0, + "step": 18079 + }, + { + "epoch": 1.9855040632549967, + "grad_norm": 2.454444646835327, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7195987701416016, + "num_tokens": 467784467.0, + "step": 18080 + }, + { + "epoch": 1.9856138809576103, + "grad_norm": 2.253037452697754, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7318687438964844, + "num_tokens": 467809037.0, + "step": 18081 + }, + { + "epoch": 1.985723698660224, + "grad_norm": 2.3736135959625244, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7351435422897339, + "num_tokens": 467833099.0, + "step": 18082 + }, + { + "epoch": 1.9858335163628378, + "grad_norm": 2.2341811656951904, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7150189876556396, + "num_tokens": 467861147.0, + "step": 18083 + }, + { + "epoch": 1.9859433340654513, + "grad_norm": 2.288645029067993, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7114482522010803, + "num_tokens": 467886383.0, + "step": 18084 + }, + { + "epoch": 1.9860531517680649, + "grad_norm": 2.7084410190582275, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7470105886459351, + "num_tokens": 467904883.0, + "step": 18085 + }, + { + "epoch": 1.9861629694706786, + "grad_norm": 2.47530460357666, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7212549448013306, + "num_tokens": 467927722.0, + "step": 18086 + }, + { + "epoch": 1.9862727871732924, + "grad_norm": 2.21567702293396, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7400346994400024, + "num_tokens": 467954533.0, + "step": 18087 + }, + { + "epoch": 1.9863826048759061, + "grad_norm": 2.0768871307373047, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.748803973197937, + "num_tokens": 467980580.0, + "step": 18088 + }, + { + "epoch": 1.9864924225785197, + "grad_norm": 2.509873390197754, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7418578267097473, + "num_tokens": 468001046.0, + "step": 18089 + }, + { + "epoch": 1.9866022402811332, + "grad_norm": 2.4616763591766357, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7343684434890747, + "num_tokens": 468022546.0, + "step": 18090 + }, + { + "epoch": 1.986712057983747, + "grad_norm": 2.490328788757324, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7200484275817871, + "num_tokens": 468045996.0, + "step": 18091 + }, + { + "epoch": 1.9868218756863607, + "grad_norm": 2.1120619773864746, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7384021282196045, + "num_tokens": 468073492.0, + "step": 18092 + }, + { + "epoch": 1.9869316933889745, + "grad_norm": 2.4688382148742676, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7080877423286438, + "num_tokens": 468097229.0, + "step": 18093 + }, + { + "epoch": 1.987041511091588, + "grad_norm": 2.121652126312256, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.6993564367294312, + "num_tokens": 468124449.0, + "step": 18094 + }, + { + "epoch": 1.9871513287942015, + "grad_norm": 2.108196973800659, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7042546272277832, + "num_tokens": 468153002.0, + "step": 18095 + }, + { + "epoch": 1.9872611464968153, + "grad_norm": 2.3444292545318604, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7277325391769409, + "num_tokens": 468175728.0, + "step": 18096 + }, + { + "epoch": 1.987370964199429, + "grad_norm": 2.0507304668426514, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.68956458568573, + "num_tokens": 468208067.0, + "step": 18097 + }, + { + "epoch": 1.9874807819020426, + "grad_norm": 2.4969396591186523, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7351341247558594, + "num_tokens": 468229495.0, + "step": 18098 + }, + { + "epoch": 1.9875905996046561, + "grad_norm": 2.207150936126709, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7312385439872742, + "num_tokens": 468256501.0, + "step": 18099 + }, + { + "epoch": 1.9877004173072699, + "grad_norm": 2.3277885913848877, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7407259941101074, + "num_tokens": 468280381.0, + "step": 18100 + }, + { + "epoch": 1.9878102350098836, + "grad_norm": 1.9917367696762085, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7125683426856995, + "num_tokens": 468311940.0, + "step": 18101 + }, + { + "epoch": 1.9879200527124974, + "grad_norm": 2.501936435699463, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.721792459487915, + "num_tokens": 468335383.0, + "step": 18102 + }, + { + "epoch": 1.988029870415111, + "grad_norm": 2.3503143787384033, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7070170640945435, + "num_tokens": 468358251.0, + "step": 18103 + }, + { + "epoch": 1.9881396881177245, + "grad_norm": 2.3087096214294434, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7505835890769958, + "num_tokens": 468381560.0, + "step": 18104 + }, + { + "epoch": 1.9882495058203382, + "grad_norm": 2.4379312992095947, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7078681588172913, + "num_tokens": 468405235.0, + "step": 18105 + }, + { + "epoch": 1.988359323522952, + "grad_norm": 2.1195006370544434, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7279796600341797, + "num_tokens": 468433779.0, + "step": 18106 + }, + { + "epoch": 1.9884691412255655, + "grad_norm": 1.921704888343811, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7324438691139221, + "num_tokens": 468465278.0, + "step": 18107 + }, + { + "epoch": 1.9885789589281793, + "grad_norm": 2.4415512084960938, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7364566326141357, + "num_tokens": 468489088.0, + "step": 18108 + }, + { + "epoch": 1.9886887766307928, + "grad_norm": 2.091428279876709, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.71091628074646, + "num_tokens": 468521256.0, + "step": 18109 + }, + { + "epoch": 1.9887985943334066, + "grad_norm": 2.4748265743255615, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7369762659072876, + "num_tokens": 468541124.0, + "step": 18110 + }, + { + "epoch": 1.9889084120360203, + "grad_norm": 2.3608546257019043, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7243036031723022, + "num_tokens": 468565961.0, + "step": 18111 + }, + { + "epoch": 1.9890182297386338, + "grad_norm": 1.996829867362976, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7109122276306152, + "num_tokens": 468595579.0, + "step": 18112 + }, + { + "epoch": 1.9891280474412474, + "grad_norm": 2.235934257507324, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7239828109741211, + "num_tokens": 468621118.0, + "step": 18113 + }, + { + "epoch": 1.9892378651438611, + "grad_norm": 2.5059006214141846, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.73163902759552, + "num_tokens": 468641545.0, + "step": 18114 + }, + { + "epoch": 1.989347682846475, + "grad_norm": 2.1376099586486816, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7195665836334229, + "num_tokens": 468668558.0, + "step": 18115 + }, + { + "epoch": 1.9894575005490887, + "grad_norm": 2.105222463607788, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7015994787216187, + "num_tokens": 468697598.0, + "step": 18116 + }, + { + "epoch": 1.9895673182517022, + "grad_norm": 2.0886058807373047, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7095394134521484, + "num_tokens": 468726401.0, + "step": 18117 + }, + { + "epoch": 1.9896771359543157, + "grad_norm": 2.473633289337158, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7091453075408936, + "num_tokens": 468748943.0, + "step": 18118 + }, + { + "epoch": 1.9897869536569295, + "grad_norm": 2.0092055797576904, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7162888050079346, + "num_tokens": 468777685.0, + "step": 18119 + }, + { + "epoch": 1.9898967713595432, + "grad_norm": 2.394756317138672, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7239670753479004, + "num_tokens": 468800301.0, + "step": 18120 + }, + { + "epoch": 1.9900065890621568, + "grad_norm": 2.104583978652954, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7247692346572876, + "num_tokens": 468826461.0, + "step": 18121 + }, + { + "epoch": 1.9901164067647705, + "grad_norm": 1.9626554250717163, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6954556107521057, + "num_tokens": 468858553.0, + "step": 18122 + }, + { + "epoch": 1.990226224467384, + "grad_norm": 1.9973386526107788, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.6999631524085999, + "num_tokens": 468888908.0, + "step": 18123 + }, + { + "epoch": 1.9903360421699978, + "grad_norm": 1.9274604320526123, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6915693283081055, + "num_tokens": 468922491.0, + "step": 18124 + }, + { + "epoch": 1.9904458598726116, + "grad_norm": 1.9839991331100464, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7105380296707153, + "num_tokens": 468954283.0, + "step": 18125 + }, + { + "epoch": 1.9905556775752251, + "grad_norm": 2.4209115505218506, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.713218092918396, + "num_tokens": 468977253.0, + "step": 18126 + }, + { + "epoch": 1.9906654952778386, + "grad_norm": 2.444835901260376, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7323799729347229, + "num_tokens": 468997728.0, + "step": 18127 + }, + { + "epoch": 1.9907753129804524, + "grad_norm": 2.285737991333008, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7229001522064209, + "num_tokens": 469022017.0, + "step": 18128 + }, + { + "epoch": 1.9908851306830662, + "grad_norm": 2.2544684410095215, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7239804267883301, + "num_tokens": 469045797.0, + "step": 18129 + }, + { + "epoch": 1.99099494838568, + "grad_norm": 2.136789321899414, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6960373520851135, + "num_tokens": 469077174.0, + "step": 18130 + }, + { + "epoch": 1.9911047660882935, + "grad_norm": 2.0869247913360596, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6940606236457825, + "num_tokens": 469104610.0, + "step": 18131 + }, + { + "epoch": 1.991214583790907, + "grad_norm": 1.9649357795715332, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7131636142730713, + "num_tokens": 469136746.0, + "step": 18132 + }, + { + "epoch": 1.9913244014935207, + "grad_norm": 2.0082578659057617, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7093030214309692, + "num_tokens": 469167149.0, + "step": 18133 + }, + { + "epoch": 1.9914342191961345, + "grad_norm": 2.0828211307525635, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7141635417938232, + "num_tokens": 469194948.0, + "step": 18134 + }, + { + "epoch": 1.991544036898748, + "grad_norm": 2.124669075012207, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7149522304534912, + "num_tokens": 469220938.0, + "step": 18135 + }, + { + "epoch": 1.9916538546013618, + "grad_norm": 1.8605234622955322, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7243062853813171, + "num_tokens": 469257117.0, + "step": 18136 + }, + { + "epoch": 1.9917636723039753, + "grad_norm": 2.1614396572113037, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.718429684638977, + "num_tokens": 469284383.0, + "step": 18137 + }, + { + "epoch": 1.991873490006589, + "grad_norm": 2.363161087036133, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7258964776992798, + "num_tokens": 469307692.0, + "step": 18138 + }, + { + "epoch": 1.9919833077092028, + "grad_norm": 2.3747096061706543, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7187424898147583, + "num_tokens": 469330147.0, + "step": 18139 + }, + { + "epoch": 1.9920931254118164, + "grad_norm": 2.2005317211151123, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7223447561264038, + "num_tokens": 469355986.0, + "step": 18140 + }, + { + "epoch": 1.99220294311443, + "grad_norm": 2.154754638671875, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7240522503852844, + "num_tokens": 469384324.0, + "step": 18141 + }, + { + "epoch": 1.9923127608170437, + "grad_norm": 1.939139485359192, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7256106734275818, + "num_tokens": 469414795.0, + "step": 18142 + }, + { + "epoch": 1.9924225785196574, + "grad_norm": 2.402238607406616, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7123122811317444, + "num_tokens": 469437569.0, + "step": 18143 + }, + { + "epoch": 1.9925323962222712, + "grad_norm": 2.32084321975708, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7442439198493958, + "num_tokens": 469461788.0, + "step": 18144 + }, + { + "epoch": 1.9926422139248847, + "grad_norm": 2.12699556350708, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.712012529373169, + "num_tokens": 469489058.0, + "step": 18145 + }, + { + "epoch": 1.9927520316274983, + "grad_norm": 2.0741608142852783, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6926367282867432, + "num_tokens": 469518606.0, + "step": 18146 + }, + { + "epoch": 1.992861849330112, + "grad_norm": 2.2781028747558594, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7232842445373535, + "num_tokens": 469542224.0, + "step": 18147 + }, + { + "epoch": 1.9929716670327258, + "grad_norm": 2.069746255874634, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7157590389251709, + "num_tokens": 469571020.0, + "step": 18148 + }, + { + "epoch": 1.9930814847353393, + "grad_norm": 2.0766022205352783, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7264102697372437, + "num_tokens": 469598688.0, + "step": 18149 + }, + { + "epoch": 1.9931913024379528, + "grad_norm": 2.0073721408843994, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.72801673412323, + "num_tokens": 469628287.0, + "step": 18150 + }, + { + "epoch": 1.9933011201405666, + "grad_norm": 1.9416531324386597, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7074329853057861, + "num_tokens": 469661848.0, + "step": 18151 + }, + { + "epoch": 1.9934109378431804, + "grad_norm": 2.0613465309143066, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7285186648368835, + "num_tokens": 469691358.0, + "step": 18152 + }, + { + "epoch": 1.993520755545794, + "grad_norm": 2.2353010177612305, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7217659950256348, + "num_tokens": 469719859.0, + "step": 18153 + }, + { + "epoch": 1.9936305732484076, + "grad_norm": 2.2425997257232666, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7071657776832581, + "num_tokens": 469744208.0, + "step": 18154 + }, + { + "epoch": 1.9937403909510212, + "grad_norm": 2.0455894470214844, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7037521600723267, + "num_tokens": 469775148.0, + "step": 18155 + }, + { + "epoch": 1.993850208653635, + "grad_norm": 2.5425305366516113, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7559879422187805, + "num_tokens": 469796709.0, + "step": 18156 + }, + { + "epoch": 1.9939600263562487, + "grad_norm": 2.5174224376678467, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7167543172836304, + "num_tokens": 469819795.0, + "step": 18157 + }, + { + "epoch": 1.9940698440588625, + "grad_norm": 2.215484380722046, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7411658763885498, + "num_tokens": 469844011.0, + "step": 18158 + }, + { + "epoch": 1.994179661761476, + "grad_norm": 2.325443744659424, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7206352949142456, + "num_tokens": 469867553.0, + "step": 18159 + }, + { + "epoch": 1.9942894794640895, + "grad_norm": 2.1650068759918213, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7249208688735962, + "num_tokens": 469896905.0, + "step": 18160 + }, + { + "epoch": 1.9943992971667033, + "grad_norm": 2.3951447010040283, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7041587829589844, + "num_tokens": 469920200.0, + "step": 18161 + }, + { + "epoch": 1.994509114869317, + "grad_norm": 2.478696346282959, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7192670702934265, + "num_tokens": 469944573.0, + "step": 18162 + }, + { + "epoch": 1.9946189325719306, + "grad_norm": 2.3162903785705566, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7220861911773682, + "num_tokens": 469967220.0, + "step": 18163 + }, + { + "epoch": 1.994728750274544, + "grad_norm": 2.533501625061035, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.731282114982605, + "num_tokens": 469987942.0, + "step": 18164 + }, + { + "epoch": 1.9948385679771579, + "grad_norm": 2.134929895401001, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7190655469894409, + "num_tokens": 470017012.0, + "step": 18165 + }, + { + "epoch": 1.9949483856797716, + "grad_norm": 2.1883647441864014, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.725799560546875, + "num_tokens": 470042843.0, + "step": 18166 + }, + { + "epoch": 1.9950582033823854, + "grad_norm": 2.1393609046936035, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.743312656879425, + "num_tokens": 470069002.0, + "step": 18167 + }, + { + "epoch": 1.995168021084999, + "grad_norm": 2.455528736114502, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7340452671051025, + "num_tokens": 470089815.0, + "step": 18168 + }, + { + "epoch": 1.9952778387876124, + "grad_norm": 1.9959136247634888, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7084417343139648, + "num_tokens": 470121345.0, + "step": 18169 + }, + { + "epoch": 1.9953876564902262, + "grad_norm": 2.091294288635254, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7330983877182007, + "num_tokens": 470148531.0, + "step": 18170 + }, + { + "epoch": 1.99549747419284, + "grad_norm": 2.300170660018921, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6994057893753052, + "num_tokens": 470174290.0, + "step": 18171 + }, + { + "epoch": 1.9956072918954535, + "grad_norm": 2.1856441497802734, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7157562971115112, + "num_tokens": 470201166.0, + "step": 18172 + }, + { + "epoch": 1.9957171095980673, + "grad_norm": 2.481158971786499, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.722055971622467, + "num_tokens": 470222186.0, + "step": 18173 + }, + { + "epoch": 1.9958269273006808, + "grad_norm": 1.9640482664108276, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7363085150718689, + "num_tokens": 470253168.0, + "step": 18174 + }, + { + "epoch": 1.9959367450032945, + "grad_norm": 2.2654566764831543, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7168625593185425, + "num_tokens": 470281135.0, + "step": 18175 + }, + { + "epoch": 1.9960465627059083, + "grad_norm": 2.212183952331543, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7158635854721069, + "num_tokens": 470307947.0, + "step": 18176 + }, + { + "epoch": 1.9961563804085218, + "grad_norm": 2.119208812713623, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7504809498786926, + "num_tokens": 470335331.0, + "step": 18177 + }, + { + "epoch": 1.9962661981111354, + "grad_norm": 2.0358285903930664, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7115769386291504, + "num_tokens": 470365413.0, + "step": 18178 + }, + { + "epoch": 1.9963760158137491, + "grad_norm": 2.0055723190307617, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7033277750015259, + "num_tokens": 470396737.0, + "step": 18179 + }, + { + "epoch": 1.9964858335163629, + "grad_norm": 2.2740304470062256, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7128844857215881, + "num_tokens": 470422654.0, + "step": 18180 + }, + { + "epoch": 1.9965956512189766, + "grad_norm": 2.1320691108703613, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7284084558486938, + "num_tokens": 470449058.0, + "step": 18181 + }, + { + "epoch": 1.9967054689215902, + "grad_norm": 2.4077746868133545, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7417744994163513, + "num_tokens": 470471733.0, + "step": 18182 + }, + { + "epoch": 1.9968152866242037, + "grad_norm": 2.419528007507324, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7184684872627258, + "num_tokens": 470494067.0, + "step": 18183 + }, + { + "epoch": 1.9969251043268175, + "grad_norm": 2.1575398445129395, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7177149057388306, + "num_tokens": 470521583.0, + "step": 18184 + }, + { + "epoch": 1.9970349220294312, + "grad_norm": 2.2857987880706787, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7156966924667358, + "num_tokens": 470545232.0, + "step": 18185 + }, + { + "epoch": 1.9971447397320448, + "grad_norm": 2.0656206607818604, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.725106954574585, + "num_tokens": 470573581.0, + "step": 18186 + }, + { + "epoch": 1.9972545574346585, + "grad_norm": 2.5085997581481934, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7277947664260864, + "num_tokens": 470594915.0, + "step": 18187 + }, + { + "epoch": 1.997364375137272, + "grad_norm": 2.2564923763275146, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7244240641593933, + "num_tokens": 470619696.0, + "step": 18188 + }, + { + "epoch": 1.9974741928398858, + "grad_norm": 2.1245880126953125, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6968845725059509, + "num_tokens": 470648134.0, + "step": 18189 + }, + { + "epoch": 1.9975840105424996, + "grad_norm": 2.244710922241211, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7338728904724121, + "num_tokens": 470672694.0, + "step": 18190 + }, + { + "epoch": 1.997693828245113, + "grad_norm": 2.161907196044922, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7556641101837158, + "num_tokens": 470698698.0, + "step": 18191 + }, + { + "epoch": 1.9978036459477266, + "grad_norm": 2.1004889011383057, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7297037243843079, + "num_tokens": 470725993.0, + "step": 18192 + }, + { + "epoch": 1.9979134636503404, + "grad_norm": 2.1902551651000977, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7149646878242493, + "num_tokens": 470751735.0, + "step": 18193 + }, + { + "epoch": 1.9980232813529542, + "grad_norm": 3.086198568344116, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7386568784713745, + "num_tokens": 470768977.0, + "step": 18194 + }, + { + "epoch": 1.998133099055568, + "grad_norm": 2.0076417922973633, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7146100997924805, + "num_tokens": 470797383.0, + "step": 18195 + }, + { + "epoch": 1.9982429167581814, + "grad_norm": 2.418694257736206, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7469961643218994, + "num_tokens": 470818262.0, + "step": 18196 + }, + { + "epoch": 1.998352734460795, + "grad_norm": 2.2322428226470947, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7085614204406738, + "num_tokens": 470845484.0, + "step": 18197 + }, + { + "epoch": 1.9984625521634087, + "grad_norm": 2.334512948989868, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7012471556663513, + "num_tokens": 470870283.0, + "step": 18198 + }, + { + "epoch": 1.9985723698660225, + "grad_norm": 2.4675724506378174, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7310022115707397, + "num_tokens": 470891092.0, + "step": 18199 + }, + { + "epoch": 1.998682187568636, + "grad_norm": 2.1709694862365723, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.72735196352005, + "num_tokens": 470916592.0, + "step": 18200 + }, + { + "epoch": 1.9987920052712496, + "grad_norm": 2.195430040359497, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7398532032966614, + "num_tokens": 470943251.0, + "step": 18201 + }, + { + "epoch": 1.9989018229738633, + "grad_norm": 2.211091995239258, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7352931499481201, + "num_tokens": 470970150.0, + "step": 18202 + }, + { + "epoch": 1.999011640676477, + "grad_norm": 2.0058977603912354, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.694603443145752, + "num_tokens": 471001184.0, + "step": 18203 + }, + { + "epoch": 1.9991214583790908, + "grad_norm": 2.305497646331787, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7247178554534912, + "num_tokens": 471025144.0, + "step": 18204 + }, + { + "epoch": 1.9992312760817044, + "grad_norm": 1.9162477254867554, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7055116891860962, + "num_tokens": 471059042.0, + "step": 18205 + }, + { + "epoch": 1.999341093784318, + "grad_norm": 1.994235873222351, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6917545795440674, + "num_tokens": 471090984.0, + "step": 18206 + }, + { + "epoch": 1.9994509114869317, + "grad_norm": 2.420888900756836, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7326894998550415, + "num_tokens": 471111508.0, + "step": 18207 + }, + { + "epoch": 1.9995607291895454, + "grad_norm": 2.268609046936035, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7100446224212646, + "num_tokens": 471136394.0, + "step": 18208 + }, + { + "epoch": 1.9996705468921592, + "grad_norm": 2.195418357849121, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7172094583511353, + "num_tokens": 471164069.0, + "step": 18209 + }, + { + "epoch": 1.9997803645947727, + "grad_norm": 2.302124500274658, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7207319736480713, + "num_tokens": 471189023.0, + "step": 18210 + }, + { + "epoch": 1.9998901822973862, + "grad_norm": 2.4794256687164307, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7356683611869812, + "num_tokens": 471209443.0, + "step": 18211 + }, + { + "epoch": 2.0, + "grad_norm": 2.2312679290771484, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7072192430496216, + "num_tokens": 471235014.0, + "step": 18212 + }, + { + "epoch": 2.0001098177026138, + "grad_norm": 2.173426628112793, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7306240797042847, + "num_tokens": 471260654.0, + "step": 18213 + }, + { + "epoch": 2.0002196354052275, + "grad_norm": 2.0597143173217773, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7293570637702942, + "num_tokens": 471286421.0, + "step": 18214 + }, + { + "epoch": 2.000329453107841, + "grad_norm": 2.2072017192840576, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7293969392776489, + "num_tokens": 471310258.0, + "step": 18215 + }, + { + "epoch": 2.0004392708104546, + "grad_norm": 2.0681965351104736, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.722846508026123, + "num_tokens": 471338723.0, + "step": 18216 + }, + { + "epoch": 2.0005490885130683, + "grad_norm": 2.0354135036468506, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7307080626487732, + "num_tokens": 471369007.0, + "step": 18217 + }, + { + "epoch": 2.000658906215682, + "grad_norm": 2.4186737537384033, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.745942234992981, + "num_tokens": 471390448.0, + "step": 18218 + }, + { + "epoch": 2.0007687239182954, + "grad_norm": 2.0117762088775635, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7208517789840698, + "num_tokens": 471420002.0, + "step": 18219 + }, + { + "epoch": 2.000878541620909, + "grad_norm": 2.489237070083618, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7270966172218323, + "num_tokens": 471442498.0, + "step": 18220 + }, + { + "epoch": 2.000988359323523, + "grad_norm": 2.5084950923919678, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.735218346118927, + "num_tokens": 471463307.0, + "step": 18221 + }, + { + "epoch": 2.0010981770261367, + "grad_norm": 2.295192003250122, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7555761337280273, + "num_tokens": 471487919.0, + "step": 18222 + }, + { + "epoch": 2.0012079947287504, + "grad_norm": 2.192849636077881, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7587074041366577, + "num_tokens": 471514828.0, + "step": 18223 + }, + { + "epoch": 2.0013178124313638, + "grad_norm": 2.4659929275512695, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7571322917938232, + "num_tokens": 471537025.0, + "step": 18224 + }, + { + "epoch": 2.0014276301339775, + "grad_norm": 2.492988109588623, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7259198427200317, + "num_tokens": 471557930.0, + "step": 18225 + }, + { + "epoch": 2.0015374478365913, + "grad_norm": 2.256666421890259, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7219468355178833, + "num_tokens": 471582641.0, + "step": 18226 + }, + { + "epoch": 2.001647265539205, + "grad_norm": 2.427921772003174, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7573072910308838, + "num_tokens": 471604738.0, + "step": 18227 + }, + { + "epoch": 2.001757083241819, + "grad_norm": 2.215484857559204, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7037914991378784, + "num_tokens": 471634425.0, + "step": 18228 + }, + { + "epoch": 2.001866900944432, + "grad_norm": 2.2351744174957275, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7314720153808594, + "num_tokens": 471660871.0, + "step": 18229 + }, + { + "epoch": 2.001976718647046, + "grad_norm": 2.1705210208892822, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7300629615783691, + "num_tokens": 471688335.0, + "step": 18230 + }, + { + "epoch": 2.0020865363496596, + "grad_norm": 2.100332260131836, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7094161510467529, + "num_tokens": 471718337.0, + "step": 18231 + }, + { + "epoch": 2.0021963540522734, + "grad_norm": 1.9536528587341309, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7160362601280212, + "num_tokens": 471751966.0, + "step": 18232 + }, + { + "epoch": 2.0023061717548867, + "grad_norm": 2.4976396560668945, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7521703839302063, + "num_tokens": 471773386.0, + "step": 18233 + }, + { + "epoch": 2.0024159894575004, + "grad_norm": 2.4614017009735107, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7351214289665222, + "num_tokens": 471795477.0, + "step": 18234 + }, + { + "epoch": 2.002525807160114, + "grad_norm": 2.0619826316833496, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7247229814529419, + "num_tokens": 471825957.0, + "step": 18235 + }, + { + "epoch": 2.002635624862728, + "grad_norm": 2.204002618789673, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7228657603263855, + "num_tokens": 471853767.0, + "step": 18236 + }, + { + "epoch": 2.0027454425653417, + "grad_norm": 2.2304821014404297, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7182474136352539, + "num_tokens": 471883082.0, + "step": 18237 + }, + { + "epoch": 2.002855260267955, + "grad_norm": 2.2194769382476807, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7381441593170166, + "num_tokens": 471910218.0, + "step": 18238 + }, + { + "epoch": 2.0029650779705688, + "grad_norm": 2.270263195037842, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7270422577857971, + "num_tokens": 471936993.0, + "step": 18239 + }, + { + "epoch": 2.0030748956731825, + "grad_norm": 2.2664363384246826, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7051222920417786, + "num_tokens": 471964226.0, + "step": 18240 + }, + { + "epoch": 2.0031847133757963, + "grad_norm": 2.411496162414551, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7324239015579224, + "num_tokens": 471986803.0, + "step": 18241 + }, + { + "epoch": 2.00329453107841, + "grad_norm": 2.461963176727295, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.730678915977478, + "num_tokens": 472010234.0, + "step": 18242 + }, + { + "epoch": 2.0034043487810234, + "grad_norm": 2.58870530128479, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7359256148338318, + "num_tokens": 472032126.0, + "step": 18243 + }, + { + "epoch": 2.003514166483637, + "grad_norm": 2.3475325107574463, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7254531383514404, + "num_tokens": 472057020.0, + "step": 18244 + }, + { + "epoch": 2.003623984186251, + "grad_norm": 2.516436815261841, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7220667004585266, + "num_tokens": 472079754.0, + "step": 18245 + }, + { + "epoch": 2.0037338018888646, + "grad_norm": 2.097775459289551, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7503405809402466, + "num_tokens": 472107357.0, + "step": 18246 + }, + { + "epoch": 2.003843619591478, + "grad_norm": 2.4013683795928955, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7399492263793945, + "num_tokens": 472132314.0, + "step": 18247 + }, + { + "epoch": 2.0039534372940917, + "grad_norm": 2.2907915115356445, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7223055362701416, + "num_tokens": 472159056.0, + "step": 18248 + }, + { + "epoch": 2.0040632549967055, + "grad_norm": 2.18155574798584, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7487958669662476, + "num_tokens": 472187610.0, + "step": 18249 + }, + { + "epoch": 2.004173072699319, + "grad_norm": 2.437145471572876, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7359728813171387, + "num_tokens": 472211027.0, + "step": 18250 + }, + { + "epoch": 2.004282890401933, + "grad_norm": 2.1542422771453857, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7284892797470093, + "num_tokens": 472240380.0, + "step": 18251 + }, + { + "epoch": 2.0043927081045463, + "grad_norm": 2.0597527027130127, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7511681318283081, + "num_tokens": 472270893.0, + "step": 18252 + }, + { + "epoch": 2.00450252580716, + "grad_norm": 2.1834967136383057, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7490110993385315, + "num_tokens": 472296851.0, + "step": 18253 + }, + { + "epoch": 2.004612343509774, + "grad_norm": 2.273242950439453, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7303585410118103, + "num_tokens": 472324270.0, + "step": 18254 + }, + { + "epoch": 2.0047221612123876, + "grad_norm": 2.216029405593872, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7400856614112854, + "num_tokens": 472350052.0, + "step": 18255 + }, + { + "epoch": 2.0048319789150013, + "grad_norm": 2.2266223430633545, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.731438398361206, + "num_tokens": 472377760.0, + "step": 18256 + }, + { + "epoch": 2.0049417966176146, + "grad_norm": 1.9236804246902466, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7235533595085144, + "num_tokens": 472413478.0, + "step": 18257 + }, + { + "epoch": 2.0050516143202284, + "grad_norm": 2.2718288898468018, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7034204006195068, + "num_tokens": 472441679.0, + "step": 18258 + }, + { + "epoch": 2.005161432022842, + "grad_norm": 2.2707741260528564, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7475602030754089, + "num_tokens": 472467174.0, + "step": 18259 + }, + { + "epoch": 2.005271249725456, + "grad_norm": 2.3261656761169434, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7416132688522339, + "num_tokens": 472491450.0, + "step": 18260 + }, + { + "epoch": 2.005381067428069, + "grad_norm": 2.5605154037475586, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7502807378768921, + "num_tokens": 472513867.0, + "step": 18261 + }, + { + "epoch": 2.005490885130683, + "grad_norm": 2.288273334503174, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7406754493713379, + "num_tokens": 472538909.0, + "step": 18262 + }, + { + "epoch": 2.0056007028332967, + "grad_norm": 1.8930599689483643, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7232592105865479, + "num_tokens": 472573516.0, + "step": 18263 + }, + { + "epoch": 2.0057105205359105, + "grad_norm": 2.1794497966766357, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7172406315803528, + "num_tokens": 472604040.0, + "step": 18264 + }, + { + "epoch": 2.0058203382385242, + "grad_norm": 2.335312843322754, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7393264770507812, + "num_tokens": 472628842.0, + "step": 18265 + }, + { + "epoch": 2.0059301559411375, + "grad_norm": 2.430297613143921, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7098554372787476, + "num_tokens": 472653793.0, + "step": 18266 + }, + { + "epoch": 2.0060399736437513, + "grad_norm": 2.252120018005371, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7059729099273682, + "num_tokens": 472682750.0, + "step": 18267 + }, + { + "epoch": 2.006149791346365, + "grad_norm": 2.2605535984039307, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7056320309638977, + "num_tokens": 472708280.0, + "step": 18268 + }, + { + "epoch": 2.006259609048979, + "grad_norm": 2.3277344703674316, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7208232879638672, + "num_tokens": 472732535.0, + "step": 18269 + }, + { + "epoch": 2.0063694267515926, + "grad_norm": 2.1260154247283936, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7110545039176941, + "num_tokens": 472761119.0, + "step": 18270 + }, + { + "epoch": 2.006479244454206, + "grad_norm": 2.6940979957580566, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7427883148193359, + "num_tokens": 472782492.0, + "step": 18271 + }, + { + "epoch": 2.0065890621568196, + "grad_norm": 2.2270376682281494, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7232188582420349, + "num_tokens": 472809592.0, + "step": 18272 + }, + { + "epoch": 2.0066988798594334, + "grad_norm": 2.1456212997436523, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7268832921981812, + "num_tokens": 472838346.0, + "step": 18273 + }, + { + "epoch": 2.006808697562047, + "grad_norm": 2.174448013305664, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7136752009391785, + "num_tokens": 472868408.0, + "step": 18274 + }, + { + "epoch": 2.0069185152646605, + "grad_norm": 2.4917759895324707, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7452215552330017, + "num_tokens": 472891999.0, + "step": 18275 + }, + { + "epoch": 2.0070283329672742, + "grad_norm": 2.6850950717926025, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7486510276794434, + "num_tokens": 472912478.0, + "step": 18276 + }, + { + "epoch": 2.007138150669888, + "grad_norm": 2.485257387161255, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7216182947158813, + "num_tokens": 472937871.0, + "step": 18277 + }, + { + "epoch": 2.0072479683725017, + "grad_norm": 2.126657724380493, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7133734226226807, + "num_tokens": 472968653.0, + "step": 18278 + }, + { + "epoch": 2.0073577860751155, + "grad_norm": 2.271819591522217, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7344794273376465, + "num_tokens": 472993966.0, + "step": 18279 + }, + { + "epoch": 2.007467603777729, + "grad_norm": 2.489849090576172, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7276813983917236, + "num_tokens": 473019198.0, + "step": 18280 + }, + { + "epoch": 2.0075774214803426, + "grad_norm": 2.203348159790039, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7236572504043579, + "num_tokens": 473047784.0, + "step": 18281 + }, + { + "epoch": 2.0076872391829563, + "grad_norm": 2.3580703735351562, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7216270565986633, + "num_tokens": 473073103.0, + "step": 18282 + }, + { + "epoch": 2.00779705688557, + "grad_norm": 2.296056032180786, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7468034625053406, + "num_tokens": 473098209.0, + "step": 18283 + }, + { + "epoch": 2.0079068745881834, + "grad_norm": 2.3730146884918213, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.6999348402023315, + "num_tokens": 473123539.0, + "step": 18284 + }, + { + "epoch": 2.008016692290797, + "grad_norm": 1.9618260860443115, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.713134765625, + "num_tokens": 473158145.0, + "step": 18285 + }, + { + "epoch": 2.008126509993411, + "grad_norm": 2.1655690670013428, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7216125726699829, + "num_tokens": 473185679.0, + "step": 18286 + }, + { + "epoch": 2.0082363276960247, + "grad_norm": 2.640528917312622, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7398817539215088, + "num_tokens": 473205197.0, + "step": 18287 + }, + { + "epoch": 2.0083461453986384, + "grad_norm": 2.3066422939300537, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7397754788398743, + "num_tokens": 473230948.0, + "step": 18288 + }, + { + "epoch": 2.0084559631012517, + "grad_norm": 2.5667688846588135, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7599597573280334, + "num_tokens": 473251055.0, + "step": 18289 + }, + { + "epoch": 2.0085657808038655, + "grad_norm": 2.444356679916382, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7349309325218201, + "num_tokens": 473274792.0, + "step": 18290 + }, + { + "epoch": 2.0086755985064793, + "grad_norm": 2.0515224933624268, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7543178796768188, + "num_tokens": 473304591.0, + "step": 18291 + }, + { + "epoch": 2.008785416209093, + "grad_norm": 2.170161485671997, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7371958494186401, + "num_tokens": 473333610.0, + "step": 18292 + }, + { + "epoch": 2.0088952339117068, + "grad_norm": 2.748169422149658, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7402912378311157, + "num_tokens": 473352052.0, + "step": 18293 + }, + { + "epoch": 2.00900505161432, + "grad_norm": 2.0740857124328613, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7135883569717407, + "num_tokens": 473384444.0, + "step": 18294 + }, + { + "epoch": 2.009114869316934, + "grad_norm": 2.0429012775421143, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7212676405906677, + "num_tokens": 473415544.0, + "step": 18295 + }, + { + "epoch": 2.0092246870195476, + "grad_norm": 2.239621162414551, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7258917093276978, + "num_tokens": 473443069.0, + "step": 18296 + }, + { + "epoch": 2.0093345047221614, + "grad_norm": 2.326580047607422, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.741073727607727, + "num_tokens": 473467650.0, + "step": 18297 + }, + { + "epoch": 2.0094443224247747, + "grad_norm": 2.105177402496338, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7414337396621704, + "num_tokens": 473495979.0, + "step": 18298 + }, + { + "epoch": 2.0095541401273884, + "grad_norm": 2.20709228515625, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7432138919830322, + "num_tokens": 473521867.0, + "step": 18299 + }, + { + "epoch": 2.009663957830002, + "grad_norm": 2.1545376777648926, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7372671365737915, + "num_tokens": 473551775.0, + "step": 18300 + }, + { + "epoch": 2.009773775532616, + "grad_norm": 2.80572772026062, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.754975438117981, + "num_tokens": 473570580.0, + "step": 18301 + }, + { + "epoch": 2.0098835932352297, + "grad_norm": 2.7306973934173584, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7408008575439453, + "num_tokens": 473589849.0, + "step": 18302 + }, + { + "epoch": 2.009993410937843, + "grad_norm": 2.0055594444274902, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7276173830032349, + "num_tokens": 473621943.0, + "step": 18303 + }, + { + "epoch": 2.0101032286404568, + "grad_norm": 2.4599618911743164, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7377172112464905, + "num_tokens": 473644412.0, + "step": 18304 + }, + { + "epoch": 2.0102130463430705, + "grad_norm": 1.89189875125885, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7434931993484497, + "num_tokens": 473681233.0, + "step": 18305 + }, + { + "epoch": 2.0103228640456843, + "grad_norm": 2.5392167568206787, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7437792420387268, + "num_tokens": 473703760.0, + "step": 18306 + }, + { + "epoch": 2.010432681748298, + "grad_norm": 2.3169267177581787, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7302259206771851, + "num_tokens": 473728373.0, + "step": 18307 + }, + { + "epoch": 2.0105424994509113, + "grad_norm": 2.221486806869507, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7291923761367798, + "num_tokens": 473754675.0, + "step": 18308 + }, + { + "epoch": 2.010652317153525, + "grad_norm": 2.4508941173553467, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7379174828529358, + "num_tokens": 473776730.0, + "step": 18309 + }, + { + "epoch": 2.010762134856139, + "grad_norm": 2.197160243988037, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7240297794342041, + "num_tokens": 473806682.0, + "step": 18310 + }, + { + "epoch": 2.0108719525587526, + "grad_norm": 2.1679911613464355, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7096536755561829, + "num_tokens": 473837602.0, + "step": 18311 + }, + { + "epoch": 2.010981770261366, + "grad_norm": 2.3226840496063232, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7375681400299072, + "num_tokens": 473861516.0, + "step": 18312 + }, + { + "epoch": 2.0110915879639797, + "grad_norm": 2.244190216064453, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7504332065582275, + "num_tokens": 473887367.0, + "step": 18313 + }, + { + "epoch": 2.0112014056665934, + "grad_norm": 2.2421000003814697, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7260897755622864, + "num_tokens": 473914911.0, + "step": 18314 + }, + { + "epoch": 2.011311223369207, + "grad_norm": 2.520918607711792, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7251312136650085, + "num_tokens": 473941329.0, + "step": 18315 + }, + { + "epoch": 2.011421041071821, + "grad_norm": 2.722280979156494, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7256626486778259, + "num_tokens": 473962147.0, + "step": 18316 + }, + { + "epoch": 2.0115308587744343, + "grad_norm": 2.3172385692596436, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7216311693191528, + "num_tokens": 473989895.0, + "step": 18317 + }, + { + "epoch": 2.011640676477048, + "grad_norm": 2.511444568634033, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7644979953765869, + "num_tokens": 474010572.0, + "step": 18318 + }, + { + "epoch": 2.011750494179662, + "grad_norm": 2.616121768951416, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.73124760389328, + "num_tokens": 474032697.0, + "step": 18319 + }, + { + "epoch": 2.0118603118822755, + "grad_norm": 2.017371892929077, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.751998782157898, + "num_tokens": 474063383.0, + "step": 18320 + }, + { + "epoch": 2.0119701295848893, + "grad_norm": 2.191988945007324, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.729427695274353, + "num_tokens": 474091568.0, + "step": 18321 + }, + { + "epoch": 2.0120799472875026, + "grad_norm": 2.3810458183288574, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7586090564727783, + "num_tokens": 474113882.0, + "step": 18322 + }, + { + "epoch": 2.0121897649901164, + "grad_norm": 2.197704792022705, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7333036661148071, + "num_tokens": 474143063.0, + "step": 18323 + }, + { + "epoch": 2.01229958269273, + "grad_norm": 2.231811285018921, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7721703052520752, + "num_tokens": 474167338.0, + "step": 18324 + }, + { + "epoch": 2.012409400395344, + "grad_norm": 2.5346148014068604, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7467650771141052, + "num_tokens": 474191108.0, + "step": 18325 + }, + { + "epoch": 2.012519218097957, + "grad_norm": 2.3074493408203125, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7321089506149292, + "num_tokens": 474216320.0, + "step": 18326 + }, + { + "epoch": 2.012629035800571, + "grad_norm": 2.0518391132354736, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7290291786193848, + "num_tokens": 474248732.0, + "step": 18327 + }, + { + "epoch": 2.0127388535031847, + "grad_norm": 2.4379119873046875, + "learning_rate": 1e-06, + "loss": 0.781, + "mean_token_accuracy": 0.7560438513755798, + "num_tokens": 474270619.0, + "step": 18328 + }, + { + "epoch": 2.0128486712057985, + "grad_norm": 2.489572286605835, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7672328352928162, + "num_tokens": 474293168.0, + "step": 18329 + }, + { + "epoch": 2.0129584889084122, + "grad_norm": 2.345395565032959, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7327457666397095, + "num_tokens": 474318595.0, + "step": 18330 + }, + { + "epoch": 2.0130683066110255, + "grad_norm": 2.227661609649658, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7363706827163696, + "num_tokens": 474346507.0, + "step": 18331 + }, + { + "epoch": 2.0131781243136393, + "grad_norm": 2.2630844116210938, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7160302400588989, + "num_tokens": 474374731.0, + "step": 18332 + }, + { + "epoch": 2.013287942016253, + "grad_norm": 2.546217203140259, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7485606074333191, + "num_tokens": 474396677.0, + "step": 18333 + }, + { + "epoch": 2.013397759718867, + "grad_norm": 2.2027804851531982, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7214044332504272, + "num_tokens": 474424538.0, + "step": 18334 + }, + { + "epoch": 2.01350757742148, + "grad_norm": 2.7274131774902344, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.755955159664154, + "num_tokens": 474444738.0, + "step": 18335 + }, + { + "epoch": 2.013617395124094, + "grad_norm": 2.6034140586853027, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7609054446220398, + "num_tokens": 474464517.0, + "step": 18336 + }, + { + "epoch": 2.0137272128267076, + "grad_norm": 2.148566246032715, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.725101113319397, + "num_tokens": 474495613.0, + "step": 18337 + }, + { + "epoch": 2.0138370305293214, + "grad_norm": 2.3731472492218018, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7080155611038208, + "num_tokens": 474521657.0, + "step": 18338 + }, + { + "epoch": 2.013946848231935, + "grad_norm": 2.2064051628112793, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7260345220565796, + "num_tokens": 474549732.0, + "step": 18339 + }, + { + "epoch": 2.0140566659345485, + "grad_norm": 2.081496238708496, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7301076650619507, + "num_tokens": 474578853.0, + "step": 18340 + }, + { + "epoch": 2.014166483637162, + "grad_norm": 2.07800030708313, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7254724502563477, + "num_tokens": 474609898.0, + "step": 18341 + }, + { + "epoch": 2.014276301339776, + "grad_norm": 2.6662681102752686, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7525275945663452, + "num_tokens": 474630096.0, + "step": 18342 + }, + { + "epoch": 2.0143861190423897, + "grad_norm": 2.373586893081665, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7519710659980774, + "num_tokens": 474655330.0, + "step": 18343 + }, + { + "epoch": 2.0144959367450035, + "grad_norm": 2.4747672080993652, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7264633774757385, + "num_tokens": 474679402.0, + "step": 18344 + }, + { + "epoch": 2.014605754447617, + "grad_norm": 2.1623470783233643, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.740963876247406, + "num_tokens": 474709822.0, + "step": 18345 + }, + { + "epoch": 2.0147155721502306, + "grad_norm": 2.307976007461548, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7583344578742981, + "num_tokens": 474734308.0, + "step": 18346 + }, + { + "epoch": 2.0148253898528443, + "grad_norm": 2.4654757976531982, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7498571872711182, + "num_tokens": 474757181.0, + "step": 18347 + }, + { + "epoch": 2.014935207555458, + "grad_norm": 2.4994313716888428, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7292096614837646, + "num_tokens": 474779652.0, + "step": 18348 + }, + { + "epoch": 2.0150450252580714, + "grad_norm": 2.5757734775543213, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7292360067367554, + "num_tokens": 474802679.0, + "step": 18349 + }, + { + "epoch": 2.015154842960685, + "grad_norm": 2.509899616241455, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7303181886672974, + "num_tokens": 474826802.0, + "step": 18350 + }, + { + "epoch": 2.015264660663299, + "grad_norm": 2.337146043777466, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7308307886123657, + "num_tokens": 474852591.0, + "step": 18351 + }, + { + "epoch": 2.0153744783659127, + "grad_norm": 2.1238760948181152, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7420897483825684, + "num_tokens": 474881174.0, + "step": 18352 + }, + { + "epoch": 2.0154842960685264, + "grad_norm": 2.0647099018096924, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7371982932090759, + "num_tokens": 474912081.0, + "step": 18353 + }, + { + "epoch": 2.0155941137711397, + "grad_norm": 2.4922683238983154, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7161492109298706, + "num_tokens": 474935198.0, + "step": 18354 + }, + { + "epoch": 2.0157039314737535, + "grad_norm": 2.517430305480957, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7207286357879639, + "num_tokens": 474958871.0, + "step": 18355 + }, + { + "epoch": 2.0158137491763672, + "grad_norm": 2.3773083686828613, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7330731153488159, + "num_tokens": 474982605.0, + "step": 18356 + }, + { + "epoch": 2.015923566878981, + "grad_norm": 2.4716343879699707, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.725121259689331, + "num_tokens": 475006223.0, + "step": 18357 + }, + { + "epoch": 2.0160333845815948, + "grad_norm": 2.1999149322509766, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7314510345458984, + "num_tokens": 475034253.0, + "step": 18358 + }, + { + "epoch": 2.016143202284208, + "grad_norm": 2.186109781265259, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7235986590385437, + "num_tokens": 475064327.0, + "step": 18359 + }, + { + "epoch": 2.016253019986822, + "grad_norm": 2.3263795375823975, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7384490370750427, + "num_tokens": 475088498.0, + "step": 18360 + }, + { + "epoch": 2.0163628376894356, + "grad_norm": 2.253722906112671, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7352797985076904, + "num_tokens": 475113631.0, + "step": 18361 + }, + { + "epoch": 2.0164726553920493, + "grad_norm": 2.121086835861206, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7236902713775635, + "num_tokens": 475144330.0, + "step": 18362 + }, + { + "epoch": 2.0165824730946627, + "grad_norm": 2.3314898014068604, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7256746292114258, + "num_tokens": 475173602.0, + "step": 18363 + }, + { + "epoch": 2.0166922907972764, + "grad_norm": 2.4628372192382812, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7384284734725952, + "num_tokens": 475197199.0, + "step": 18364 + }, + { + "epoch": 2.01680210849989, + "grad_norm": 2.0063226222991943, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7541495561599731, + "num_tokens": 475228261.0, + "step": 18365 + }, + { + "epoch": 2.016911926202504, + "grad_norm": 2.294393539428711, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7212598323822021, + "num_tokens": 475255246.0, + "step": 18366 + }, + { + "epoch": 2.0170217439051177, + "grad_norm": 2.466740608215332, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7572462558746338, + "num_tokens": 475278080.0, + "step": 18367 + }, + { + "epoch": 2.017131561607731, + "grad_norm": 2.744663953781128, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7487834692001343, + "num_tokens": 475297692.0, + "step": 18368 + }, + { + "epoch": 2.0172413793103448, + "grad_norm": 2.372546434402466, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7067499160766602, + "num_tokens": 475323342.0, + "step": 18369 + }, + { + "epoch": 2.0173511970129585, + "grad_norm": 2.3810014724731445, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7354875206947327, + "num_tokens": 475346994.0, + "step": 18370 + }, + { + "epoch": 2.0174610147155723, + "grad_norm": 2.3093173503875732, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.733834445476532, + "num_tokens": 475372315.0, + "step": 18371 + }, + { + "epoch": 2.017570832418186, + "grad_norm": 2.329059362411499, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7482275366783142, + "num_tokens": 475396438.0, + "step": 18372 + }, + { + "epoch": 2.0176806501207993, + "grad_norm": 2.3589539527893066, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7185519933700562, + "num_tokens": 475422077.0, + "step": 18373 + }, + { + "epoch": 2.017790467823413, + "grad_norm": 2.2858593463897705, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7388992309570312, + "num_tokens": 475447617.0, + "step": 18374 + }, + { + "epoch": 2.017900285526027, + "grad_norm": 2.488846778869629, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7415758967399597, + "num_tokens": 475471261.0, + "step": 18375 + }, + { + "epoch": 2.0180101032286406, + "grad_norm": 2.6142091751098633, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7400106191635132, + "num_tokens": 475492817.0, + "step": 18376 + }, + { + "epoch": 2.018119920931254, + "grad_norm": 2.3055379390716553, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7254669070243835, + "num_tokens": 475519515.0, + "step": 18377 + }, + { + "epoch": 2.0182297386338677, + "grad_norm": 2.2562077045440674, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7323511242866516, + "num_tokens": 475546017.0, + "step": 18378 + }, + { + "epoch": 2.0183395563364814, + "grad_norm": 2.207839250564575, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7127395272254944, + "num_tokens": 475576814.0, + "step": 18379 + }, + { + "epoch": 2.018449374039095, + "grad_norm": 2.263031005859375, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7193447947502136, + "num_tokens": 475604395.0, + "step": 18380 + }, + { + "epoch": 2.018559191741709, + "grad_norm": 2.2053651809692383, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.714584469795227, + "num_tokens": 475632406.0, + "step": 18381 + }, + { + "epoch": 2.0186690094443223, + "grad_norm": 2.038848638534546, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7215219736099243, + "num_tokens": 475663063.0, + "step": 18382 + }, + { + "epoch": 2.018778827146936, + "grad_norm": 2.6420416831970215, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7408456802368164, + "num_tokens": 475683086.0, + "step": 18383 + }, + { + "epoch": 2.0188886448495498, + "grad_norm": 2.7672624588012695, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7460219860076904, + "num_tokens": 475701641.0, + "step": 18384 + }, + { + "epoch": 2.0189984625521635, + "grad_norm": 2.4357478618621826, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7204681634902954, + "num_tokens": 475725651.0, + "step": 18385 + }, + { + "epoch": 2.0191082802547773, + "grad_norm": 2.096823215484619, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7308072447776794, + "num_tokens": 475755963.0, + "step": 18386 + }, + { + "epoch": 2.0192180979573906, + "grad_norm": 2.3326854705810547, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7180931568145752, + "num_tokens": 475782500.0, + "step": 18387 + }, + { + "epoch": 2.0193279156600044, + "grad_norm": 2.2968006134033203, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7254235744476318, + "num_tokens": 475810085.0, + "step": 18388 + }, + { + "epoch": 2.019437733362618, + "grad_norm": 2.5878422260284424, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7377675771713257, + "num_tokens": 475831769.0, + "step": 18389 + }, + { + "epoch": 2.019547551065232, + "grad_norm": 2.2201781272888184, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7195556163787842, + "num_tokens": 475860288.0, + "step": 18390 + }, + { + "epoch": 2.019657368767845, + "grad_norm": 2.4592645168304443, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7564834356307983, + "num_tokens": 475881887.0, + "step": 18391 + }, + { + "epoch": 2.019767186470459, + "grad_norm": 2.5398833751678467, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7193740606307983, + "num_tokens": 475907087.0, + "step": 18392 + }, + { + "epoch": 2.0198770041730727, + "grad_norm": 2.5312790870666504, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7286281585693359, + "num_tokens": 475930532.0, + "step": 18393 + }, + { + "epoch": 2.0199868218756865, + "grad_norm": 2.301351308822632, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7459771633148193, + "num_tokens": 475956755.0, + "step": 18394 + }, + { + "epoch": 2.0200966395783, + "grad_norm": 2.094149589538574, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7330358028411865, + "num_tokens": 475988953.0, + "step": 18395 + }, + { + "epoch": 2.0202064572809135, + "grad_norm": 2.6413094997406006, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7460498213768005, + "num_tokens": 476010678.0, + "step": 18396 + }, + { + "epoch": 2.0203162749835273, + "grad_norm": 2.293426036834717, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7419350147247314, + "num_tokens": 476038599.0, + "step": 18397 + }, + { + "epoch": 2.020426092686141, + "grad_norm": 2.2363524436950684, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7201098203659058, + "num_tokens": 476067170.0, + "step": 18398 + }, + { + "epoch": 2.020535910388755, + "grad_norm": 2.3435781002044678, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7384253144264221, + "num_tokens": 476093353.0, + "step": 18399 + }, + { + "epoch": 2.020645728091368, + "grad_norm": 2.1946768760681152, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7291707396507263, + "num_tokens": 476120892.0, + "step": 18400 + }, + { + "epoch": 2.020755545793982, + "grad_norm": 2.2548935413360596, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7285040616989136, + "num_tokens": 476148830.0, + "step": 18401 + }, + { + "epoch": 2.0208653634965956, + "grad_norm": 2.251495122909546, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7068072557449341, + "num_tokens": 476176269.0, + "step": 18402 + }, + { + "epoch": 2.0209751811992094, + "grad_norm": 2.276771306991577, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7074393630027771, + "num_tokens": 476203739.0, + "step": 18403 + }, + { + "epoch": 2.021084998901823, + "grad_norm": 2.5369462966918945, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7399230599403381, + "num_tokens": 476227125.0, + "step": 18404 + }, + { + "epoch": 2.0211948166044365, + "grad_norm": 2.332413911819458, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7255039215087891, + "num_tokens": 476253997.0, + "step": 18405 + }, + { + "epoch": 2.02130463430705, + "grad_norm": 2.391188621520996, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7781402468681335, + "num_tokens": 476276686.0, + "step": 18406 + }, + { + "epoch": 2.021414452009664, + "grad_norm": 2.0295772552490234, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7229298949241638, + "num_tokens": 476307945.0, + "step": 18407 + }, + { + "epoch": 2.0215242697122777, + "grad_norm": 1.8993171453475952, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7185765504837036, + "num_tokens": 476344153.0, + "step": 18408 + }, + { + "epoch": 2.0216340874148915, + "grad_norm": 2.2159390449523926, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7502577304840088, + "num_tokens": 476372534.0, + "step": 18409 + }, + { + "epoch": 2.021743905117505, + "grad_norm": 2.2487595081329346, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.722293496131897, + "num_tokens": 476400143.0, + "step": 18410 + }, + { + "epoch": 2.0218537228201185, + "grad_norm": 2.5568370819091797, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7543597221374512, + "num_tokens": 476421887.0, + "step": 18411 + }, + { + "epoch": 2.0219635405227323, + "grad_norm": 2.2459607124328613, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.73292076587677, + "num_tokens": 476448408.0, + "step": 18412 + }, + { + "epoch": 2.022073358225346, + "grad_norm": 2.856984853744507, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7453109622001648, + "num_tokens": 476467945.0, + "step": 18413 + }, + { + "epoch": 2.0221831759279594, + "grad_norm": 1.9819382429122925, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7150210738182068, + "num_tokens": 476504064.0, + "step": 18414 + }, + { + "epoch": 2.022292993630573, + "grad_norm": 2.421964406967163, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7410244345664978, + "num_tokens": 476528563.0, + "step": 18415 + }, + { + "epoch": 2.022402811333187, + "grad_norm": 2.5537965297698975, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.734675943851471, + "num_tokens": 476551278.0, + "step": 18416 + }, + { + "epoch": 2.0225126290358006, + "grad_norm": 2.6809518337249756, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7541829943656921, + "num_tokens": 476571059.0, + "step": 18417 + }, + { + "epoch": 2.0226224467384144, + "grad_norm": 2.2072219848632812, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7368276715278625, + "num_tokens": 476601332.0, + "step": 18418 + }, + { + "epoch": 2.0227322644410277, + "grad_norm": 2.74995756149292, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7577942609786987, + "num_tokens": 476620721.0, + "step": 18419 + }, + { + "epoch": 2.0228420821436415, + "grad_norm": 2.347778081893921, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7351251840591431, + "num_tokens": 476647984.0, + "step": 18420 + }, + { + "epoch": 2.0229518998462552, + "grad_norm": 2.2892143726348877, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.73729407787323, + "num_tokens": 476675219.0, + "step": 18421 + }, + { + "epoch": 2.023061717548869, + "grad_norm": 2.0869743824005127, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7167050242424011, + "num_tokens": 476707059.0, + "step": 18422 + }, + { + "epoch": 2.0231715352514827, + "grad_norm": 2.188969135284424, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7462148070335388, + "num_tokens": 476734933.0, + "step": 18423 + }, + { + "epoch": 2.023281352954096, + "grad_norm": 2.4138436317443848, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.74822598695755, + "num_tokens": 476758364.0, + "step": 18424 + }, + { + "epoch": 2.02339117065671, + "grad_norm": 2.178802490234375, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7171856760978699, + "num_tokens": 476788896.0, + "step": 18425 + }, + { + "epoch": 2.0235009883593236, + "grad_norm": 2.4084694385528564, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7107461094856262, + "num_tokens": 476814130.0, + "step": 18426 + }, + { + "epoch": 2.0236108060619373, + "grad_norm": 2.2946743965148926, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.742776095867157, + "num_tokens": 476839446.0, + "step": 18427 + }, + { + "epoch": 2.0237206237645506, + "grad_norm": 2.1698639392852783, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7152031660079956, + "num_tokens": 476871147.0, + "step": 18428 + }, + { + "epoch": 2.0238304414671644, + "grad_norm": 2.2693309783935547, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7249976396560669, + "num_tokens": 476897450.0, + "step": 18429 + }, + { + "epoch": 2.023940259169778, + "grad_norm": 2.525437831878662, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7216337323188782, + "num_tokens": 476921078.0, + "step": 18430 + }, + { + "epoch": 2.024050076872392, + "grad_norm": 2.390435218811035, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7316023707389832, + "num_tokens": 476944763.0, + "step": 18431 + }, + { + "epoch": 2.0241598945750057, + "grad_norm": 2.748257875442505, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.7457400560379028, + "num_tokens": 476964532.0, + "step": 18432 + }, + { + "epoch": 2.024269712277619, + "grad_norm": 2.642573833465576, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7447865009307861, + "num_tokens": 476984769.0, + "step": 18433 + }, + { + "epoch": 2.0243795299802327, + "grad_norm": 2.3757143020629883, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7217347025871277, + "num_tokens": 477010814.0, + "step": 18434 + }, + { + "epoch": 2.0244893476828465, + "grad_norm": 2.3041226863861084, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7319155931472778, + "num_tokens": 477037902.0, + "step": 18435 + }, + { + "epoch": 2.0245991653854603, + "grad_norm": 2.3786590099334717, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7364709973335266, + "num_tokens": 477062536.0, + "step": 18436 + }, + { + "epoch": 2.024708983088074, + "grad_norm": 2.376595973968506, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7386186122894287, + "num_tokens": 477088327.0, + "step": 18437 + }, + { + "epoch": 2.0248188007906873, + "grad_norm": 2.3939473628997803, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7571866512298584, + "num_tokens": 477111255.0, + "step": 18438 + }, + { + "epoch": 2.024928618493301, + "grad_norm": 2.1103761196136475, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7372379302978516, + "num_tokens": 477140742.0, + "step": 18439 + }, + { + "epoch": 2.025038436195915, + "grad_norm": 2.3276097774505615, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7270675301551819, + "num_tokens": 477167026.0, + "step": 18440 + }, + { + "epoch": 2.0251482538985286, + "grad_norm": 2.556856393814087, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7296015024185181, + "num_tokens": 477189984.0, + "step": 18441 + }, + { + "epoch": 2.025258071601142, + "grad_norm": 2.3502161502838135, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6982194781303406, + "num_tokens": 477216833.0, + "step": 18442 + }, + { + "epoch": 2.0253678893037557, + "grad_norm": 2.584372043609619, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7276995182037354, + "num_tokens": 477239306.0, + "step": 18443 + }, + { + "epoch": 2.0254777070063694, + "grad_norm": 2.6070001125335693, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7341266870498657, + "num_tokens": 477261024.0, + "step": 18444 + }, + { + "epoch": 2.025587524708983, + "grad_norm": 2.2724404335021973, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7364197969436646, + "num_tokens": 477287270.0, + "step": 18445 + }, + { + "epoch": 2.025697342411597, + "grad_norm": 2.2898054122924805, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7349278330802917, + "num_tokens": 477315531.0, + "step": 18446 + }, + { + "epoch": 2.0258071601142102, + "grad_norm": 2.2935378551483154, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7374153137207031, + "num_tokens": 477340295.0, + "step": 18447 + }, + { + "epoch": 2.025916977816824, + "grad_norm": 2.498563051223755, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7438837289810181, + "num_tokens": 477362513.0, + "step": 18448 + }, + { + "epoch": 2.0260267955194378, + "grad_norm": 2.1752798557281494, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7067215442657471, + "num_tokens": 477390622.0, + "step": 18449 + }, + { + "epoch": 2.0261366132220515, + "grad_norm": 2.343625068664551, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7356560826301575, + "num_tokens": 477415906.0, + "step": 18450 + }, + { + "epoch": 2.0262464309246653, + "grad_norm": 2.246807336807251, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7341349124908447, + "num_tokens": 477442984.0, + "step": 18451 + }, + { + "epoch": 2.0263562486272786, + "grad_norm": 2.2410309314727783, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7433063983917236, + "num_tokens": 477467160.0, + "step": 18452 + }, + { + "epoch": 2.0264660663298923, + "grad_norm": 2.117377281188965, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7392284274101257, + "num_tokens": 477494952.0, + "step": 18453 + }, + { + "epoch": 2.026575884032506, + "grad_norm": 2.2552475929260254, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7239980101585388, + "num_tokens": 477523578.0, + "step": 18454 + }, + { + "epoch": 2.02668570173512, + "grad_norm": 2.34604549407959, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7392150163650513, + "num_tokens": 477547502.0, + "step": 18455 + }, + { + "epoch": 2.026795519437733, + "grad_norm": 2.695267677307129, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7194057703018188, + "num_tokens": 477568762.0, + "step": 18456 + }, + { + "epoch": 2.026905337140347, + "grad_norm": 2.517422914505005, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.744247555732727, + "num_tokens": 477591095.0, + "step": 18457 + }, + { + "epoch": 2.0270151548429607, + "grad_norm": 2.3681697845458984, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7363064289093018, + "num_tokens": 477615159.0, + "step": 18458 + }, + { + "epoch": 2.0271249725455744, + "grad_norm": 2.157278299331665, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7026516199111938, + "num_tokens": 477646241.0, + "step": 18459 + }, + { + "epoch": 2.027234790248188, + "grad_norm": 2.0188066959381104, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7010284662246704, + "num_tokens": 477678701.0, + "step": 18460 + }, + { + "epoch": 2.0273446079508015, + "grad_norm": 2.186182737350464, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7099722623825073, + "num_tokens": 477708276.0, + "step": 18461 + }, + { + "epoch": 2.0274544256534153, + "grad_norm": 2.209888458251953, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7293126583099365, + "num_tokens": 477734701.0, + "step": 18462 + }, + { + "epoch": 2.027564243356029, + "grad_norm": 2.241185188293457, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7260593175888062, + "num_tokens": 477761108.0, + "step": 18463 + }, + { + "epoch": 2.027674061058643, + "grad_norm": 2.2393712997436523, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.747006893157959, + "num_tokens": 477787164.0, + "step": 18464 + }, + { + "epoch": 2.027783878761256, + "grad_norm": 2.4558658599853516, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7309900522232056, + "num_tokens": 477811238.0, + "step": 18465 + }, + { + "epoch": 2.02789369646387, + "grad_norm": 2.1337409019470215, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7282491326332092, + "num_tokens": 477840378.0, + "step": 18466 + }, + { + "epoch": 2.0280035141664836, + "grad_norm": 2.345294713973999, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7185981273651123, + "num_tokens": 477866312.0, + "step": 18467 + }, + { + "epoch": 2.0281133318690974, + "grad_norm": 2.40580415725708, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7483586072921753, + "num_tokens": 477889462.0, + "step": 18468 + }, + { + "epoch": 2.028223149571711, + "grad_norm": 2.3612515926361084, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7324953079223633, + "num_tokens": 477912523.0, + "step": 18469 + }, + { + "epoch": 2.0283329672743244, + "grad_norm": 2.3177239894866943, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7459748983383179, + "num_tokens": 477938584.0, + "step": 18470 + }, + { + "epoch": 2.028442784976938, + "grad_norm": 2.144442081451416, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7246320843696594, + "num_tokens": 477969511.0, + "step": 18471 + }, + { + "epoch": 2.028552602679552, + "grad_norm": 2.3598852157592773, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7146221399307251, + "num_tokens": 477996399.0, + "step": 18472 + }, + { + "epoch": 2.0286624203821657, + "grad_norm": 2.52152419090271, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7406066656112671, + "num_tokens": 478019229.0, + "step": 18473 + }, + { + "epoch": 2.0287722380847795, + "grad_norm": 2.5822978019714355, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7464938759803772, + "num_tokens": 478040209.0, + "step": 18474 + }, + { + "epoch": 2.028882055787393, + "grad_norm": 2.3823304176330566, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7479733228683472, + "num_tokens": 478062881.0, + "step": 18475 + }, + { + "epoch": 2.0289918734900065, + "grad_norm": 2.462517738342285, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7497603893280029, + "num_tokens": 478085248.0, + "step": 18476 + }, + { + "epoch": 2.0291016911926203, + "grad_norm": 2.259176731109619, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7295324206352234, + "num_tokens": 478111876.0, + "step": 18477 + }, + { + "epoch": 2.029211508895234, + "grad_norm": 2.204153299331665, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7340961694717407, + "num_tokens": 478140060.0, + "step": 18478 + }, + { + "epoch": 2.0293213265978474, + "grad_norm": 2.0238423347473145, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7125629186630249, + "num_tokens": 478173027.0, + "step": 18479 + }, + { + "epoch": 2.029431144300461, + "grad_norm": 2.4731552600860596, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7345497608184814, + "num_tokens": 478197596.0, + "step": 18480 + }, + { + "epoch": 2.029540962003075, + "grad_norm": 2.2303895950317383, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7437382936477661, + "num_tokens": 478227789.0, + "step": 18481 + }, + { + "epoch": 2.0296507797056886, + "grad_norm": 2.0912530422210693, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7228546738624573, + "num_tokens": 478259865.0, + "step": 18482 + }, + { + "epoch": 2.0297605974083024, + "grad_norm": 2.7126152515411377, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7421236038208008, + "num_tokens": 478280288.0, + "step": 18483 + }, + { + "epoch": 2.0298704151109157, + "grad_norm": 2.6196908950805664, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7331053018569946, + "num_tokens": 478301159.0, + "step": 18484 + }, + { + "epoch": 2.0299802328135295, + "grad_norm": 2.4135501384735107, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.737981915473938, + "num_tokens": 478326072.0, + "step": 18485 + }, + { + "epoch": 2.030090050516143, + "grad_norm": 2.09450364112854, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.6937423944473267, + "num_tokens": 478362978.0, + "step": 18486 + }, + { + "epoch": 2.030199868218757, + "grad_norm": 2.094870090484619, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7337262630462646, + "num_tokens": 478395259.0, + "step": 18487 + }, + { + "epoch": 2.0303096859213707, + "grad_norm": 2.2331793308258057, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7220958471298218, + "num_tokens": 478424449.0, + "step": 18488 + }, + { + "epoch": 2.030419503623984, + "grad_norm": 2.102461576461792, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.71666419506073, + "num_tokens": 478454349.0, + "step": 18489 + }, + { + "epoch": 2.030529321326598, + "grad_norm": 2.6475226879119873, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7215016484260559, + "num_tokens": 478476519.0, + "step": 18490 + }, + { + "epoch": 2.0306391390292116, + "grad_norm": 2.438326120376587, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7286301851272583, + "num_tokens": 478501584.0, + "step": 18491 + }, + { + "epoch": 2.0307489567318253, + "grad_norm": 2.3459227085113525, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7265008687973022, + "num_tokens": 478527028.0, + "step": 18492 + }, + { + "epoch": 2.0308587744344386, + "grad_norm": 2.1351592540740967, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7295882701873779, + "num_tokens": 478556497.0, + "step": 18493 + }, + { + "epoch": 2.0309685921370524, + "grad_norm": 2.3619110584259033, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7380272150039673, + "num_tokens": 478581600.0, + "step": 18494 + }, + { + "epoch": 2.031078409839666, + "grad_norm": 2.360659122467041, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7198150157928467, + "num_tokens": 478608224.0, + "step": 18495 + }, + { + "epoch": 2.03118822754228, + "grad_norm": 2.6286566257476807, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7443945407867432, + "num_tokens": 478630025.0, + "step": 18496 + }, + { + "epoch": 2.0312980452448937, + "grad_norm": 2.2535226345062256, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7283030152320862, + "num_tokens": 478655944.0, + "step": 18497 + }, + { + "epoch": 2.031407862947507, + "grad_norm": 2.3081719875335693, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7194170951843262, + "num_tokens": 478682722.0, + "step": 18498 + }, + { + "epoch": 2.0315176806501207, + "grad_norm": 2.2182319164276123, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7222568988800049, + "num_tokens": 478712650.0, + "step": 18499 + }, + { + "epoch": 2.0316274983527345, + "grad_norm": 1.8173408508300781, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7180633544921875, + "num_tokens": 478750751.0, + "step": 18500 + }, + { + "epoch": 2.0317373160553482, + "grad_norm": 2.224332571029663, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7048324942588806, + "num_tokens": 478782213.0, + "step": 18501 + }, + { + "epoch": 2.031847133757962, + "grad_norm": 2.5613059997558594, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7255611419677734, + "num_tokens": 478806595.0, + "step": 18502 + }, + { + "epoch": 2.0319569514605753, + "grad_norm": 1.9637073278427124, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.708343505859375, + "num_tokens": 478842045.0, + "step": 18503 + }, + { + "epoch": 2.032066769163189, + "grad_norm": 2.2695863246917725, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7250843644142151, + "num_tokens": 478870674.0, + "step": 18504 + }, + { + "epoch": 2.032176586865803, + "grad_norm": 2.671567678451538, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.737449049949646, + "num_tokens": 478891731.0, + "step": 18505 + }, + { + "epoch": 2.0322864045684166, + "grad_norm": 2.1392321586608887, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7253834009170532, + "num_tokens": 478920241.0, + "step": 18506 + }, + { + "epoch": 2.03239622227103, + "grad_norm": 2.07600474357605, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7109951972961426, + "num_tokens": 478951766.0, + "step": 18507 + }, + { + "epoch": 2.0325060399736437, + "grad_norm": 2.0875887870788574, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.721265435218811, + "num_tokens": 478984028.0, + "step": 18508 + }, + { + "epoch": 2.0326158576762574, + "grad_norm": 2.231119155883789, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7192307710647583, + "num_tokens": 479013463.0, + "step": 18509 + }, + { + "epoch": 2.032725675378871, + "grad_norm": 2.3683700561523438, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7111121416091919, + "num_tokens": 479038882.0, + "step": 18510 + }, + { + "epoch": 2.032835493081485, + "grad_norm": 2.472907781600952, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7198151350021362, + "num_tokens": 479063048.0, + "step": 18511 + }, + { + "epoch": 2.0329453107840982, + "grad_norm": 2.1848013401031494, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7169351577758789, + "num_tokens": 479092793.0, + "step": 18512 + }, + { + "epoch": 2.033055128486712, + "grad_norm": 2.542630195617676, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7395969033241272, + "num_tokens": 479114094.0, + "step": 18513 + }, + { + "epoch": 2.0331649461893258, + "grad_norm": 2.2764947414398193, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7447440028190613, + "num_tokens": 479139288.0, + "step": 18514 + }, + { + "epoch": 2.0332747638919395, + "grad_norm": 2.4725327491760254, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7376477718353271, + "num_tokens": 479161789.0, + "step": 18515 + }, + { + "epoch": 2.033384581594553, + "grad_norm": 2.323632001876831, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7208949327468872, + "num_tokens": 479187593.0, + "step": 18516 + }, + { + "epoch": 2.0334943992971666, + "grad_norm": 2.242614269256592, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.735127866268158, + "num_tokens": 479215416.0, + "step": 18517 + }, + { + "epoch": 2.0336042169997803, + "grad_norm": 2.2503514289855957, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7283638119697571, + "num_tokens": 479242278.0, + "step": 18518 + }, + { + "epoch": 2.033714034702394, + "grad_norm": 2.1503913402557373, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7191161513328552, + "num_tokens": 479270308.0, + "step": 18519 + }, + { + "epoch": 2.033823852405008, + "grad_norm": 2.2758524417877197, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7416947484016418, + "num_tokens": 479297337.0, + "step": 18520 + }, + { + "epoch": 2.033933670107621, + "grad_norm": 2.180974006652832, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7198513746261597, + "num_tokens": 479326319.0, + "step": 18521 + }, + { + "epoch": 2.034043487810235, + "grad_norm": 2.274623155593872, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7351808547973633, + "num_tokens": 479353862.0, + "step": 18522 + }, + { + "epoch": 2.0341533055128487, + "grad_norm": 2.243659734725952, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7309914827346802, + "num_tokens": 479380234.0, + "step": 18523 + }, + { + "epoch": 2.0342631232154624, + "grad_norm": 2.638878583908081, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7422969937324524, + "num_tokens": 479400222.0, + "step": 18524 + }, + { + "epoch": 2.034372940918076, + "grad_norm": 2.2905259132385254, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7339869141578674, + "num_tokens": 479425746.0, + "step": 18525 + }, + { + "epoch": 2.0344827586206895, + "grad_norm": 2.4648849964141846, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7428439259529114, + "num_tokens": 479449819.0, + "step": 18526 + }, + { + "epoch": 2.0345925763233033, + "grad_norm": 2.363178014755249, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7257112264633179, + "num_tokens": 479475323.0, + "step": 18527 + }, + { + "epoch": 2.034702394025917, + "grad_norm": 2.133254289627075, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7111871242523193, + "num_tokens": 479505738.0, + "step": 18528 + }, + { + "epoch": 2.0348122117285308, + "grad_norm": 2.573438882827759, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7394348978996277, + "num_tokens": 479527435.0, + "step": 18529 + }, + { + "epoch": 2.034922029431144, + "grad_norm": 2.83610200881958, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.767095685005188, + "num_tokens": 479545792.0, + "step": 18530 + }, + { + "epoch": 2.035031847133758, + "grad_norm": 2.3031020164489746, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7381474375724792, + "num_tokens": 479570974.0, + "step": 18531 + }, + { + "epoch": 2.0351416648363716, + "grad_norm": 2.724950075149536, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.764310896396637, + "num_tokens": 479588853.0, + "step": 18532 + }, + { + "epoch": 2.0352514825389854, + "grad_norm": 2.7598330974578857, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7438632249832153, + "num_tokens": 479609539.0, + "step": 18533 + }, + { + "epoch": 2.035361300241599, + "grad_norm": 2.2808475494384766, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7408340573310852, + "num_tokens": 479635985.0, + "step": 18534 + }, + { + "epoch": 2.0354711179442124, + "grad_norm": 2.26009202003479, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7258273363113403, + "num_tokens": 479662486.0, + "step": 18535 + }, + { + "epoch": 2.035580935646826, + "grad_norm": 2.5088624954223633, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7274450063705444, + "num_tokens": 479686596.0, + "step": 18536 + }, + { + "epoch": 2.03569075334944, + "grad_norm": 1.9832559823989868, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7408062815666199, + "num_tokens": 479718032.0, + "step": 18537 + }, + { + "epoch": 2.0358005710520537, + "grad_norm": 2.479694366455078, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7327539324760437, + "num_tokens": 479743117.0, + "step": 18538 + }, + { + "epoch": 2.0359103887546675, + "grad_norm": 2.1651928424835205, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.754165768623352, + "num_tokens": 479772035.0, + "step": 18539 + }, + { + "epoch": 2.0360202064572808, + "grad_norm": 2.4367825984954834, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7335367202758789, + "num_tokens": 479797667.0, + "step": 18540 + }, + { + "epoch": 2.0361300241598945, + "grad_norm": 2.1237964630126953, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7045856714248657, + "num_tokens": 479825969.0, + "step": 18541 + }, + { + "epoch": 2.0362398418625083, + "grad_norm": 2.4190287590026855, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7334675788879395, + "num_tokens": 479851139.0, + "step": 18542 + }, + { + "epoch": 2.036349659565122, + "grad_norm": 2.322486162185669, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.724023699760437, + "num_tokens": 479876346.0, + "step": 18543 + }, + { + "epoch": 2.0364594772677354, + "grad_norm": 2.8480782508850098, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7323817610740662, + "num_tokens": 479897002.0, + "step": 18544 + }, + { + "epoch": 2.036569294970349, + "grad_norm": 2.519080400466919, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.735675036907196, + "num_tokens": 479919893.0, + "step": 18545 + }, + { + "epoch": 2.036679112672963, + "grad_norm": 2.168947458267212, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7326877117156982, + "num_tokens": 479949082.0, + "step": 18546 + }, + { + "epoch": 2.0367889303755766, + "grad_norm": 2.600144624710083, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7280938625335693, + "num_tokens": 479972061.0, + "step": 18547 + }, + { + "epoch": 2.0368987480781904, + "grad_norm": 2.190420150756836, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7591397166252136, + "num_tokens": 480000751.0, + "step": 18548 + }, + { + "epoch": 2.0370085657808037, + "grad_norm": 2.2088980674743652, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7301474213600159, + "num_tokens": 480029222.0, + "step": 18549 + }, + { + "epoch": 2.0371183834834174, + "grad_norm": 2.4964606761932373, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7509130239486694, + "num_tokens": 480053003.0, + "step": 18550 + }, + { + "epoch": 2.037228201186031, + "grad_norm": 2.195544719696045, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7387392520904541, + "num_tokens": 480080069.0, + "step": 18551 + }, + { + "epoch": 2.037338018888645, + "grad_norm": 2.4720358848571777, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7331362962722778, + "num_tokens": 480103242.0, + "step": 18552 + }, + { + "epoch": 2.0374478365912587, + "grad_norm": 2.4330883026123047, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7290233373641968, + "num_tokens": 480128365.0, + "step": 18553 + }, + { + "epoch": 2.037557654293872, + "grad_norm": 2.6006617546081543, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7516194581985474, + "num_tokens": 480150859.0, + "step": 18554 + }, + { + "epoch": 2.037667471996486, + "grad_norm": 2.374194622039795, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7351935505867004, + "num_tokens": 480176966.0, + "step": 18555 + }, + { + "epoch": 2.0377772896990995, + "grad_norm": 2.721763849258423, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7452586889266968, + "num_tokens": 480196250.0, + "step": 18556 + }, + { + "epoch": 2.0378871074017133, + "grad_norm": 2.403830051422119, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.75106281042099, + "num_tokens": 480218428.0, + "step": 18557 + }, + { + "epoch": 2.0379969251043266, + "grad_norm": 2.179722547531128, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7445144653320312, + "num_tokens": 480247440.0, + "step": 18558 + }, + { + "epoch": 2.0381067428069404, + "grad_norm": 2.535693645477295, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7548163533210754, + "num_tokens": 480270897.0, + "step": 18559 + }, + { + "epoch": 2.038216560509554, + "grad_norm": 2.195164680480957, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7457078099250793, + "num_tokens": 480299102.0, + "step": 18560 + }, + { + "epoch": 2.038326378212168, + "grad_norm": 2.6677918434143066, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7205827236175537, + "num_tokens": 480322035.0, + "step": 18561 + }, + { + "epoch": 2.0384361959147816, + "grad_norm": 2.341660737991333, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7480447888374329, + "num_tokens": 480346328.0, + "step": 18562 + }, + { + "epoch": 2.038546013617395, + "grad_norm": 2.5318093299865723, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7382340431213379, + "num_tokens": 480370514.0, + "step": 18563 + }, + { + "epoch": 2.0386558313200087, + "grad_norm": 2.4790256023406982, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7486841678619385, + "num_tokens": 480391909.0, + "step": 18564 + }, + { + "epoch": 2.0387656490226225, + "grad_norm": 2.510110855102539, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7407835721969604, + "num_tokens": 480415314.0, + "step": 18565 + }, + { + "epoch": 2.0388754667252362, + "grad_norm": 2.1810269355773926, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7215024828910828, + "num_tokens": 480444693.0, + "step": 18566 + }, + { + "epoch": 2.03898528442785, + "grad_norm": 2.3590657711029053, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7260888814926147, + "num_tokens": 480472164.0, + "step": 18567 + }, + { + "epoch": 2.0390951021304633, + "grad_norm": 2.0077433586120605, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.758662223815918, + "num_tokens": 480504905.0, + "step": 18568 + }, + { + "epoch": 2.039204919833077, + "grad_norm": 2.4604992866516113, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.72669517993927, + "num_tokens": 480530076.0, + "step": 18569 + }, + { + "epoch": 2.039314737535691, + "grad_norm": 2.0270698070526123, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7224408388137817, + "num_tokens": 480560966.0, + "step": 18570 + }, + { + "epoch": 2.0394245552383046, + "grad_norm": 2.269622802734375, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.710719883441925, + "num_tokens": 480586988.0, + "step": 18571 + }, + { + "epoch": 2.039534372940918, + "grad_norm": 2.284491539001465, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7083616852760315, + "num_tokens": 480615259.0, + "step": 18572 + }, + { + "epoch": 2.0396441906435316, + "grad_norm": 2.0550777912139893, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7177370190620422, + "num_tokens": 480646526.0, + "step": 18573 + }, + { + "epoch": 2.0397540083461454, + "grad_norm": 2.7801525592803955, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7290664911270142, + "num_tokens": 480667027.0, + "step": 18574 + }, + { + "epoch": 2.039863826048759, + "grad_norm": 2.470767021179199, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7373789548873901, + "num_tokens": 480690476.0, + "step": 18575 + }, + { + "epoch": 2.039973643751373, + "grad_norm": 2.268472194671631, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7133361101150513, + "num_tokens": 480719978.0, + "step": 18576 + }, + { + "epoch": 2.0400834614539862, + "grad_norm": 2.081698179244995, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7397941946983337, + "num_tokens": 480749163.0, + "step": 18577 + }, + { + "epoch": 2.0401932791566, + "grad_norm": 2.624476432800293, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7525820136070251, + "num_tokens": 480769114.0, + "step": 18578 + }, + { + "epoch": 2.0403030968592137, + "grad_norm": 2.36055064201355, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7667927742004395, + "num_tokens": 480792241.0, + "step": 18579 + }, + { + "epoch": 2.0404129145618275, + "grad_norm": 2.3801867961883545, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7154684066772461, + "num_tokens": 480817932.0, + "step": 18580 + }, + { + "epoch": 2.0405227322644413, + "grad_norm": 2.5993621349334717, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7445964813232422, + "num_tokens": 480839867.0, + "step": 18581 + }, + { + "epoch": 2.0406325499670546, + "grad_norm": 2.3510348796844482, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7471553683280945, + "num_tokens": 480863797.0, + "step": 18582 + }, + { + "epoch": 2.0407423676696683, + "grad_norm": 2.129230260848999, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7410451173782349, + "num_tokens": 480894772.0, + "step": 18583 + }, + { + "epoch": 2.040852185372282, + "grad_norm": 2.542874336242676, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7358366250991821, + "num_tokens": 480917888.0, + "step": 18584 + }, + { + "epoch": 2.040962003074896, + "grad_norm": 2.5442216396331787, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7575815916061401, + "num_tokens": 480937887.0, + "step": 18585 + }, + { + "epoch": 2.041071820777509, + "grad_norm": 2.5714731216430664, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7476933002471924, + "num_tokens": 480959462.0, + "step": 18586 + }, + { + "epoch": 2.041181638480123, + "grad_norm": 2.2350218296051025, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7184703946113586, + "num_tokens": 480987958.0, + "step": 18587 + }, + { + "epoch": 2.0412914561827367, + "grad_norm": 2.4361467361450195, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7212610244750977, + "num_tokens": 481011290.0, + "step": 18588 + }, + { + "epoch": 2.0414012738853504, + "grad_norm": 2.391418695449829, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7422706484794617, + "num_tokens": 481036578.0, + "step": 18589 + }, + { + "epoch": 2.041511091587964, + "grad_norm": 2.2056424617767334, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7171671390533447, + "num_tokens": 481065866.0, + "step": 18590 + }, + { + "epoch": 2.0416209092905775, + "grad_norm": 2.476468324661255, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7027743458747864, + "num_tokens": 481092331.0, + "step": 18591 + }, + { + "epoch": 2.0417307269931912, + "grad_norm": 2.5269086360931396, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7267231345176697, + "num_tokens": 481116641.0, + "step": 18592 + }, + { + "epoch": 2.041840544695805, + "grad_norm": 2.098876953125, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7200164794921875, + "num_tokens": 481147880.0, + "step": 18593 + }, + { + "epoch": 2.0419503623984188, + "grad_norm": 2.1642980575561523, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.72990882396698, + "num_tokens": 481177967.0, + "step": 18594 + }, + { + "epoch": 2.042060180101032, + "grad_norm": 2.6117446422576904, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7350823283195496, + "num_tokens": 481199113.0, + "step": 18595 + }, + { + "epoch": 2.042169997803646, + "grad_norm": 2.2524895668029785, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7343693971633911, + "num_tokens": 481224513.0, + "step": 18596 + }, + { + "epoch": 2.0422798155062596, + "grad_norm": 2.563633441925049, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7346712350845337, + "num_tokens": 481247538.0, + "step": 18597 + }, + { + "epoch": 2.0423896332088733, + "grad_norm": 2.7345097064971924, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7432239055633545, + "num_tokens": 481268354.0, + "step": 18598 + }, + { + "epoch": 2.042499450911487, + "grad_norm": 2.399049758911133, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.715948224067688, + "num_tokens": 481292886.0, + "step": 18599 + }, + { + "epoch": 2.0426092686141004, + "grad_norm": 2.3341450691223145, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7349287271499634, + "num_tokens": 481318875.0, + "step": 18600 + }, + { + "epoch": 2.042719086316714, + "grad_norm": 2.0105819702148438, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7167120575904846, + "num_tokens": 481350672.0, + "step": 18601 + }, + { + "epoch": 2.042828904019328, + "grad_norm": 2.718824625015259, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7121397256851196, + "num_tokens": 481373624.0, + "step": 18602 + }, + { + "epoch": 2.0429387217219417, + "grad_norm": 2.7551279067993164, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7700009942054749, + "num_tokens": 481392991.0, + "step": 18603 + }, + { + "epoch": 2.0430485394245554, + "grad_norm": 2.6860098838806152, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7343717217445374, + "num_tokens": 481413218.0, + "step": 18604 + }, + { + "epoch": 2.0431583571271688, + "grad_norm": 2.161515712738037, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7133284211158752, + "num_tokens": 481441656.0, + "step": 18605 + }, + { + "epoch": 2.0432681748297825, + "grad_norm": 2.8416941165924072, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7447268962860107, + "num_tokens": 481459766.0, + "step": 18606 + }, + { + "epoch": 2.0433779925323963, + "grad_norm": 2.0952250957489014, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.739216685295105, + "num_tokens": 481486822.0, + "step": 18607 + }, + { + "epoch": 2.04348781023501, + "grad_norm": 2.5124893188476562, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7481191158294678, + "num_tokens": 481508810.0, + "step": 18608 + }, + { + "epoch": 2.0435976279376233, + "grad_norm": 2.2650201320648193, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7207469940185547, + "num_tokens": 481535761.0, + "step": 18609 + }, + { + "epoch": 2.043707445640237, + "grad_norm": 1.9948451519012451, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7595323324203491, + "num_tokens": 481567869.0, + "step": 18610 + }, + { + "epoch": 2.043817263342851, + "grad_norm": 2.1974689960479736, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7138364911079407, + "num_tokens": 481596756.0, + "step": 18611 + }, + { + "epoch": 2.0439270810454646, + "grad_norm": 2.2747883796691895, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6998904943466187, + "num_tokens": 481625764.0, + "step": 18612 + }, + { + "epoch": 2.0440368987480784, + "grad_norm": 2.201178789138794, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7201823592185974, + "num_tokens": 481654419.0, + "step": 18613 + }, + { + "epoch": 2.0441467164506917, + "grad_norm": 2.6546452045440674, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7430605888366699, + "num_tokens": 481673690.0, + "step": 18614 + }, + { + "epoch": 2.0442565341533054, + "grad_norm": 2.3325493335723877, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7341809272766113, + "num_tokens": 481700537.0, + "step": 18615 + }, + { + "epoch": 2.044366351855919, + "grad_norm": 2.547658920288086, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7486249208450317, + "num_tokens": 481721716.0, + "step": 18616 + }, + { + "epoch": 2.044476169558533, + "grad_norm": 2.523442029953003, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7363017797470093, + "num_tokens": 481744694.0, + "step": 18617 + }, + { + "epoch": 2.0445859872611467, + "grad_norm": 2.198103904724121, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7330735921859741, + "num_tokens": 481773619.0, + "step": 18618 + }, + { + "epoch": 2.04469580496376, + "grad_norm": 2.523336410522461, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7445694208145142, + "num_tokens": 481796860.0, + "step": 18619 + }, + { + "epoch": 2.044805622666374, + "grad_norm": 2.169394016265869, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7314289808273315, + "num_tokens": 481824971.0, + "step": 18620 + }, + { + "epoch": 2.0449154403689875, + "grad_norm": 2.3880388736724854, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7347739934921265, + "num_tokens": 481851380.0, + "step": 18621 + }, + { + "epoch": 2.0450252580716013, + "grad_norm": 2.4916651248931885, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7468751668930054, + "num_tokens": 481875181.0, + "step": 18622 + }, + { + "epoch": 2.0451350757742146, + "grad_norm": 2.4512157440185547, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7112874984741211, + "num_tokens": 481901916.0, + "step": 18623 + }, + { + "epoch": 2.0452448934768284, + "grad_norm": 2.359471321105957, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7194072008132935, + "num_tokens": 481929392.0, + "step": 18624 + }, + { + "epoch": 2.045354711179442, + "grad_norm": 2.47239351272583, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.731734573841095, + "num_tokens": 481955024.0, + "step": 18625 + }, + { + "epoch": 2.045464528882056, + "grad_norm": 2.268317461013794, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7353217601776123, + "num_tokens": 481982868.0, + "step": 18626 + }, + { + "epoch": 2.0455743465846696, + "grad_norm": 2.2166569232940674, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7163722515106201, + "num_tokens": 482010241.0, + "step": 18627 + }, + { + "epoch": 2.045684164287283, + "grad_norm": 2.166607141494751, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7135088443756104, + "num_tokens": 482038554.0, + "step": 18628 + }, + { + "epoch": 2.0457939819898967, + "grad_norm": 2.048839569091797, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.719656229019165, + "num_tokens": 482071571.0, + "step": 18629 + }, + { + "epoch": 2.0459037996925105, + "grad_norm": 2.4076406955718994, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7128989100456238, + "num_tokens": 482094918.0, + "step": 18630 + }, + { + "epoch": 2.046013617395124, + "grad_norm": 2.9005179405212402, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7236760854721069, + "num_tokens": 482113080.0, + "step": 18631 + }, + { + "epoch": 2.046123435097738, + "grad_norm": 2.284095287322998, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7408344745635986, + "num_tokens": 482139261.0, + "step": 18632 + }, + { + "epoch": 2.0462332528003513, + "grad_norm": 2.7555088996887207, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7416957020759583, + "num_tokens": 482159933.0, + "step": 18633 + }, + { + "epoch": 2.046343070502965, + "grad_norm": 2.4579062461853027, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7263538837432861, + "num_tokens": 482184511.0, + "step": 18634 + }, + { + "epoch": 2.046452888205579, + "grad_norm": 2.1582698822021484, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7248479127883911, + "num_tokens": 482214353.0, + "step": 18635 + }, + { + "epoch": 2.0465627059081926, + "grad_norm": 2.34930157661438, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.7574859857559204, + "num_tokens": 482238019.0, + "step": 18636 + }, + { + "epoch": 2.046672523610806, + "grad_norm": 2.412693500518799, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7328324317932129, + "num_tokens": 482263259.0, + "step": 18637 + }, + { + "epoch": 2.0467823413134196, + "grad_norm": 2.051144599914551, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7331789135932922, + "num_tokens": 482297651.0, + "step": 18638 + }, + { + "epoch": 2.0468921590160334, + "grad_norm": 2.3032288551330566, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.6979958415031433, + "num_tokens": 482326127.0, + "step": 18639 + }, + { + "epoch": 2.047001976718647, + "grad_norm": 2.1478543281555176, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7300374507904053, + "num_tokens": 482355648.0, + "step": 18640 + }, + { + "epoch": 2.047111794421261, + "grad_norm": 1.8772424459457397, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7396096587181091, + "num_tokens": 482393914.0, + "step": 18641 + }, + { + "epoch": 2.047221612123874, + "grad_norm": 2.336951494216919, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7191649675369263, + "num_tokens": 482419669.0, + "step": 18642 + }, + { + "epoch": 2.047331429826488, + "grad_norm": 2.557838201522827, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7327024936676025, + "num_tokens": 482441513.0, + "step": 18643 + }, + { + "epoch": 2.0474412475291017, + "grad_norm": 2.440523862838745, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7163147926330566, + "num_tokens": 482467869.0, + "step": 18644 + }, + { + "epoch": 2.0475510652317155, + "grad_norm": 2.2281391620635986, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7395882606506348, + "num_tokens": 482496206.0, + "step": 18645 + }, + { + "epoch": 2.047660882934329, + "grad_norm": 2.4147632122039795, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7168235778808594, + "num_tokens": 482521093.0, + "step": 18646 + }, + { + "epoch": 2.0477707006369426, + "grad_norm": 2.4838125705718994, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7351728677749634, + "num_tokens": 482545337.0, + "step": 18647 + }, + { + "epoch": 2.0478805183395563, + "grad_norm": 2.259692668914795, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7600417137145996, + "num_tokens": 482572245.0, + "step": 18648 + }, + { + "epoch": 2.04799033604217, + "grad_norm": 2.3642067909240723, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7422040700912476, + "num_tokens": 482600882.0, + "step": 18649 + }, + { + "epoch": 2.048100153744784, + "grad_norm": 2.3881237506866455, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.744074285030365, + "num_tokens": 482626059.0, + "step": 18650 + }, + { + "epoch": 2.048209971447397, + "grad_norm": 2.1950185298919678, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.708570122718811, + "num_tokens": 482654852.0, + "step": 18651 + }, + { + "epoch": 2.048319789150011, + "grad_norm": 2.31815242767334, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7496705055236816, + "num_tokens": 482681155.0, + "step": 18652 + }, + { + "epoch": 2.0484296068526247, + "grad_norm": 2.718853712081909, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7501632571220398, + "num_tokens": 482699738.0, + "step": 18653 + }, + { + "epoch": 2.0485394245552384, + "grad_norm": 2.1684207916259766, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7201159000396729, + "num_tokens": 482729568.0, + "step": 18654 + }, + { + "epoch": 2.048649242257852, + "grad_norm": 2.4274163246154785, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7453541159629822, + "num_tokens": 482752127.0, + "step": 18655 + }, + { + "epoch": 2.0487590599604655, + "grad_norm": 2.489166259765625, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.7367650270462036, + "num_tokens": 482775429.0, + "step": 18656 + }, + { + "epoch": 2.0488688776630792, + "grad_norm": 2.2280261516571045, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.716981053352356, + "num_tokens": 482803518.0, + "step": 18657 + }, + { + "epoch": 2.048978695365693, + "grad_norm": 2.3146400451660156, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7245259284973145, + "num_tokens": 482829798.0, + "step": 18658 + }, + { + "epoch": 2.0490885130683067, + "grad_norm": 2.3243038654327393, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7368825674057007, + "num_tokens": 482858683.0, + "step": 18659 + }, + { + "epoch": 2.04919833077092, + "grad_norm": 2.238677740097046, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7250489592552185, + "num_tokens": 482884621.0, + "step": 18660 + }, + { + "epoch": 2.049308148473534, + "grad_norm": 2.2973852157592773, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7336652278900146, + "num_tokens": 482910212.0, + "step": 18661 + }, + { + "epoch": 2.0494179661761476, + "grad_norm": 2.14109206199646, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.745879054069519, + "num_tokens": 482938617.0, + "step": 18662 + }, + { + "epoch": 2.0495277838787613, + "grad_norm": 2.159390926361084, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7332372069358826, + "num_tokens": 482968113.0, + "step": 18663 + }, + { + "epoch": 2.049637601581375, + "grad_norm": 2.232231616973877, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7111910581588745, + "num_tokens": 482998324.0, + "step": 18664 + }, + { + "epoch": 2.0497474192839884, + "grad_norm": 2.2069358825683594, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.716869592666626, + "num_tokens": 483025136.0, + "step": 18665 + }, + { + "epoch": 2.049857236986602, + "grad_norm": 2.456721305847168, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7402057647705078, + "num_tokens": 483048026.0, + "step": 18666 + }, + { + "epoch": 2.049967054689216, + "grad_norm": 2.633995294570923, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7557808756828308, + "num_tokens": 483069609.0, + "step": 18667 + }, + { + "epoch": 2.0500768723918297, + "grad_norm": 2.4323928356170654, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7392220497131348, + "num_tokens": 483092995.0, + "step": 18668 + }, + { + "epoch": 2.0501866900944434, + "grad_norm": 2.2157392501831055, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7278216481208801, + "num_tokens": 483119666.0, + "step": 18669 + }, + { + "epoch": 2.0502965077970567, + "grad_norm": 2.088301420211792, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7599570751190186, + "num_tokens": 483150595.0, + "step": 18670 + }, + { + "epoch": 2.0504063254996705, + "grad_norm": 2.4276466369628906, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7430435419082642, + "num_tokens": 483173866.0, + "step": 18671 + }, + { + "epoch": 2.0505161432022843, + "grad_norm": 2.6377906799316406, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7234678268432617, + "num_tokens": 483195512.0, + "step": 18672 + }, + { + "epoch": 2.050625960904898, + "grad_norm": 2.1956381797790527, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7477233409881592, + "num_tokens": 483224278.0, + "step": 18673 + }, + { + "epoch": 2.0507357786075113, + "grad_norm": 2.2294726371765137, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.736388087272644, + "num_tokens": 483251606.0, + "step": 18674 + }, + { + "epoch": 2.050845596310125, + "grad_norm": 2.3867099285125732, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7254159450531006, + "num_tokens": 483276901.0, + "step": 18675 + }, + { + "epoch": 2.050955414012739, + "grad_norm": 2.253046989440918, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7436749935150146, + "num_tokens": 483304570.0, + "step": 18676 + }, + { + "epoch": 2.0510652317153526, + "grad_norm": 2.2201497554779053, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7571382522583008, + "num_tokens": 483330328.0, + "step": 18677 + }, + { + "epoch": 2.0511750494179664, + "grad_norm": 2.367417097091675, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.704418957233429, + "num_tokens": 483356872.0, + "step": 18678 + }, + { + "epoch": 2.0512848671205797, + "grad_norm": 2.395880699157715, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.721849262714386, + "num_tokens": 483382817.0, + "step": 18679 + }, + { + "epoch": 2.0513946848231934, + "grad_norm": 2.1697232723236084, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7546745538711548, + "num_tokens": 483411726.0, + "step": 18680 + }, + { + "epoch": 2.051504502525807, + "grad_norm": 2.2820136547088623, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.735441267490387, + "num_tokens": 483440186.0, + "step": 18681 + }, + { + "epoch": 2.051614320228421, + "grad_norm": 2.5297083854675293, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.739648163318634, + "num_tokens": 483463864.0, + "step": 18682 + }, + { + "epoch": 2.0517241379310347, + "grad_norm": 2.305935859680176, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7334482073783875, + "num_tokens": 483488716.0, + "step": 18683 + }, + { + "epoch": 2.051833955633648, + "grad_norm": 2.2005271911621094, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7171010971069336, + "num_tokens": 483516897.0, + "step": 18684 + }, + { + "epoch": 2.0519437733362618, + "grad_norm": 2.558290958404541, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7390257120132446, + "num_tokens": 483539481.0, + "step": 18685 + }, + { + "epoch": 2.0520535910388755, + "grad_norm": 2.465897560119629, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7501518726348877, + "num_tokens": 483562160.0, + "step": 18686 + }, + { + "epoch": 2.0521634087414893, + "grad_norm": 2.4772064685821533, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7257851958274841, + "num_tokens": 483587201.0, + "step": 18687 + }, + { + "epoch": 2.0522732264441026, + "grad_norm": 2.4897990226745605, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7378244996070862, + "num_tokens": 483610822.0, + "step": 18688 + }, + { + "epoch": 2.0523830441467164, + "grad_norm": 2.5713870525360107, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7288110852241516, + "num_tokens": 483632577.0, + "step": 18689 + }, + { + "epoch": 2.05249286184933, + "grad_norm": 2.280243396759033, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7215249538421631, + "num_tokens": 483659643.0, + "step": 18690 + }, + { + "epoch": 2.052602679551944, + "grad_norm": 2.417649269104004, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7313578128814697, + "num_tokens": 483685053.0, + "step": 18691 + }, + { + "epoch": 2.0527124972545576, + "grad_norm": 2.550724983215332, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7300845384597778, + "num_tokens": 483708837.0, + "step": 18692 + }, + { + "epoch": 2.052822314957171, + "grad_norm": 2.401387929916382, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7341588735580444, + "num_tokens": 483732545.0, + "step": 18693 + }, + { + "epoch": 2.0529321326597847, + "grad_norm": 2.4099669456481934, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7477062940597534, + "num_tokens": 483755112.0, + "step": 18694 + }, + { + "epoch": 2.0530419503623984, + "grad_norm": 2.546489715576172, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7420657873153687, + "num_tokens": 483777305.0, + "step": 18695 + }, + { + "epoch": 2.053151768065012, + "grad_norm": 2.3988986015319824, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7605098485946655, + "num_tokens": 483800500.0, + "step": 18696 + }, + { + "epoch": 2.0532615857676255, + "grad_norm": 2.307098150253296, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7347455620765686, + "num_tokens": 483826162.0, + "step": 18697 + }, + { + "epoch": 2.0533714034702393, + "grad_norm": 2.05930233001709, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7339007258415222, + "num_tokens": 483855322.0, + "step": 18698 + }, + { + "epoch": 2.053481221172853, + "grad_norm": 2.629650115966797, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.738325834274292, + "num_tokens": 483877944.0, + "step": 18699 + }, + { + "epoch": 2.053591038875467, + "grad_norm": 2.2585484981536865, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7371603846549988, + "num_tokens": 483904372.0, + "step": 18700 + }, + { + "epoch": 2.0537008565780805, + "grad_norm": 2.419823408126831, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7053800821304321, + "num_tokens": 483933983.0, + "step": 18701 + }, + { + "epoch": 2.053810674280694, + "grad_norm": 2.3900766372680664, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7248784303665161, + "num_tokens": 483960201.0, + "step": 18702 + }, + { + "epoch": 2.0539204919833076, + "grad_norm": 2.1232738494873047, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7160775661468506, + "num_tokens": 483990681.0, + "step": 18703 + }, + { + "epoch": 2.0540303096859214, + "grad_norm": 2.206179141998291, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7367392778396606, + "num_tokens": 484019494.0, + "step": 18704 + }, + { + "epoch": 2.054140127388535, + "grad_norm": 2.398786783218384, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7316849231719971, + "num_tokens": 484044720.0, + "step": 18705 + }, + { + "epoch": 2.054249945091149, + "grad_norm": 2.235989809036255, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7310301065444946, + "num_tokens": 484071022.0, + "step": 18706 + }, + { + "epoch": 2.054359762793762, + "grad_norm": 2.5803942680358887, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7605248689651489, + "num_tokens": 484089798.0, + "step": 18707 + }, + { + "epoch": 2.054469580496376, + "grad_norm": 2.3637166023254395, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7416746616363525, + "num_tokens": 484115280.0, + "step": 18708 + }, + { + "epoch": 2.0545793981989897, + "grad_norm": 2.541710376739502, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7254441380500793, + "num_tokens": 484139932.0, + "step": 18709 + }, + { + "epoch": 2.0546892159016035, + "grad_norm": 2.0325756072998047, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7528144717216492, + "num_tokens": 484170720.0, + "step": 18710 + }, + { + "epoch": 2.054799033604217, + "grad_norm": 2.1493289470672607, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.724305272102356, + "num_tokens": 484199463.0, + "step": 18711 + }, + { + "epoch": 2.0549088513068305, + "grad_norm": 2.55143141746521, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7469114065170288, + "num_tokens": 484223309.0, + "step": 18712 + }, + { + "epoch": 2.0550186690094443, + "grad_norm": 2.367900848388672, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7136681079864502, + "num_tokens": 484252023.0, + "step": 18713 + }, + { + "epoch": 2.055128486712058, + "grad_norm": 2.49243426322937, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7345038652420044, + "num_tokens": 484276029.0, + "step": 18714 + }, + { + "epoch": 2.055238304414672, + "grad_norm": 2.2509119510650635, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7151854038238525, + "num_tokens": 484305587.0, + "step": 18715 + }, + { + "epoch": 2.055348122117285, + "grad_norm": 2.8451287746429443, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7588162422180176, + "num_tokens": 484324145.0, + "step": 18716 + }, + { + "epoch": 2.055457939819899, + "grad_norm": 2.036843776702881, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.725811243057251, + "num_tokens": 484356125.0, + "step": 18717 + }, + { + "epoch": 2.0555677575225126, + "grad_norm": 2.2718658447265625, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7269824743270874, + "num_tokens": 484385600.0, + "step": 18718 + }, + { + "epoch": 2.0556775752251264, + "grad_norm": 2.4592604637145996, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7567079067230225, + "num_tokens": 484410154.0, + "step": 18719 + }, + { + "epoch": 2.05578739292774, + "grad_norm": 2.518428087234497, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7481183409690857, + "num_tokens": 484433103.0, + "step": 18720 + }, + { + "epoch": 2.0558972106303535, + "grad_norm": 2.2319540977478027, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.731170654296875, + "num_tokens": 484461954.0, + "step": 18721 + }, + { + "epoch": 2.0560070283329672, + "grad_norm": 2.363687038421631, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7620292901992798, + "num_tokens": 484484988.0, + "step": 18722 + }, + { + "epoch": 2.056116846035581, + "grad_norm": 2.103234052658081, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7279150485992432, + "num_tokens": 484516484.0, + "step": 18723 + }, + { + "epoch": 2.0562266637381947, + "grad_norm": 2.4363157749176025, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7280635833740234, + "num_tokens": 484539532.0, + "step": 18724 + }, + { + "epoch": 2.056336481440808, + "grad_norm": 2.5785984992980957, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7307522296905518, + "num_tokens": 484564198.0, + "step": 18725 + }, + { + "epoch": 2.056446299143422, + "grad_norm": 2.3160476684570312, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7225714921951294, + "num_tokens": 484590622.0, + "step": 18726 + }, + { + "epoch": 2.0565561168460356, + "grad_norm": 2.213759660720825, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.6996039152145386, + "num_tokens": 484619910.0, + "step": 18727 + }, + { + "epoch": 2.0566659345486493, + "grad_norm": 2.6385340690612793, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7397918701171875, + "num_tokens": 484640093.0, + "step": 18728 + }, + { + "epoch": 2.056775752251263, + "grad_norm": 2.3169047832489014, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7213478088378906, + "num_tokens": 484669591.0, + "step": 18729 + }, + { + "epoch": 2.0568855699538764, + "grad_norm": 2.398012161254883, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7310332655906677, + "num_tokens": 484693323.0, + "step": 18730 + }, + { + "epoch": 2.05699538765649, + "grad_norm": 2.6141140460968018, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7633610963821411, + "num_tokens": 484713895.0, + "step": 18731 + }, + { + "epoch": 2.057105205359104, + "grad_norm": 2.6327340602874756, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7720149755477905, + "num_tokens": 484734883.0, + "step": 18732 + }, + { + "epoch": 2.0572150230617177, + "grad_norm": 2.464003562927246, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7366743087768555, + "num_tokens": 484757425.0, + "step": 18733 + }, + { + "epoch": 2.0573248407643314, + "grad_norm": 2.306619882583618, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7354074716567993, + "num_tokens": 484782516.0, + "step": 18734 + }, + { + "epoch": 2.0574346584669447, + "grad_norm": 2.673413038253784, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.744749903678894, + "num_tokens": 484805520.0, + "step": 18735 + }, + { + "epoch": 2.0575444761695585, + "grad_norm": 2.3156378269195557, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7085862159729004, + "num_tokens": 484833321.0, + "step": 18736 + }, + { + "epoch": 2.0576542938721722, + "grad_norm": 2.076763391494751, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.6969606876373291, + "num_tokens": 484868316.0, + "step": 18737 + }, + { + "epoch": 2.057764111574786, + "grad_norm": 2.438643455505371, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7396677732467651, + "num_tokens": 484892628.0, + "step": 18738 + }, + { + "epoch": 2.0578739292773993, + "grad_norm": 2.206517219543457, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7273174524307251, + "num_tokens": 484919840.0, + "step": 18739 + }, + { + "epoch": 2.057983746980013, + "grad_norm": 2.121526002883911, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7395810484886169, + "num_tokens": 484949046.0, + "step": 18740 + }, + { + "epoch": 2.058093564682627, + "grad_norm": 2.454540729522705, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7226695418357849, + "num_tokens": 484973166.0, + "step": 18741 + }, + { + "epoch": 2.0582033823852406, + "grad_norm": 2.434541702270508, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7244418263435364, + "num_tokens": 484997745.0, + "step": 18742 + }, + { + "epoch": 2.0583132000878543, + "grad_norm": 2.1130642890930176, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.733301043510437, + "num_tokens": 485030912.0, + "step": 18743 + }, + { + "epoch": 2.0584230177904677, + "grad_norm": 2.2092158794403076, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7339285612106323, + "num_tokens": 485059141.0, + "step": 18744 + }, + { + "epoch": 2.0585328354930814, + "grad_norm": 2.4078023433685303, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7451930046081543, + "num_tokens": 485083742.0, + "step": 18745 + }, + { + "epoch": 2.058642653195695, + "grad_norm": 2.6744959354400635, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7428227663040161, + "num_tokens": 485105674.0, + "step": 18746 + }, + { + "epoch": 2.058752470898309, + "grad_norm": 2.20801043510437, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7319110631942749, + "num_tokens": 485132428.0, + "step": 18747 + }, + { + "epoch": 2.0588622886009227, + "grad_norm": 2.5058810710906982, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7424310445785522, + "num_tokens": 485155653.0, + "step": 18748 + }, + { + "epoch": 2.058972106303536, + "grad_norm": 2.320258855819702, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7300875782966614, + "num_tokens": 485183472.0, + "step": 18749 + }, + { + "epoch": 2.0590819240061498, + "grad_norm": 2.293308734893799, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7280658483505249, + "num_tokens": 485208921.0, + "step": 18750 + }, + { + "epoch": 2.0591917417087635, + "grad_norm": 2.025343656539917, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.721252977848053, + "num_tokens": 485243201.0, + "step": 18751 + }, + { + "epoch": 2.0593015594113773, + "grad_norm": 2.140763998031616, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7324388027191162, + "num_tokens": 485274135.0, + "step": 18752 + }, + { + "epoch": 2.0594113771139906, + "grad_norm": 2.2937533855438232, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7258091568946838, + "num_tokens": 485303106.0, + "step": 18753 + }, + { + "epoch": 2.0595211948166043, + "grad_norm": 2.392894744873047, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7378696203231812, + "num_tokens": 485326573.0, + "step": 18754 + }, + { + "epoch": 2.059631012519218, + "grad_norm": 2.266231060028076, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7350684404373169, + "num_tokens": 485354403.0, + "step": 18755 + }, + { + "epoch": 2.059740830221832, + "grad_norm": 2.3896431922912598, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7229090332984924, + "num_tokens": 485381116.0, + "step": 18756 + }, + { + "epoch": 2.0598506479244456, + "grad_norm": 2.4296445846557617, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7480459809303284, + "num_tokens": 485404852.0, + "step": 18757 + }, + { + "epoch": 2.059960465627059, + "grad_norm": 2.365154266357422, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7478751540184021, + "num_tokens": 485429368.0, + "step": 18758 + }, + { + "epoch": 2.0600702833296727, + "grad_norm": 2.027170181274414, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7191982865333557, + "num_tokens": 485462570.0, + "step": 18759 + }, + { + "epoch": 2.0601801010322864, + "grad_norm": 2.3092639446258545, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7515159845352173, + "num_tokens": 485486324.0, + "step": 18760 + }, + { + "epoch": 2.0602899187349, + "grad_norm": 2.3565194606781006, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7382062673568726, + "num_tokens": 485514079.0, + "step": 18761 + }, + { + "epoch": 2.060399736437514, + "grad_norm": 2.3426425457000732, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7152684926986694, + "num_tokens": 485539712.0, + "step": 18762 + }, + { + "epoch": 2.0605095541401273, + "grad_norm": 2.350651502609253, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7185617685317993, + "num_tokens": 485565998.0, + "step": 18763 + }, + { + "epoch": 2.060619371842741, + "grad_norm": 2.550342082977295, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7226120829582214, + "num_tokens": 485590371.0, + "step": 18764 + }, + { + "epoch": 2.060729189545355, + "grad_norm": 2.4836978912353516, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7342606782913208, + "num_tokens": 485613661.0, + "step": 18765 + }, + { + "epoch": 2.0608390072479685, + "grad_norm": 2.3503613471984863, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7359123826026917, + "num_tokens": 485640582.0, + "step": 18766 + }, + { + "epoch": 2.060948824950582, + "grad_norm": 2.3434841632843018, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.717305600643158, + "num_tokens": 485667177.0, + "step": 18767 + }, + { + "epoch": 2.0610586426531956, + "grad_norm": 2.1995885372161865, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7395268678665161, + "num_tokens": 485692337.0, + "step": 18768 + }, + { + "epoch": 2.0611684603558094, + "grad_norm": 2.4245121479034424, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7231118083000183, + "num_tokens": 485715895.0, + "step": 18769 + }, + { + "epoch": 2.061278278058423, + "grad_norm": 2.417320728302002, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7248781323432922, + "num_tokens": 485740056.0, + "step": 18770 + }, + { + "epoch": 2.061388095761037, + "grad_norm": 2.297755002975464, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7377065420150757, + "num_tokens": 485765531.0, + "step": 18771 + }, + { + "epoch": 2.06149791346365, + "grad_norm": 2.2901158332824707, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7350007891654968, + "num_tokens": 485790828.0, + "step": 18772 + }, + { + "epoch": 2.061607731166264, + "grad_norm": 2.1635780334472656, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7147116661071777, + "num_tokens": 485821307.0, + "step": 18773 + }, + { + "epoch": 2.0617175488688777, + "grad_norm": 2.4935057163238525, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7482905387878418, + "num_tokens": 485842368.0, + "step": 18774 + }, + { + "epoch": 2.0618273665714915, + "grad_norm": 2.236614942550659, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7327288389205933, + "num_tokens": 485869451.0, + "step": 18775 + }, + { + "epoch": 2.0619371842741048, + "grad_norm": 2.3537585735321045, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7276591658592224, + "num_tokens": 485894433.0, + "step": 18776 + }, + { + "epoch": 2.0620470019767185, + "grad_norm": 2.649134874343872, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7455617785453796, + "num_tokens": 485915047.0, + "step": 18777 + }, + { + "epoch": 2.0621568196793323, + "grad_norm": 2.2689123153686523, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7283426523208618, + "num_tokens": 485942877.0, + "step": 18778 + }, + { + "epoch": 2.062266637381946, + "grad_norm": 2.4476804733276367, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7337765693664551, + "num_tokens": 485966534.0, + "step": 18779 + }, + { + "epoch": 2.06237645508456, + "grad_norm": 2.8216490745544434, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7480404376983643, + "num_tokens": 485985634.0, + "step": 18780 + }, + { + "epoch": 2.062486272787173, + "grad_norm": 2.5954577922821045, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7338625192642212, + "num_tokens": 486012183.0, + "step": 18781 + }, + { + "epoch": 2.062596090489787, + "grad_norm": 2.2533576488494873, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7235246896743774, + "num_tokens": 486041381.0, + "step": 18782 + }, + { + "epoch": 2.0627059081924006, + "grad_norm": 2.310166597366333, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7408009767532349, + "num_tokens": 486067899.0, + "step": 18783 + }, + { + "epoch": 2.0628157258950144, + "grad_norm": 2.524134874343872, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7351678013801575, + "num_tokens": 486090785.0, + "step": 18784 + }, + { + "epoch": 2.062925543597628, + "grad_norm": 2.2620584964752197, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7295723557472229, + "num_tokens": 486117247.0, + "step": 18785 + }, + { + "epoch": 2.0630353613002415, + "grad_norm": 2.416694402694702, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7349475622177124, + "num_tokens": 486142138.0, + "step": 18786 + }, + { + "epoch": 2.063145179002855, + "grad_norm": 2.887303352355957, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7492944598197937, + "num_tokens": 486162436.0, + "step": 18787 + }, + { + "epoch": 2.063254996705469, + "grad_norm": 2.655831813812256, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7544575929641724, + "num_tokens": 486183170.0, + "step": 18788 + }, + { + "epoch": 2.0633648144080827, + "grad_norm": 2.0840680599212646, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.715923547744751, + "num_tokens": 486219250.0, + "step": 18789 + }, + { + "epoch": 2.063474632110696, + "grad_norm": 2.3126964569091797, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7219851016998291, + "num_tokens": 486248913.0, + "step": 18790 + }, + { + "epoch": 2.06358444981331, + "grad_norm": 2.560772180557251, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7422236204147339, + "num_tokens": 486270415.0, + "step": 18791 + }, + { + "epoch": 2.0636942675159236, + "grad_norm": 2.2936434745788574, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7112714052200317, + "num_tokens": 486298050.0, + "step": 18792 + }, + { + "epoch": 2.0638040852185373, + "grad_norm": 2.418509006500244, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7235876321792603, + "num_tokens": 486321495.0, + "step": 18793 + }, + { + "epoch": 2.063913902921151, + "grad_norm": 2.3263463973999023, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7414239645004272, + "num_tokens": 486346931.0, + "step": 18794 + }, + { + "epoch": 2.0640237206237644, + "grad_norm": 2.3999598026275635, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.720180332660675, + "num_tokens": 486374258.0, + "step": 18795 + }, + { + "epoch": 2.064133538326378, + "grad_norm": 2.303520917892456, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7237263321876526, + "num_tokens": 486401074.0, + "step": 18796 + }, + { + "epoch": 2.064243356028992, + "grad_norm": 2.3983752727508545, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7395859360694885, + "num_tokens": 486425411.0, + "step": 18797 + }, + { + "epoch": 2.0643531737316057, + "grad_norm": 2.434098243713379, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7322235107421875, + "num_tokens": 486450354.0, + "step": 18798 + }, + { + "epoch": 2.0644629914342194, + "grad_norm": 2.1528749465942383, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.743294894695282, + "num_tokens": 486479627.0, + "step": 18799 + }, + { + "epoch": 2.0645728091368327, + "grad_norm": 2.4633984565734863, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7347574830055237, + "num_tokens": 486502869.0, + "step": 18800 + }, + { + "epoch": 2.0646826268394465, + "grad_norm": 2.0625391006469727, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7182456254959106, + "num_tokens": 486534327.0, + "step": 18801 + }, + { + "epoch": 2.0647924445420602, + "grad_norm": 2.3965845108032227, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7183769941329956, + "num_tokens": 486559490.0, + "step": 18802 + }, + { + "epoch": 2.064902262244674, + "grad_norm": 2.6465134620666504, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7361599206924438, + "num_tokens": 486579601.0, + "step": 18803 + }, + { + "epoch": 2.0650120799472873, + "grad_norm": 2.6165413856506348, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7577811479568481, + "num_tokens": 486600957.0, + "step": 18804 + }, + { + "epoch": 2.065121897649901, + "grad_norm": 2.3852903842926025, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7383723258972168, + "num_tokens": 486627356.0, + "step": 18805 + }, + { + "epoch": 2.065231715352515, + "grad_norm": 2.3224880695343018, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7365827560424805, + "num_tokens": 486653002.0, + "step": 18806 + }, + { + "epoch": 2.0653415330551286, + "grad_norm": 2.4230659008026123, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7469077110290527, + "num_tokens": 486677465.0, + "step": 18807 + }, + { + "epoch": 2.0654513507577423, + "grad_norm": 2.4225265979766846, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7253196239471436, + "num_tokens": 486704398.0, + "step": 18808 + }, + { + "epoch": 2.0655611684603556, + "grad_norm": 2.1869447231292725, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.727995753288269, + "num_tokens": 486735319.0, + "step": 18809 + }, + { + "epoch": 2.0656709861629694, + "grad_norm": 2.6128432750701904, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7263227701187134, + "num_tokens": 486756914.0, + "step": 18810 + }, + { + "epoch": 2.065780803865583, + "grad_norm": 2.3185250759124756, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7419064044952393, + "num_tokens": 486781933.0, + "step": 18811 + }, + { + "epoch": 2.065890621568197, + "grad_norm": 2.301830291748047, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7305065393447876, + "num_tokens": 486808660.0, + "step": 18812 + }, + { + "epoch": 2.0660004392708107, + "grad_norm": 2.0706820487976074, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7157560586929321, + "num_tokens": 486844039.0, + "step": 18813 + }, + { + "epoch": 2.066110256973424, + "grad_norm": 1.9730744361877441, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7244814038276672, + "num_tokens": 486878633.0, + "step": 18814 + }, + { + "epoch": 2.0662200746760377, + "grad_norm": 2.2305898666381836, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7401999235153198, + "num_tokens": 486907779.0, + "step": 18815 + }, + { + "epoch": 2.0663298923786515, + "grad_norm": 2.2529616355895996, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7200467586517334, + "num_tokens": 486938039.0, + "step": 18816 + }, + { + "epoch": 2.0664397100812653, + "grad_norm": 2.226754903793335, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7118764519691467, + "num_tokens": 486967462.0, + "step": 18817 + }, + { + "epoch": 2.0665495277838786, + "grad_norm": 2.227201223373413, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7463517189025879, + "num_tokens": 486994806.0, + "step": 18818 + }, + { + "epoch": 2.0666593454864923, + "grad_norm": 2.3535284996032715, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7134535312652588, + "num_tokens": 487021312.0, + "step": 18819 + }, + { + "epoch": 2.066769163189106, + "grad_norm": 2.321348190307617, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7362064123153687, + "num_tokens": 487048531.0, + "step": 18820 + }, + { + "epoch": 2.06687898089172, + "grad_norm": 2.3668112754821777, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.735377311706543, + "num_tokens": 487074614.0, + "step": 18821 + }, + { + "epoch": 2.0669887985943336, + "grad_norm": 2.4610445499420166, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7332009077072144, + "num_tokens": 487097613.0, + "step": 18822 + }, + { + "epoch": 2.067098616296947, + "grad_norm": 2.3785746097564697, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7254608273506165, + "num_tokens": 487124192.0, + "step": 18823 + }, + { + "epoch": 2.0672084339995607, + "grad_norm": 2.059234857559204, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.6994332671165466, + "num_tokens": 487157659.0, + "step": 18824 + }, + { + "epoch": 2.0673182517021744, + "grad_norm": 2.182032823562622, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7560818791389465, + "num_tokens": 487186665.0, + "step": 18825 + }, + { + "epoch": 2.067428069404788, + "grad_norm": 2.2611567974090576, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7548925876617432, + "num_tokens": 487213021.0, + "step": 18826 + }, + { + "epoch": 2.0675378871074015, + "grad_norm": 2.4523398876190186, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7272048592567444, + "num_tokens": 487236826.0, + "step": 18827 + }, + { + "epoch": 2.0676477048100153, + "grad_norm": 2.514298677444458, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7646772861480713, + "num_tokens": 487257249.0, + "step": 18828 + }, + { + "epoch": 2.067757522512629, + "grad_norm": 2.473181962966919, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7398064732551575, + "num_tokens": 487279719.0, + "step": 18829 + }, + { + "epoch": 2.0678673402152428, + "grad_norm": 2.2075252532958984, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7290636301040649, + "num_tokens": 487305998.0, + "step": 18830 + }, + { + "epoch": 2.0679771579178565, + "grad_norm": 2.1654515266418457, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7479960918426514, + "num_tokens": 487335168.0, + "step": 18831 + }, + { + "epoch": 2.06808697562047, + "grad_norm": 2.404115915298462, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.735293447971344, + "num_tokens": 487359653.0, + "step": 18832 + }, + { + "epoch": 2.0681967933230836, + "grad_norm": 2.6159567832946777, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7509992122650146, + "num_tokens": 487379403.0, + "step": 18833 + }, + { + "epoch": 2.0683066110256974, + "grad_norm": 2.438934564590454, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7336803078651428, + "num_tokens": 487403805.0, + "step": 18834 + }, + { + "epoch": 2.068416428728311, + "grad_norm": 2.3113110065460205, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7346470952033997, + "num_tokens": 487429302.0, + "step": 18835 + }, + { + "epoch": 2.068526246430925, + "grad_norm": 2.291367292404175, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7586747407913208, + "num_tokens": 487454107.0, + "step": 18836 + }, + { + "epoch": 2.068636064133538, + "grad_norm": 2.2737512588500977, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7396823763847351, + "num_tokens": 487480299.0, + "step": 18837 + }, + { + "epoch": 2.068745881836152, + "grad_norm": 2.3054919242858887, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7230162620544434, + "num_tokens": 487507207.0, + "step": 18838 + }, + { + "epoch": 2.0688556995387657, + "grad_norm": 2.375375509262085, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7462680339813232, + "num_tokens": 487532758.0, + "step": 18839 + }, + { + "epoch": 2.0689655172413794, + "grad_norm": 2.573544502258301, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.747779905796051, + "num_tokens": 487555940.0, + "step": 18840 + }, + { + "epoch": 2.0690753349439928, + "grad_norm": 2.5014750957489014, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7235928773880005, + "num_tokens": 487581254.0, + "step": 18841 + }, + { + "epoch": 2.0691851526466065, + "grad_norm": 2.7573864459991455, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.753986120223999, + "num_tokens": 487600889.0, + "step": 18842 + }, + { + "epoch": 2.0692949703492203, + "grad_norm": 2.6673383712768555, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7521985173225403, + "num_tokens": 487620106.0, + "step": 18843 + }, + { + "epoch": 2.069404788051834, + "grad_norm": 2.0408244132995605, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7521486878395081, + "num_tokens": 487652019.0, + "step": 18844 + }, + { + "epoch": 2.069514605754448, + "grad_norm": 2.08915114402771, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7333569526672363, + "num_tokens": 487682072.0, + "step": 18845 + }, + { + "epoch": 2.069624423457061, + "grad_norm": 2.0789318084716797, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7277772426605225, + "num_tokens": 487712303.0, + "step": 18846 + }, + { + "epoch": 2.069734241159675, + "grad_norm": 2.552393913269043, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7445359230041504, + "num_tokens": 487733598.0, + "step": 18847 + }, + { + "epoch": 2.0698440588622886, + "grad_norm": 2.35713267326355, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7399143576622009, + "num_tokens": 487758116.0, + "step": 18848 + }, + { + "epoch": 2.0699538765649024, + "grad_norm": 2.3158278465270996, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7177593111991882, + "num_tokens": 487784928.0, + "step": 18849 + }, + { + "epoch": 2.070063694267516, + "grad_norm": 2.469183921813965, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.734055757522583, + "num_tokens": 487806999.0, + "step": 18850 + }, + { + "epoch": 2.0701735119701294, + "grad_norm": 2.436229944229126, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7509279251098633, + "num_tokens": 487829487.0, + "step": 18851 + }, + { + "epoch": 2.070283329672743, + "grad_norm": 2.26100492477417, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.726729691028595, + "num_tokens": 487857701.0, + "step": 18852 + }, + { + "epoch": 2.070393147375357, + "grad_norm": 2.191826820373535, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7247098684310913, + "num_tokens": 487886019.0, + "step": 18853 + }, + { + "epoch": 2.0705029650779707, + "grad_norm": 2.6323869228363037, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7532615065574646, + "num_tokens": 487906054.0, + "step": 18854 + }, + { + "epoch": 2.070612782780584, + "grad_norm": 2.2100517749786377, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.743051290512085, + "num_tokens": 487932732.0, + "step": 18855 + }, + { + "epoch": 2.070722600483198, + "grad_norm": 2.5207788944244385, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7216688394546509, + "num_tokens": 487954950.0, + "step": 18856 + }, + { + "epoch": 2.0708324181858115, + "grad_norm": 1.9231268167495728, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.728622555732727, + "num_tokens": 487989789.0, + "step": 18857 + }, + { + "epoch": 2.0709422358884253, + "grad_norm": 2.447063684463501, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7327526807785034, + "num_tokens": 488013438.0, + "step": 18858 + }, + { + "epoch": 2.071052053591039, + "grad_norm": 2.6744041442871094, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7417832612991333, + "num_tokens": 488033967.0, + "step": 18859 + }, + { + "epoch": 2.0711618712936524, + "grad_norm": 2.6429617404937744, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7357262372970581, + "num_tokens": 488055618.0, + "step": 18860 + }, + { + "epoch": 2.071271688996266, + "grad_norm": 2.4643707275390625, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7368267774581909, + "num_tokens": 488077824.0, + "step": 18861 + }, + { + "epoch": 2.07138150669888, + "grad_norm": 2.432002305984497, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7222264409065247, + "num_tokens": 488103374.0, + "step": 18862 + }, + { + "epoch": 2.0714913244014936, + "grad_norm": 2.204454183578491, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.745819091796875, + "num_tokens": 488129776.0, + "step": 18863 + }, + { + "epoch": 2.0716011421041074, + "grad_norm": 2.40100359916687, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7235199809074402, + "num_tokens": 488154567.0, + "step": 18864 + }, + { + "epoch": 2.0717109598067207, + "grad_norm": 2.651473045349121, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.740621030330658, + "num_tokens": 488175092.0, + "step": 18865 + }, + { + "epoch": 2.0718207775093345, + "grad_norm": 2.3684520721435547, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7087857723236084, + "num_tokens": 488202329.0, + "step": 18866 + }, + { + "epoch": 2.0719305952119482, + "grad_norm": 2.367637872695923, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.727705717086792, + "num_tokens": 488229448.0, + "step": 18867 + }, + { + "epoch": 2.072040412914562, + "grad_norm": 2.8395867347717285, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7389796376228333, + "num_tokens": 488248341.0, + "step": 18868 + }, + { + "epoch": 2.0721502306171753, + "grad_norm": 2.2826504707336426, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7166857123374939, + "num_tokens": 488276403.0, + "step": 18869 + }, + { + "epoch": 2.072260048319789, + "grad_norm": 2.4113047122955322, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7201204895973206, + "num_tokens": 488300829.0, + "step": 18870 + }, + { + "epoch": 2.072369866022403, + "grad_norm": 2.2021734714508057, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7214711904525757, + "num_tokens": 488330090.0, + "step": 18871 + }, + { + "epoch": 2.0724796837250166, + "grad_norm": 2.1959333419799805, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7342422604560852, + "num_tokens": 488357430.0, + "step": 18872 + }, + { + "epoch": 2.0725895014276303, + "grad_norm": 2.382539987564087, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.732453465461731, + "num_tokens": 488382389.0, + "step": 18873 + }, + { + "epoch": 2.0726993191302436, + "grad_norm": 2.437087297439575, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7454327344894409, + "num_tokens": 488405819.0, + "step": 18874 + }, + { + "epoch": 2.0728091368328574, + "grad_norm": 2.456159830093384, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7294277548789978, + "num_tokens": 488428902.0, + "step": 18875 + }, + { + "epoch": 2.072918954535471, + "grad_norm": 2.4702064990997314, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7509647607803345, + "num_tokens": 488451267.0, + "step": 18876 + }, + { + "epoch": 2.073028772238085, + "grad_norm": 2.1977651119232178, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.7372166514396667, + "num_tokens": 488477786.0, + "step": 18877 + }, + { + "epoch": 2.073138589940698, + "grad_norm": 2.4205970764160156, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7475195527076721, + "num_tokens": 488501779.0, + "step": 18878 + }, + { + "epoch": 2.073248407643312, + "grad_norm": 2.4151363372802734, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.729597806930542, + "num_tokens": 488525927.0, + "step": 18879 + }, + { + "epoch": 2.0733582253459257, + "grad_norm": 2.578364372253418, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7338581085205078, + "num_tokens": 488549374.0, + "step": 18880 + }, + { + "epoch": 2.0734680430485395, + "grad_norm": 2.138594627380371, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7167662978172302, + "num_tokens": 488581234.0, + "step": 18881 + }, + { + "epoch": 2.0735778607511532, + "grad_norm": 2.4702916145324707, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7387934327125549, + "num_tokens": 488604300.0, + "step": 18882 + }, + { + "epoch": 2.0736876784537666, + "grad_norm": 2.4372315406799316, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7279906868934631, + "num_tokens": 488628493.0, + "step": 18883 + }, + { + "epoch": 2.0737974961563803, + "grad_norm": 2.5945522785186768, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.757705569267273, + "num_tokens": 488649579.0, + "step": 18884 + }, + { + "epoch": 2.073907313858994, + "grad_norm": 2.2227206230163574, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7190936803817749, + "num_tokens": 488679453.0, + "step": 18885 + }, + { + "epoch": 2.074017131561608, + "grad_norm": 2.1780946254730225, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7384564876556396, + "num_tokens": 488706913.0, + "step": 18886 + }, + { + "epoch": 2.0741269492642216, + "grad_norm": 2.2477383613586426, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7378068566322327, + "num_tokens": 488733613.0, + "step": 18887 + }, + { + "epoch": 2.074236766966835, + "grad_norm": 2.2481889724731445, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.722792387008667, + "num_tokens": 488760801.0, + "step": 18888 + }, + { + "epoch": 2.0743465846694487, + "grad_norm": 2.2842392921447754, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7230937480926514, + "num_tokens": 488788687.0, + "step": 18889 + }, + { + "epoch": 2.0744564023720624, + "grad_norm": 2.4068195819854736, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7364711761474609, + "num_tokens": 488814072.0, + "step": 18890 + }, + { + "epoch": 2.074566220074676, + "grad_norm": 2.2294187545776367, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7580639123916626, + "num_tokens": 488841485.0, + "step": 18891 + }, + { + "epoch": 2.07467603777729, + "grad_norm": 2.268648862838745, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7403630614280701, + "num_tokens": 488867353.0, + "step": 18892 + }, + { + "epoch": 2.0747858554799032, + "grad_norm": 2.34097957611084, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7036752700805664, + "num_tokens": 488896078.0, + "step": 18893 + }, + { + "epoch": 2.074895673182517, + "grad_norm": 2.7040388584136963, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7324276566505432, + "num_tokens": 488917064.0, + "step": 18894 + }, + { + "epoch": 2.0750054908851308, + "grad_norm": 2.202828884124756, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7248210906982422, + "num_tokens": 488945488.0, + "step": 18895 + }, + { + "epoch": 2.0751153085877445, + "grad_norm": 2.3496930599212646, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7384904623031616, + "num_tokens": 488968188.0, + "step": 18896 + }, + { + "epoch": 2.075225126290358, + "grad_norm": 2.7965989112854004, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7426167726516724, + "num_tokens": 488989368.0, + "step": 18897 + }, + { + "epoch": 2.0753349439929716, + "grad_norm": 2.2659804821014404, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7233730554580688, + "num_tokens": 489019061.0, + "step": 18898 + }, + { + "epoch": 2.0754447616955853, + "grad_norm": 2.2084221839904785, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7177691459655762, + "num_tokens": 489049187.0, + "step": 18899 + }, + { + "epoch": 2.075554579398199, + "grad_norm": 2.4920520782470703, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7425843477249146, + "num_tokens": 489072262.0, + "step": 18900 + }, + { + "epoch": 2.075664397100813, + "grad_norm": 2.2826385498046875, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7365124225616455, + "num_tokens": 489098065.0, + "step": 18901 + }, + { + "epoch": 2.075774214803426, + "grad_norm": 2.465906858444214, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7338833808898926, + "num_tokens": 489122738.0, + "step": 18902 + }, + { + "epoch": 2.07588403250604, + "grad_norm": 2.3847668170928955, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7201579809188843, + "num_tokens": 489149658.0, + "step": 18903 + }, + { + "epoch": 2.0759938502086537, + "grad_norm": 2.591538190841675, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7457699775695801, + "num_tokens": 489170115.0, + "step": 18904 + }, + { + "epoch": 2.0761036679112674, + "grad_norm": 2.2555384635925293, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7098526358604431, + "num_tokens": 489198899.0, + "step": 18905 + }, + { + "epoch": 2.0762134856138807, + "grad_norm": 2.6185302734375, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.731630802154541, + "num_tokens": 489220994.0, + "step": 18906 + }, + { + "epoch": 2.0763233033164945, + "grad_norm": 2.1996560096740723, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7263650894165039, + "num_tokens": 489247684.0, + "step": 18907 + }, + { + "epoch": 2.0764331210191083, + "grad_norm": 2.278517961502075, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7423173189163208, + "num_tokens": 489274000.0, + "step": 18908 + }, + { + "epoch": 2.076542938721722, + "grad_norm": 2.220656394958496, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7341325283050537, + "num_tokens": 489301138.0, + "step": 18909 + }, + { + "epoch": 2.0766527564243358, + "grad_norm": 2.5760881900787354, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7370026111602783, + "num_tokens": 489323949.0, + "step": 18910 + }, + { + "epoch": 2.076762574126949, + "grad_norm": 2.3484439849853516, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7307097911834717, + "num_tokens": 489348177.0, + "step": 18911 + }, + { + "epoch": 2.076872391829563, + "grad_norm": 2.520003318786621, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7250419855117798, + "num_tokens": 489370653.0, + "step": 18912 + }, + { + "epoch": 2.0769822095321766, + "grad_norm": 2.4905154705047607, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7231990694999695, + "num_tokens": 489396076.0, + "step": 18913 + }, + { + "epoch": 2.0770920272347904, + "grad_norm": 2.1814868450164795, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7631877660751343, + "num_tokens": 489422322.0, + "step": 18914 + }, + { + "epoch": 2.077201844937404, + "grad_norm": 2.210193395614624, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7181035280227661, + "num_tokens": 489452241.0, + "step": 18915 + }, + { + "epoch": 2.0773116626400174, + "grad_norm": 2.6374008655548096, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7429794073104858, + "num_tokens": 489471717.0, + "step": 18916 + }, + { + "epoch": 2.077421480342631, + "grad_norm": 2.5457966327667236, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7530375123023987, + "num_tokens": 489493599.0, + "step": 18917 + }, + { + "epoch": 2.077531298045245, + "grad_norm": 2.4372308254241943, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7142230868339539, + "num_tokens": 489517991.0, + "step": 18918 + }, + { + "epoch": 2.0776411157478587, + "grad_norm": 2.0723414421081543, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7071365118026733, + "num_tokens": 489550331.0, + "step": 18919 + }, + { + "epoch": 2.077750933450472, + "grad_norm": 2.2202136516571045, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7218332886695862, + "num_tokens": 489576755.0, + "step": 18920 + }, + { + "epoch": 2.0778607511530858, + "grad_norm": 2.2692747116088867, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7416113615036011, + "num_tokens": 489603232.0, + "step": 18921 + }, + { + "epoch": 2.0779705688556995, + "grad_norm": 2.1412832736968994, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7107259035110474, + "num_tokens": 489633861.0, + "step": 18922 + }, + { + "epoch": 2.0780803865583133, + "grad_norm": 2.57073712348938, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7299734950065613, + "num_tokens": 489656715.0, + "step": 18923 + }, + { + "epoch": 2.078190204260927, + "grad_norm": 2.6889288425445557, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7355298399925232, + "num_tokens": 489678745.0, + "step": 18924 + }, + { + "epoch": 2.0783000219635404, + "grad_norm": 2.2770445346832275, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7264411449432373, + "num_tokens": 489705042.0, + "step": 18925 + }, + { + "epoch": 2.078409839666154, + "grad_norm": 2.292536735534668, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7302500009536743, + "num_tokens": 489732389.0, + "step": 18926 + }, + { + "epoch": 2.078519657368768, + "grad_norm": 2.430292844772339, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7050686478614807, + "num_tokens": 489756253.0, + "step": 18927 + }, + { + "epoch": 2.0786294750713816, + "grad_norm": 2.4971015453338623, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7490618824958801, + "num_tokens": 489779374.0, + "step": 18928 + }, + { + "epoch": 2.078739292773995, + "grad_norm": 2.5049703121185303, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7267825603485107, + "num_tokens": 489802145.0, + "step": 18929 + }, + { + "epoch": 2.0788491104766087, + "grad_norm": 2.093522071838379, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7153345346450806, + "num_tokens": 489832481.0, + "step": 18930 + }, + { + "epoch": 2.0789589281792225, + "grad_norm": 2.4397451877593994, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7210636138916016, + "num_tokens": 489855325.0, + "step": 18931 + }, + { + "epoch": 2.079068745881836, + "grad_norm": 2.3329267501831055, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7304554581642151, + "num_tokens": 489880338.0, + "step": 18932 + }, + { + "epoch": 2.07917856358445, + "grad_norm": 2.4622621536254883, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7407383918762207, + "num_tokens": 489904091.0, + "step": 18933 + }, + { + "epoch": 2.0792883812870633, + "grad_norm": 2.5309438705444336, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7274104356765747, + "num_tokens": 489928676.0, + "step": 18934 + }, + { + "epoch": 2.079398198989677, + "grad_norm": 2.205425977706909, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.727622389793396, + "num_tokens": 489956846.0, + "step": 18935 + }, + { + "epoch": 2.079508016692291, + "grad_norm": 2.4712331295013428, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7274942398071289, + "num_tokens": 489981568.0, + "step": 18936 + }, + { + "epoch": 2.0796178343949046, + "grad_norm": 2.2782480716705322, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7451185584068298, + "num_tokens": 490007835.0, + "step": 18937 + }, + { + "epoch": 2.0797276520975183, + "grad_norm": 2.5375959873199463, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.7853999137878418, + "num_tokens": 490027937.0, + "step": 18938 + }, + { + "epoch": 2.0798374698001316, + "grad_norm": 2.059180736541748, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7171276211738586, + "num_tokens": 490063001.0, + "step": 18939 + }, + { + "epoch": 2.0799472875027454, + "grad_norm": 2.355597496032715, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7429295778274536, + "num_tokens": 490088863.0, + "step": 18940 + }, + { + "epoch": 2.080057105205359, + "grad_norm": 2.516820192337036, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7615468502044678, + "num_tokens": 490110820.0, + "step": 18941 + }, + { + "epoch": 2.080166922907973, + "grad_norm": 2.1623446941375732, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7187605500221252, + "num_tokens": 490140847.0, + "step": 18942 + }, + { + "epoch": 2.0802767406105866, + "grad_norm": 2.325925588607788, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7059347629547119, + "num_tokens": 490166679.0, + "step": 18943 + }, + { + "epoch": 2.0803865583132, + "grad_norm": 2.6634676456451416, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7441576719284058, + "num_tokens": 490187894.0, + "step": 18944 + }, + { + "epoch": 2.0804963760158137, + "grad_norm": 1.9087917804718018, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7284473776817322, + "num_tokens": 490221796.0, + "step": 18945 + }, + { + "epoch": 2.0806061937184275, + "grad_norm": 2.5451371669769287, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.732151985168457, + "num_tokens": 490243502.0, + "step": 18946 + }, + { + "epoch": 2.0807160114210412, + "grad_norm": 2.2196035385131836, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7470777034759521, + "num_tokens": 490271915.0, + "step": 18947 + }, + { + "epoch": 2.0808258291236545, + "grad_norm": 2.2475109100341797, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7352399826049805, + "num_tokens": 490300156.0, + "step": 18948 + }, + { + "epoch": 2.0809356468262683, + "grad_norm": 2.2580389976501465, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7029992341995239, + "num_tokens": 490330308.0, + "step": 18949 + }, + { + "epoch": 2.081045464528882, + "grad_norm": 2.214810371398926, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7152990102767944, + "num_tokens": 490358907.0, + "step": 18950 + }, + { + "epoch": 2.081155282231496, + "grad_norm": 2.440408945083618, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.722783088684082, + "num_tokens": 490383541.0, + "step": 18951 + }, + { + "epoch": 2.0812650999341096, + "grad_norm": 2.5171449184417725, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7373009920120239, + "num_tokens": 490409125.0, + "step": 18952 + }, + { + "epoch": 2.081374917636723, + "grad_norm": 2.4435412883758545, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7392532825469971, + "num_tokens": 490432306.0, + "step": 18953 + }, + { + "epoch": 2.0814847353393366, + "grad_norm": 2.5847911834716797, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7241040468215942, + "num_tokens": 490454009.0, + "step": 18954 + }, + { + "epoch": 2.0815945530419504, + "grad_norm": 2.257991075515747, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7174696922302246, + "num_tokens": 490482565.0, + "step": 18955 + }, + { + "epoch": 2.081704370744564, + "grad_norm": 1.973402500152588, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7072784900665283, + "num_tokens": 490519834.0, + "step": 18956 + }, + { + "epoch": 2.0818141884471775, + "grad_norm": 2.3148560523986816, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7563936710357666, + "num_tokens": 490546074.0, + "step": 18957 + }, + { + "epoch": 2.0819240061497912, + "grad_norm": 2.515465021133423, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7620489001274109, + "num_tokens": 490568911.0, + "step": 18958 + }, + { + "epoch": 2.082033823852405, + "grad_norm": 2.2863309383392334, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7169404625892639, + "num_tokens": 490594591.0, + "step": 18959 + }, + { + "epoch": 2.0821436415550187, + "grad_norm": 2.3411991596221924, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7350990772247314, + "num_tokens": 490619546.0, + "step": 18960 + }, + { + "epoch": 2.0822534592576325, + "grad_norm": 2.319532871246338, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7332190871238708, + "num_tokens": 490648240.0, + "step": 18961 + }, + { + "epoch": 2.082363276960246, + "grad_norm": 2.323240280151367, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7287797331809998, + "num_tokens": 490672451.0, + "step": 18962 + }, + { + "epoch": 2.0824730946628596, + "grad_norm": 2.383366346359253, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7454603314399719, + "num_tokens": 490698015.0, + "step": 18963 + }, + { + "epoch": 2.0825829123654733, + "grad_norm": 2.396268844604492, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7499613761901855, + "num_tokens": 490721407.0, + "step": 18964 + }, + { + "epoch": 2.082692730068087, + "grad_norm": 2.4822328090667725, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7553458213806152, + "num_tokens": 490743600.0, + "step": 18965 + }, + { + "epoch": 2.082802547770701, + "grad_norm": 2.4428961277008057, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7354158759117126, + "num_tokens": 490766922.0, + "step": 18966 + }, + { + "epoch": 2.082912365473314, + "grad_norm": 2.695646047592163, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7494141459465027, + "num_tokens": 490787071.0, + "step": 18967 + }, + { + "epoch": 2.083022183175928, + "grad_norm": 2.009143829345703, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7331740856170654, + "num_tokens": 490818928.0, + "step": 18968 + }, + { + "epoch": 2.0831320008785417, + "grad_norm": 2.282255172729492, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7235956192016602, + "num_tokens": 490845146.0, + "step": 18969 + }, + { + "epoch": 2.0832418185811554, + "grad_norm": 2.4150309562683105, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7241420745849609, + "num_tokens": 490870347.0, + "step": 18970 + }, + { + "epoch": 2.0833516362837687, + "grad_norm": 2.4190711975097656, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7249634265899658, + "num_tokens": 490897708.0, + "step": 18971 + }, + { + "epoch": 2.0834614539863825, + "grad_norm": 2.2391769886016846, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7202407121658325, + "num_tokens": 490926444.0, + "step": 18972 + }, + { + "epoch": 2.0835712716889963, + "grad_norm": 2.3952887058258057, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7451893091201782, + "num_tokens": 490951897.0, + "step": 18973 + }, + { + "epoch": 2.08368108939161, + "grad_norm": 1.9807488918304443, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7061543464660645, + "num_tokens": 490987745.0, + "step": 18974 + }, + { + "epoch": 2.0837909070942238, + "grad_norm": 2.1448888778686523, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7154706120491028, + "num_tokens": 491018623.0, + "step": 18975 + }, + { + "epoch": 2.083900724796837, + "grad_norm": 2.4320228099823, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7362792491912842, + "num_tokens": 491042741.0, + "step": 18976 + }, + { + "epoch": 2.084010542499451, + "grad_norm": 2.2552967071533203, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7316425442695618, + "num_tokens": 491068750.0, + "step": 18977 + }, + { + "epoch": 2.0841203602020646, + "grad_norm": 2.421053171157837, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.713947057723999, + "num_tokens": 491092239.0, + "step": 18978 + }, + { + "epoch": 2.0842301779046783, + "grad_norm": 2.5319509506225586, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7513239979743958, + "num_tokens": 491114475.0, + "step": 18979 + }, + { + "epoch": 2.084339995607292, + "grad_norm": 2.6980509757995605, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7439447641372681, + "num_tokens": 491134611.0, + "step": 18980 + }, + { + "epoch": 2.0844498133099054, + "grad_norm": 2.6355745792388916, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.736943244934082, + "num_tokens": 491155132.0, + "step": 18981 + }, + { + "epoch": 2.084559631012519, + "grad_norm": 2.5193302631378174, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7439954280853271, + "num_tokens": 491178243.0, + "step": 18982 + }, + { + "epoch": 2.084669448715133, + "grad_norm": 2.1353726387023926, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7069935202598572, + "num_tokens": 491209504.0, + "step": 18983 + }, + { + "epoch": 2.0847792664177467, + "grad_norm": 2.5492610931396484, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7288684248924255, + "num_tokens": 491234330.0, + "step": 18984 + }, + { + "epoch": 2.08488908412036, + "grad_norm": 2.198760509490967, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7345269322395325, + "num_tokens": 491262831.0, + "step": 18985 + }, + { + "epoch": 2.0849989018229738, + "grad_norm": 2.3046765327453613, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7103171348571777, + "num_tokens": 491290913.0, + "step": 18986 + }, + { + "epoch": 2.0851087195255875, + "grad_norm": 2.558823823928833, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7172631025314331, + "num_tokens": 491314614.0, + "step": 18987 + }, + { + "epoch": 2.0852185372282013, + "grad_norm": 2.2070670127868652, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7176434993743896, + "num_tokens": 491344133.0, + "step": 18988 + }, + { + "epoch": 2.085328354930815, + "grad_norm": 2.3775360584259033, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7310588359832764, + "num_tokens": 491370174.0, + "step": 18989 + }, + { + "epoch": 2.0854381726334283, + "grad_norm": 2.350329875946045, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7407666444778442, + "num_tokens": 491396116.0, + "step": 18990 + }, + { + "epoch": 2.085547990336042, + "grad_norm": 2.120150566101074, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7036694288253784, + "num_tokens": 491430346.0, + "step": 18991 + }, + { + "epoch": 2.085657808038656, + "grad_norm": 2.289433002471924, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7319533824920654, + "num_tokens": 491458979.0, + "step": 18992 + }, + { + "epoch": 2.0857676257412696, + "grad_norm": 2.4442718029022217, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7469890117645264, + "num_tokens": 491484176.0, + "step": 18993 + }, + { + "epoch": 2.0858774434438834, + "grad_norm": 2.3625376224517822, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7321062088012695, + "num_tokens": 491509964.0, + "step": 18994 + }, + { + "epoch": 2.0859872611464967, + "grad_norm": 2.6081080436706543, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7314895391464233, + "num_tokens": 491531663.0, + "step": 18995 + }, + { + "epoch": 2.0860970788491104, + "grad_norm": 2.7185933589935303, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.720402717590332, + "num_tokens": 491554738.0, + "step": 18996 + }, + { + "epoch": 2.086206896551724, + "grad_norm": 2.4494283199310303, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7424294352531433, + "num_tokens": 491578738.0, + "step": 18997 + }, + { + "epoch": 2.086316714254338, + "grad_norm": 2.1965339183807373, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7137938737869263, + "num_tokens": 491606169.0, + "step": 18998 + }, + { + "epoch": 2.0864265319569513, + "grad_norm": 2.1913821697235107, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7400534749031067, + "num_tokens": 491634519.0, + "step": 18999 + }, + { + "epoch": 2.086536349659565, + "grad_norm": 2.2847025394439697, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7055248618125916, + "num_tokens": 491664802.0, + "step": 19000 + }, + { + "epoch": 2.086646167362179, + "grad_norm": 1.9348238706588745, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7424504160881042, + "num_tokens": 491699769.0, + "step": 19001 + }, + { + "epoch": 2.0867559850647925, + "grad_norm": 2.3601768016815186, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7310360670089722, + "num_tokens": 491726122.0, + "step": 19002 + }, + { + "epoch": 2.0868658027674063, + "grad_norm": 2.42368221282959, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7431052327156067, + "num_tokens": 491751062.0, + "step": 19003 + }, + { + "epoch": 2.0869756204700196, + "grad_norm": 2.3446476459503174, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.737115204334259, + "num_tokens": 491777880.0, + "step": 19004 + }, + { + "epoch": 2.0870854381726334, + "grad_norm": 2.318549156188965, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7288229465484619, + "num_tokens": 491803602.0, + "step": 19005 + }, + { + "epoch": 2.087195255875247, + "grad_norm": 2.093395948410034, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7319045066833496, + "num_tokens": 491833549.0, + "step": 19006 + }, + { + "epoch": 2.087305073577861, + "grad_norm": 2.5777132511138916, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7086378931999207, + "num_tokens": 491855771.0, + "step": 19007 + }, + { + "epoch": 2.087414891280474, + "grad_norm": 2.179647922515869, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7303597927093506, + "num_tokens": 491883597.0, + "step": 19008 + }, + { + "epoch": 2.087524708983088, + "grad_norm": 2.1731808185577393, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7189255952835083, + "num_tokens": 491912746.0, + "step": 19009 + }, + { + "epoch": 2.0876345266857017, + "grad_norm": 2.4123406410217285, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7370456457138062, + "num_tokens": 491937029.0, + "step": 19010 + }, + { + "epoch": 2.0877443443883155, + "grad_norm": 2.4993557929992676, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7426153421401978, + "num_tokens": 491959920.0, + "step": 19011 + }, + { + "epoch": 2.087854162090929, + "grad_norm": 2.158505916595459, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7447985410690308, + "num_tokens": 491989044.0, + "step": 19012 + }, + { + "epoch": 2.0879639797935425, + "grad_norm": 2.427485704421997, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7316931486129761, + "num_tokens": 492013024.0, + "step": 19013 + }, + { + "epoch": 2.0880737974961563, + "grad_norm": 2.1233417987823486, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7294525504112244, + "num_tokens": 492044906.0, + "step": 19014 + }, + { + "epoch": 2.08818361519877, + "grad_norm": 2.314067840576172, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7359391450881958, + "num_tokens": 492070927.0, + "step": 19015 + }, + { + "epoch": 2.088293432901384, + "grad_norm": 2.4191477298736572, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7281607389450073, + "num_tokens": 492095050.0, + "step": 19016 + }, + { + "epoch": 2.0884032506039976, + "grad_norm": 2.646360158920288, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7331025004386902, + "num_tokens": 492115775.0, + "step": 19017 + }, + { + "epoch": 2.088513068306611, + "grad_norm": 2.4782986640930176, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7273485064506531, + "num_tokens": 492139040.0, + "step": 19018 + }, + { + "epoch": 2.0886228860092246, + "grad_norm": 2.345822334289551, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7245581746101379, + "num_tokens": 492167280.0, + "step": 19019 + }, + { + "epoch": 2.0887327037118384, + "grad_norm": 2.3995227813720703, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7264533638954163, + "num_tokens": 492193315.0, + "step": 19020 + }, + { + "epoch": 2.088842521414452, + "grad_norm": 2.46968936920166, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.728126049041748, + "num_tokens": 492217710.0, + "step": 19021 + }, + { + "epoch": 2.0889523391170655, + "grad_norm": 2.730152130126953, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7238925695419312, + "num_tokens": 492238561.0, + "step": 19022 + }, + { + "epoch": 2.089062156819679, + "grad_norm": 2.519401788711548, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7294438481330872, + "num_tokens": 492261352.0, + "step": 19023 + }, + { + "epoch": 2.089171974522293, + "grad_norm": 2.445664644241333, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.730165958404541, + "num_tokens": 492286699.0, + "step": 19024 + }, + { + "epoch": 2.0892817922249067, + "grad_norm": 2.5319302082061768, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7437869310379028, + "num_tokens": 492309374.0, + "step": 19025 + }, + { + "epoch": 2.0893916099275205, + "grad_norm": 2.0514469146728516, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7203649282455444, + "num_tokens": 492342546.0, + "step": 19026 + }, + { + "epoch": 2.089501427630134, + "grad_norm": 2.2887661457061768, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7508653402328491, + "num_tokens": 492369335.0, + "step": 19027 + }, + { + "epoch": 2.0896112453327476, + "grad_norm": 2.510263681411743, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7252813577651978, + "num_tokens": 492393485.0, + "step": 19028 + }, + { + "epoch": 2.0897210630353613, + "grad_norm": 2.6511788368225098, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7305066585540771, + "num_tokens": 492417059.0, + "step": 19029 + }, + { + "epoch": 2.089830880737975, + "grad_norm": 2.719839572906494, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7315381765365601, + "num_tokens": 492438264.0, + "step": 19030 + }, + { + "epoch": 2.089940698440589, + "grad_norm": 2.420259952545166, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.717858612537384, + "num_tokens": 492463019.0, + "step": 19031 + }, + { + "epoch": 2.090050516143202, + "grad_norm": 2.313823938369751, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7293367385864258, + "num_tokens": 492489882.0, + "step": 19032 + }, + { + "epoch": 2.090160333845816, + "grad_norm": 2.3821911811828613, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7459090948104858, + "num_tokens": 492514387.0, + "step": 19033 + }, + { + "epoch": 2.0902701515484297, + "grad_norm": 2.654259443283081, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7538326382637024, + "num_tokens": 492535709.0, + "step": 19034 + }, + { + "epoch": 2.0903799692510434, + "grad_norm": 2.451392889022827, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7327756881713867, + "num_tokens": 492558357.0, + "step": 19035 + }, + { + "epoch": 2.0904897869536567, + "grad_norm": 2.3877813816070557, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7221936583518982, + "num_tokens": 492585376.0, + "step": 19036 + }, + { + "epoch": 2.0905996046562705, + "grad_norm": 2.1563684940338135, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7146196365356445, + "num_tokens": 492614595.0, + "step": 19037 + }, + { + "epoch": 2.0907094223588842, + "grad_norm": 2.7593817710876465, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.7457472681999207, + "num_tokens": 492632727.0, + "step": 19038 + }, + { + "epoch": 2.090819240061498, + "grad_norm": 2.1034016609191895, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7145190238952637, + "num_tokens": 492662994.0, + "step": 19039 + }, + { + "epoch": 2.0909290577641118, + "grad_norm": 2.3308186531066895, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7299386262893677, + "num_tokens": 492690017.0, + "step": 19040 + }, + { + "epoch": 2.091038875466725, + "grad_norm": 2.2368574142456055, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7333871722221375, + "num_tokens": 492718107.0, + "step": 19041 + }, + { + "epoch": 2.091148693169339, + "grad_norm": 2.555755615234375, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7349823117256165, + "num_tokens": 492741305.0, + "step": 19042 + }, + { + "epoch": 2.0912585108719526, + "grad_norm": 2.300901174545288, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7314372062683105, + "num_tokens": 492766787.0, + "step": 19043 + }, + { + "epoch": 2.0913683285745663, + "grad_norm": 2.336014986038208, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7331777811050415, + "num_tokens": 492793789.0, + "step": 19044 + }, + { + "epoch": 2.09147814627718, + "grad_norm": 2.254117965698242, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.744157075881958, + "num_tokens": 492820917.0, + "step": 19045 + }, + { + "epoch": 2.0915879639797934, + "grad_norm": 2.179084300994873, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7123988270759583, + "num_tokens": 492849776.0, + "step": 19046 + }, + { + "epoch": 2.091697781682407, + "grad_norm": 2.190873146057129, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7428632974624634, + "num_tokens": 492877582.0, + "step": 19047 + }, + { + "epoch": 2.091807599385021, + "grad_norm": 2.3472044467926025, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7056560516357422, + "num_tokens": 492906048.0, + "step": 19048 + }, + { + "epoch": 2.0919174170876347, + "grad_norm": 2.5434770584106445, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7288430333137512, + "num_tokens": 492928817.0, + "step": 19049 + }, + { + "epoch": 2.092027234790248, + "grad_norm": 2.5537867546081543, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7386317253112793, + "num_tokens": 492951138.0, + "step": 19050 + }, + { + "epoch": 2.0921370524928617, + "grad_norm": 1.9613269567489624, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.739819347858429, + "num_tokens": 492985746.0, + "step": 19051 + }, + { + "epoch": 2.0922468701954755, + "grad_norm": 2.404550552368164, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7382534146308899, + "num_tokens": 493008945.0, + "step": 19052 + }, + { + "epoch": 2.0923566878980893, + "grad_norm": 2.499070167541504, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.738396167755127, + "num_tokens": 493032172.0, + "step": 19053 + }, + { + "epoch": 2.092466505600703, + "grad_norm": 2.0177268981933594, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7197838425636292, + "num_tokens": 493064895.0, + "step": 19054 + }, + { + "epoch": 2.0925763233033163, + "grad_norm": 2.2713921070098877, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7019383907318115, + "num_tokens": 493094111.0, + "step": 19055 + }, + { + "epoch": 2.09268614100593, + "grad_norm": 2.222564935684204, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7361568212509155, + "num_tokens": 493120831.0, + "step": 19056 + }, + { + "epoch": 2.092795958708544, + "grad_norm": 2.2410149574279785, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7344325184822083, + "num_tokens": 493148749.0, + "step": 19057 + }, + { + "epoch": 2.0929057764111576, + "grad_norm": 2.0567450523376465, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7224279642105103, + "num_tokens": 493187015.0, + "step": 19058 + }, + { + "epoch": 2.093015594113771, + "grad_norm": 2.5561161041259766, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7243702411651611, + "num_tokens": 493211688.0, + "step": 19059 + }, + { + "epoch": 2.0931254118163847, + "grad_norm": 2.31152606010437, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7364675998687744, + "num_tokens": 493238035.0, + "step": 19060 + }, + { + "epoch": 2.0932352295189984, + "grad_norm": 2.2253475189208984, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7474836111068726, + "num_tokens": 493265714.0, + "step": 19061 + }, + { + "epoch": 2.093345047221612, + "grad_norm": 2.215989828109741, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.751040518283844, + "num_tokens": 493294297.0, + "step": 19062 + }, + { + "epoch": 2.093454864924226, + "grad_norm": 2.505880832672119, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7552483081817627, + "num_tokens": 493315677.0, + "step": 19063 + }, + { + "epoch": 2.0935646826268393, + "grad_norm": 2.3938872814178467, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7153868079185486, + "num_tokens": 493342213.0, + "step": 19064 + }, + { + "epoch": 2.093674500329453, + "grad_norm": 2.469219207763672, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7315068244934082, + "num_tokens": 493365307.0, + "step": 19065 + }, + { + "epoch": 2.0937843180320668, + "grad_norm": 2.1157143115997314, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7214085459709167, + "num_tokens": 493396239.0, + "step": 19066 + }, + { + "epoch": 2.0938941357346805, + "grad_norm": 2.8160288333892822, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7329264879226685, + "num_tokens": 493416150.0, + "step": 19067 + }, + { + "epoch": 2.0940039534372943, + "grad_norm": 2.4454970359802246, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7114049196243286, + "num_tokens": 493440906.0, + "step": 19068 + }, + { + "epoch": 2.0941137711399076, + "grad_norm": 2.255073070526123, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7457276582717896, + "num_tokens": 493467869.0, + "step": 19069 + }, + { + "epoch": 2.0942235888425214, + "grad_norm": 2.1908183097839355, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7256002426147461, + "num_tokens": 493496899.0, + "step": 19070 + }, + { + "epoch": 2.094333406545135, + "grad_norm": 2.6941885948181152, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.745092511177063, + "num_tokens": 493518423.0, + "step": 19071 + }, + { + "epoch": 2.094443224247749, + "grad_norm": 2.3812742233276367, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7273403406143188, + "num_tokens": 493543772.0, + "step": 19072 + }, + { + "epoch": 2.0945530419503626, + "grad_norm": 2.3237314224243164, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7454549670219421, + "num_tokens": 493569302.0, + "step": 19073 + }, + { + "epoch": 2.094662859652976, + "grad_norm": 2.794008493423462, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7374544143676758, + "num_tokens": 493590965.0, + "step": 19074 + }, + { + "epoch": 2.0947726773555897, + "grad_norm": 2.3211214542388916, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7124789953231812, + "num_tokens": 493618059.0, + "step": 19075 + }, + { + "epoch": 2.0948824950582035, + "grad_norm": 2.297637462615967, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7377520203590393, + "num_tokens": 493644737.0, + "step": 19076 + }, + { + "epoch": 2.094992312760817, + "grad_norm": 2.3775200843811035, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7184854745864868, + "num_tokens": 493670418.0, + "step": 19077 + }, + { + "epoch": 2.0951021304634305, + "grad_norm": 2.0904529094696045, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7459315061569214, + "num_tokens": 493700392.0, + "step": 19078 + }, + { + "epoch": 2.0952119481660443, + "grad_norm": 2.1918346881866455, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7470287084579468, + "num_tokens": 493727488.0, + "step": 19079 + }, + { + "epoch": 2.095321765868658, + "grad_norm": 2.4274327754974365, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7521325945854187, + "num_tokens": 493750634.0, + "step": 19080 + }, + { + "epoch": 2.095431583571272, + "grad_norm": 3.2265207767486572, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7546367049217224, + "num_tokens": 493766122.0, + "step": 19081 + }, + { + "epoch": 2.0955414012738856, + "grad_norm": 2.282623767852783, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7063970565795898, + "num_tokens": 493792108.0, + "step": 19082 + }, + { + "epoch": 2.095651218976499, + "grad_norm": 2.2155978679656982, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7346670627593994, + "num_tokens": 493818761.0, + "step": 19083 + }, + { + "epoch": 2.0957610366791126, + "grad_norm": 2.4626829624176025, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7413005232810974, + "num_tokens": 493842579.0, + "step": 19084 + }, + { + "epoch": 2.0958708543817264, + "grad_norm": 2.511599540710449, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7262914776802063, + "num_tokens": 493866585.0, + "step": 19085 + }, + { + "epoch": 2.09598067208434, + "grad_norm": 2.4279282093048096, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7321361303329468, + "num_tokens": 493891549.0, + "step": 19086 + }, + { + "epoch": 2.0960904897869534, + "grad_norm": 2.3750789165496826, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7116333246231079, + "num_tokens": 493917048.0, + "step": 19087 + }, + { + "epoch": 2.096200307489567, + "grad_norm": 2.805717945098877, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7328884601593018, + "num_tokens": 493936071.0, + "step": 19088 + }, + { + "epoch": 2.096310125192181, + "grad_norm": 2.581636667251587, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7346876859664917, + "num_tokens": 493957816.0, + "step": 19089 + }, + { + "epoch": 2.0964199428947947, + "grad_norm": 2.2687671184539795, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7159932255744934, + "num_tokens": 493983395.0, + "step": 19090 + }, + { + "epoch": 2.0965297605974085, + "grad_norm": 2.3097684383392334, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7117358446121216, + "num_tokens": 494012211.0, + "step": 19091 + }, + { + "epoch": 2.096639578300022, + "grad_norm": 2.2942991256713867, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7212023735046387, + "num_tokens": 494038658.0, + "step": 19092 + }, + { + "epoch": 2.0967493960026355, + "grad_norm": 2.359825849533081, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7329162955284119, + "num_tokens": 494064052.0, + "step": 19093 + }, + { + "epoch": 2.0968592137052493, + "grad_norm": 2.3774492740631104, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7336434721946716, + "num_tokens": 494089375.0, + "step": 19094 + }, + { + "epoch": 2.096969031407863, + "grad_norm": 2.2066266536712646, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7294697761535645, + "num_tokens": 494118483.0, + "step": 19095 + }, + { + "epoch": 2.097078849110477, + "grad_norm": 2.5638575553894043, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7114853858947754, + "num_tokens": 494141958.0, + "step": 19096 + }, + { + "epoch": 2.09718866681309, + "grad_norm": 2.172774076461792, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7277069091796875, + "num_tokens": 494170703.0, + "step": 19097 + }, + { + "epoch": 2.097298484515704, + "grad_norm": 2.503840684890747, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7218230962753296, + "num_tokens": 494194926.0, + "step": 19098 + }, + { + "epoch": 2.0974083022183176, + "grad_norm": 2.353283643722534, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.751765251159668, + "num_tokens": 494220067.0, + "step": 19099 + }, + { + "epoch": 2.0975181199209314, + "grad_norm": 2.2641870975494385, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7447854280471802, + "num_tokens": 494246743.0, + "step": 19100 + }, + { + "epoch": 2.0976279376235447, + "grad_norm": 2.634860038757324, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7229674458503723, + "num_tokens": 494267627.0, + "step": 19101 + }, + { + "epoch": 2.0977377553261585, + "grad_norm": 2.0535459518432617, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7223571538925171, + "num_tokens": 494298287.0, + "step": 19102 + }, + { + "epoch": 2.0978475730287722, + "grad_norm": 2.366511344909668, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7284320592880249, + "num_tokens": 494323848.0, + "step": 19103 + }, + { + "epoch": 2.097957390731386, + "grad_norm": 2.4332966804504395, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7389369010925293, + "num_tokens": 494348007.0, + "step": 19104 + }, + { + "epoch": 2.0980672084339997, + "grad_norm": 2.476259708404541, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7329091429710388, + "num_tokens": 494372682.0, + "step": 19105 + }, + { + "epoch": 2.098177026136613, + "grad_norm": 2.291518449783325, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7447618246078491, + "num_tokens": 494398639.0, + "step": 19106 + }, + { + "epoch": 2.098286843839227, + "grad_norm": 2.6634979248046875, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7302855253219604, + "num_tokens": 494420579.0, + "step": 19107 + }, + { + "epoch": 2.0983966615418406, + "grad_norm": 2.5396804809570312, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7259521484375, + "num_tokens": 494444564.0, + "step": 19108 + }, + { + "epoch": 2.0985064792444543, + "grad_norm": 2.445876359939575, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7243013381958008, + "num_tokens": 494468730.0, + "step": 19109 + }, + { + "epoch": 2.0986162969470676, + "grad_norm": 2.4463248252868652, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7286937236785889, + "num_tokens": 494494472.0, + "step": 19110 + }, + { + "epoch": 2.0987261146496814, + "grad_norm": 2.2124133110046387, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7191873788833618, + "num_tokens": 494523174.0, + "step": 19111 + }, + { + "epoch": 2.098835932352295, + "grad_norm": 2.5660877227783203, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7344027757644653, + "num_tokens": 494545274.0, + "step": 19112 + }, + { + "epoch": 2.098945750054909, + "grad_norm": 2.5190699100494385, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7368063926696777, + "num_tokens": 494566831.0, + "step": 19113 + }, + { + "epoch": 2.0990555677575227, + "grad_norm": 2.0558688640594482, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7232552766799927, + "num_tokens": 494598634.0, + "step": 19114 + }, + { + "epoch": 2.099165385460136, + "grad_norm": 2.381453037261963, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7208947539329529, + "num_tokens": 494624358.0, + "step": 19115 + }, + { + "epoch": 2.0992752031627497, + "grad_norm": 2.2624659538269043, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7364857196807861, + "num_tokens": 494652414.0, + "step": 19116 + }, + { + "epoch": 2.0993850208653635, + "grad_norm": 2.3409650325775146, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.750996470451355, + "num_tokens": 494677253.0, + "step": 19117 + }, + { + "epoch": 2.0994948385679773, + "grad_norm": 2.6922154426574707, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7489864230155945, + "num_tokens": 494697568.0, + "step": 19118 + }, + { + "epoch": 2.099604656270591, + "grad_norm": 2.291364908218384, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7295345067977905, + "num_tokens": 494723570.0, + "step": 19119 + }, + { + "epoch": 2.0997144739732043, + "grad_norm": 2.640676736831665, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7447922229766846, + "num_tokens": 494744675.0, + "step": 19120 + }, + { + "epoch": 2.099824291675818, + "grad_norm": 2.1469168663024902, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.732912540435791, + "num_tokens": 494771771.0, + "step": 19121 + }, + { + "epoch": 2.099934109378432, + "grad_norm": 2.520106554031372, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7233769297599792, + "num_tokens": 494796506.0, + "step": 19122 + }, + { + "epoch": 2.1000439270810456, + "grad_norm": 2.0271835327148438, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7397714853286743, + "num_tokens": 494829970.0, + "step": 19123 + }, + { + "epoch": 2.1001537447836593, + "grad_norm": 2.3956472873687744, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7187983393669128, + "num_tokens": 494856312.0, + "step": 19124 + }, + { + "epoch": 2.1002635624862727, + "grad_norm": 2.1368460655212402, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7437983751296997, + "num_tokens": 494887483.0, + "step": 19125 + }, + { + "epoch": 2.1003733801888864, + "grad_norm": 2.4368836879730225, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7393089532852173, + "num_tokens": 494910484.0, + "step": 19126 + }, + { + "epoch": 2.1004831978915, + "grad_norm": 2.1930785179138184, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7508776187896729, + "num_tokens": 494942311.0, + "step": 19127 + }, + { + "epoch": 2.100593015594114, + "grad_norm": 2.6236226558685303, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7597498893737793, + "num_tokens": 494963926.0, + "step": 19128 + }, + { + "epoch": 2.1007028332967272, + "grad_norm": 2.124906301498413, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7659777998924255, + "num_tokens": 494992019.0, + "step": 19129 + }, + { + "epoch": 2.100812650999341, + "grad_norm": 1.9588611125946045, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7097642421722412, + "num_tokens": 495028204.0, + "step": 19130 + }, + { + "epoch": 2.1009224687019548, + "grad_norm": 2.7349119186401367, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7411420345306396, + "num_tokens": 495048069.0, + "step": 19131 + }, + { + "epoch": 2.1010322864045685, + "grad_norm": 2.20627498626709, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7527053356170654, + "num_tokens": 495075347.0, + "step": 19132 + }, + { + "epoch": 2.1011421041071823, + "grad_norm": 2.4821860790252686, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7587323784828186, + "num_tokens": 495099759.0, + "step": 19133 + }, + { + "epoch": 2.1012519218097956, + "grad_norm": 2.2235798835754395, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7304320931434631, + "num_tokens": 495128353.0, + "step": 19134 + }, + { + "epoch": 2.1013617395124093, + "grad_norm": 2.7568962574005127, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7482268810272217, + "num_tokens": 495148321.0, + "step": 19135 + }, + { + "epoch": 2.101471557215023, + "grad_norm": 2.418076515197754, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7278839945793152, + "num_tokens": 495173397.0, + "step": 19136 + }, + { + "epoch": 2.101581374917637, + "grad_norm": 2.452714204788208, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.718448281288147, + "num_tokens": 495197877.0, + "step": 19137 + }, + { + "epoch": 2.10169119262025, + "grad_norm": 2.194721221923828, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7247093915939331, + "num_tokens": 495227587.0, + "step": 19138 + }, + { + "epoch": 2.101801010322864, + "grad_norm": 2.3083226680755615, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7239699363708496, + "num_tokens": 495253735.0, + "step": 19139 + }, + { + "epoch": 2.1019108280254777, + "grad_norm": 2.572859287261963, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7349981069564819, + "num_tokens": 495275860.0, + "step": 19140 + }, + { + "epoch": 2.1020206457280914, + "grad_norm": 2.6645007133483887, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.758088231086731, + "num_tokens": 495295552.0, + "step": 19141 + }, + { + "epoch": 2.102130463430705, + "grad_norm": 2.563270092010498, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7370105385780334, + "num_tokens": 495318799.0, + "step": 19142 + }, + { + "epoch": 2.1022402811333185, + "grad_norm": 2.6546523571014404, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.742334246635437, + "num_tokens": 495340919.0, + "step": 19143 + }, + { + "epoch": 2.1023500988359323, + "grad_norm": 2.2827396392822266, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7304307818412781, + "num_tokens": 495366914.0, + "step": 19144 + }, + { + "epoch": 2.102459916538546, + "grad_norm": 2.9597561359405518, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7399054169654846, + "num_tokens": 495384879.0, + "step": 19145 + }, + { + "epoch": 2.10256973424116, + "grad_norm": 2.4974989891052246, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7240612506866455, + "num_tokens": 495407837.0, + "step": 19146 + }, + { + "epoch": 2.1026795519437735, + "grad_norm": 2.304309368133545, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7442660331726074, + "num_tokens": 495434520.0, + "step": 19147 + }, + { + "epoch": 2.102789369646387, + "grad_norm": 2.8989999294281006, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7464790344238281, + "num_tokens": 495452239.0, + "step": 19148 + }, + { + "epoch": 2.1028991873490006, + "grad_norm": 2.3084137439727783, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7272511124610901, + "num_tokens": 495478247.0, + "step": 19149 + }, + { + "epoch": 2.1030090050516144, + "grad_norm": 2.2281062602996826, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7223187685012817, + "num_tokens": 495507686.0, + "step": 19150 + }, + { + "epoch": 2.103118822754228, + "grad_norm": 2.070035696029663, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7223438024520874, + "num_tokens": 495542033.0, + "step": 19151 + }, + { + "epoch": 2.1032286404568414, + "grad_norm": 2.4455254077911377, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7545397281646729, + "num_tokens": 495568759.0, + "step": 19152 + }, + { + "epoch": 2.103338458159455, + "grad_norm": 2.437594175338745, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7312221527099609, + "num_tokens": 495591660.0, + "step": 19153 + }, + { + "epoch": 2.103448275862069, + "grad_norm": 2.2434680461883545, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7437555193901062, + "num_tokens": 495618271.0, + "step": 19154 + }, + { + "epoch": 2.1035580935646827, + "grad_norm": 2.562347412109375, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7167468070983887, + "num_tokens": 495642256.0, + "step": 19155 + }, + { + "epoch": 2.1036679112672965, + "grad_norm": 2.6500940322875977, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7578191757202148, + "num_tokens": 495662454.0, + "step": 19156 + }, + { + "epoch": 2.1037777289699098, + "grad_norm": 2.2491421699523926, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7207005023956299, + "num_tokens": 495690054.0, + "step": 19157 + }, + { + "epoch": 2.1038875466725235, + "grad_norm": 2.424367904663086, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7430566549301147, + "num_tokens": 495713834.0, + "step": 19158 + }, + { + "epoch": 2.1039973643751373, + "grad_norm": 2.2318809032440186, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7582341432571411, + "num_tokens": 495740116.0, + "step": 19159 + }, + { + "epoch": 2.104107182077751, + "grad_norm": 2.5794036388397217, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7389001250267029, + "num_tokens": 495762199.0, + "step": 19160 + }, + { + "epoch": 2.104216999780365, + "grad_norm": 2.2689125537872314, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7459281086921692, + "num_tokens": 495787374.0, + "step": 19161 + }, + { + "epoch": 2.104326817482978, + "grad_norm": 2.078281879425049, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7245705127716064, + "num_tokens": 495820269.0, + "step": 19162 + }, + { + "epoch": 2.104436635185592, + "grad_norm": 2.2455015182495117, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7274957895278931, + "num_tokens": 495848752.0, + "step": 19163 + }, + { + "epoch": 2.1045464528882056, + "grad_norm": 2.573925495147705, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7300609350204468, + "num_tokens": 495871746.0, + "step": 19164 + }, + { + "epoch": 2.1046562705908194, + "grad_norm": 2.6346311569213867, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7217385768890381, + "num_tokens": 495894868.0, + "step": 19165 + }, + { + "epoch": 2.1047660882934327, + "grad_norm": 2.1839680671691895, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7146645188331604, + "num_tokens": 495923860.0, + "step": 19166 + }, + { + "epoch": 2.1048759059960465, + "grad_norm": 2.0160484313964844, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7365846633911133, + "num_tokens": 495957197.0, + "step": 19167 + }, + { + "epoch": 2.10498572369866, + "grad_norm": 2.2657406330108643, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7211313843727112, + "num_tokens": 495985904.0, + "step": 19168 + }, + { + "epoch": 2.105095541401274, + "grad_norm": 2.1510229110717773, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.6941436529159546, + "num_tokens": 496015293.0, + "step": 19169 + }, + { + "epoch": 2.1052053591038877, + "grad_norm": 2.3038406372070312, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7298576235771179, + "num_tokens": 496042205.0, + "step": 19170 + }, + { + "epoch": 2.105315176806501, + "grad_norm": 2.343755006790161, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.72663414478302, + "num_tokens": 496067972.0, + "step": 19171 + }, + { + "epoch": 2.105424994509115, + "grad_norm": 2.4234254360198975, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7495344281196594, + "num_tokens": 496092607.0, + "step": 19172 + }, + { + "epoch": 2.1055348122117286, + "grad_norm": 2.18613862991333, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7286428809165955, + "num_tokens": 496122886.0, + "step": 19173 + }, + { + "epoch": 2.1056446299143423, + "grad_norm": 2.542045831680298, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7163974642753601, + "num_tokens": 496146924.0, + "step": 19174 + }, + { + "epoch": 2.105754447616956, + "grad_norm": 2.607855796813965, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.73842853307724, + "num_tokens": 496166899.0, + "step": 19175 + }, + { + "epoch": 2.1058642653195694, + "grad_norm": 2.2390518188476562, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7202503681182861, + "num_tokens": 496196377.0, + "step": 19176 + }, + { + "epoch": 2.105974083022183, + "grad_norm": 2.3282554149627686, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7266278862953186, + "num_tokens": 496223225.0, + "step": 19177 + }, + { + "epoch": 2.106083900724797, + "grad_norm": 2.287813425064087, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7151533961296082, + "num_tokens": 496250112.0, + "step": 19178 + }, + { + "epoch": 2.1061937184274107, + "grad_norm": 2.0692191123962402, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7412064671516418, + "num_tokens": 496279922.0, + "step": 19179 + }, + { + "epoch": 2.106303536130024, + "grad_norm": 2.4774417877197266, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7129688262939453, + "num_tokens": 496304436.0, + "step": 19180 + }, + { + "epoch": 2.1064133538326377, + "grad_norm": 2.1594340801239014, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7212680578231812, + "num_tokens": 496334696.0, + "step": 19181 + }, + { + "epoch": 2.1065231715352515, + "grad_norm": 2.4904520511627197, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7380905151367188, + "num_tokens": 496358059.0, + "step": 19182 + }, + { + "epoch": 2.1066329892378652, + "grad_norm": 2.104861259460449, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7263839840888977, + "num_tokens": 496387740.0, + "step": 19183 + }, + { + "epoch": 2.106742806940479, + "grad_norm": 2.1946144104003906, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7200807929039001, + "num_tokens": 496415509.0, + "step": 19184 + }, + { + "epoch": 2.1068526246430923, + "grad_norm": 2.7375454902648926, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7336263656616211, + "num_tokens": 496437011.0, + "step": 19185 + }, + { + "epoch": 2.106962442345706, + "grad_norm": 2.050875663757324, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7239546775817871, + "num_tokens": 496469315.0, + "step": 19186 + }, + { + "epoch": 2.10707226004832, + "grad_norm": 2.352360248565674, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7242940068244934, + "num_tokens": 496496872.0, + "step": 19187 + }, + { + "epoch": 2.1071820777509336, + "grad_norm": 2.1867494583129883, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7434666752815247, + "num_tokens": 496526371.0, + "step": 19188 + }, + { + "epoch": 2.107291895453547, + "grad_norm": 2.3238723278045654, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7283421754837036, + "num_tokens": 496552103.0, + "step": 19189 + }, + { + "epoch": 2.1074017131561606, + "grad_norm": 2.2604799270629883, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.718929648399353, + "num_tokens": 496579357.0, + "step": 19190 + }, + { + "epoch": 2.1075115308587744, + "grad_norm": 2.4303479194641113, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7128639817237854, + "num_tokens": 496605039.0, + "step": 19191 + }, + { + "epoch": 2.107621348561388, + "grad_norm": 2.8898634910583496, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7582321166992188, + "num_tokens": 496621766.0, + "step": 19192 + }, + { + "epoch": 2.107731166264002, + "grad_norm": 2.2776856422424316, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7404129505157471, + "num_tokens": 496649137.0, + "step": 19193 + }, + { + "epoch": 2.1078409839666152, + "grad_norm": 2.3854644298553467, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7352485060691833, + "num_tokens": 496675600.0, + "step": 19194 + }, + { + "epoch": 2.107950801669229, + "grad_norm": 2.1882872581481934, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7317308187484741, + "num_tokens": 496702492.0, + "step": 19195 + }, + { + "epoch": 2.1080606193718427, + "grad_norm": 2.3508236408233643, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7476789355278015, + "num_tokens": 496729085.0, + "step": 19196 + }, + { + "epoch": 2.1081704370744565, + "grad_norm": 2.439133405685425, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7295122146606445, + "num_tokens": 496752094.0, + "step": 19197 + }, + { + "epoch": 2.1082802547770703, + "grad_norm": 2.0532829761505127, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7027596235275269, + "num_tokens": 496785434.0, + "step": 19198 + }, + { + "epoch": 2.1083900724796836, + "grad_norm": 2.155325174331665, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7204341292381287, + "num_tokens": 496814915.0, + "step": 19199 + }, + { + "epoch": 2.1084998901822973, + "grad_norm": 2.091170072555542, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7383296489715576, + "num_tokens": 496842686.0, + "step": 19200 + }, + { + "epoch": 2.108609707884911, + "grad_norm": 2.617676258087158, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7356765270233154, + "num_tokens": 496863892.0, + "step": 19201 + }, + { + "epoch": 2.108719525587525, + "grad_norm": 2.3771607875823975, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7349324226379395, + "num_tokens": 496886522.0, + "step": 19202 + }, + { + "epoch": 2.108829343290138, + "grad_norm": 2.4821009635925293, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7425668239593506, + "num_tokens": 496909244.0, + "step": 19203 + }, + { + "epoch": 2.108939160992752, + "grad_norm": 2.4485878944396973, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7216843366622925, + "num_tokens": 496935758.0, + "step": 19204 + }, + { + "epoch": 2.1090489786953657, + "grad_norm": 2.3998146057128906, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7376068830490112, + "num_tokens": 496960666.0, + "step": 19205 + }, + { + "epoch": 2.1091587963979794, + "grad_norm": 2.2065160274505615, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7266184091567993, + "num_tokens": 496987818.0, + "step": 19206 + }, + { + "epoch": 2.109268614100593, + "grad_norm": 2.399042844772339, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7334251403808594, + "num_tokens": 497013415.0, + "step": 19207 + }, + { + "epoch": 2.1093784318032065, + "grad_norm": 2.6115379333496094, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7390444278717041, + "num_tokens": 497034957.0, + "step": 19208 + }, + { + "epoch": 2.1094882495058203, + "grad_norm": 2.506073236465454, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7405230402946472, + "num_tokens": 497057299.0, + "step": 19209 + }, + { + "epoch": 2.109598067208434, + "grad_norm": 2.5607292652130127, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.724022626876831, + "num_tokens": 497083160.0, + "step": 19210 + }, + { + "epoch": 2.1097078849110478, + "grad_norm": 2.3786637783050537, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.726524829864502, + "num_tokens": 497110314.0, + "step": 19211 + }, + { + "epoch": 2.1098177026136615, + "grad_norm": 2.431668758392334, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7494207620620728, + "num_tokens": 497133958.0, + "step": 19212 + }, + { + "epoch": 2.109927520316275, + "grad_norm": 2.4325602054595947, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7376300096511841, + "num_tokens": 497156841.0, + "step": 19213 + }, + { + "epoch": 2.1100373380188886, + "grad_norm": 2.3885252475738525, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7518163919448853, + "num_tokens": 497181270.0, + "step": 19214 + }, + { + "epoch": 2.1101471557215024, + "grad_norm": 2.3815574645996094, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7354374527931213, + "num_tokens": 497206080.0, + "step": 19215 + }, + { + "epoch": 2.110256973424116, + "grad_norm": 2.145008087158203, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7180215120315552, + "num_tokens": 497236514.0, + "step": 19216 + }, + { + "epoch": 2.1103667911267294, + "grad_norm": 2.179809331893921, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7189825177192688, + "num_tokens": 497265054.0, + "step": 19217 + }, + { + "epoch": 2.110476608829343, + "grad_norm": 2.7165777683258057, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7476685047149658, + "num_tokens": 497284782.0, + "step": 19218 + }, + { + "epoch": 2.110586426531957, + "grad_norm": 2.163800001144409, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7340102195739746, + "num_tokens": 497315170.0, + "step": 19219 + }, + { + "epoch": 2.1106962442345707, + "grad_norm": 2.3963277339935303, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7158231139183044, + "num_tokens": 497342680.0, + "step": 19220 + }, + { + "epoch": 2.1108060619371845, + "grad_norm": 2.5833494663238525, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7200675010681152, + "num_tokens": 497365246.0, + "step": 19221 + }, + { + "epoch": 2.1109158796397978, + "grad_norm": 2.6097967624664307, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7243908643722534, + "num_tokens": 497386614.0, + "step": 19222 + }, + { + "epoch": 2.1110256973424115, + "grad_norm": 2.5637311935424805, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7349430322647095, + "num_tokens": 497408631.0, + "step": 19223 + }, + { + "epoch": 2.1111355150450253, + "grad_norm": 2.2859458923339844, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7528733611106873, + "num_tokens": 497435262.0, + "step": 19224 + }, + { + "epoch": 2.111245332747639, + "grad_norm": 2.2552852630615234, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7352685928344727, + "num_tokens": 497462587.0, + "step": 19225 + }, + { + "epoch": 2.111355150450253, + "grad_norm": 2.4921963214874268, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7380605936050415, + "num_tokens": 497486300.0, + "step": 19226 + }, + { + "epoch": 2.111464968152866, + "grad_norm": 2.531087636947632, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7415648698806763, + "num_tokens": 497507945.0, + "step": 19227 + }, + { + "epoch": 2.11157478585548, + "grad_norm": 2.124319076538086, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7063387632369995, + "num_tokens": 497540400.0, + "step": 19228 + }, + { + "epoch": 2.1116846035580936, + "grad_norm": 2.656034231185913, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.736878514289856, + "num_tokens": 497562841.0, + "step": 19229 + }, + { + "epoch": 2.1117944212607074, + "grad_norm": 2.4559881687164307, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7196525931358337, + "num_tokens": 497586063.0, + "step": 19230 + }, + { + "epoch": 2.1119042389633207, + "grad_norm": 2.3541135787963867, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7406094074249268, + "num_tokens": 497611524.0, + "step": 19231 + }, + { + "epoch": 2.1120140566659344, + "grad_norm": 2.348233938217163, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.737176775932312, + "num_tokens": 497636834.0, + "step": 19232 + }, + { + "epoch": 2.112123874368548, + "grad_norm": 2.4136247634887695, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7386167645454407, + "num_tokens": 497662050.0, + "step": 19233 + }, + { + "epoch": 2.112233692071162, + "grad_norm": 2.176382541656494, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7320245504379272, + "num_tokens": 497691458.0, + "step": 19234 + }, + { + "epoch": 2.1123435097737757, + "grad_norm": 2.5111520290374756, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.73781818151474, + "num_tokens": 497713683.0, + "step": 19235 + }, + { + "epoch": 2.112453327476389, + "grad_norm": 2.366924524307251, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7327494621276855, + "num_tokens": 497740788.0, + "step": 19236 + }, + { + "epoch": 2.112563145179003, + "grad_norm": 2.2704057693481445, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7252612113952637, + "num_tokens": 497767673.0, + "step": 19237 + }, + { + "epoch": 2.1126729628816165, + "grad_norm": 2.710273265838623, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.749937891960144, + "num_tokens": 497787761.0, + "step": 19238 + }, + { + "epoch": 2.1127827805842303, + "grad_norm": 2.2320337295532227, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.727155864238739, + "num_tokens": 497818542.0, + "step": 19239 + }, + { + "epoch": 2.1128925982868436, + "grad_norm": 2.3502585887908936, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7187923192977905, + "num_tokens": 497845783.0, + "step": 19240 + }, + { + "epoch": 2.1130024159894574, + "grad_norm": 2.1449360847473145, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7263627648353577, + "num_tokens": 497876838.0, + "step": 19241 + }, + { + "epoch": 2.113112233692071, + "grad_norm": 2.146090269088745, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7205657362937927, + "num_tokens": 497906561.0, + "step": 19242 + }, + { + "epoch": 2.113222051394685, + "grad_norm": 2.212026357650757, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7078593969345093, + "num_tokens": 497936046.0, + "step": 19243 + }, + { + "epoch": 2.1133318690972986, + "grad_norm": 2.5666861534118652, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7286648750305176, + "num_tokens": 497959831.0, + "step": 19244 + }, + { + "epoch": 2.113441686799912, + "grad_norm": 2.3720993995666504, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.732885479927063, + "num_tokens": 497986435.0, + "step": 19245 + }, + { + "epoch": 2.1135515045025257, + "grad_norm": 2.187023401260376, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7242377400398254, + "num_tokens": 498016562.0, + "step": 19246 + }, + { + "epoch": 2.1136613222051395, + "grad_norm": 2.295243978500366, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7411198019981384, + "num_tokens": 498042669.0, + "step": 19247 + }, + { + "epoch": 2.1137711399077532, + "grad_norm": 2.8000800609588623, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7275179624557495, + "num_tokens": 498062798.0, + "step": 19248 + }, + { + "epoch": 2.113880957610367, + "grad_norm": 2.239569902420044, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7217483520507812, + "num_tokens": 498090075.0, + "step": 19249 + }, + { + "epoch": 2.1139907753129803, + "grad_norm": 2.1569769382476807, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.721592903137207, + "num_tokens": 498119591.0, + "step": 19250 + }, + { + "epoch": 2.114100593015594, + "grad_norm": 1.9693602323532104, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7277819514274597, + "num_tokens": 498154592.0, + "step": 19251 + }, + { + "epoch": 2.114210410718208, + "grad_norm": 2.1211724281311035, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7160278558731079, + "num_tokens": 498185943.0, + "step": 19252 + }, + { + "epoch": 2.1143202284208216, + "grad_norm": 2.4150454998016357, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7204570770263672, + "num_tokens": 498212773.0, + "step": 19253 + }, + { + "epoch": 2.1144300461234353, + "grad_norm": 2.161807060241699, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7374507784843445, + "num_tokens": 498240260.0, + "step": 19254 + }, + { + "epoch": 2.1145398638260486, + "grad_norm": 2.450698137283325, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7306280732154846, + "num_tokens": 498264336.0, + "step": 19255 + }, + { + "epoch": 2.1146496815286624, + "grad_norm": 2.2310256958007812, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7233489751815796, + "num_tokens": 498293781.0, + "step": 19256 + }, + { + "epoch": 2.114759499231276, + "grad_norm": 2.1875572204589844, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.723395586013794, + "num_tokens": 498322343.0, + "step": 19257 + }, + { + "epoch": 2.11486931693389, + "grad_norm": 2.2636241912841797, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7429266571998596, + "num_tokens": 498352780.0, + "step": 19258 + }, + { + "epoch": 2.114979134636503, + "grad_norm": 2.2478199005126953, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7406428456306458, + "num_tokens": 498377441.0, + "step": 19259 + }, + { + "epoch": 2.115088952339117, + "grad_norm": 2.743748426437378, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7318316698074341, + "num_tokens": 498397880.0, + "step": 19260 + }, + { + "epoch": 2.1151987700417307, + "grad_norm": 2.209219455718994, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7171662449836731, + "num_tokens": 498426227.0, + "step": 19261 + }, + { + "epoch": 2.1153085877443445, + "grad_norm": 2.3769760131835938, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.721051812171936, + "num_tokens": 498453888.0, + "step": 19262 + }, + { + "epoch": 2.1154184054469582, + "grad_norm": 2.3741815090179443, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7306879162788391, + "num_tokens": 498482431.0, + "step": 19263 + }, + { + "epoch": 2.1155282231495716, + "grad_norm": 2.149582862854004, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7154819965362549, + "num_tokens": 498517225.0, + "step": 19264 + }, + { + "epoch": 2.1156380408521853, + "grad_norm": 2.3205201625823975, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7369420528411865, + "num_tokens": 498542827.0, + "step": 19265 + }, + { + "epoch": 2.115747858554799, + "grad_norm": 2.1135010719299316, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6995644569396973, + "num_tokens": 498574854.0, + "step": 19266 + }, + { + "epoch": 2.115857676257413, + "grad_norm": 2.510058641433716, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7292138934135437, + "num_tokens": 498600075.0, + "step": 19267 + }, + { + "epoch": 2.115967493960026, + "grad_norm": 2.264284133911133, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.72243332862854, + "num_tokens": 498627775.0, + "step": 19268 + }, + { + "epoch": 2.11607731166264, + "grad_norm": 2.345562696456909, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7253823280334473, + "num_tokens": 498652204.0, + "step": 19269 + }, + { + "epoch": 2.1161871293652537, + "grad_norm": 2.7134666442871094, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7447562217712402, + "num_tokens": 498672745.0, + "step": 19270 + }, + { + "epoch": 2.1162969470678674, + "grad_norm": 2.4745185375213623, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7367583513259888, + "num_tokens": 498695867.0, + "step": 19271 + }, + { + "epoch": 2.116406764770481, + "grad_norm": 2.7906556129455566, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7341533899307251, + "num_tokens": 498715297.0, + "step": 19272 + }, + { + "epoch": 2.1165165824730945, + "grad_norm": 2.4258933067321777, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7327907085418701, + "num_tokens": 498739354.0, + "step": 19273 + }, + { + "epoch": 2.1166264001757082, + "grad_norm": 2.4968252182006836, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7462766170501709, + "num_tokens": 498762108.0, + "step": 19274 + }, + { + "epoch": 2.116736217878322, + "grad_norm": 2.2964236736297607, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7362070083618164, + "num_tokens": 498789026.0, + "step": 19275 + }, + { + "epoch": 2.1168460355809358, + "grad_norm": 2.2828235626220703, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7281395196914673, + "num_tokens": 498814879.0, + "step": 19276 + }, + { + "epoch": 2.1169558532835495, + "grad_norm": 2.5870752334594727, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7451393604278564, + "num_tokens": 498836829.0, + "step": 19277 + }, + { + "epoch": 2.117065670986163, + "grad_norm": 2.4902594089508057, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7264639139175415, + "num_tokens": 498861658.0, + "step": 19278 + }, + { + "epoch": 2.1171754886887766, + "grad_norm": 2.8984298706054688, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7297296524047852, + "num_tokens": 498880637.0, + "step": 19279 + }, + { + "epoch": 2.1172853063913903, + "grad_norm": 2.4065194129943848, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7181053161621094, + "num_tokens": 498907284.0, + "step": 19280 + }, + { + "epoch": 2.117395124094004, + "grad_norm": 2.2818214893341064, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7306168079376221, + "num_tokens": 498936942.0, + "step": 19281 + }, + { + "epoch": 2.1175049417966174, + "grad_norm": 2.0876963138580322, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.751350462436676, + "num_tokens": 498966621.0, + "step": 19282 + }, + { + "epoch": 2.117614759499231, + "grad_norm": 2.417630434036255, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7364836931228638, + "num_tokens": 498990140.0, + "step": 19283 + }, + { + "epoch": 2.117724577201845, + "grad_norm": 2.5864148139953613, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7324045300483704, + "num_tokens": 499011949.0, + "step": 19284 + }, + { + "epoch": 2.1178343949044587, + "grad_norm": 2.2439756393432617, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7350670099258423, + "num_tokens": 499038072.0, + "step": 19285 + }, + { + "epoch": 2.1179442126070724, + "grad_norm": 2.570662498474121, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7661797404289246, + "num_tokens": 499058058.0, + "step": 19286 + }, + { + "epoch": 2.1180540303096858, + "grad_norm": 2.153071165084839, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7183027267456055, + "num_tokens": 499090487.0, + "step": 19287 + }, + { + "epoch": 2.1181638480122995, + "grad_norm": 2.342982053756714, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7185560464859009, + "num_tokens": 499117531.0, + "step": 19288 + }, + { + "epoch": 2.1182736657149133, + "grad_norm": 2.238185167312622, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7438342571258545, + "num_tokens": 499146062.0, + "step": 19289 + }, + { + "epoch": 2.118383483417527, + "grad_norm": 2.2188453674316406, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.711563229560852, + "num_tokens": 499178273.0, + "step": 19290 + }, + { + "epoch": 2.118493301120141, + "grad_norm": 2.164400815963745, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7326275110244751, + "num_tokens": 499209831.0, + "step": 19291 + }, + { + "epoch": 2.118603118822754, + "grad_norm": 2.299553155899048, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7247501611709595, + "num_tokens": 499236931.0, + "step": 19292 + }, + { + "epoch": 2.118712936525368, + "grad_norm": 2.059286117553711, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7217117547988892, + "num_tokens": 499268058.0, + "step": 19293 + }, + { + "epoch": 2.1188227542279816, + "grad_norm": 2.309917688369751, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7500689029693604, + "num_tokens": 499294168.0, + "step": 19294 + }, + { + "epoch": 2.1189325719305954, + "grad_norm": 2.2975902557373047, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7271385192871094, + "num_tokens": 499321024.0, + "step": 19295 + }, + { + "epoch": 2.1190423896332087, + "grad_norm": 2.082505702972412, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7507649660110474, + "num_tokens": 499352640.0, + "step": 19296 + }, + { + "epoch": 2.1191522073358224, + "grad_norm": 2.367145299911499, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7273260354995728, + "num_tokens": 499379641.0, + "step": 19297 + }, + { + "epoch": 2.119262025038436, + "grad_norm": 2.1087892055511475, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7206266522407532, + "num_tokens": 499411356.0, + "step": 19298 + }, + { + "epoch": 2.11937184274105, + "grad_norm": 2.697037935256958, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7455558776855469, + "num_tokens": 499431949.0, + "step": 19299 + }, + { + "epoch": 2.1194816604436637, + "grad_norm": 2.9111666679382324, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7368093132972717, + "num_tokens": 499448969.0, + "step": 19300 + }, + { + "epoch": 2.119591478146277, + "grad_norm": 2.357417345046997, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7321020364761353, + "num_tokens": 499474514.0, + "step": 19301 + }, + { + "epoch": 2.1197012958488908, + "grad_norm": 2.509427547454834, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7371687889099121, + "num_tokens": 499497833.0, + "step": 19302 + }, + { + "epoch": 2.1198111135515045, + "grad_norm": 2.1432385444641113, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7436787486076355, + "num_tokens": 499527147.0, + "step": 19303 + }, + { + "epoch": 2.1199209312541183, + "grad_norm": 2.3874893188476562, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7221384048461914, + "num_tokens": 499553289.0, + "step": 19304 + }, + { + "epoch": 2.120030748956732, + "grad_norm": 2.6037862300872803, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7594047784805298, + "num_tokens": 499575098.0, + "step": 19305 + }, + { + "epoch": 2.1201405666593454, + "grad_norm": 2.164720058441162, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7148417234420776, + "num_tokens": 499605573.0, + "step": 19306 + }, + { + "epoch": 2.120250384361959, + "grad_norm": 2.404083728790283, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7335854768753052, + "num_tokens": 499630419.0, + "step": 19307 + }, + { + "epoch": 2.120360202064573, + "grad_norm": 2.3423454761505127, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7392363548278809, + "num_tokens": 499655863.0, + "step": 19308 + }, + { + "epoch": 2.1204700197671866, + "grad_norm": 2.0276119709014893, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7151663303375244, + "num_tokens": 499689025.0, + "step": 19309 + }, + { + "epoch": 2.1205798374698, + "grad_norm": 2.1598782539367676, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7386084198951721, + "num_tokens": 499717713.0, + "step": 19310 + }, + { + "epoch": 2.1206896551724137, + "grad_norm": 2.5684947967529297, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7529393434524536, + "num_tokens": 499739851.0, + "step": 19311 + }, + { + "epoch": 2.1207994728750275, + "grad_norm": 2.0934784412384033, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7361348271369934, + "num_tokens": 499770583.0, + "step": 19312 + }, + { + "epoch": 2.120909290577641, + "grad_norm": 2.3195106983184814, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7495593428611755, + "num_tokens": 499798119.0, + "step": 19313 + }, + { + "epoch": 2.121019108280255, + "grad_norm": 2.1994240283966064, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7031792402267456, + "num_tokens": 499828541.0, + "step": 19314 + }, + { + "epoch": 2.1211289259828683, + "grad_norm": 2.549468517303467, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7560346722602844, + "num_tokens": 499849585.0, + "step": 19315 + }, + { + "epoch": 2.121238743685482, + "grad_norm": 2.428642988204956, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7411695718765259, + "num_tokens": 499871411.0, + "step": 19316 + }, + { + "epoch": 2.121348561388096, + "grad_norm": 2.5935328006744385, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7288771867752075, + "num_tokens": 499895292.0, + "step": 19317 + }, + { + "epoch": 2.1214583790907096, + "grad_norm": 2.2224392890930176, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7206910252571106, + "num_tokens": 499925480.0, + "step": 19318 + }, + { + "epoch": 2.121568196793323, + "grad_norm": 2.6120998859405518, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7251962423324585, + "num_tokens": 499948452.0, + "step": 19319 + }, + { + "epoch": 2.1216780144959366, + "grad_norm": 2.373622417449951, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7372751235961914, + "num_tokens": 499972771.0, + "step": 19320 + }, + { + "epoch": 2.1217878321985504, + "grad_norm": 2.7287018299102783, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7384483218193054, + "num_tokens": 499993166.0, + "step": 19321 + }, + { + "epoch": 2.121897649901164, + "grad_norm": 2.1823740005493164, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.7424978017807007, + "num_tokens": 500024026.0, + "step": 19322 + }, + { + "epoch": 2.122007467603778, + "grad_norm": 2.322993516921997, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7325060963630676, + "num_tokens": 500052901.0, + "step": 19323 + }, + { + "epoch": 2.122117285306391, + "grad_norm": 2.4672281742095947, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.728181004524231, + "num_tokens": 500076201.0, + "step": 19324 + }, + { + "epoch": 2.122227103009005, + "grad_norm": 2.245250940322876, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.715407133102417, + "num_tokens": 500102261.0, + "step": 19325 + }, + { + "epoch": 2.1223369207116187, + "grad_norm": 2.7614951133728027, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7534415125846863, + "num_tokens": 500121653.0, + "step": 19326 + }, + { + "epoch": 2.1224467384142325, + "grad_norm": 2.2339158058166504, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7341136932373047, + "num_tokens": 500150513.0, + "step": 19327 + }, + { + "epoch": 2.1225565561168462, + "grad_norm": 2.335655689239502, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7255276441574097, + "num_tokens": 500176494.0, + "step": 19328 + }, + { + "epoch": 2.1226663738194596, + "grad_norm": 2.702937364578247, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7523163557052612, + "num_tokens": 500197664.0, + "step": 19329 + }, + { + "epoch": 2.1227761915220733, + "grad_norm": 2.3086910247802734, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7151801586151123, + "num_tokens": 500225618.0, + "step": 19330 + }, + { + "epoch": 2.122886009224687, + "grad_norm": 2.3992271423339844, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7343518733978271, + "num_tokens": 500250309.0, + "step": 19331 + }, + { + "epoch": 2.122995826927301, + "grad_norm": 2.413687229156494, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7398713827133179, + "num_tokens": 500274450.0, + "step": 19332 + }, + { + "epoch": 2.123105644629914, + "grad_norm": 2.1220107078552246, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7236322164535522, + "num_tokens": 500305517.0, + "step": 19333 + }, + { + "epoch": 2.123215462332528, + "grad_norm": 2.4071409702301025, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7665692567825317, + "num_tokens": 500329715.0, + "step": 19334 + }, + { + "epoch": 2.1233252800351416, + "grad_norm": 2.529726266860962, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7477381825447083, + "num_tokens": 500352933.0, + "step": 19335 + }, + { + "epoch": 2.1234350977377554, + "grad_norm": 2.5001063346862793, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7300078868865967, + "num_tokens": 500374961.0, + "step": 19336 + }, + { + "epoch": 2.123544915440369, + "grad_norm": 2.2239749431610107, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7388722896575928, + "num_tokens": 500404202.0, + "step": 19337 + }, + { + "epoch": 2.1236547331429825, + "grad_norm": 2.2599434852600098, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7336493730545044, + "num_tokens": 500431137.0, + "step": 19338 + }, + { + "epoch": 2.1237645508455962, + "grad_norm": 2.3003880977630615, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7467154264450073, + "num_tokens": 500456704.0, + "step": 19339 + }, + { + "epoch": 2.12387436854821, + "grad_norm": 2.560368061065674, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7401594519615173, + "num_tokens": 500479314.0, + "step": 19340 + }, + { + "epoch": 2.1239841862508237, + "grad_norm": 2.435270309448242, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7297062277793884, + "num_tokens": 500502601.0, + "step": 19341 + }, + { + "epoch": 2.1240940039534375, + "grad_norm": 2.6094844341278076, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7234697341918945, + "num_tokens": 500525637.0, + "step": 19342 + }, + { + "epoch": 2.124203821656051, + "grad_norm": 2.264030933380127, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7244662046432495, + "num_tokens": 500551665.0, + "step": 19343 + }, + { + "epoch": 2.1243136393586646, + "grad_norm": 2.3308863639831543, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7218981385231018, + "num_tokens": 500580210.0, + "step": 19344 + }, + { + "epoch": 2.1244234570612783, + "grad_norm": 2.3913307189941406, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7185534834861755, + "num_tokens": 500605288.0, + "step": 19345 + }, + { + "epoch": 2.124533274763892, + "grad_norm": 2.0781795978546143, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7203102111816406, + "num_tokens": 500634790.0, + "step": 19346 + }, + { + "epoch": 2.1246430924665054, + "grad_norm": 2.364814281463623, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7111456394195557, + "num_tokens": 500661686.0, + "step": 19347 + }, + { + "epoch": 2.124752910169119, + "grad_norm": 2.696383237838745, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7468467354774475, + "num_tokens": 500682856.0, + "step": 19348 + }, + { + "epoch": 2.124862727871733, + "grad_norm": 2.140773057937622, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7232903838157654, + "num_tokens": 500713087.0, + "step": 19349 + }, + { + "epoch": 2.1249725455743467, + "grad_norm": 2.543307304382324, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7359164953231812, + "num_tokens": 500736800.0, + "step": 19350 + }, + { + "epoch": 2.1250823632769604, + "grad_norm": 2.468320846557617, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.744770348072052, + "num_tokens": 500760036.0, + "step": 19351 + }, + { + "epoch": 2.1251921809795737, + "grad_norm": 2.4585084915161133, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.733334481716156, + "num_tokens": 500786040.0, + "step": 19352 + }, + { + "epoch": 2.1253019986821875, + "grad_norm": 2.2763824462890625, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7325937747955322, + "num_tokens": 500813124.0, + "step": 19353 + }, + { + "epoch": 2.1254118163848013, + "grad_norm": 2.4011940956115723, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7412152290344238, + "num_tokens": 500838213.0, + "step": 19354 + }, + { + "epoch": 2.125521634087415, + "grad_norm": 2.425565481185913, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7372000217437744, + "num_tokens": 500863071.0, + "step": 19355 + }, + { + "epoch": 2.1256314517900288, + "grad_norm": 2.3551511764526367, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7287523746490479, + "num_tokens": 500889884.0, + "step": 19356 + }, + { + "epoch": 2.125741269492642, + "grad_norm": 2.238588333129883, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7495630979537964, + "num_tokens": 500917979.0, + "step": 19357 + }, + { + "epoch": 2.125851087195256, + "grad_norm": 2.3262133598327637, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7269062399864197, + "num_tokens": 500944795.0, + "step": 19358 + }, + { + "epoch": 2.1259609048978696, + "grad_norm": 2.2008750438690186, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7256890535354614, + "num_tokens": 500973310.0, + "step": 19359 + }, + { + "epoch": 2.1260707226004834, + "grad_norm": 2.258878231048584, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7332537174224854, + "num_tokens": 501001309.0, + "step": 19360 + }, + { + "epoch": 2.1261805403030967, + "grad_norm": 2.2222774028778076, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7135562300682068, + "num_tokens": 501031692.0, + "step": 19361 + }, + { + "epoch": 2.1262903580057104, + "grad_norm": 2.5258307456970215, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7319819331169128, + "num_tokens": 501055722.0, + "step": 19362 + }, + { + "epoch": 2.126400175708324, + "grad_norm": 2.4385156631469727, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7383427619934082, + "num_tokens": 501079954.0, + "step": 19363 + }, + { + "epoch": 2.126509993410938, + "grad_norm": 2.4861185550689697, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7420946359634399, + "num_tokens": 501102979.0, + "step": 19364 + }, + { + "epoch": 2.1266198111135517, + "grad_norm": 2.2466254234313965, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7358482480049133, + "num_tokens": 501128962.0, + "step": 19365 + }, + { + "epoch": 2.126729628816165, + "grad_norm": 2.4808506965637207, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7375898957252502, + "num_tokens": 501153358.0, + "step": 19366 + }, + { + "epoch": 2.1268394465187788, + "grad_norm": 2.0673303604125977, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7328211069107056, + "num_tokens": 501183138.0, + "step": 19367 + }, + { + "epoch": 2.1269492642213925, + "grad_norm": 2.7875988483428955, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.739078164100647, + "num_tokens": 501202590.0, + "step": 19368 + }, + { + "epoch": 2.1270590819240063, + "grad_norm": 2.5419774055480957, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7442560791969299, + "num_tokens": 501224546.0, + "step": 19369 + }, + { + "epoch": 2.1271688996266196, + "grad_norm": 2.522451639175415, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.7444847226142883, + "num_tokens": 501246890.0, + "step": 19370 + }, + { + "epoch": 2.1272787173292333, + "grad_norm": 2.543064832687378, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7448699474334717, + "num_tokens": 501271218.0, + "step": 19371 + }, + { + "epoch": 2.127388535031847, + "grad_norm": 2.2404417991638184, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7221422791481018, + "num_tokens": 501298334.0, + "step": 19372 + }, + { + "epoch": 2.127498352734461, + "grad_norm": 2.6899242401123047, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7435657978057861, + "num_tokens": 501318225.0, + "step": 19373 + }, + { + "epoch": 2.1276081704370746, + "grad_norm": 2.5632145404815674, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7506788969039917, + "num_tokens": 501337679.0, + "step": 19374 + }, + { + "epoch": 2.127717988139688, + "grad_norm": 2.6413962841033936, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7560548186302185, + "num_tokens": 501359800.0, + "step": 19375 + }, + { + "epoch": 2.1278278058423017, + "grad_norm": 2.3650622367858887, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7207732796669006, + "num_tokens": 501384640.0, + "step": 19376 + }, + { + "epoch": 2.1279376235449154, + "grad_norm": 2.4081063270568848, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7394295930862427, + "num_tokens": 501408346.0, + "step": 19377 + }, + { + "epoch": 2.128047441247529, + "grad_norm": 2.385411262512207, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7394589185714722, + "num_tokens": 501434397.0, + "step": 19378 + }, + { + "epoch": 2.128157258950143, + "grad_norm": 2.602559804916382, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7304299473762512, + "num_tokens": 501458928.0, + "step": 19379 + }, + { + "epoch": 2.1282670766527563, + "grad_norm": 2.541853189468384, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7319226264953613, + "num_tokens": 501482911.0, + "step": 19380 + }, + { + "epoch": 2.12837689435537, + "grad_norm": 2.2728612422943115, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7290551662445068, + "num_tokens": 501511662.0, + "step": 19381 + }, + { + "epoch": 2.128486712057984, + "grad_norm": 2.156412124633789, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7298253178596497, + "num_tokens": 501540540.0, + "step": 19382 + }, + { + "epoch": 2.1285965297605975, + "grad_norm": 2.192124605178833, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7094484567642212, + "num_tokens": 501570794.0, + "step": 19383 + }, + { + "epoch": 2.1287063474632113, + "grad_norm": 2.3229000568389893, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7146258354187012, + "num_tokens": 501598342.0, + "step": 19384 + }, + { + "epoch": 2.1288161651658246, + "grad_norm": 2.6757893562316895, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7500893473625183, + "num_tokens": 501617992.0, + "step": 19385 + }, + { + "epoch": 2.1289259828684384, + "grad_norm": 2.5123748779296875, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.726452112197876, + "num_tokens": 501640359.0, + "step": 19386 + }, + { + "epoch": 2.129035800571052, + "grad_norm": 2.329479455947876, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7367309927940369, + "num_tokens": 501666252.0, + "step": 19387 + }, + { + "epoch": 2.129145618273666, + "grad_norm": 2.16495680809021, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7085023522377014, + "num_tokens": 501696625.0, + "step": 19388 + }, + { + "epoch": 2.129255435976279, + "grad_norm": 2.4147894382476807, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7094290256500244, + "num_tokens": 501721953.0, + "step": 19389 + }, + { + "epoch": 2.129365253678893, + "grad_norm": 2.0798702239990234, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7482253313064575, + "num_tokens": 501751758.0, + "step": 19390 + }, + { + "epoch": 2.1294750713815067, + "grad_norm": 2.3125500679016113, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7560712099075317, + "num_tokens": 501777933.0, + "step": 19391 + }, + { + "epoch": 2.1295848890841205, + "grad_norm": 2.3052818775177, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7051982283592224, + "num_tokens": 501805384.0, + "step": 19392 + }, + { + "epoch": 2.1296947067867342, + "grad_norm": 2.775606393814087, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7535755634307861, + "num_tokens": 501825156.0, + "step": 19393 + }, + { + "epoch": 2.1298045244893475, + "grad_norm": 2.5612382888793945, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7560815811157227, + "num_tokens": 501847690.0, + "step": 19394 + }, + { + "epoch": 2.1299143421919613, + "grad_norm": 2.5920569896698, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7365991473197937, + "num_tokens": 501871064.0, + "step": 19395 + }, + { + "epoch": 2.130024159894575, + "grad_norm": 2.2292912006378174, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.718632698059082, + "num_tokens": 501898573.0, + "step": 19396 + }, + { + "epoch": 2.130133977597189, + "grad_norm": 2.2385942935943604, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7308832406997681, + "num_tokens": 501926272.0, + "step": 19397 + }, + { + "epoch": 2.130243795299802, + "grad_norm": 2.275151014328003, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7045066356658936, + "num_tokens": 501954207.0, + "step": 19398 + }, + { + "epoch": 2.130353613002416, + "grad_norm": 2.6389780044555664, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7319700717926025, + "num_tokens": 501977226.0, + "step": 19399 + }, + { + "epoch": 2.1304634307050296, + "grad_norm": 2.323991298675537, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.748160183429718, + "num_tokens": 502001304.0, + "step": 19400 + }, + { + "epoch": 2.1305732484076434, + "grad_norm": 2.5514955520629883, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7395415306091309, + "num_tokens": 502023302.0, + "step": 19401 + }, + { + "epoch": 2.130683066110257, + "grad_norm": 2.3189783096313477, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7336621284484863, + "num_tokens": 502048565.0, + "step": 19402 + }, + { + "epoch": 2.1307928838128705, + "grad_norm": 2.538632392883301, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7228068113327026, + "num_tokens": 502071693.0, + "step": 19403 + }, + { + "epoch": 2.130902701515484, + "grad_norm": 2.1473336219787598, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7046657800674438, + "num_tokens": 502102292.0, + "step": 19404 + }, + { + "epoch": 2.131012519218098, + "grad_norm": 2.575073003768921, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7479034662246704, + "num_tokens": 502124153.0, + "step": 19405 + }, + { + "epoch": 2.1311223369207117, + "grad_norm": 2.380046844482422, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7457390427589417, + "num_tokens": 502149127.0, + "step": 19406 + }, + { + "epoch": 2.1312321546233255, + "grad_norm": 2.2500460147857666, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7419439554214478, + "num_tokens": 502175688.0, + "step": 19407 + }, + { + "epoch": 2.131341972325939, + "grad_norm": 2.895198106765747, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7372283339500427, + "num_tokens": 502194729.0, + "step": 19408 + }, + { + "epoch": 2.1314517900285526, + "grad_norm": 2.2208902835845947, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.733896017074585, + "num_tokens": 502223616.0, + "step": 19409 + }, + { + "epoch": 2.1315616077311663, + "grad_norm": 2.5771450996398926, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7358554005622864, + "num_tokens": 502246377.0, + "step": 19410 + }, + { + "epoch": 2.13167142543378, + "grad_norm": 2.341745615005493, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7153284549713135, + "num_tokens": 502274348.0, + "step": 19411 + }, + { + "epoch": 2.1317812431363934, + "grad_norm": 2.1660537719726562, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7388943433761597, + "num_tokens": 502304932.0, + "step": 19412 + }, + { + "epoch": 2.131891060839007, + "grad_norm": 2.0975167751312256, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7265639305114746, + "num_tokens": 502334295.0, + "step": 19413 + }, + { + "epoch": 2.132000878541621, + "grad_norm": 2.2085111141204834, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7314083576202393, + "num_tokens": 502361123.0, + "step": 19414 + }, + { + "epoch": 2.1321106962442347, + "grad_norm": 2.427309513092041, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7409325838088989, + "num_tokens": 502385868.0, + "step": 19415 + }, + { + "epoch": 2.1322205139468484, + "grad_norm": 2.348073959350586, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7649611830711365, + "num_tokens": 502411193.0, + "step": 19416 + }, + { + "epoch": 2.1323303316494617, + "grad_norm": 2.4342164993286133, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7326783537864685, + "num_tokens": 502436624.0, + "step": 19417 + }, + { + "epoch": 2.1324401493520755, + "grad_norm": 2.3645434379577637, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7135951519012451, + "num_tokens": 502463703.0, + "step": 19418 + }, + { + "epoch": 2.1325499670546892, + "grad_norm": 2.3364222049713135, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7286177277565002, + "num_tokens": 502490220.0, + "step": 19419 + }, + { + "epoch": 2.132659784757303, + "grad_norm": 2.0136213302612305, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7603980302810669, + "num_tokens": 502520829.0, + "step": 19420 + }, + { + "epoch": 2.1327696024599163, + "grad_norm": 2.2189245223999023, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7192659974098206, + "num_tokens": 502548809.0, + "step": 19421 + }, + { + "epoch": 2.13287942016253, + "grad_norm": 2.309069871902466, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7362844944000244, + "num_tokens": 502572556.0, + "step": 19422 + }, + { + "epoch": 2.132989237865144, + "grad_norm": 2.262251853942871, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7200194001197815, + "num_tokens": 502601085.0, + "step": 19423 + }, + { + "epoch": 2.1330990555677576, + "grad_norm": 2.3625051975250244, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7282518744468689, + "num_tokens": 502626949.0, + "step": 19424 + }, + { + "epoch": 2.1332088732703713, + "grad_norm": 2.619856834411621, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7535571455955505, + "num_tokens": 502648130.0, + "step": 19425 + }, + { + "epoch": 2.1333186909729847, + "grad_norm": 2.4138998985290527, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7045555710792542, + "num_tokens": 502672859.0, + "step": 19426 + }, + { + "epoch": 2.1334285086755984, + "grad_norm": 2.1306569576263428, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7114335894584656, + "num_tokens": 502704382.0, + "step": 19427 + }, + { + "epoch": 2.133538326378212, + "grad_norm": 2.2129855155944824, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7453501224517822, + "num_tokens": 502734070.0, + "step": 19428 + }, + { + "epoch": 2.133648144080826, + "grad_norm": 3.0242209434509277, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7481477856636047, + "num_tokens": 502752470.0, + "step": 19429 + }, + { + "epoch": 2.1337579617834397, + "grad_norm": 2.4208133220672607, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7229750156402588, + "num_tokens": 502777691.0, + "step": 19430 + }, + { + "epoch": 2.133867779486053, + "grad_norm": 2.4772698879241943, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7259116172790527, + "num_tokens": 502802280.0, + "step": 19431 + }, + { + "epoch": 2.1339775971886668, + "grad_norm": 2.0134353637695312, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.722881555557251, + "num_tokens": 502836765.0, + "step": 19432 + }, + { + "epoch": 2.1340874148912805, + "grad_norm": 2.156390428543091, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7112921476364136, + "num_tokens": 502865359.0, + "step": 19433 + }, + { + "epoch": 2.1341972325938943, + "grad_norm": 2.358841896057129, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7294620275497437, + "num_tokens": 502890557.0, + "step": 19434 + }, + { + "epoch": 2.134307050296508, + "grad_norm": 2.189730644226074, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7333883047103882, + "num_tokens": 502919872.0, + "step": 19435 + }, + { + "epoch": 2.1344168679991213, + "grad_norm": 2.5781338214874268, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7090224623680115, + "num_tokens": 502945931.0, + "step": 19436 + }, + { + "epoch": 2.134526685701735, + "grad_norm": 2.275266647338867, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7298176288604736, + "num_tokens": 502973651.0, + "step": 19437 + }, + { + "epoch": 2.134636503404349, + "grad_norm": 2.409959077835083, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7239230871200562, + "num_tokens": 502999893.0, + "step": 19438 + }, + { + "epoch": 2.1347463211069626, + "grad_norm": 2.8953840732574463, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7488769292831421, + "num_tokens": 503019470.0, + "step": 19439 + }, + { + "epoch": 2.134856138809576, + "grad_norm": 2.489741802215576, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7293987274169922, + "num_tokens": 503043003.0, + "step": 19440 + }, + { + "epoch": 2.1349659565121897, + "grad_norm": 2.183696985244751, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7189127802848816, + "num_tokens": 503072935.0, + "step": 19441 + }, + { + "epoch": 2.1350757742148034, + "grad_norm": 2.3858895301818848, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7597566246986389, + "num_tokens": 503097451.0, + "step": 19442 + }, + { + "epoch": 2.135185591917417, + "grad_norm": 2.419189691543579, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7336395978927612, + "num_tokens": 503122261.0, + "step": 19443 + }, + { + "epoch": 2.135295409620031, + "grad_norm": 2.6748220920562744, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7157483100891113, + "num_tokens": 503152427.0, + "step": 19444 + }, + { + "epoch": 2.1354052273226443, + "grad_norm": 2.439887762069702, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7127720713615417, + "num_tokens": 503177395.0, + "step": 19445 + }, + { + "epoch": 2.135515045025258, + "grad_norm": 2.5185298919677734, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7336554527282715, + "num_tokens": 503200906.0, + "step": 19446 + }, + { + "epoch": 2.1356248627278718, + "grad_norm": 2.301643133163452, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7259537577629089, + "num_tokens": 503224808.0, + "step": 19447 + }, + { + "epoch": 2.1357346804304855, + "grad_norm": 2.3190195560455322, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7183220982551575, + "num_tokens": 503252145.0, + "step": 19448 + }, + { + "epoch": 2.135844498133099, + "grad_norm": 2.301039695739746, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7487420439720154, + "num_tokens": 503280156.0, + "step": 19449 + }, + { + "epoch": 2.1359543158357126, + "grad_norm": 2.4090607166290283, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7525057196617126, + "num_tokens": 503305478.0, + "step": 19450 + }, + { + "epoch": 2.1360641335383264, + "grad_norm": 2.421988010406494, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7241544723510742, + "num_tokens": 503328600.0, + "step": 19451 + }, + { + "epoch": 2.13617395124094, + "grad_norm": 2.647050142288208, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7264405488967896, + "num_tokens": 503351610.0, + "step": 19452 + }, + { + "epoch": 2.136283768943554, + "grad_norm": 2.7697153091430664, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7581000328063965, + "num_tokens": 503370742.0, + "step": 19453 + }, + { + "epoch": 2.136393586646167, + "grad_norm": 2.2894654273986816, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7379837036132812, + "num_tokens": 503397469.0, + "step": 19454 + }, + { + "epoch": 2.136503404348781, + "grad_norm": 1.9811830520629883, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.738187849521637, + "num_tokens": 503431983.0, + "step": 19455 + }, + { + "epoch": 2.1366132220513947, + "grad_norm": 2.2574033737182617, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7405112981796265, + "num_tokens": 503460751.0, + "step": 19456 + }, + { + "epoch": 2.1367230397540085, + "grad_norm": 2.3806304931640625, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7297123670578003, + "num_tokens": 503485362.0, + "step": 19457 + }, + { + "epoch": 2.136832857456622, + "grad_norm": 2.408906936645508, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.751986026763916, + "num_tokens": 503508116.0, + "step": 19458 + }, + { + "epoch": 2.1369426751592355, + "grad_norm": 2.3661630153656006, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7289643287658691, + "num_tokens": 503535310.0, + "step": 19459 + }, + { + "epoch": 2.1370524928618493, + "grad_norm": 2.277008056640625, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.6971727013587952, + "num_tokens": 503565033.0, + "step": 19460 + }, + { + "epoch": 2.137162310564463, + "grad_norm": 2.591966390609741, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7567663192749023, + "num_tokens": 503585580.0, + "step": 19461 + }, + { + "epoch": 2.137272128267077, + "grad_norm": 2.3931243419647217, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7394347190856934, + "num_tokens": 503610320.0, + "step": 19462 + }, + { + "epoch": 2.13738194596969, + "grad_norm": 2.3070056438446045, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.738807201385498, + "num_tokens": 503636171.0, + "step": 19463 + }, + { + "epoch": 2.137491763672304, + "grad_norm": 2.1214711666107178, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.723802387714386, + "num_tokens": 503667294.0, + "step": 19464 + }, + { + "epoch": 2.1376015813749176, + "grad_norm": 2.3799235820770264, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7457464933395386, + "num_tokens": 503693320.0, + "step": 19465 + }, + { + "epoch": 2.1377113990775314, + "grad_norm": 2.38706374168396, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7424485087394714, + "num_tokens": 503718333.0, + "step": 19466 + }, + { + "epoch": 2.137821216780145, + "grad_norm": 2.02986216545105, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7371208667755127, + "num_tokens": 503750800.0, + "step": 19467 + }, + { + "epoch": 2.1379310344827585, + "grad_norm": 2.2357866764068604, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.750335693359375, + "num_tokens": 503776981.0, + "step": 19468 + }, + { + "epoch": 2.138040852185372, + "grad_norm": 2.1192774772644043, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7390041351318359, + "num_tokens": 503805743.0, + "step": 19469 + }, + { + "epoch": 2.138150669887986, + "grad_norm": 2.4663031101226807, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7568725943565369, + "num_tokens": 503828450.0, + "step": 19470 + }, + { + "epoch": 2.1382604875905997, + "grad_norm": 2.5567831993103027, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7403062582015991, + "num_tokens": 503851585.0, + "step": 19471 + }, + { + "epoch": 2.138370305293213, + "grad_norm": 2.326162099838257, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7405281066894531, + "num_tokens": 503877200.0, + "step": 19472 + }, + { + "epoch": 2.138480122995827, + "grad_norm": 2.622314453125, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7301807403564453, + "num_tokens": 503901070.0, + "step": 19473 + }, + { + "epoch": 2.1385899406984406, + "grad_norm": 2.4762532711029053, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7213805913925171, + "num_tokens": 503924246.0, + "step": 19474 + }, + { + "epoch": 2.1386997584010543, + "grad_norm": 2.152104616165161, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7195228934288025, + "num_tokens": 503954041.0, + "step": 19475 + }, + { + "epoch": 2.138809576103668, + "grad_norm": 2.221627950668335, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7332030534744263, + "num_tokens": 503981062.0, + "step": 19476 + }, + { + "epoch": 2.1389193938062814, + "grad_norm": 2.713988780975342, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7364616990089417, + "num_tokens": 504002281.0, + "step": 19477 + }, + { + "epoch": 2.139029211508895, + "grad_norm": 2.1528029441833496, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7413408756256104, + "num_tokens": 504032604.0, + "step": 19478 + }, + { + "epoch": 2.139139029211509, + "grad_norm": 2.133572816848755, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7214751243591309, + "num_tokens": 504061406.0, + "step": 19479 + }, + { + "epoch": 2.1392488469141226, + "grad_norm": 2.573638439178467, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.740735650062561, + "num_tokens": 504082777.0, + "step": 19480 + }, + { + "epoch": 2.1393586646167364, + "grad_norm": 2.4011683464050293, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7255768775939941, + "num_tokens": 504107983.0, + "step": 19481 + }, + { + "epoch": 2.1394684823193497, + "grad_norm": 2.181941032409668, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7036875486373901, + "num_tokens": 504139751.0, + "step": 19482 + }, + { + "epoch": 2.1395783000219635, + "grad_norm": 2.4271657466888428, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.731523871421814, + "num_tokens": 504163631.0, + "step": 19483 + }, + { + "epoch": 2.1396881177245772, + "grad_norm": 2.466205596923828, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7197546362876892, + "num_tokens": 504188728.0, + "step": 19484 + }, + { + "epoch": 2.139797935427191, + "grad_norm": 2.4103667736053467, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7276307344436646, + "num_tokens": 504214656.0, + "step": 19485 + }, + { + "epoch": 2.1399077531298047, + "grad_norm": 2.617013454437256, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7657817602157593, + "num_tokens": 504236576.0, + "step": 19486 + }, + { + "epoch": 2.140017570832418, + "grad_norm": 2.2571346759796143, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7369982004165649, + "num_tokens": 504263587.0, + "step": 19487 + }, + { + "epoch": 2.140127388535032, + "grad_norm": 2.4630489349365234, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7407717704772949, + "num_tokens": 504286875.0, + "step": 19488 + }, + { + "epoch": 2.1402372062376456, + "grad_norm": 2.4086244106292725, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7466803193092346, + "num_tokens": 504310774.0, + "step": 19489 + }, + { + "epoch": 2.1403470239402593, + "grad_norm": 2.579482316970825, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7535445690155029, + "num_tokens": 504332683.0, + "step": 19490 + }, + { + "epoch": 2.1404568416428726, + "grad_norm": 2.3670828342437744, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7026200890541077, + "num_tokens": 504359380.0, + "step": 19491 + }, + { + "epoch": 2.1405666593454864, + "grad_norm": 2.213212490081787, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7311021685600281, + "num_tokens": 504389134.0, + "step": 19492 + }, + { + "epoch": 2.1406764770481, + "grad_norm": 2.3942601680755615, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7225199341773987, + "num_tokens": 504414439.0, + "step": 19493 + }, + { + "epoch": 2.140786294750714, + "grad_norm": 2.57061505317688, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7215074300765991, + "num_tokens": 504437166.0, + "step": 19494 + }, + { + "epoch": 2.1408961124533277, + "grad_norm": 2.427183151245117, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.732703685760498, + "num_tokens": 504460330.0, + "step": 19495 + }, + { + "epoch": 2.141005930155941, + "grad_norm": 2.2565319538116455, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7131797075271606, + "num_tokens": 504487957.0, + "step": 19496 + }, + { + "epoch": 2.1411157478585547, + "grad_norm": 2.176762342453003, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7319575548171997, + "num_tokens": 504516567.0, + "step": 19497 + }, + { + "epoch": 2.1412255655611685, + "grad_norm": 2.2553842067718506, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7283796072006226, + "num_tokens": 504545127.0, + "step": 19498 + }, + { + "epoch": 2.1413353832637823, + "grad_norm": 2.8988661766052246, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7467252016067505, + "num_tokens": 504563497.0, + "step": 19499 + }, + { + "epoch": 2.1414452009663956, + "grad_norm": 2.412369966506958, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7484734654426575, + "num_tokens": 504587343.0, + "step": 19500 + }, + { + "epoch": 2.1415550186690093, + "grad_norm": 2.110299825668335, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7282977104187012, + "num_tokens": 504620265.0, + "step": 19501 + }, + { + "epoch": 2.141664836371623, + "grad_norm": 2.778728485107422, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7690826654434204, + "num_tokens": 504638830.0, + "step": 19502 + }, + { + "epoch": 2.141774654074237, + "grad_norm": 2.5851409435272217, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7304403781890869, + "num_tokens": 504661914.0, + "step": 19503 + }, + { + "epoch": 2.1418844717768506, + "grad_norm": 2.5196661949157715, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7282645106315613, + "num_tokens": 504689335.0, + "step": 19504 + }, + { + "epoch": 2.141994289479464, + "grad_norm": 2.11901593208313, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7269762754440308, + "num_tokens": 504718675.0, + "step": 19505 + }, + { + "epoch": 2.1421041071820777, + "grad_norm": 2.5112318992614746, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7436774373054504, + "num_tokens": 504741613.0, + "step": 19506 + }, + { + "epoch": 2.1422139248846914, + "grad_norm": 2.2973392009735107, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.731845498085022, + "num_tokens": 504769859.0, + "step": 19507 + }, + { + "epoch": 2.142323742587305, + "grad_norm": 2.246140480041504, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7490361928939819, + "num_tokens": 504797627.0, + "step": 19508 + }, + { + "epoch": 2.142433560289919, + "grad_norm": 2.444493055343628, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7455683946609497, + "num_tokens": 504820414.0, + "step": 19509 + }, + { + "epoch": 2.1425433779925322, + "grad_norm": 2.3625640869140625, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7292073965072632, + "num_tokens": 504844428.0, + "step": 19510 + }, + { + "epoch": 2.142653195695146, + "grad_norm": 2.4465880393981934, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7016356587409973, + "num_tokens": 504871808.0, + "step": 19511 + }, + { + "epoch": 2.1427630133977598, + "grad_norm": 2.403965473175049, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7268372774124146, + "num_tokens": 504897352.0, + "step": 19512 + }, + { + "epoch": 2.1428728311003735, + "grad_norm": 2.5297350883483887, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7413340210914612, + "num_tokens": 504920529.0, + "step": 19513 + }, + { + "epoch": 2.142982648802987, + "grad_norm": 2.5338196754455566, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7351388931274414, + "num_tokens": 504943383.0, + "step": 19514 + }, + { + "epoch": 2.1430924665056006, + "grad_norm": 2.225715160369873, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7331308126449585, + "num_tokens": 504972413.0, + "step": 19515 + }, + { + "epoch": 2.1432022842082143, + "grad_norm": 2.7302165031433105, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7367957830429077, + "num_tokens": 504994458.0, + "step": 19516 + }, + { + "epoch": 2.143312101910828, + "grad_norm": 2.3638479709625244, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7394397258758545, + "num_tokens": 505022774.0, + "step": 19517 + }, + { + "epoch": 2.143421919613442, + "grad_norm": 2.462374448776245, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7304931879043579, + "num_tokens": 505046288.0, + "step": 19518 + }, + { + "epoch": 2.143531737316055, + "grad_norm": 2.290710687637329, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7342416644096375, + "num_tokens": 505073053.0, + "step": 19519 + }, + { + "epoch": 2.143641555018669, + "grad_norm": 2.2492454051971436, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.728230357170105, + "num_tokens": 505101192.0, + "step": 19520 + }, + { + "epoch": 2.1437513727212827, + "grad_norm": 2.4206645488739014, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7277073860168457, + "num_tokens": 505127222.0, + "step": 19521 + }, + { + "epoch": 2.1438611904238964, + "grad_norm": 2.2594246864318848, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7031123042106628, + "num_tokens": 505155887.0, + "step": 19522 + }, + { + "epoch": 2.1439710081265098, + "grad_norm": 2.2696638107299805, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7262209057807922, + "num_tokens": 505184600.0, + "step": 19523 + }, + { + "epoch": 2.1440808258291235, + "grad_norm": 2.5271506309509277, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7226511836051941, + "num_tokens": 505208945.0, + "step": 19524 + }, + { + "epoch": 2.1441906435317373, + "grad_norm": 2.6291286945343018, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7576936483383179, + "num_tokens": 505232164.0, + "step": 19525 + }, + { + "epoch": 2.144300461234351, + "grad_norm": 2.275783061981201, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7161976099014282, + "num_tokens": 505260252.0, + "step": 19526 + }, + { + "epoch": 2.144410278936965, + "grad_norm": 2.4894051551818848, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7488545179367065, + "num_tokens": 505282177.0, + "step": 19527 + }, + { + "epoch": 2.144520096639578, + "grad_norm": 2.5546793937683105, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7337380647659302, + "num_tokens": 505305190.0, + "step": 19528 + }, + { + "epoch": 2.144629914342192, + "grad_norm": 2.4979047775268555, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7397555708885193, + "num_tokens": 505329180.0, + "step": 19529 + }, + { + "epoch": 2.1447397320448056, + "grad_norm": 2.132798194885254, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.74254310131073, + "num_tokens": 505358431.0, + "step": 19530 + }, + { + "epoch": 2.1448495497474194, + "grad_norm": 2.1338770389556885, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7182869911193848, + "num_tokens": 505388397.0, + "step": 19531 + }, + { + "epoch": 2.144959367450033, + "grad_norm": 2.1911580562591553, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.732190728187561, + "num_tokens": 505420726.0, + "step": 19532 + }, + { + "epoch": 2.1450691851526464, + "grad_norm": 2.3438754081726074, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7494572401046753, + "num_tokens": 505447288.0, + "step": 19533 + }, + { + "epoch": 2.14517900285526, + "grad_norm": 2.028337240219116, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7251518368721008, + "num_tokens": 505481821.0, + "step": 19534 + }, + { + "epoch": 2.145288820557874, + "grad_norm": 2.162307024002075, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7267141938209534, + "num_tokens": 505510728.0, + "step": 19535 + }, + { + "epoch": 2.1453986382604877, + "grad_norm": 2.601879835128784, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7184333801269531, + "num_tokens": 505533522.0, + "step": 19536 + }, + { + "epoch": 2.1455084559631015, + "grad_norm": 2.2341930866241455, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7133393287658691, + "num_tokens": 505561078.0, + "step": 19537 + }, + { + "epoch": 2.145618273665715, + "grad_norm": 2.0155012607574463, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7103583812713623, + "num_tokens": 505593858.0, + "step": 19538 + }, + { + "epoch": 2.1457280913683285, + "grad_norm": 2.4812331199645996, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7281434535980225, + "num_tokens": 505619141.0, + "step": 19539 + }, + { + "epoch": 2.1458379090709423, + "grad_norm": 2.2692484855651855, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7518560290336609, + "num_tokens": 505645105.0, + "step": 19540 + }, + { + "epoch": 2.145947726773556, + "grad_norm": 2.251394748687744, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7131059169769287, + "num_tokens": 505675618.0, + "step": 19541 + }, + { + "epoch": 2.1460575444761694, + "grad_norm": 2.4106669425964355, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.721574068069458, + "num_tokens": 505701316.0, + "step": 19542 + }, + { + "epoch": 2.146167362178783, + "grad_norm": 2.283986806869507, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.719092607498169, + "num_tokens": 505728786.0, + "step": 19543 + }, + { + "epoch": 2.146277179881397, + "grad_norm": 2.315584897994995, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7453843355178833, + "num_tokens": 505754729.0, + "step": 19544 + }, + { + "epoch": 2.1463869975840106, + "grad_norm": 2.1358296871185303, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7099517583847046, + "num_tokens": 505786922.0, + "step": 19545 + }, + { + "epoch": 2.1464968152866244, + "grad_norm": 2.241806745529175, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7294967770576477, + "num_tokens": 505816400.0, + "step": 19546 + }, + { + "epoch": 2.1466066329892377, + "grad_norm": 2.3130993843078613, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7643011212348938, + "num_tokens": 505839740.0, + "step": 19547 + }, + { + "epoch": 2.1467164506918515, + "grad_norm": 2.206906795501709, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7369612455368042, + "num_tokens": 505868143.0, + "step": 19548 + }, + { + "epoch": 2.146826268394465, + "grad_norm": 2.4067776203155518, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7252688407897949, + "num_tokens": 505894277.0, + "step": 19549 + }, + { + "epoch": 2.146936086097079, + "grad_norm": 2.4441606998443604, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.736038327217102, + "num_tokens": 505919504.0, + "step": 19550 + }, + { + "epoch": 2.1470459037996923, + "grad_norm": 2.5572922229766846, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7325271368026733, + "num_tokens": 505942117.0, + "step": 19551 + }, + { + "epoch": 2.147155721502306, + "grad_norm": 2.5436484813690186, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7270689010620117, + "num_tokens": 505966337.0, + "step": 19552 + }, + { + "epoch": 2.14726553920492, + "grad_norm": 2.4958040714263916, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7202735543251038, + "num_tokens": 505991024.0, + "step": 19553 + }, + { + "epoch": 2.1473753569075336, + "grad_norm": 2.207218885421753, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7277264595031738, + "num_tokens": 506018116.0, + "step": 19554 + }, + { + "epoch": 2.1474851746101473, + "grad_norm": 2.3569955825805664, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7274928092956543, + "num_tokens": 506042599.0, + "step": 19555 + }, + { + "epoch": 2.1475949923127606, + "grad_norm": 2.2329537868499756, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.6971592307090759, + "num_tokens": 506071190.0, + "step": 19556 + }, + { + "epoch": 2.1477048100153744, + "grad_norm": 2.3459315299987793, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7361724376678467, + "num_tokens": 506101070.0, + "step": 19557 + }, + { + "epoch": 2.147814627717988, + "grad_norm": 2.4605653285980225, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7548571228981018, + "num_tokens": 506123347.0, + "step": 19558 + }, + { + "epoch": 2.147924445420602, + "grad_norm": 2.548327684402466, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.726263165473938, + "num_tokens": 506146118.0, + "step": 19559 + }, + { + "epoch": 2.1480342631232157, + "grad_norm": 2.3506851196289062, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7187438607215881, + "num_tokens": 506174054.0, + "step": 19560 + }, + { + "epoch": 2.148144080825829, + "grad_norm": 2.464930534362793, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7445464134216309, + "num_tokens": 506197243.0, + "step": 19561 + }, + { + "epoch": 2.1482538985284427, + "grad_norm": 2.0660619735717773, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7153834104537964, + "num_tokens": 506229973.0, + "step": 19562 + }, + { + "epoch": 2.1483637162310565, + "grad_norm": 2.6780238151550293, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7356570959091187, + "num_tokens": 506250745.0, + "step": 19563 + }, + { + "epoch": 2.1484735339336702, + "grad_norm": 2.0979721546173096, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7207667827606201, + "num_tokens": 506282315.0, + "step": 19564 + }, + { + "epoch": 2.148583351636284, + "grad_norm": 2.020047426223755, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.720247745513916, + "num_tokens": 506316308.0, + "step": 19565 + }, + { + "epoch": 2.1486931693388973, + "grad_norm": 2.1792714595794678, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7226734161376953, + "num_tokens": 506345180.0, + "step": 19566 + }, + { + "epoch": 2.148802987041511, + "grad_norm": 2.444647789001465, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7399585247039795, + "num_tokens": 506369076.0, + "step": 19567 + }, + { + "epoch": 2.148912804744125, + "grad_norm": 2.4756040573120117, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7183388471603394, + "num_tokens": 506393479.0, + "step": 19568 + }, + { + "epoch": 2.1490226224467386, + "grad_norm": 2.3262875080108643, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7333158254623413, + "num_tokens": 506419932.0, + "step": 19569 + }, + { + "epoch": 2.149132440149352, + "grad_norm": 2.2652945518493652, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7254080772399902, + "num_tokens": 506448953.0, + "step": 19570 + }, + { + "epoch": 2.1492422578519657, + "grad_norm": 2.231269598007202, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7328197956085205, + "num_tokens": 506478305.0, + "step": 19571 + }, + { + "epoch": 2.1493520755545794, + "grad_norm": 2.2259104251861572, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7331042885780334, + "num_tokens": 506503441.0, + "step": 19572 + }, + { + "epoch": 2.149461893257193, + "grad_norm": 2.5113155841827393, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7161413431167603, + "num_tokens": 506527930.0, + "step": 19573 + }, + { + "epoch": 2.149571710959807, + "grad_norm": 2.2025864124298096, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.72376948595047, + "num_tokens": 506557797.0, + "step": 19574 + }, + { + "epoch": 2.1496815286624202, + "grad_norm": 2.285209894180298, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7443717122077942, + "num_tokens": 506584377.0, + "step": 19575 + }, + { + "epoch": 2.149791346365034, + "grad_norm": 2.964507818222046, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7349809408187866, + "num_tokens": 506603088.0, + "step": 19576 + }, + { + "epoch": 2.1499011640676478, + "grad_norm": 2.3612101078033447, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7141664028167725, + "num_tokens": 506630461.0, + "step": 19577 + }, + { + "epoch": 2.1500109817702615, + "grad_norm": 2.350301504135132, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7640469074249268, + "num_tokens": 506654651.0, + "step": 19578 + }, + { + "epoch": 2.150120799472875, + "grad_norm": 2.133547782897949, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7118837833404541, + "num_tokens": 506685387.0, + "step": 19579 + }, + { + "epoch": 2.1502306171754886, + "grad_norm": 2.2332489490509033, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7572752833366394, + "num_tokens": 506710343.0, + "step": 19580 + }, + { + "epoch": 2.1503404348781023, + "grad_norm": 2.2745699882507324, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7557841539382935, + "num_tokens": 506736132.0, + "step": 19581 + }, + { + "epoch": 2.150450252580716, + "grad_norm": 2.6423380374908447, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7336961627006531, + "num_tokens": 506757653.0, + "step": 19582 + }, + { + "epoch": 2.15056007028333, + "grad_norm": 2.680100679397583, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7423326969146729, + "num_tokens": 506777449.0, + "step": 19583 + }, + { + "epoch": 2.150669887985943, + "grad_norm": 2.3141260147094727, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7127854228019714, + "num_tokens": 506804917.0, + "step": 19584 + }, + { + "epoch": 2.150779705688557, + "grad_norm": 2.2598793506622314, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7450833320617676, + "num_tokens": 506831845.0, + "step": 19585 + }, + { + "epoch": 2.1508895233911707, + "grad_norm": 2.497227191925049, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7282806634902954, + "num_tokens": 506858898.0, + "step": 19586 + }, + { + "epoch": 2.1509993410937844, + "grad_norm": 2.6660664081573486, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7352735996246338, + "num_tokens": 506881919.0, + "step": 19587 + }, + { + "epoch": 2.151109158796398, + "grad_norm": 2.410407543182373, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7440395355224609, + "num_tokens": 506907384.0, + "step": 19588 + }, + { + "epoch": 2.1512189764990115, + "grad_norm": 2.454000234603882, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7223340272903442, + "num_tokens": 506931758.0, + "step": 19589 + }, + { + "epoch": 2.1513287942016253, + "grad_norm": 2.670696973800659, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.756294846534729, + "num_tokens": 506950804.0, + "step": 19590 + }, + { + "epoch": 2.151438611904239, + "grad_norm": 2.589036703109741, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7370743155479431, + "num_tokens": 506973831.0, + "step": 19591 + }, + { + "epoch": 2.1515484296068528, + "grad_norm": 2.300816059112549, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7113334536552429, + "num_tokens": 507002361.0, + "step": 19592 + }, + { + "epoch": 2.151658247309466, + "grad_norm": 2.1810758113861084, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7355543971061707, + "num_tokens": 507031774.0, + "step": 19593 + }, + { + "epoch": 2.15176806501208, + "grad_norm": 2.350433826446533, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7077125310897827, + "num_tokens": 507056790.0, + "step": 19594 + }, + { + "epoch": 2.1518778827146936, + "grad_norm": 2.4206886291503906, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7205044031143188, + "num_tokens": 507082694.0, + "step": 19595 + }, + { + "epoch": 2.1519877004173074, + "grad_norm": 2.7762575149536133, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7456950545310974, + "num_tokens": 507102802.0, + "step": 19596 + }, + { + "epoch": 2.152097518119921, + "grad_norm": 2.340991497039795, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7547760009765625, + "num_tokens": 507126400.0, + "step": 19597 + }, + { + "epoch": 2.1522073358225344, + "grad_norm": 2.268075704574585, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7382751107215881, + "num_tokens": 507153282.0, + "step": 19598 + }, + { + "epoch": 2.152317153525148, + "grad_norm": 2.706103801727295, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7520115375518799, + "num_tokens": 507174702.0, + "step": 19599 + }, + { + "epoch": 2.152426971227762, + "grad_norm": 2.246161699295044, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7195647358894348, + "num_tokens": 507202298.0, + "step": 19600 + }, + { + "epoch": 2.1525367889303757, + "grad_norm": 2.388241767883301, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7172162532806396, + "num_tokens": 507227837.0, + "step": 19601 + }, + { + "epoch": 2.152646606632989, + "grad_norm": 2.226025104522705, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7235568761825562, + "num_tokens": 507255427.0, + "step": 19602 + }, + { + "epoch": 2.1527564243356028, + "grad_norm": 2.252653121948242, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7313037514686584, + "num_tokens": 507282890.0, + "step": 19603 + }, + { + "epoch": 2.1528662420382165, + "grad_norm": 2.547907829284668, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7482632994651794, + "num_tokens": 507306909.0, + "step": 19604 + }, + { + "epoch": 2.1529760597408303, + "grad_norm": 1.980870246887207, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7471776008605957, + "num_tokens": 507338753.0, + "step": 19605 + }, + { + "epoch": 2.153085877443444, + "grad_norm": 2.595458984375, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.720531702041626, + "num_tokens": 507360979.0, + "step": 19606 + }, + { + "epoch": 2.1531956951460574, + "grad_norm": 2.1102302074432373, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7336927652359009, + "num_tokens": 507390687.0, + "step": 19607 + }, + { + "epoch": 2.153305512848671, + "grad_norm": 2.3610262870788574, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.704580545425415, + "num_tokens": 507415851.0, + "step": 19608 + }, + { + "epoch": 2.153415330551285, + "grad_norm": 2.803683042526245, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.735356867313385, + "num_tokens": 507436547.0, + "step": 19609 + }, + { + "epoch": 2.1535251482538986, + "grad_norm": 2.584474802017212, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7693996429443359, + "num_tokens": 507456399.0, + "step": 19610 + }, + { + "epoch": 2.1536349659565124, + "grad_norm": 2.343996047973633, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7103986740112305, + "num_tokens": 507483644.0, + "step": 19611 + }, + { + "epoch": 2.1537447836591257, + "grad_norm": 2.4530911445617676, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7424299716949463, + "num_tokens": 507507416.0, + "step": 19612 + }, + { + "epoch": 2.1538546013617395, + "grad_norm": 2.2835299968719482, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.6994858980178833, + "num_tokens": 507534168.0, + "step": 19613 + }, + { + "epoch": 2.153964419064353, + "grad_norm": 2.2633614540100098, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7360019087791443, + "num_tokens": 507560904.0, + "step": 19614 + }, + { + "epoch": 2.154074236766967, + "grad_norm": 2.6268880367279053, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7548769116401672, + "num_tokens": 507583246.0, + "step": 19615 + }, + { + "epoch": 2.1541840544695807, + "grad_norm": 2.3849329948425293, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7424962520599365, + "num_tokens": 507608366.0, + "step": 19616 + }, + { + "epoch": 2.154293872172194, + "grad_norm": 2.2757349014282227, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7218042612075806, + "num_tokens": 507637581.0, + "step": 19617 + }, + { + "epoch": 2.154403689874808, + "grad_norm": 2.7310245037078857, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7377505898475647, + "num_tokens": 507657453.0, + "step": 19618 + }, + { + "epoch": 2.1545135075774215, + "grad_norm": 2.3078219890594482, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.713127076625824, + "num_tokens": 507685092.0, + "step": 19619 + }, + { + "epoch": 2.1546233252800353, + "grad_norm": 2.3506953716278076, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7391407489776611, + "num_tokens": 507710248.0, + "step": 19620 + }, + { + "epoch": 2.1547331429826486, + "grad_norm": 2.3961238861083984, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7610225081443787, + "num_tokens": 507735881.0, + "step": 19621 + }, + { + "epoch": 2.1548429606852624, + "grad_norm": 2.4002983570098877, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7221033573150635, + "num_tokens": 507762092.0, + "step": 19622 + }, + { + "epoch": 2.154952778387876, + "grad_norm": 2.2439002990722656, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7164472937583923, + "num_tokens": 507789685.0, + "step": 19623 + }, + { + "epoch": 2.15506259609049, + "grad_norm": 2.443413019180298, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7155054807662964, + "num_tokens": 507815097.0, + "step": 19624 + }, + { + "epoch": 2.1551724137931036, + "grad_norm": 2.4492104053497314, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.737882137298584, + "num_tokens": 507839109.0, + "step": 19625 + }, + { + "epoch": 2.155282231495717, + "grad_norm": 2.43965220451355, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.705201268196106, + "num_tokens": 507864532.0, + "step": 19626 + }, + { + "epoch": 2.1553920491983307, + "grad_norm": 2.7136237621307373, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7329679727554321, + "num_tokens": 507886816.0, + "step": 19627 + }, + { + "epoch": 2.1555018669009445, + "grad_norm": 2.3864643573760986, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.747296929359436, + "num_tokens": 507910570.0, + "step": 19628 + }, + { + "epoch": 2.1556116846035582, + "grad_norm": 2.6303186416625977, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7526334524154663, + "num_tokens": 507930748.0, + "step": 19629 + }, + { + "epoch": 2.1557215023061715, + "grad_norm": 2.5139591693878174, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7215997576713562, + "num_tokens": 507954800.0, + "step": 19630 + }, + { + "epoch": 2.1558313200087853, + "grad_norm": 2.7553324699401855, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7402328252792358, + "num_tokens": 507974285.0, + "step": 19631 + }, + { + "epoch": 2.155941137711399, + "grad_norm": 2.2406952381134033, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7433191537857056, + "num_tokens": 508001138.0, + "step": 19632 + }, + { + "epoch": 2.156050955414013, + "grad_norm": 2.2922873497009277, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7361465096473694, + "num_tokens": 508027055.0, + "step": 19633 + }, + { + "epoch": 2.1561607731166266, + "grad_norm": 2.255333662033081, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7149204015731812, + "num_tokens": 508055242.0, + "step": 19634 + }, + { + "epoch": 2.15627059081924, + "grad_norm": 2.2484328746795654, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7120705842971802, + "num_tokens": 508084001.0, + "step": 19635 + }, + { + "epoch": 2.1563804085218536, + "grad_norm": 2.249706983566284, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7346580028533936, + "num_tokens": 508111763.0, + "step": 19636 + }, + { + "epoch": 2.1564902262244674, + "grad_norm": 2.3361759185791016, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7424161434173584, + "num_tokens": 508137475.0, + "step": 19637 + }, + { + "epoch": 2.156600043927081, + "grad_norm": 2.1962196826934814, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7295303344726562, + "num_tokens": 508165164.0, + "step": 19638 + }, + { + "epoch": 2.156709861629695, + "grad_norm": 2.8882861137390137, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7562521696090698, + "num_tokens": 508183366.0, + "step": 19639 + }, + { + "epoch": 2.1568196793323082, + "grad_norm": 2.3516788482666016, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.717739462852478, + "num_tokens": 508210922.0, + "step": 19640 + }, + { + "epoch": 2.156929497034922, + "grad_norm": 2.4440064430236816, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7277625799179077, + "num_tokens": 508233612.0, + "step": 19641 + }, + { + "epoch": 2.1570393147375357, + "grad_norm": 2.206888198852539, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7176257967948914, + "num_tokens": 508262678.0, + "step": 19642 + }, + { + "epoch": 2.1571491324401495, + "grad_norm": 2.275648832321167, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7184692621231079, + "num_tokens": 508290696.0, + "step": 19643 + }, + { + "epoch": 2.157258950142763, + "grad_norm": 2.1335339546203613, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7430711984634399, + "num_tokens": 508321748.0, + "step": 19644 + }, + { + "epoch": 2.1573687678453766, + "grad_norm": 2.3869950771331787, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7333981990814209, + "num_tokens": 508345438.0, + "step": 19645 + }, + { + "epoch": 2.1574785855479903, + "grad_norm": 2.308570384979248, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.717420220375061, + "num_tokens": 508372099.0, + "step": 19646 + }, + { + "epoch": 2.157588403250604, + "grad_norm": 2.3225433826446533, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7363853454589844, + "num_tokens": 508398974.0, + "step": 19647 + }, + { + "epoch": 2.157698220953218, + "grad_norm": 2.239872694015503, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7271802425384521, + "num_tokens": 508428562.0, + "step": 19648 + }, + { + "epoch": 2.157808038655831, + "grad_norm": 2.3720459938049316, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.712250292301178, + "num_tokens": 508456487.0, + "step": 19649 + }, + { + "epoch": 2.157917856358445, + "grad_norm": 2.5388500690460205, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7382142543792725, + "num_tokens": 508481014.0, + "step": 19650 + }, + { + "epoch": 2.1580276740610587, + "grad_norm": 2.3990745544433594, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7343083024024963, + "num_tokens": 508506398.0, + "step": 19651 + }, + { + "epoch": 2.1581374917636724, + "grad_norm": 2.5745849609375, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7413414716720581, + "num_tokens": 508530440.0, + "step": 19652 + }, + { + "epoch": 2.1582473094662857, + "grad_norm": 2.5378692150115967, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7225991487503052, + "num_tokens": 508553552.0, + "step": 19653 + }, + { + "epoch": 2.1583571271688995, + "grad_norm": 2.3430521488189697, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7474509477615356, + "num_tokens": 508579793.0, + "step": 19654 + }, + { + "epoch": 2.1584669448715132, + "grad_norm": 2.0224897861480713, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7157464623451233, + "num_tokens": 508614825.0, + "step": 19655 + }, + { + "epoch": 2.158576762574127, + "grad_norm": 2.673334836959839, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7495279312133789, + "num_tokens": 508636198.0, + "step": 19656 + }, + { + "epoch": 2.1586865802767408, + "grad_norm": 2.1795787811279297, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7236648797988892, + "num_tokens": 508665107.0, + "step": 19657 + }, + { + "epoch": 2.158796397979354, + "grad_norm": 2.1874146461486816, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7331476807594299, + "num_tokens": 508692624.0, + "step": 19658 + }, + { + "epoch": 2.158906215681968, + "grad_norm": 2.416865587234497, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7362151145935059, + "num_tokens": 508716756.0, + "step": 19659 + }, + { + "epoch": 2.1590160333845816, + "grad_norm": 2.2750186920166016, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7229355573654175, + "num_tokens": 508745627.0, + "step": 19660 + }, + { + "epoch": 2.1591258510871953, + "grad_norm": 2.5035009384155273, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7179757952690125, + "num_tokens": 508770712.0, + "step": 19661 + }, + { + "epoch": 2.159235668789809, + "grad_norm": 2.163628101348877, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7016638517379761, + "num_tokens": 508800724.0, + "step": 19662 + }, + { + "epoch": 2.1593454864924224, + "grad_norm": 2.2717087268829346, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7344992756843567, + "num_tokens": 508826714.0, + "step": 19663 + }, + { + "epoch": 2.159455304195036, + "grad_norm": 2.350327968597412, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.728103756904602, + "num_tokens": 508852337.0, + "step": 19664 + }, + { + "epoch": 2.15956512189765, + "grad_norm": 2.793919563293457, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.727908730506897, + "num_tokens": 508872900.0, + "step": 19665 + }, + { + "epoch": 2.1596749396002637, + "grad_norm": 2.4018137454986572, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7409625053405762, + "num_tokens": 508898461.0, + "step": 19666 + }, + { + "epoch": 2.1597847573028774, + "grad_norm": 2.2777044773101807, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.734247088432312, + "num_tokens": 508925970.0, + "step": 19667 + }, + { + "epoch": 2.1598945750054908, + "grad_norm": 2.1644396781921387, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7222872972488403, + "num_tokens": 508954639.0, + "step": 19668 + }, + { + "epoch": 2.1600043927081045, + "grad_norm": 2.0744924545288086, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7158344388008118, + "num_tokens": 508987944.0, + "step": 19669 + }, + { + "epoch": 2.1601142104107183, + "grad_norm": 2.957415819168091, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7406978607177734, + "num_tokens": 509006625.0, + "step": 19670 + }, + { + "epoch": 2.160224028113332, + "grad_norm": 2.1331660747528076, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.723910927772522, + "num_tokens": 509036342.0, + "step": 19671 + }, + { + "epoch": 2.1603338458159453, + "grad_norm": 2.2075603008270264, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7171697020530701, + "num_tokens": 509064156.0, + "step": 19672 + }, + { + "epoch": 2.160443663518559, + "grad_norm": 2.4282259941101074, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7330613136291504, + "num_tokens": 509088825.0, + "step": 19673 + }, + { + "epoch": 2.160553481221173, + "grad_norm": 2.1320641040802, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7232265472412109, + "num_tokens": 509120868.0, + "step": 19674 + }, + { + "epoch": 2.1606632989237866, + "grad_norm": 2.158785343170166, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.709606945514679, + "num_tokens": 509150995.0, + "step": 19675 + }, + { + "epoch": 2.1607731166264004, + "grad_norm": 2.2533669471740723, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7461733818054199, + "num_tokens": 509177385.0, + "step": 19676 + }, + { + "epoch": 2.1608829343290137, + "grad_norm": 2.5287554264068604, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6980721950531006, + "num_tokens": 509203424.0, + "step": 19677 + }, + { + "epoch": 2.1609927520316274, + "grad_norm": 2.3285200595855713, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7182630300521851, + "num_tokens": 509228899.0, + "step": 19678 + }, + { + "epoch": 2.161102569734241, + "grad_norm": 2.3321244716644287, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.74012690782547, + "num_tokens": 509256290.0, + "step": 19679 + }, + { + "epoch": 2.161212387436855, + "grad_norm": 2.2477872371673584, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7320882678031921, + "num_tokens": 509283170.0, + "step": 19680 + }, + { + "epoch": 2.1613222051394683, + "grad_norm": 2.196338653564453, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7182950973510742, + "num_tokens": 509312279.0, + "step": 19681 + }, + { + "epoch": 2.161432022842082, + "grad_norm": 2.3375015258789062, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7427771091461182, + "num_tokens": 509338350.0, + "step": 19682 + }, + { + "epoch": 2.161541840544696, + "grad_norm": 2.2262802124023438, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.738330066204071, + "num_tokens": 509366366.0, + "step": 19683 + }, + { + "epoch": 2.1616516582473095, + "grad_norm": 2.7643468379974365, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.778861939907074, + "num_tokens": 509383927.0, + "step": 19684 + }, + { + "epoch": 2.1617614759499233, + "grad_norm": 2.4833250045776367, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7166500687599182, + "num_tokens": 509411164.0, + "step": 19685 + }, + { + "epoch": 2.1618712936525366, + "grad_norm": 2.5258524417877197, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7208839654922485, + "num_tokens": 509436736.0, + "step": 19686 + }, + { + "epoch": 2.1619811113551504, + "grad_norm": 2.4678916931152344, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.708969235420227, + "num_tokens": 509462765.0, + "step": 19687 + }, + { + "epoch": 2.162090929057764, + "grad_norm": 2.522866725921631, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7474063634872437, + "num_tokens": 509485486.0, + "step": 19688 + }, + { + "epoch": 2.162200746760378, + "grad_norm": 2.3511948585510254, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7554453015327454, + "num_tokens": 509511127.0, + "step": 19689 + }, + { + "epoch": 2.1623105644629916, + "grad_norm": 2.3123228549957275, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7425186038017273, + "num_tokens": 509536626.0, + "step": 19690 + }, + { + "epoch": 2.162420382165605, + "grad_norm": 2.380594491958618, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7523861527442932, + "num_tokens": 509560467.0, + "step": 19691 + }, + { + "epoch": 2.1625301998682187, + "grad_norm": 2.4568421840667725, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.715798020362854, + "num_tokens": 509587425.0, + "step": 19692 + }, + { + "epoch": 2.1626400175708325, + "grad_norm": 2.5430710315704346, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7439193725585938, + "num_tokens": 509610781.0, + "step": 19693 + }, + { + "epoch": 2.162749835273446, + "grad_norm": 2.5153379440307617, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7297298908233643, + "num_tokens": 509635032.0, + "step": 19694 + }, + { + "epoch": 2.16285965297606, + "grad_norm": 2.248288631439209, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7281800508499146, + "num_tokens": 509663336.0, + "step": 19695 + }, + { + "epoch": 2.1629694706786733, + "grad_norm": 2.725754499435425, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7106132507324219, + "num_tokens": 509687814.0, + "step": 19696 + }, + { + "epoch": 2.163079288381287, + "grad_norm": 2.3807241916656494, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7182238101959229, + "num_tokens": 509713456.0, + "step": 19697 + }, + { + "epoch": 2.163189106083901, + "grad_norm": 2.7816450595855713, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7554354667663574, + "num_tokens": 509735769.0, + "step": 19698 + }, + { + "epoch": 2.1632989237865146, + "grad_norm": 2.4814157485961914, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7261924743652344, + "num_tokens": 509761707.0, + "step": 19699 + }, + { + "epoch": 2.163408741489128, + "grad_norm": 2.038839817047119, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7339589595794678, + "num_tokens": 509793233.0, + "step": 19700 + }, + { + "epoch": 2.1635185591917416, + "grad_norm": 2.174034357070923, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7258325815200806, + "num_tokens": 509820851.0, + "step": 19701 + }, + { + "epoch": 2.1636283768943554, + "grad_norm": 2.188779592514038, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7281323671340942, + "num_tokens": 509848138.0, + "step": 19702 + }, + { + "epoch": 2.163738194596969, + "grad_norm": 2.329712152481079, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7244558930397034, + "num_tokens": 509875119.0, + "step": 19703 + }, + { + "epoch": 2.1638480122995825, + "grad_norm": 2.3770411014556885, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7314794063568115, + "num_tokens": 509902039.0, + "step": 19704 + }, + { + "epoch": 2.163957830002196, + "grad_norm": 2.3459954261779785, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7377867698669434, + "num_tokens": 509926147.0, + "step": 19705 + }, + { + "epoch": 2.16406764770481, + "grad_norm": 2.7461040019989014, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7237175703048706, + "num_tokens": 509949592.0, + "step": 19706 + }, + { + "epoch": 2.1641774654074237, + "grad_norm": 2.322157621383667, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7357637882232666, + "num_tokens": 509975177.0, + "step": 19707 + }, + { + "epoch": 2.1642872831100375, + "grad_norm": 2.432054042816162, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7389422655105591, + "num_tokens": 509998908.0, + "step": 19708 + }, + { + "epoch": 2.164397100812651, + "grad_norm": 2.4213707447052, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7396833896636963, + "num_tokens": 510024977.0, + "step": 19709 + }, + { + "epoch": 2.1645069185152646, + "grad_norm": 2.178675651550293, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7388604879379272, + "num_tokens": 510054957.0, + "step": 19710 + }, + { + "epoch": 2.1646167362178783, + "grad_norm": 2.3281424045562744, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7363460659980774, + "num_tokens": 510081631.0, + "step": 19711 + }, + { + "epoch": 2.164726553920492, + "grad_norm": 2.3865275382995605, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7438706159591675, + "num_tokens": 510107994.0, + "step": 19712 + }, + { + "epoch": 2.164836371623106, + "grad_norm": 2.2480673789978027, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7077420949935913, + "num_tokens": 510136020.0, + "step": 19713 + }, + { + "epoch": 2.164946189325719, + "grad_norm": 2.4393885135650635, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7400957942008972, + "num_tokens": 510161525.0, + "step": 19714 + }, + { + "epoch": 2.165056007028333, + "grad_norm": 2.268050193786621, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7134339213371277, + "num_tokens": 510188589.0, + "step": 19715 + }, + { + "epoch": 2.1651658247309467, + "grad_norm": 2.1376547813415527, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.73475581407547, + "num_tokens": 510216497.0, + "step": 19716 + }, + { + "epoch": 2.1652756424335604, + "grad_norm": 2.3088200092315674, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7392667531967163, + "num_tokens": 510240815.0, + "step": 19717 + }, + { + "epoch": 2.165385460136174, + "grad_norm": 2.4968812465667725, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7331988215446472, + "num_tokens": 510263707.0, + "step": 19718 + }, + { + "epoch": 2.1654952778387875, + "grad_norm": 2.3451180458068848, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7604029178619385, + "num_tokens": 510287514.0, + "step": 19719 + }, + { + "epoch": 2.1656050955414012, + "grad_norm": 2.1969738006591797, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.750430703163147, + "num_tokens": 510315369.0, + "step": 19720 + }, + { + "epoch": 2.165714913244015, + "grad_norm": 2.5027756690979004, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7436800003051758, + "num_tokens": 510339585.0, + "step": 19721 + }, + { + "epoch": 2.1658247309466288, + "grad_norm": 2.3230738639831543, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7136274576187134, + "num_tokens": 510367976.0, + "step": 19722 + }, + { + "epoch": 2.165934548649242, + "grad_norm": 2.584348678588867, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7519272565841675, + "num_tokens": 510389909.0, + "step": 19723 + }, + { + "epoch": 2.166044366351856, + "grad_norm": 2.554410457611084, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7234352827072144, + "num_tokens": 510412666.0, + "step": 19724 + }, + { + "epoch": 2.1661541840544696, + "grad_norm": 2.7990527153015137, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7556200623512268, + "num_tokens": 510431793.0, + "step": 19725 + }, + { + "epoch": 2.1662640017570833, + "grad_norm": 2.443356513977051, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.739277720451355, + "num_tokens": 510457084.0, + "step": 19726 + }, + { + "epoch": 2.166373819459697, + "grad_norm": 2.3683981895446777, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7301205396652222, + "num_tokens": 510483651.0, + "step": 19727 + }, + { + "epoch": 2.1664836371623104, + "grad_norm": 2.4949569702148438, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7414639592170715, + "num_tokens": 510508228.0, + "step": 19728 + }, + { + "epoch": 2.166593454864924, + "grad_norm": 2.277341604232788, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7178840041160583, + "num_tokens": 510535082.0, + "step": 19729 + }, + { + "epoch": 2.166703272567538, + "grad_norm": 2.2767884731292725, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7495458722114563, + "num_tokens": 510561521.0, + "step": 19730 + }, + { + "epoch": 2.1668130902701517, + "grad_norm": 2.458425521850586, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7396801710128784, + "num_tokens": 510586160.0, + "step": 19731 + }, + { + "epoch": 2.166922907972765, + "grad_norm": 2.4735734462738037, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7359074354171753, + "num_tokens": 510611053.0, + "step": 19732 + }, + { + "epoch": 2.1670327256753787, + "grad_norm": 2.325432538986206, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7219500541687012, + "num_tokens": 510639136.0, + "step": 19733 + }, + { + "epoch": 2.1671425433779925, + "grad_norm": 2.26812481880188, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6990183591842651, + "num_tokens": 510667618.0, + "step": 19734 + }, + { + "epoch": 2.1672523610806063, + "grad_norm": 2.5619473457336426, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7537922859191895, + "num_tokens": 510690341.0, + "step": 19735 + }, + { + "epoch": 2.16736217878322, + "grad_norm": 2.2054948806762695, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7344073057174683, + "num_tokens": 510717531.0, + "step": 19736 + }, + { + "epoch": 2.1674719964858333, + "grad_norm": 1.9941998720169067, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.701727569103241, + "num_tokens": 510750234.0, + "step": 19737 + }, + { + "epoch": 2.167581814188447, + "grad_norm": 2.1937973499298096, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7290741205215454, + "num_tokens": 510779786.0, + "step": 19738 + }, + { + "epoch": 2.167691631891061, + "grad_norm": 2.528813362121582, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7401437163352966, + "num_tokens": 510802666.0, + "step": 19739 + }, + { + "epoch": 2.1678014495936746, + "grad_norm": 2.4402523040771484, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7336418628692627, + "num_tokens": 510825954.0, + "step": 19740 + }, + { + "epoch": 2.1679112672962884, + "grad_norm": 2.659188985824585, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7597106695175171, + "num_tokens": 510847338.0, + "step": 19741 + }, + { + "epoch": 2.1680210849989017, + "grad_norm": 2.437037706375122, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7193283438682556, + "num_tokens": 510873730.0, + "step": 19742 + }, + { + "epoch": 2.1681309027015154, + "grad_norm": 2.2527973651885986, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.716015100479126, + "num_tokens": 510903789.0, + "step": 19743 + }, + { + "epoch": 2.168240720404129, + "grad_norm": 2.2808432579040527, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7763121128082275, + "num_tokens": 510929730.0, + "step": 19744 + }, + { + "epoch": 2.168350538106743, + "grad_norm": 2.406832456588745, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7451636791229248, + "num_tokens": 510955077.0, + "step": 19745 + }, + { + "epoch": 2.1684603558093567, + "grad_norm": 2.394160270690918, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.746353268623352, + "num_tokens": 510979271.0, + "step": 19746 + }, + { + "epoch": 2.16857017351197, + "grad_norm": 2.01798152923584, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7139288187026978, + "num_tokens": 511012254.0, + "step": 19747 + }, + { + "epoch": 2.1686799912145838, + "grad_norm": 2.745429039001465, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7423242330551147, + "num_tokens": 511033036.0, + "step": 19748 + }, + { + "epoch": 2.1687898089171975, + "grad_norm": 2.380499839782715, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7492510080337524, + "num_tokens": 511058387.0, + "step": 19749 + }, + { + "epoch": 2.1688996266198113, + "grad_norm": 2.233731269836426, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.731382429599762, + "num_tokens": 511084828.0, + "step": 19750 + }, + { + "epoch": 2.1690094443224246, + "grad_norm": 2.300732374191284, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7315142154693604, + "num_tokens": 511112041.0, + "step": 19751 + }, + { + "epoch": 2.1691192620250384, + "grad_norm": 2.1564016342163086, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7185664176940918, + "num_tokens": 511143885.0, + "step": 19752 + }, + { + "epoch": 2.169229079727652, + "grad_norm": 2.4419243335723877, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7256760597229004, + "num_tokens": 511168625.0, + "step": 19753 + }, + { + "epoch": 2.169338897430266, + "grad_norm": 2.3054161071777344, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7310951948165894, + "num_tokens": 511194843.0, + "step": 19754 + }, + { + "epoch": 2.1694487151328796, + "grad_norm": 2.6538305282592773, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7242845296859741, + "num_tokens": 511216030.0, + "step": 19755 + }, + { + "epoch": 2.169558532835493, + "grad_norm": 2.41525936126709, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7196720838546753, + "num_tokens": 511241673.0, + "step": 19756 + }, + { + "epoch": 2.1696683505381067, + "grad_norm": 2.42091703414917, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.725836992263794, + "num_tokens": 511265301.0, + "step": 19757 + }, + { + "epoch": 2.1697781682407205, + "grad_norm": 2.7254607677459717, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7395617961883545, + "num_tokens": 511284891.0, + "step": 19758 + }, + { + "epoch": 2.169887985943334, + "grad_norm": 2.3676841259002686, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7199289798736572, + "num_tokens": 511310615.0, + "step": 19759 + }, + { + "epoch": 2.1699978036459475, + "grad_norm": 2.12007212638855, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7399230599403381, + "num_tokens": 511339053.0, + "step": 19760 + }, + { + "epoch": 2.1701076213485613, + "grad_norm": 2.0225830078125, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.713061511516571, + "num_tokens": 511371657.0, + "step": 19761 + }, + { + "epoch": 2.170217439051175, + "grad_norm": 2.2723169326782227, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7477726340293884, + "num_tokens": 511400108.0, + "step": 19762 + }, + { + "epoch": 2.170327256753789, + "grad_norm": 2.421428680419922, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7591195702552795, + "num_tokens": 511422134.0, + "step": 19763 + }, + { + "epoch": 2.1704370744564025, + "grad_norm": 2.13527774810791, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7329879999160767, + "num_tokens": 511452620.0, + "step": 19764 + }, + { + "epoch": 2.170546892159016, + "grad_norm": 2.4932899475097656, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.731624186038971, + "num_tokens": 511477119.0, + "step": 19765 + }, + { + "epoch": 2.1706567098616296, + "grad_norm": 2.155801296234131, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7271602153778076, + "num_tokens": 511506474.0, + "step": 19766 + }, + { + "epoch": 2.1707665275642434, + "grad_norm": 2.4806108474731445, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7559026479721069, + "num_tokens": 511529501.0, + "step": 19767 + }, + { + "epoch": 2.170876345266857, + "grad_norm": 2.288632869720459, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7405010461807251, + "num_tokens": 511558257.0, + "step": 19768 + }, + { + "epoch": 2.170986162969471, + "grad_norm": 2.3874800205230713, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.729401171207428, + "num_tokens": 511583088.0, + "step": 19769 + }, + { + "epoch": 2.171095980672084, + "grad_norm": 2.190793752670288, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7250978350639343, + "num_tokens": 511612614.0, + "step": 19770 + }, + { + "epoch": 2.171205798374698, + "grad_norm": 2.612797737121582, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7421053647994995, + "num_tokens": 511633909.0, + "step": 19771 + }, + { + "epoch": 2.1713156160773117, + "grad_norm": 2.308586835861206, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7247543931007385, + "num_tokens": 511659968.0, + "step": 19772 + }, + { + "epoch": 2.1714254337799255, + "grad_norm": 2.5490691661834717, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7426018118858337, + "num_tokens": 511682095.0, + "step": 19773 + }, + { + "epoch": 2.171535251482539, + "grad_norm": 2.2246956825256348, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7214686870574951, + "num_tokens": 511710990.0, + "step": 19774 + }, + { + "epoch": 2.1716450691851525, + "grad_norm": 2.3201262950897217, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.748536229133606, + "num_tokens": 511739624.0, + "step": 19775 + }, + { + "epoch": 2.1717548868877663, + "grad_norm": 2.3992793560028076, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7130302786827087, + "num_tokens": 511764140.0, + "step": 19776 + }, + { + "epoch": 2.17186470459038, + "grad_norm": 2.289271354675293, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7406527996063232, + "num_tokens": 511791977.0, + "step": 19777 + }, + { + "epoch": 2.171974522292994, + "grad_norm": 2.4933886528015137, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7355448007583618, + "num_tokens": 511815727.0, + "step": 19778 + }, + { + "epoch": 2.172084339995607, + "grad_norm": 2.1484477519989014, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.716668963432312, + "num_tokens": 511844926.0, + "step": 19779 + }, + { + "epoch": 2.172194157698221, + "grad_norm": 2.0669751167297363, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7455434203147888, + "num_tokens": 511877010.0, + "step": 19780 + }, + { + "epoch": 2.1723039754008346, + "grad_norm": 2.5701277256011963, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7356576323509216, + "num_tokens": 511897821.0, + "step": 19781 + }, + { + "epoch": 2.1724137931034484, + "grad_norm": 2.31492280960083, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7243896722793579, + "num_tokens": 511925977.0, + "step": 19782 + }, + { + "epoch": 2.1725236108060617, + "grad_norm": 2.395122528076172, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7378542423248291, + "num_tokens": 511949715.0, + "step": 19783 + }, + { + "epoch": 2.1726334285086755, + "grad_norm": 2.3177108764648438, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7527615427970886, + "num_tokens": 511977205.0, + "step": 19784 + }, + { + "epoch": 2.1727432462112892, + "grad_norm": 2.364368200302124, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7360794544219971, + "num_tokens": 512002980.0, + "step": 19785 + }, + { + "epoch": 2.172853063913903, + "grad_norm": 2.57592511177063, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7469994425773621, + "num_tokens": 512023265.0, + "step": 19786 + }, + { + "epoch": 2.1729628816165167, + "grad_norm": 2.2371249198913574, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7323126792907715, + "num_tokens": 512049915.0, + "step": 19787 + }, + { + "epoch": 2.17307269931913, + "grad_norm": 2.164518356323242, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7324604392051697, + "num_tokens": 512078755.0, + "step": 19788 + }, + { + "epoch": 2.173182517021744, + "grad_norm": 2.0618629455566406, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7423151731491089, + "num_tokens": 512109347.0, + "step": 19789 + }, + { + "epoch": 2.1732923347243576, + "grad_norm": 2.1976821422576904, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7382810711860657, + "num_tokens": 512137702.0, + "step": 19790 + }, + { + "epoch": 2.1734021524269713, + "grad_norm": 2.3100056648254395, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7351397275924683, + "num_tokens": 512163873.0, + "step": 19791 + }, + { + "epoch": 2.173511970129585, + "grad_norm": 2.4959399700164795, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7360097169876099, + "num_tokens": 512187004.0, + "step": 19792 + }, + { + "epoch": 2.1736217878321984, + "grad_norm": 2.8759920597076416, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.747596263885498, + "num_tokens": 512206001.0, + "step": 19793 + }, + { + "epoch": 2.173731605534812, + "grad_norm": 2.0745491981506348, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7159082889556885, + "num_tokens": 512239300.0, + "step": 19794 + }, + { + "epoch": 2.173841423237426, + "grad_norm": 2.1744325160980225, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7389543056488037, + "num_tokens": 512267649.0, + "step": 19795 + }, + { + "epoch": 2.1739512409400397, + "grad_norm": 2.596195936203003, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7679831981658936, + "num_tokens": 512287428.0, + "step": 19796 + }, + { + "epoch": 2.1740610586426534, + "grad_norm": 2.3896443843841553, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7380119562149048, + "num_tokens": 512312435.0, + "step": 19797 + }, + { + "epoch": 2.1741708763452667, + "grad_norm": 2.366251230239868, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7456958889961243, + "num_tokens": 512336657.0, + "step": 19798 + }, + { + "epoch": 2.1742806940478805, + "grad_norm": 2.372713088989258, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7164920568466187, + "num_tokens": 512363372.0, + "step": 19799 + }, + { + "epoch": 2.1743905117504942, + "grad_norm": 2.345475435256958, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7287319302558899, + "num_tokens": 512391054.0, + "step": 19800 + }, + { + "epoch": 2.174500329453108, + "grad_norm": 2.5917112827301025, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7427136898040771, + "num_tokens": 512413378.0, + "step": 19801 + }, + { + "epoch": 2.1746101471557213, + "grad_norm": 2.5309512615203857, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7380134463310242, + "num_tokens": 512436309.0, + "step": 19802 + }, + { + "epoch": 2.174719964858335, + "grad_norm": 2.542524576187134, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7377479076385498, + "num_tokens": 512458612.0, + "step": 19803 + }, + { + "epoch": 2.174829782560949, + "grad_norm": 2.474177598953247, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7174774408340454, + "num_tokens": 512483455.0, + "step": 19804 + }, + { + "epoch": 2.1749396002635626, + "grad_norm": 2.3081727027893066, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7314199805259705, + "num_tokens": 512509908.0, + "step": 19805 + }, + { + "epoch": 2.1750494179661763, + "grad_norm": 2.4991462230682373, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7242382764816284, + "num_tokens": 512534717.0, + "step": 19806 + }, + { + "epoch": 2.1751592356687897, + "grad_norm": 2.5482165813446045, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7439431548118591, + "num_tokens": 512557093.0, + "step": 19807 + }, + { + "epoch": 2.1752690533714034, + "grad_norm": 2.3419551849365234, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7153184413909912, + "num_tokens": 512584806.0, + "step": 19808 + }, + { + "epoch": 2.175378871074017, + "grad_norm": 2.3341009616851807, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7114583253860474, + "num_tokens": 512610521.0, + "step": 19809 + }, + { + "epoch": 2.175488688776631, + "grad_norm": 2.1166458129882812, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7550274133682251, + "num_tokens": 512638067.0, + "step": 19810 + }, + { + "epoch": 2.1755985064792442, + "grad_norm": 2.612821102142334, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7644491195678711, + "num_tokens": 512657400.0, + "step": 19811 + }, + { + "epoch": 2.175708324181858, + "grad_norm": 2.3550729751586914, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7402581572532654, + "num_tokens": 512683831.0, + "step": 19812 + }, + { + "epoch": 2.1758181418844718, + "grad_norm": 2.2931909561157227, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7313776016235352, + "num_tokens": 512711437.0, + "step": 19813 + }, + { + "epoch": 2.1759279595870855, + "grad_norm": 2.204143524169922, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7210838794708252, + "num_tokens": 512739807.0, + "step": 19814 + }, + { + "epoch": 2.1760377772896993, + "grad_norm": 2.043963670730591, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7428317666053772, + "num_tokens": 512772090.0, + "step": 19815 + }, + { + "epoch": 2.1761475949923126, + "grad_norm": 2.540607452392578, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7249952554702759, + "num_tokens": 512796315.0, + "step": 19816 + }, + { + "epoch": 2.1762574126949263, + "grad_norm": 2.2828073501586914, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7172713279724121, + "num_tokens": 512823980.0, + "step": 19817 + }, + { + "epoch": 2.17636723039754, + "grad_norm": 2.1432971954345703, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7448346614837646, + "num_tokens": 512853809.0, + "step": 19818 + }, + { + "epoch": 2.176477048100154, + "grad_norm": 2.490488290786743, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7448216080665588, + "num_tokens": 512877992.0, + "step": 19819 + }, + { + "epoch": 2.1765868658027676, + "grad_norm": 2.575228691101074, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7456679940223694, + "num_tokens": 512900783.0, + "step": 19820 + }, + { + "epoch": 2.176696683505381, + "grad_norm": 2.3194522857666016, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7291247844696045, + "num_tokens": 512926616.0, + "step": 19821 + }, + { + "epoch": 2.1768065012079947, + "grad_norm": 2.493623971939087, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7550768852233887, + "num_tokens": 512950669.0, + "step": 19822 + }, + { + "epoch": 2.1769163189106084, + "grad_norm": 2.413861036300659, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7228312492370605, + "num_tokens": 512975107.0, + "step": 19823 + }, + { + "epoch": 2.177026136613222, + "grad_norm": 2.2905514240264893, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.747593879699707, + "num_tokens": 513001713.0, + "step": 19824 + }, + { + "epoch": 2.1771359543158355, + "grad_norm": 2.607593536376953, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7146742939949036, + "num_tokens": 513026129.0, + "step": 19825 + }, + { + "epoch": 2.1772457720184493, + "grad_norm": 2.460142135620117, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.756598711013794, + "num_tokens": 513049616.0, + "step": 19826 + }, + { + "epoch": 2.177355589721063, + "grad_norm": 2.4231224060058594, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7237067818641663, + "num_tokens": 513076361.0, + "step": 19827 + }, + { + "epoch": 2.177465407423677, + "grad_norm": 2.352731704711914, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7272021770477295, + "num_tokens": 513103893.0, + "step": 19828 + }, + { + "epoch": 2.1775752251262905, + "grad_norm": 2.648754358291626, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7303566932678223, + "num_tokens": 513125256.0, + "step": 19829 + }, + { + "epoch": 2.177685042828904, + "grad_norm": 2.295224666595459, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7049199342727661, + "num_tokens": 513152505.0, + "step": 19830 + }, + { + "epoch": 2.1777948605315176, + "grad_norm": 2.0654430389404297, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7416389584541321, + "num_tokens": 513182326.0, + "step": 19831 + }, + { + "epoch": 2.1779046782341314, + "grad_norm": 2.4416351318359375, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7627895474433899, + "num_tokens": 513204562.0, + "step": 19832 + }, + { + "epoch": 2.178014495936745, + "grad_norm": 2.1418275833129883, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7588024735450745, + "num_tokens": 513231469.0, + "step": 19833 + }, + { + "epoch": 2.1781243136393584, + "grad_norm": 2.1295440196990967, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.73960942029953, + "num_tokens": 513260543.0, + "step": 19834 + }, + { + "epoch": 2.178234131341972, + "grad_norm": 2.5566246509552, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7533277273178101, + "num_tokens": 513283046.0, + "step": 19835 + }, + { + "epoch": 2.178343949044586, + "grad_norm": 2.423464775085449, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6951056718826294, + "num_tokens": 513309369.0, + "step": 19836 + }, + { + "epoch": 2.1784537667471997, + "grad_norm": 2.353101968765259, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.6988776326179504, + "num_tokens": 513337836.0, + "step": 19837 + }, + { + "epoch": 2.1785635844498135, + "grad_norm": 2.049318313598633, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6919996738433838, + "num_tokens": 513370115.0, + "step": 19838 + }, + { + "epoch": 2.1786734021524268, + "grad_norm": 2.2879631519317627, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7227034568786621, + "num_tokens": 513398324.0, + "step": 19839 + }, + { + "epoch": 2.1787832198550405, + "grad_norm": 2.2490477561950684, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7325295805931091, + "num_tokens": 513427101.0, + "step": 19840 + }, + { + "epoch": 2.1788930375576543, + "grad_norm": 2.419475555419922, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7514805793762207, + "num_tokens": 513450935.0, + "step": 19841 + }, + { + "epoch": 2.179002855260268, + "grad_norm": 2.3778724670410156, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7282016277313232, + "num_tokens": 513476918.0, + "step": 19842 + }, + { + "epoch": 2.179112672962882, + "grad_norm": 2.1621181964874268, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7310188412666321, + "num_tokens": 513508803.0, + "step": 19843 + }, + { + "epoch": 2.179222490665495, + "grad_norm": 2.4981284141540527, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7298693656921387, + "num_tokens": 513533374.0, + "step": 19844 + }, + { + "epoch": 2.179332308368109, + "grad_norm": 2.51446533203125, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7446452379226685, + "num_tokens": 513557068.0, + "step": 19845 + }, + { + "epoch": 2.1794421260707226, + "grad_norm": 2.481123924255371, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7413781881332397, + "num_tokens": 513579110.0, + "step": 19846 + }, + { + "epoch": 2.1795519437733364, + "grad_norm": 2.3213372230529785, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7269824147224426, + "num_tokens": 513604626.0, + "step": 19847 + }, + { + "epoch": 2.17966176147595, + "grad_norm": 2.306591510772705, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7174170613288879, + "num_tokens": 513630768.0, + "step": 19848 + }, + { + "epoch": 2.1797715791785635, + "grad_norm": 2.259222984313965, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7455840706825256, + "num_tokens": 513656694.0, + "step": 19849 + }, + { + "epoch": 2.179881396881177, + "grad_norm": 2.2827069759368896, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7190719842910767, + "num_tokens": 513685769.0, + "step": 19850 + }, + { + "epoch": 2.179991214583791, + "grad_norm": 2.0428287982940674, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7054660320281982, + "num_tokens": 513719014.0, + "step": 19851 + }, + { + "epoch": 2.1801010322864047, + "grad_norm": 2.4948482513427734, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.723091185092926, + "num_tokens": 513745490.0, + "step": 19852 + }, + { + "epoch": 2.180210849989018, + "grad_norm": 2.242720127105713, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7213521003723145, + "num_tokens": 513774431.0, + "step": 19853 + }, + { + "epoch": 2.180320667691632, + "grad_norm": 2.2912938594818115, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7325072884559631, + "num_tokens": 513799223.0, + "step": 19854 + }, + { + "epoch": 2.1804304853942456, + "grad_norm": 2.2686984539031982, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7239563465118408, + "num_tokens": 513826121.0, + "step": 19855 + }, + { + "epoch": 2.1805403030968593, + "grad_norm": 2.0641016960144043, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7205720543861389, + "num_tokens": 513857479.0, + "step": 19856 + }, + { + "epoch": 2.180650120799473, + "grad_norm": 2.6719837188720703, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7213643789291382, + "num_tokens": 513878498.0, + "step": 19857 + }, + { + "epoch": 2.1807599385020864, + "grad_norm": 2.6973061561584473, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7355043888092041, + "num_tokens": 513899898.0, + "step": 19858 + }, + { + "epoch": 2.1808697562047, + "grad_norm": 2.162015438079834, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7205286026000977, + "num_tokens": 513929478.0, + "step": 19859 + }, + { + "epoch": 2.180979573907314, + "grad_norm": 2.228720188140869, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7272073030471802, + "num_tokens": 513956905.0, + "step": 19860 + }, + { + "epoch": 2.1810893916099277, + "grad_norm": 2.1906073093414307, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7453773617744446, + "num_tokens": 513984325.0, + "step": 19861 + }, + { + "epoch": 2.181199209312541, + "grad_norm": 2.2128546237945557, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7093176245689392, + "num_tokens": 514013885.0, + "step": 19862 + }, + { + "epoch": 2.1813090270151547, + "grad_norm": 2.0677239894866943, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7232528924942017, + "num_tokens": 514045712.0, + "step": 19863 + }, + { + "epoch": 2.1814188447177685, + "grad_norm": 2.6636219024658203, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7370943427085876, + "num_tokens": 514067777.0, + "step": 19864 + }, + { + "epoch": 2.1815286624203822, + "grad_norm": 3.0673274993896484, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7579007148742676, + "num_tokens": 514084239.0, + "step": 19865 + }, + { + "epoch": 2.181638480122996, + "grad_norm": 2.3400635719299316, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7146424055099487, + "num_tokens": 514111071.0, + "step": 19866 + }, + { + "epoch": 2.1817482978256093, + "grad_norm": 2.2669870853424072, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.747246503829956, + "num_tokens": 514139515.0, + "step": 19867 + }, + { + "epoch": 2.181858115528223, + "grad_norm": 2.3368780612945557, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7181662321090698, + "num_tokens": 514166181.0, + "step": 19868 + }, + { + "epoch": 2.181967933230837, + "grad_norm": 2.1993680000305176, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7436745762825012, + "num_tokens": 514194972.0, + "step": 19869 + }, + { + "epoch": 2.1820777509334506, + "grad_norm": 2.262207269668579, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7204281091690063, + "num_tokens": 514224534.0, + "step": 19870 + }, + { + "epoch": 2.1821875686360643, + "grad_norm": 2.0749735832214355, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7436491250991821, + "num_tokens": 514256824.0, + "step": 19871 + }, + { + "epoch": 2.1822973863386776, + "grad_norm": 2.209925889968872, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.739008903503418, + "num_tokens": 514284891.0, + "step": 19872 + }, + { + "epoch": 2.1824072040412914, + "grad_norm": 2.6294443607330322, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7467751502990723, + "num_tokens": 514306297.0, + "step": 19873 + }, + { + "epoch": 2.182517021743905, + "grad_norm": 2.3770391941070557, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7411602139472961, + "num_tokens": 514331575.0, + "step": 19874 + }, + { + "epoch": 2.182626839446519, + "grad_norm": 2.4583706855773926, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7314450144767761, + "num_tokens": 514355046.0, + "step": 19875 + }, + { + "epoch": 2.1827366571491327, + "grad_norm": 2.32004451751709, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7321217060089111, + "num_tokens": 514381567.0, + "step": 19876 + }, + { + "epoch": 2.182846474851746, + "grad_norm": 2.4410297870635986, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7434847354888916, + "num_tokens": 514404830.0, + "step": 19877 + }, + { + "epoch": 2.1829562925543597, + "grad_norm": 2.138190984725952, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7374112606048584, + "num_tokens": 514434722.0, + "step": 19878 + }, + { + "epoch": 2.1830661102569735, + "grad_norm": 2.563690662384033, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.723709225654602, + "num_tokens": 514457611.0, + "step": 19879 + }, + { + "epoch": 2.1831759279595873, + "grad_norm": 2.8258414268493652, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7477052211761475, + "num_tokens": 514477830.0, + "step": 19880 + }, + { + "epoch": 2.1832857456622006, + "grad_norm": 2.2677977085113525, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7436885237693787, + "num_tokens": 514503871.0, + "step": 19881 + }, + { + "epoch": 2.1833955633648143, + "grad_norm": 2.632230043411255, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7332932949066162, + "num_tokens": 514525892.0, + "step": 19882 + }, + { + "epoch": 2.183505381067428, + "grad_norm": 2.1086361408233643, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7204355001449585, + "num_tokens": 514557113.0, + "step": 19883 + }, + { + "epoch": 2.183615198770042, + "grad_norm": 2.11657452583313, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7357776165008545, + "num_tokens": 514588219.0, + "step": 19884 + }, + { + "epoch": 2.1837250164726556, + "grad_norm": 2.2183403968811035, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7413305640220642, + "num_tokens": 514615084.0, + "step": 19885 + }, + { + "epoch": 2.183834834175269, + "grad_norm": 2.325010061264038, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7352259755134583, + "num_tokens": 514639856.0, + "step": 19886 + }, + { + "epoch": 2.1839446518778827, + "grad_norm": 2.9087467193603516, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.747254490852356, + "num_tokens": 514659269.0, + "step": 19887 + }, + { + "epoch": 2.1840544695804964, + "grad_norm": 2.492135763168335, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7302062511444092, + "num_tokens": 514682620.0, + "step": 19888 + }, + { + "epoch": 2.18416428728311, + "grad_norm": 2.694760322570801, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7216516137123108, + "num_tokens": 514705976.0, + "step": 19889 + }, + { + "epoch": 2.1842741049857235, + "grad_norm": 2.511305570602417, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7567242383956909, + "num_tokens": 514727483.0, + "step": 19890 + }, + { + "epoch": 2.1843839226883373, + "grad_norm": 2.522313117980957, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7285879254341125, + "num_tokens": 514750353.0, + "step": 19891 + }, + { + "epoch": 2.184493740390951, + "grad_norm": 2.2206172943115234, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7223702073097229, + "num_tokens": 514776854.0, + "step": 19892 + }, + { + "epoch": 2.1846035580935648, + "grad_norm": 2.0918469429016113, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.732262134552002, + "num_tokens": 514809145.0, + "step": 19893 + }, + { + "epoch": 2.1847133757961785, + "grad_norm": 2.186736583709717, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7302569150924683, + "num_tokens": 514837614.0, + "step": 19894 + }, + { + "epoch": 2.184823193498792, + "grad_norm": 2.385073661804199, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7024571895599365, + "num_tokens": 514863122.0, + "step": 19895 + }, + { + "epoch": 2.1849330112014056, + "grad_norm": 2.5296528339385986, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7473660707473755, + "num_tokens": 514884835.0, + "step": 19896 + }, + { + "epoch": 2.1850428289040194, + "grad_norm": 2.3546407222747803, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7381139993667603, + "num_tokens": 514910027.0, + "step": 19897 + }, + { + "epoch": 2.185152646606633, + "grad_norm": 2.390702724456787, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7581179738044739, + "num_tokens": 514934175.0, + "step": 19898 + }, + { + "epoch": 2.185262464309247, + "grad_norm": 2.5950820446014404, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.728344202041626, + "num_tokens": 514956012.0, + "step": 19899 + }, + { + "epoch": 2.18537228201186, + "grad_norm": 2.331270694732666, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7348021864891052, + "num_tokens": 514982460.0, + "step": 19900 + }, + { + "epoch": 2.185482099714474, + "grad_norm": 2.5467076301574707, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7507122755050659, + "num_tokens": 515005156.0, + "step": 19901 + }, + { + "epoch": 2.1855919174170877, + "grad_norm": 2.439420223236084, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7273004651069641, + "num_tokens": 515030699.0, + "step": 19902 + }, + { + "epoch": 2.1857017351197014, + "grad_norm": 2.3391737937927246, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7304607629776001, + "num_tokens": 515055876.0, + "step": 19903 + }, + { + "epoch": 2.1858115528223148, + "grad_norm": 2.179565668106079, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7153472900390625, + "num_tokens": 515086888.0, + "step": 19904 + }, + { + "epoch": 2.1859213705249285, + "grad_norm": 2.261683940887451, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7396674156188965, + "num_tokens": 515115200.0, + "step": 19905 + }, + { + "epoch": 2.1860311882275423, + "grad_norm": 2.314120292663574, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7310535907745361, + "num_tokens": 515141586.0, + "step": 19906 + }, + { + "epoch": 2.186141005930156, + "grad_norm": 2.3603928089141846, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7347322702407837, + "num_tokens": 515167298.0, + "step": 19907 + }, + { + "epoch": 2.18625082363277, + "grad_norm": 2.5180656909942627, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7359089255332947, + "num_tokens": 515190325.0, + "step": 19908 + }, + { + "epoch": 2.186360641335383, + "grad_norm": 2.3910274505615234, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7462574243545532, + "num_tokens": 515215158.0, + "step": 19909 + }, + { + "epoch": 2.186470459037997, + "grad_norm": 2.161149263381958, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7429699897766113, + "num_tokens": 515244882.0, + "step": 19910 + }, + { + "epoch": 2.1865802767406106, + "grad_norm": 2.679985761642456, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7394771575927734, + "num_tokens": 515265951.0, + "step": 19911 + }, + { + "epoch": 2.1866900944432244, + "grad_norm": 2.648972988128662, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7259962558746338, + "num_tokens": 515287808.0, + "step": 19912 + }, + { + "epoch": 2.1867999121458377, + "grad_norm": 2.478502035140991, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7366887331008911, + "num_tokens": 515312532.0, + "step": 19913 + }, + { + "epoch": 2.1869097298484514, + "grad_norm": 2.9696362018585205, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7482218146324158, + "num_tokens": 515330522.0, + "step": 19914 + }, + { + "epoch": 2.187019547551065, + "grad_norm": 2.6015846729278564, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7096962928771973, + "num_tokens": 515353629.0, + "step": 19915 + }, + { + "epoch": 2.187129365253679, + "grad_norm": 2.6082732677459717, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7391613721847534, + "num_tokens": 515377622.0, + "step": 19916 + }, + { + "epoch": 2.1872391829562927, + "grad_norm": 2.4908509254455566, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7273692488670349, + "num_tokens": 515402491.0, + "step": 19917 + }, + { + "epoch": 2.187349000658906, + "grad_norm": 2.4883227348327637, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7539496421813965, + "num_tokens": 515427669.0, + "step": 19918 + }, + { + "epoch": 2.18745881836152, + "grad_norm": 2.3040592670440674, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7044914960861206, + "num_tokens": 515455641.0, + "step": 19919 + }, + { + "epoch": 2.1875686360641335, + "grad_norm": 2.437957525253296, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7462035417556763, + "num_tokens": 515480433.0, + "step": 19920 + }, + { + "epoch": 2.1876784537667473, + "grad_norm": 2.3039684295654297, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7339785099029541, + "num_tokens": 515506270.0, + "step": 19921 + }, + { + "epoch": 2.187788271469361, + "grad_norm": 2.437699794769287, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7336145639419556, + "num_tokens": 515532306.0, + "step": 19922 + }, + { + "epoch": 2.1878980891719744, + "grad_norm": 2.3657190799713135, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7197101712226868, + "num_tokens": 515556984.0, + "step": 19923 + }, + { + "epoch": 2.188007906874588, + "grad_norm": 2.469797134399414, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7316235303878784, + "num_tokens": 515581961.0, + "step": 19924 + }, + { + "epoch": 2.188117724577202, + "grad_norm": 3.0666303634643555, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7393745183944702, + "num_tokens": 515599473.0, + "step": 19925 + }, + { + "epoch": 2.1882275422798156, + "grad_norm": 2.3453075885772705, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7164610624313354, + "num_tokens": 515623550.0, + "step": 19926 + }, + { + "epoch": 2.1883373599824294, + "grad_norm": 2.3931772708892822, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7630138397216797, + "num_tokens": 515647512.0, + "step": 19927 + }, + { + "epoch": 2.1884471776850427, + "grad_norm": 2.762117385864258, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7613101005554199, + "num_tokens": 515666805.0, + "step": 19928 + }, + { + "epoch": 2.1885569953876565, + "grad_norm": 2.5278778076171875, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7436329126358032, + "num_tokens": 515691097.0, + "step": 19929 + }, + { + "epoch": 2.1886668130902702, + "grad_norm": 2.1471829414367676, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7031478881835938, + "num_tokens": 515721818.0, + "step": 19930 + }, + { + "epoch": 2.188776630792884, + "grad_norm": 2.1681389808654785, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7112256288528442, + "num_tokens": 515754030.0, + "step": 19931 + }, + { + "epoch": 2.1888864484954973, + "grad_norm": 2.332122325897217, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7329506874084473, + "num_tokens": 515780601.0, + "step": 19932 + }, + { + "epoch": 2.188996266198111, + "grad_norm": 2.4286913871765137, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7347828149795532, + "num_tokens": 515803712.0, + "step": 19933 + }, + { + "epoch": 2.189106083900725, + "grad_norm": 2.449465274810791, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7522832155227661, + "num_tokens": 515826421.0, + "step": 19934 + }, + { + "epoch": 2.1892159016033386, + "grad_norm": 2.3721745014190674, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7140834331512451, + "num_tokens": 515852185.0, + "step": 19935 + }, + { + "epoch": 2.1893257193059523, + "grad_norm": 2.3274292945861816, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7211989164352417, + "num_tokens": 515880345.0, + "step": 19936 + }, + { + "epoch": 2.1894355370085656, + "grad_norm": 2.182267904281616, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7278040647506714, + "num_tokens": 515909406.0, + "step": 19937 + }, + { + "epoch": 2.1895453547111794, + "grad_norm": 2.1651155948638916, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7356619238853455, + "num_tokens": 515938440.0, + "step": 19938 + }, + { + "epoch": 2.189655172413793, + "grad_norm": 2.583157539367676, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7313347458839417, + "num_tokens": 515960058.0, + "step": 19939 + }, + { + "epoch": 2.189764990116407, + "grad_norm": 2.7875585556030273, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7369529008865356, + "num_tokens": 515980139.0, + "step": 19940 + }, + { + "epoch": 2.18987480781902, + "grad_norm": 2.6157801151275635, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7346783876419067, + "num_tokens": 516000716.0, + "step": 19941 + }, + { + "epoch": 2.189984625521634, + "grad_norm": 2.9802417755126953, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7435760498046875, + "num_tokens": 516019694.0, + "step": 19942 + }, + { + "epoch": 2.1900944432242477, + "grad_norm": 2.191436767578125, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.715083122253418, + "num_tokens": 516049134.0, + "step": 19943 + }, + { + "epoch": 2.1902042609268615, + "grad_norm": 2.2743890285491943, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7216575145721436, + "num_tokens": 516076668.0, + "step": 19944 + }, + { + "epoch": 2.1903140786294752, + "grad_norm": 2.454252004623413, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.736845850944519, + "num_tokens": 516101344.0, + "step": 19945 + }, + { + "epoch": 2.1904238963320886, + "grad_norm": 2.392563581466675, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7174221277236938, + "num_tokens": 516127699.0, + "step": 19946 + }, + { + "epoch": 2.1905337140347023, + "grad_norm": 2.3159399032592773, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7415060997009277, + "num_tokens": 516153496.0, + "step": 19947 + }, + { + "epoch": 2.190643531737316, + "grad_norm": 2.491025447845459, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7118551731109619, + "num_tokens": 516179007.0, + "step": 19948 + }, + { + "epoch": 2.19075334943993, + "grad_norm": 2.355191469192505, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7323682904243469, + "num_tokens": 516203928.0, + "step": 19949 + }, + { + "epoch": 2.1908631671425436, + "grad_norm": 2.483485460281372, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7411107420921326, + "num_tokens": 516227878.0, + "step": 19950 + }, + { + "epoch": 2.190972984845157, + "grad_norm": 2.32602596282959, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7256227731704712, + "num_tokens": 516254967.0, + "step": 19951 + }, + { + "epoch": 2.1910828025477707, + "grad_norm": 2.460841655731201, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.75472491979599, + "num_tokens": 516277432.0, + "step": 19952 + }, + { + "epoch": 2.1911926202503844, + "grad_norm": 2.668769598007202, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7242566347122192, + "num_tokens": 516300329.0, + "step": 19953 + }, + { + "epoch": 2.191302437952998, + "grad_norm": 2.3265209197998047, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7137357592582703, + "num_tokens": 516328649.0, + "step": 19954 + }, + { + "epoch": 2.1914122556556115, + "grad_norm": 2.3884801864624023, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7249809503555298, + "num_tokens": 516353113.0, + "step": 19955 + }, + { + "epoch": 2.1915220733582252, + "grad_norm": 2.094905138015747, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7407630085945129, + "num_tokens": 516383463.0, + "step": 19956 + }, + { + "epoch": 2.191631891060839, + "grad_norm": 2.33048677444458, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7506046295166016, + "num_tokens": 516407754.0, + "step": 19957 + }, + { + "epoch": 2.1917417087634528, + "grad_norm": 2.633296489715576, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7510161995887756, + "num_tokens": 516428975.0, + "step": 19958 + }, + { + "epoch": 2.1918515264660665, + "grad_norm": 2.861419439315796, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7491977214813232, + "num_tokens": 516448000.0, + "step": 19959 + }, + { + "epoch": 2.19196134416868, + "grad_norm": 2.187830924987793, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7136052250862122, + "num_tokens": 516475023.0, + "step": 19960 + }, + { + "epoch": 2.1920711618712936, + "grad_norm": 2.3561198711395264, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7396537065505981, + "num_tokens": 516501229.0, + "step": 19961 + }, + { + "epoch": 2.1921809795739073, + "grad_norm": 2.23347806930542, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7329325675964355, + "num_tokens": 516532095.0, + "step": 19962 + }, + { + "epoch": 2.192290797276521, + "grad_norm": 2.442833662033081, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7311393618583679, + "num_tokens": 516555590.0, + "step": 19963 + }, + { + "epoch": 2.1924006149791344, + "grad_norm": 2.187739610671997, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7251006364822388, + "num_tokens": 516586018.0, + "step": 19964 + }, + { + "epoch": 2.192510432681748, + "grad_norm": 2.5125937461853027, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7200834155082703, + "num_tokens": 516610101.0, + "step": 19965 + }, + { + "epoch": 2.192620250384362, + "grad_norm": 2.434722900390625, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7331485748291016, + "num_tokens": 516634371.0, + "step": 19966 + }, + { + "epoch": 2.1927300680869757, + "grad_norm": 2.390928030014038, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7482736110687256, + "num_tokens": 516659537.0, + "step": 19967 + }, + { + "epoch": 2.1928398857895894, + "grad_norm": 2.4460151195526123, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7374664545059204, + "num_tokens": 516684554.0, + "step": 19968 + }, + { + "epoch": 2.1929497034922028, + "grad_norm": 2.5748369693756104, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7554140090942383, + "num_tokens": 516706127.0, + "step": 19969 + }, + { + "epoch": 2.1930595211948165, + "grad_norm": 2.4046735763549805, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7371887564659119, + "num_tokens": 516729918.0, + "step": 19970 + }, + { + "epoch": 2.1931693388974303, + "grad_norm": 2.327430486679077, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7470928430557251, + "num_tokens": 516754609.0, + "step": 19971 + }, + { + "epoch": 2.193279156600044, + "grad_norm": 2.550454616546631, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.716462254524231, + "num_tokens": 516779618.0, + "step": 19972 + }, + { + "epoch": 2.193388974302658, + "grad_norm": 2.3829643726348877, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7406232357025146, + "num_tokens": 516803725.0, + "step": 19973 + }, + { + "epoch": 2.193498792005271, + "grad_norm": 2.3325600624084473, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.75135737657547, + "num_tokens": 516829951.0, + "step": 19974 + }, + { + "epoch": 2.193608609707885, + "grad_norm": 2.1155645847320557, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7047521471977234, + "num_tokens": 516861823.0, + "step": 19975 + }, + { + "epoch": 2.1937184274104986, + "grad_norm": 2.3258328437805176, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.723194420337677, + "num_tokens": 516888694.0, + "step": 19976 + }, + { + "epoch": 2.1938282451131124, + "grad_norm": 2.3061556816101074, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7532863616943359, + "num_tokens": 516913859.0, + "step": 19977 + }, + { + "epoch": 2.193938062815726, + "grad_norm": 2.75107741355896, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7361592054367065, + "num_tokens": 516934106.0, + "step": 19978 + }, + { + "epoch": 2.1940478805183394, + "grad_norm": 2.2280733585357666, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7399110794067383, + "num_tokens": 516962220.0, + "step": 19979 + }, + { + "epoch": 2.194157698220953, + "grad_norm": 2.2321081161499023, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7272984981536865, + "num_tokens": 516992097.0, + "step": 19980 + }, + { + "epoch": 2.194267515923567, + "grad_norm": 2.0886638164520264, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7320396900177002, + "num_tokens": 517024985.0, + "step": 19981 + }, + { + "epoch": 2.1943773336261807, + "grad_norm": 2.4187965393066406, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7311633825302124, + "num_tokens": 517049686.0, + "step": 19982 + }, + { + "epoch": 2.194487151328794, + "grad_norm": 2.5872251987457275, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7322707176208496, + "num_tokens": 517073345.0, + "step": 19983 + }, + { + "epoch": 2.1945969690314078, + "grad_norm": 2.2907631397247314, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.734856367111206, + "num_tokens": 517099659.0, + "step": 19984 + }, + { + "epoch": 2.1947067867340215, + "grad_norm": 2.15008282661438, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7233394980430603, + "num_tokens": 517127778.0, + "step": 19985 + }, + { + "epoch": 2.1948166044366353, + "grad_norm": 2.7304835319519043, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7649765610694885, + "num_tokens": 517147665.0, + "step": 19986 + }, + { + "epoch": 2.194926422139249, + "grad_norm": 2.253185749053955, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7188634276390076, + "num_tokens": 517173900.0, + "step": 19987 + }, + { + "epoch": 2.1950362398418624, + "grad_norm": 2.3003013134002686, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7425668835639954, + "num_tokens": 517201170.0, + "step": 19988 + }, + { + "epoch": 2.195146057544476, + "grad_norm": 2.221167802810669, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.6927773356437683, + "num_tokens": 517230427.0, + "step": 19989 + }, + { + "epoch": 2.19525587524709, + "grad_norm": 2.0825273990631104, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7147567868232727, + "num_tokens": 517265792.0, + "step": 19990 + }, + { + "epoch": 2.1953656929497036, + "grad_norm": 2.6154773235321045, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7377020120620728, + "num_tokens": 517287592.0, + "step": 19991 + }, + { + "epoch": 2.195475510652317, + "grad_norm": 2.523538827896118, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7531511187553406, + "num_tokens": 517310886.0, + "step": 19992 + }, + { + "epoch": 2.1955853283549307, + "grad_norm": 2.6014487743377686, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7446272969245911, + "num_tokens": 517332572.0, + "step": 19993 + }, + { + "epoch": 2.1956951460575445, + "grad_norm": 2.593005418777466, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7360491752624512, + "num_tokens": 517355634.0, + "step": 19994 + }, + { + "epoch": 2.195804963760158, + "grad_norm": 2.096191167831421, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7025829553604126, + "num_tokens": 517385468.0, + "step": 19995 + }, + { + "epoch": 2.195914781462772, + "grad_norm": 2.014437675476074, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7230169773101807, + "num_tokens": 517417204.0, + "step": 19996 + }, + { + "epoch": 2.1960245991653853, + "grad_norm": 2.439868927001953, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7577000856399536, + "num_tokens": 517441033.0, + "step": 19997 + }, + { + "epoch": 2.196134416867999, + "grad_norm": 2.308192491531372, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.713335394859314, + "num_tokens": 517469258.0, + "step": 19998 + }, + { + "epoch": 2.196244234570613, + "grad_norm": 2.34995174407959, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7091468572616577, + "num_tokens": 517495697.0, + "step": 19999 + }, + { + "epoch": 2.1963540522732266, + "grad_norm": 2.1202170848846436, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7315699458122253, + "num_tokens": 517526328.0, + "step": 20000 + }, + { + "epoch": 2.1964638699758403, + "grad_norm": 2.2678472995758057, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7053359746932983, + "num_tokens": 517556477.0, + "step": 20001 + }, + { + "epoch": 2.1965736876784536, + "grad_norm": 2.2972183227539062, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7295420169830322, + "num_tokens": 517583923.0, + "step": 20002 + }, + { + "epoch": 2.1966835053810674, + "grad_norm": 2.0973150730133057, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7126146554946899, + "num_tokens": 517618071.0, + "step": 20003 + }, + { + "epoch": 2.196793323083681, + "grad_norm": 2.5526840686798096, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7647470235824585, + "num_tokens": 517640507.0, + "step": 20004 + }, + { + "epoch": 2.196903140786295, + "grad_norm": 2.3733270168304443, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7392505407333374, + "num_tokens": 517665619.0, + "step": 20005 + }, + { + "epoch": 2.197012958488908, + "grad_norm": 2.1633412837982178, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7435289025306702, + "num_tokens": 517693362.0, + "step": 20006 + }, + { + "epoch": 2.197122776191522, + "grad_norm": 2.4764254093170166, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7308914661407471, + "num_tokens": 517716288.0, + "step": 20007 + }, + { + "epoch": 2.1972325938941357, + "grad_norm": 2.338665246963501, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.713029146194458, + "num_tokens": 517741774.0, + "step": 20008 + }, + { + "epoch": 2.1973424115967495, + "grad_norm": 2.177506685256958, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7154501676559448, + "num_tokens": 517773664.0, + "step": 20009 + }, + { + "epoch": 2.1974522292993632, + "grad_norm": 2.5738189220428467, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7423450946807861, + "num_tokens": 517793926.0, + "step": 20010 + }, + { + "epoch": 2.1975620470019765, + "grad_norm": 2.1619629859924316, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7210153341293335, + "num_tokens": 517823769.0, + "step": 20011 + }, + { + "epoch": 2.1976718647045903, + "grad_norm": 2.489919900894165, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7346488833427429, + "num_tokens": 517849202.0, + "step": 20012 + }, + { + "epoch": 2.197781682407204, + "grad_norm": 2.3297674655914307, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7217007875442505, + "num_tokens": 517877265.0, + "step": 20013 + }, + { + "epoch": 2.197891500109818, + "grad_norm": 2.435077428817749, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.735811710357666, + "num_tokens": 517900377.0, + "step": 20014 + }, + { + "epoch": 2.198001317812431, + "grad_norm": 2.149721145629883, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7575259208679199, + "num_tokens": 517927383.0, + "step": 20015 + }, + { + "epoch": 2.198111135515045, + "grad_norm": 2.1509902477264404, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7111358642578125, + "num_tokens": 517957042.0, + "step": 20016 + }, + { + "epoch": 2.1982209532176586, + "grad_norm": 2.3769261837005615, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7195579409599304, + "num_tokens": 517982093.0, + "step": 20017 + }, + { + "epoch": 2.1983307709202724, + "grad_norm": 2.307223081588745, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7514575123786926, + "num_tokens": 518006726.0, + "step": 20018 + }, + { + "epoch": 2.198440588622886, + "grad_norm": 2.1514997482299805, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7274571061134338, + "num_tokens": 518038182.0, + "step": 20019 + }, + { + "epoch": 2.1985504063254995, + "grad_norm": 2.4184420108795166, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7588027715682983, + "num_tokens": 518061637.0, + "step": 20020 + }, + { + "epoch": 2.1986602240281132, + "grad_norm": 2.2309458255767822, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7199814319610596, + "num_tokens": 518091519.0, + "step": 20021 + }, + { + "epoch": 2.198770041730727, + "grad_norm": 2.4117653369903564, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7105878591537476, + "num_tokens": 518116665.0, + "step": 20022 + }, + { + "epoch": 2.1988798594333407, + "grad_norm": 2.105078935623169, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7182744741439819, + "num_tokens": 518151433.0, + "step": 20023 + }, + { + "epoch": 2.1989896771359545, + "grad_norm": 2.754542827606201, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.715628981590271, + "num_tokens": 518172320.0, + "step": 20024 + }, + { + "epoch": 2.199099494838568, + "grad_norm": 2.6453144550323486, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7630801200866699, + "num_tokens": 518192686.0, + "step": 20025 + }, + { + "epoch": 2.1992093125411816, + "grad_norm": 2.354755401611328, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7191421389579773, + "num_tokens": 518219581.0, + "step": 20026 + }, + { + "epoch": 2.1993191302437953, + "grad_norm": 2.2512497901916504, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.730605959892273, + "num_tokens": 518247489.0, + "step": 20027 + }, + { + "epoch": 2.199428947946409, + "grad_norm": 2.602464437484741, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7381188869476318, + "num_tokens": 518269276.0, + "step": 20028 + }, + { + "epoch": 2.199538765649023, + "grad_norm": 2.264202117919922, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7494958639144897, + "num_tokens": 518297160.0, + "step": 20029 + }, + { + "epoch": 2.199648583351636, + "grad_norm": 2.39517879486084, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7135578393936157, + "num_tokens": 518324758.0, + "step": 20030 + }, + { + "epoch": 2.19975840105425, + "grad_norm": 2.3627443313598633, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7460793852806091, + "num_tokens": 518351349.0, + "step": 20031 + }, + { + "epoch": 2.1998682187568637, + "grad_norm": 2.202491283416748, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.742523193359375, + "num_tokens": 518379043.0, + "step": 20032 + }, + { + "epoch": 2.1999780364594774, + "grad_norm": 2.200291633605957, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7268683910369873, + "num_tokens": 518408717.0, + "step": 20033 + }, + { + "epoch": 2.2000878541620907, + "grad_norm": 2.5528528690338135, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7257636785507202, + "num_tokens": 518433853.0, + "step": 20034 + }, + { + "epoch": 2.2001976718647045, + "grad_norm": 2.6094915866851807, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7165151834487915, + "num_tokens": 518457724.0, + "step": 20035 + }, + { + "epoch": 2.2003074895673183, + "grad_norm": 2.2249510288238525, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7539240121841431, + "num_tokens": 518485575.0, + "step": 20036 + }, + { + "epoch": 2.200417307269932, + "grad_norm": 2.2524540424346924, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7136508226394653, + "num_tokens": 518517005.0, + "step": 20037 + }, + { + "epoch": 2.2005271249725458, + "grad_norm": 2.465956687927246, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.714911937713623, + "num_tokens": 518541236.0, + "step": 20038 + }, + { + "epoch": 2.200636942675159, + "grad_norm": 2.630819082260132, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7401354312896729, + "num_tokens": 518565424.0, + "step": 20039 + }, + { + "epoch": 2.200746760377773, + "grad_norm": 2.347147226333618, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.749351978302002, + "num_tokens": 518589389.0, + "step": 20040 + }, + { + "epoch": 2.2008565780803866, + "grad_norm": 2.3879246711730957, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7299255132675171, + "num_tokens": 518614095.0, + "step": 20041 + }, + { + "epoch": 2.2009663957830004, + "grad_norm": 2.4587416648864746, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.726931631565094, + "num_tokens": 518638191.0, + "step": 20042 + }, + { + "epoch": 2.2010762134856137, + "grad_norm": 2.6638729572296143, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7349562644958496, + "num_tokens": 518660484.0, + "step": 20043 + }, + { + "epoch": 2.2011860311882274, + "grad_norm": 2.2646028995513916, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7159498333930969, + "num_tokens": 518687926.0, + "step": 20044 + }, + { + "epoch": 2.201295848890841, + "grad_norm": 2.147318124771118, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7161967158317566, + "num_tokens": 518719279.0, + "step": 20045 + }, + { + "epoch": 2.201405666593455, + "grad_norm": 2.04778790473938, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.720389723777771, + "num_tokens": 518753491.0, + "step": 20046 + }, + { + "epoch": 2.2015154842960687, + "grad_norm": 2.214024305343628, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7232468128204346, + "num_tokens": 518780926.0, + "step": 20047 + }, + { + "epoch": 2.201625301998682, + "grad_norm": 2.483421802520752, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7081504464149475, + "num_tokens": 518805478.0, + "step": 20048 + }, + { + "epoch": 2.2017351197012958, + "grad_norm": 2.3823342323303223, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7327920198440552, + "num_tokens": 518830675.0, + "step": 20049 + }, + { + "epoch": 2.2018449374039095, + "grad_norm": 2.624194622039795, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7512567639350891, + "num_tokens": 518854521.0, + "step": 20050 + }, + { + "epoch": 2.2019547551065233, + "grad_norm": 2.310941696166992, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7224844694137573, + "num_tokens": 518882182.0, + "step": 20051 + }, + { + "epoch": 2.202064572809137, + "grad_norm": 2.2386555671691895, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7302218675613403, + "num_tokens": 518911153.0, + "step": 20052 + }, + { + "epoch": 2.2021743905117503, + "grad_norm": 2.274777412414551, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.722619891166687, + "num_tokens": 518938259.0, + "step": 20053 + }, + { + "epoch": 2.202284208214364, + "grad_norm": 2.3960797786712646, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7677038311958313, + "num_tokens": 518960702.0, + "step": 20054 + }, + { + "epoch": 2.202394025916978, + "grad_norm": 2.2515852451324463, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7150589227676392, + "num_tokens": 518989298.0, + "step": 20055 + }, + { + "epoch": 2.2025038436195916, + "grad_norm": 2.3725814819335938, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7123222351074219, + "num_tokens": 519014972.0, + "step": 20056 + }, + { + "epoch": 2.2026136613222054, + "grad_norm": 2.611003875732422, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7405387163162231, + "num_tokens": 519038925.0, + "step": 20057 + }, + { + "epoch": 2.2027234790248187, + "grad_norm": 2.3243210315704346, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.730496883392334, + "num_tokens": 519064807.0, + "step": 20058 + }, + { + "epoch": 2.2028332967274324, + "grad_norm": 2.5404248237609863, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7583935260772705, + "num_tokens": 519086927.0, + "step": 20059 + }, + { + "epoch": 2.202943114430046, + "grad_norm": 2.3461296558380127, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7581912279129028, + "num_tokens": 519111766.0, + "step": 20060 + }, + { + "epoch": 2.20305293213266, + "grad_norm": 2.200575828552246, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7149142622947693, + "num_tokens": 519140981.0, + "step": 20061 + }, + { + "epoch": 2.2031627498352733, + "grad_norm": 2.420490026473999, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7192692756652832, + "num_tokens": 519165772.0, + "step": 20062 + }, + { + "epoch": 2.203272567537887, + "grad_norm": 2.4094154834747314, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7497052550315857, + "num_tokens": 519191829.0, + "step": 20063 + }, + { + "epoch": 2.203382385240501, + "grad_norm": 2.476006031036377, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7148377299308777, + "num_tokens": 519216179.0, + "step": 20064 + }, + { + "epoch": 2.2034922029431145, + "grad_norm": 2.1384503841400146, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.749962329864502, + "num_tokens": 519245075.0, + "step": 20065 + }, + { + "epoch": 2.2036020206457283, + "grad_norm": 2.272041082382202, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7355793714523315, + "num_tokens": 519272624.0, + "step": 20066 + }, + { + "epoch": 2.2037118383483416, + "grad_norm": 2.406501054763794, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.728126049041748, + "num_tokens": 519297407.0, + "step": 20067 + }, + { + "epoch": 2.2038216560509554, + "grad_norm": 1.9989091157913208, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7074024677276611, + "num_tokens": 519335419.0, + "step": 20068 + }, + { + "epoch": 2.203931473753569, + "grad_norm": 2.6415481567382812, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7548576593399048, + "num_tokens": 519355629.0, + "step": 20069 + }, + { + "epoch": 2.204041291456183, + "grad_norm": 2.374305009841919, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7347836494445801, + "num_tokens": 519381439.0, + "step": 20070 + }, + { + "epoch": 2.204151109158796, + "grad_norm": 2.4363300800323486, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.720974326133728, + "num_tokens": 519404641.0, + "step": 20071 + }, + { + "epoch": 2.20426092686141, + "grad_norm": 2.184495449066162, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7245967388153076, + "num_tokens": 519432333.0, + "step": 20072 + }, + { + "epoch": 2.2043707445640237, + "grad_norm": 2.228395462036133, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7201253175735474, + "num_tokens": 519462964.0, + "step": 20073 + }, + { + "epoch": 2.2044805622666375, + "grad_norm": 2.368408203125, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7369650602340698, + "num_tokens": 519487752.0, + "step": 20074 + }, + { + "epoch": 2.2045903799692512, + "grad_norm": 2.2267236709594727, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7262935638427734, + "num_tokens": 519516482.0, + "step": 20075 + }, + { + "epoch": 2.2047001976718645, + "grad_norm": 2.448026418685913, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7512487769126892, + "num_tokens": 519540401.0, + "step": 20076 + }, + { + "epoch": 2.2048100153744783, + "grad_norm": 2.4920241832733154, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7190378308296204, + "num_tokens": 519564891.0, + "step": 20077 + }, + { + "epoch": 2.204919833077092, + "grad_norm": 2.1181397438049316, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7187519073486328, + "num_tokens": 519595185.0, + "step": 20078 + }, + { + "epoch": 2.205029650779706, + "grad_norm": 2.3644001483917236, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7285470962524414, + "num_tokens": 519620293.0, + "step": 20079 + }, + { + "epoch": 2.2051394684823196, + "grad_norm": 2.5351250171661377, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7423220872879028, + "num_tokens": 519643636.0, + "step": 20080 + }, + { + "epoch": 2.205249286184933, + "grad_norm": 2.1631014347076416, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7189151048660278, + "num_tokens": 519671611.0, + "step": 20081 + }, + { + "epoch": 2.2053591038875466, + "grad_norm": 2.016845703125, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.716081976890564, + "num_tokens": 519706511.0, + "step": 20082 + }, + { + "epoch": 2.2054689215901604, + "grad_norm": 2.495194435119629, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7430853843688965, + "num_tokens": 519729225.0, + "step": 20083 + }, + { + "epoch": 2.205578739292774, + "grad_norm": 2.168138027191162, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7294724583625793, + "num_tokens": 519757241.0, + "step": 20084 + }, + { + "epoch": 2.2056885569953875, + "grad_norm": 2.2785871028900146, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7368420958518982, + "num_tokens": 519783692.0, + "step": 20085 + }, + { + "epoch": 2.205798374698001, + "grad_norm": 2.4901578426361084, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.753908634185791, + "num_tokens": 519807339.0, + "step": 20086 + }, + { + "epoch": 2.205908192400615, + "grad_norm": 2.50382661819458, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7444178462028503, + "num_tokens": 519830355.0, + "step": 20087 + }, + { + "epoch": 2.2060180101032287, + "grad_norm": 2.1773033142089844, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7309142351150513, + "num_tokens": 519859643.0, + "step": 20088 + }, + { + "epoch": 2.2061278278058425, + "grad_norm": 2.4975850582122803, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7434580326080322, + "num_tokens": 519883210.0, + "step": 20089 + }, + { + "epoch": 2.206237645508456, + "grad_norm": 2.647643566131592, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7573056817054749, + "num_tokens": 519903342.0, + "step": 20090 + }, + { + "epoch": 2.2063474632110696, + "grad_norm": 2.1919920444488525, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7382341623306274, + "num_tokens": 519930300.0, + "step": 20091 + }, + { + "epoch": 2.2064572809136833, + "grad_norm": 2.174315929412842, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7226975560188293, + "num_tokens": 519960474.0, + "step": 20092 + }, + { + "epoch": 2.206567098616297, + "grad_norm": 2.2440760135650635, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7404787540435791, + "num_tokens": 519986070.0, + "step": 20093 + }, + { + "epoch": 2.2066769163189104, + "grad_norm": 2.4518651962280273, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7424450516700745, + "num_tokens": 520011594.0, + "step": 20094 + }, + { + "epoch": 2.206786734021524, + "grad_norm": 2.4333853721618652, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.716688334941864, + "num_tokens": 520037705.0, + "step": 20095 + }, + { + "epoch": 2.206896551724138, + "grad_norm": 2.7410829067230225, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7427041530609131, + "num_tokens": 520056695.0, + "step": 20096 + }, + { + "epoch": 2.2070063694267517, + "grad_norm": 2.500019073486328, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7429749965667725, + "num_tokens": 520079533.0, + "step": 20097 + }, + { + "epoch": 2.2071161871293654, + "grad_norm": 2.5386946201324463, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7298747301101685, + "num_tokens": 520104959.0, + "step": 20098 + }, + { + "epoch": 2.2072260048319787, + "grad_norm": 2.6136436462402344, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7493009567260742, + "num_tokens": 520126584.0, + "step": 20099 + }, + { + "epoch": 2.2073358225345925, + "grad_norm": 2.4398698806762695, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7352540493011475, + "num_tokens": 520149976.0, + "step": 20100 + }, + { + "epoch": 2.2074456402372062, + "grad_norm": 2.408902883529663, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7342161536216736, + "num_tokens": 520177916.0, + "step": 20101 + }, + { + "epoch": 2.20755545793982, + "grad_norm": 2.2823123931884766, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.726643979549408, + "num_tokens": 520205584.0, + "step": 20102 + }, + { + "epoch": 2.2076652756424338, + "grad_norm": 2.3053407669067383, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7253461480140686, + "num_tokens": 520234122.0, + "step": 20103 + }, + { + "epoch": 2.207775093345047, + "grad_norm": 2.5369935035705566, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7103439569473267, + "num_tokens": 520258717.0, + "step": 20104 + }, + { + "epoch": 2.207884911047661, + "grad_norm": 2.4201536178588867, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7431003451347351, + "num_tokens": 520284535.0, + "step": 20105 + }, + { + "epoch": 2.2079947287502746, + "grad_norm": 2.2959794998168945, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7534202337265015, + "num_tokens": 520310647.0, + "step": 20106 + }, + { + "epoch": 2.2081045464528883, + "grad_norm": 2.0388388633728027, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7106839418411255, + "num_tokens": 520344625.0, + "step": 20107 + }, + { + "epoch": 2.208214364155502, + "grad_norm": 2.330681800842285, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7382336854934692, + "num_tokens": 520368818.0, + "step": 20108 + }, + { + "epoch": 2.2083241818581154, + "grad_norm": 2.148176431655884, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7330448627471924, + "num_tokens": 520396062.0, + "step": 20109 + }, + { + "epoch": 2.208433999560729, + "grad_norm": 2.2985501289367676, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7291572093963623, + "num_tokens": 520423499.0, + "step": 20110 + }, + { + "epoch": 2.208543817263343, + "grad_norm": 1.8982104063034058, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7182652354240417, + "num_tokens": 520460909.0, + "step": 20111 + }, + { + "epoch": 2.2086536349659567, + "grad_norm": 2.238849639892578, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7394190430641174, + "num_tokens": 520487116.0, + "step": 20112 + }, + { + "epoch": 2.20876345266857, + "grad_norm": 2.2032811641693115, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7109368443489075, + "num_tokens": 520516152.0, + "step": 20113 + }, + { + "epoch": 2.2088732703711838, + "grad_norm": 2.448488235473633, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7472246885299683, + "num_tokens": 520539944.0, + "step": 20114 + }, + { + "epoch": 2.2089830880737975, + "grad_norm": 2.1863162517547607, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7363806366920471, + "num_tokens": 520567663.0, + "step": 20115 + }, + { + "epoch": 2.2090929057764113, + "grad_norm": 2.6366329193115234, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7542635202407837, + "num_tokens": 520590056.0, + "step": 20116 + }, + { + "epoch": 2.209202723479025, + "grad_norm": 2.5455121994018555, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.742754340171814, + "num_tokens": 520612078.0, + "step": 20117 + }, + { + "epoch": 2.2093125411816383, + "grad_norm": 2.7812910079956055, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7369634509086609, + "num_tokens": 520631927.0, + "step": 20118 + }, + { + "epoch": 2.209422358884252, + "grad_norm": 2.399841070175171, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7271888256072998, + "num_tokens": 520658703.0, + "step": 20119 + }, + { + "epoch": 2.209532176586866, + "grad_norm": 2.7032384872436523, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.741909921169281, + "num_tokens": 520677388.0, + "step": 20120 + }, + { + "epoch": 2.2096419942894796, + "grad_norm": 2.2192060947418213, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7145525813102722, + "num_tokens": 520706453.0, + "step": 20121 + }, + { + "epoch": 2.209751811992093, + "grad_norm": 2.2269203662872314, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7326242327690125, + "num_tokens": 520736876.0, + "step": 20122 + }, + { + "epoch": 2.2098616296947067, + "grad_norm": 2.536959171295166, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.6986228227615356, + "num_tokens": 520762241.0, + "step": 20123 + }, + { + "epoch": 2.2099714473973204, + "grad_norm": 2.5091376304626465, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7338836193084717, + "num_tokens": 520785181.0, + "step": 20124 + }, + { + "epoch": 2.210081265099934, + "grad_norm": 2.2588613033294678, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7274912595748901, + "num_tokens": 520812759.0, + "step": 20125 + }, + { + "epoch": 2.210191082802548, + "grad_norm": 2.3840460777282715, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7443156242370605, + "num_tokens": 520839324.0, + "step": 20126 + }, + { + "epoch": 2.2103009005051613, + "grad_norm": 2.3398334980010986, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7342403531074524, + "num_tokens": 520867844.0, + "step": 20127 + }, + { + "epoch": 2.210410718207775, + "grad_norm": 2.252765655517578, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7132124304771423, + "num_tokens": 520897337.0, + "step": 20128 + }, + { + "epoch": 2.2105205359103888, + "grad_norm": 2.383268117904663, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7559753656387329, + "num_tokens": 520922710.0, + "step": 20129 + }, + { + "epoch": 2.2106303536130025, + "grad_norm": 2.4014029502868652, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7215116024017334, + "num_tokens": 520948579.0, + "step": 20130 + }, + { + "epoch": 2.2107401713156163, + "grad_norm": 2.312157392501831, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7209263443946838, + "num_tokens": 520976304.0, + "step": 20131 + }, + { + "epoch": 2.2108499890182296, + "grad_norm": 2.2181332111358643, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.6960899829864502, + "num_tokens": 521005547.0, + "step": 20132 + }, + { + "epoch": 2.2109598067208434, + "grad_norm": 2.4346306324005127, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.73094242811203, + "num_tokens": 521032555.0, + "step": 20133 + }, + { + "epoch": 2.211069624423457, + "grad_norm": 2.4360599517822266, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7114744782447815, + "num_tokens": 521059832.0, + "step": 20134 + }, + { + "epoch": 2.211179442126071, + "grad_norm": 2.245039463043213, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7197165489196777, + "num_tokens": 521088278.0, + "step": 20135 + }, + { + "epoch": 2.211289259828684, + "grad_norm": 2.5297210216522217, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7190923690795898, + "num_tokens": 521111519.0, + "step": 20136 + }, + { + "epoch": 2.211399077531298, + "grad_norm": 2.219780683517456, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7377018332481384, + "num_tokens": 521139870.0, + "step": 20137 + }, + { + "epoch": 2.2115088952339117, + "grad_norm": 2.4058127403259277, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7334798574447632, + "num_tokens": 521167763.0, + "step": 20138 + }, + { + "epoch": 2.2116187129365255, + "grad_norm": 2.7121572494506836, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7306139469146729, + "num_tokens": 521193950.0, + "step": 20139 + }, + { + "epoch": 2.211728530639139, + "grad_norm": 2.1669750213623047, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7245681285858154, + "num_tokens": 521227519.0, + "step": 20140 + }, + { + "epoch": 2.2118383483417525, + "grad_norm": 2.821842670440674, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7393425107002258, + "num_tokens": 521247804.0, + "step": 20141 + }, + { + "epoch": 2.2119481660443663, + "grad_norm": 2.6524875164031982, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7216547131538391, + "num_tokens": 521269010.0, + "step": 20142 + }, + { + "epoch": 2.21205798374698, + "grad_norm": 2.453324556350708, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7249466180801392, + "num_tokens": 521293769.0, + "step": 20143 + }, + { + "epoch": 2.212167801449594, + "grad_norm": 2.4304099082946777, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.740707278251648, + "num_tokens": 521318842.0, + "step": 20144 + }, + { + "epoch": 2.212277619152207, + "grad_norm": 2.296663522720337, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7283741235733032, + "num_tokens": 521346698.0, + "step": 20145 + }, + { + "epoch": 2.212387436854821, + "grad_norm": 2.4529433250427246, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7435489892959595, + "num_tokens": 521371578.0, + "step": 20146 + }, + { + "epoch": 2.2124972545574346, + "grad_norm": 2.344808578491211, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7447011470794678, + "num_tokens": 521397721.0, + "step": 20147 + }, + { + "epoch": 2.2126070722600484, + "grad_norm": 2.5748634338378906, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7304849624633789, + "num_tokens": 521418959.0, + "step": 20148 + }, + { + "epoch": 2.212716889962662, + "grad_norm": 2.259058713912964, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7234377861022949, + "num_tokens": 521447973.0, + "step": 20149 + }, + { + "epoch": 2.2128267076652754, + "grad_norm": 2.216052293777466, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7286015748977661, + "num_tokens": 521475234.0, + "step": 20150 + }, + { + "epoch": 2.212936525367889, + "grad_norm": 2.4127421379089355, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7439367175102234, + "num_tokens": 521499190.0, + "step": 20151 + }, + { + "epoch": 2.213046343070503, + "grad_norm": 2.382537364959717, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7311630249023438, + "num_tokens": 521525491.0, + "step": 20152 + }, + { + "epoch": 2.2131561607731167, + "grad_norm": 2.5426926612854004, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7240263223648071, + "num_tokens": 521548624.0, + "step": 20153 + }, + { + "epoch": 2.2132659784757305, + "grad_norm": 2.0943541526794434, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7141714096069336, + "num_tokens": 521579307.0, + "step": 20154 + }, + { + "epoch": 2.213375796178344, + "grad_norm": 2.485182046890259, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7538783550262451, + "num_tokens": 521602114.0, + "step": 20155 + }, + { + "epoch": 2.2134856138809575, + "grad_norm": 2.434323310852051, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7525585293769836, + "num_tokens": 521625068.0, + "step": 20156 + }, + { + "epoch": 2.2135954315835713, + "grad_norm": 2.3417088985443115, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7391414046287537, + "num_tokens": 521649794.0, + "step": 20157 + }, + { + "epoch": 2.213705249286185, + "grad_norm": 2.780264139175415, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7324538826942444, + "num_tokens": 521669433.0, + "step": 20158 + }, + { + "epoch": 2.213815066988799, + "grad_norm": 2.6650946140289307, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7249108552932739, + "num_tokens": 521691755.0, + "step": 20159 + }, + { + "epoch": 2.213924884691412, + "grad_norm": 2.324131727218628, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.732262372970581, + "num_tokens": 521717300.0, + "step": 20160 + }, + { + "epoch": 2.214034702394026, + "grad_norm": 2.275294780731201, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7034901976585388, + "num_tokens": 521746826.0, + "step": 20161 + }, + { + "epoch": 2.2141445200966396, + "grad_norm": 2.6713032722473145, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7441632747650146, + "num_tokens": 521767377.0, + "step": 20162 + }, + { + "epoch": 2.2142543377992534, + "grad_norm": 2.3363614082336426, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7335698008537292, + "num_tokens": 521792448.0, + "step": 20163 + }, + { + "epoch": 2.2143641555018667, + "grad_norm": 2.3678810596466064, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7342178225517273, + "num_tokens": 521817314.0, + "step": 20164 + }, + { + "epoch": 2.2144739732044805, + "grad_norm": 2.497299909591675, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7236469984054565, + "num_tokens": 521844677.0, + "step": 20165 + }, + { + "epoch": 2.2145837909070942, + "grad_norm": 2.6384735107421875, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7601586580276489, + "num_tokens": 521865563.0, + "step": 20166 + }, + { + "epoch": 2.214693608609708, + "grad_norm": 2.7372803688049316, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.740685760974884, + "num_tokens": 521885758.0, + "step": 20167 + }, + { + "epoch": 2.2148034263123217, + "grad_norm": 2.618685722351074, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7654377818107605, + "num_tokens": 521907417.0, + "step": 20168 + }, + { + "epoch": 2.214913244014935, + "grad_norm": 2.298137664794922, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7285605669021606, + "num_tokens": 521934415.0, + "step": 20169 + }, + { + "epoch": 2.215023061717549, + "grad_norm": 2.6399528980255127, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7226449251174927, + "num_tokens": 521956542.0, + "step": 20170 + }, + { + "epoch": 2.2151328794201626, + "grad_norm": 2.403937816619873, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7488877773284912, + "num_tokens": 521981406.0, + "step": 20171 + }, + { + "epoch": 2.2152426971227763, + "grad_norm": 2.3602962493896484, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7265287637710571, + "num_tokens": 522006913.0, + "step": 20172 + }, + { + "epoch": 2.2153525148253896, + "grad_norm": 2.3546299934387207, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.719746470451355, + "num_tokens": 522032649.0, + "step": 20173 + }, + { + "epoch": 2.2154623325280034, + "grad_norm": 2.6536636352539062, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7383586764335632, + "num_tokens": 522054575.0, + "step": 20174 + }, + { + "epoch": 2.215572150230617, + "grad_norm": 1.9861273765563965, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7028176784515381, + "num_tokens": 522088974.0, + "step": 20175 + }, + { + "epoch": 2.215681967933231, + "grad_norm": 2.5281031131744385, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7304339408874512, + "num_tokens": 522112488.0, + "step": 20176 + }, + { + "epoch": 2.2157917856358447, + "grad_norm": 2.6129262447357178, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7562936544418335, + "num_tokens": 522133063.0, + "step": 20177 + }, + { + "epoch": 2.215901603338458, + "grad_norm": 2.255983591079712, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7200981378555298, + "num_tokens": 522161195.0, + "step": 20178 + }, + { + "epoch": 2.2160114210410717, + "grad_norm": 2.265075206756592, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7332878708839417, + "num_tokens": 522188462.0, + "step": 20179 + }, + { + "epoch": 2.2161212387436855, + "grad_norm": 2.1933300495147705, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7163070440292358, + "num_tokens": 522218385.0, + "step": 20180 + }, + { + "epoch": 2.2162310564462993, + "grad_norm": 2.085167169570923, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7389943599700928, + "num_tokens": 522250015.0, + "step": 20181 + }, + { + "epoch": 2.216340874148913, + "grad_norm": 2.2916862964630127, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7279375195503235, + "num_tokens": 522276403.0, + "step": 20182 + }, + { + "epoch": 2.2164506918515263, + "grad_norm": 2.127824306488037, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7163374423980713, + "num_tokens": 522309096.0, + "step": 20183 + }, + { + "epoch": 2.21656050955414, + "grad_norm": 2.2628960609436035, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7112729549407959, + "num_tokens": 522337840.0, + "step": 20184 + }, + { + "epoch": 2.216670327256754, + "grad_norm": 2.361258029937744, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7654961347579956, + "num_tokens": 522361322.0, + "step": 20185 + }, + { + "epoch": 2.2167801449593676, + "grad_norm": 2.6105668544769287, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7519972324371338, + "num_tokens": 522382718.0, + "step": 20186 + }, + { + "epoch": 2.216889962661981, + "grad_norm": 2.385956287384033, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7525346279144287, + "num_tokens": 522407048.0, + "step": 20187 + }, + { + "epoch": 2.2169997803645947, + "grad_norm": 2.2619550228118896, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7722739577293396, + "num_tokens": 522432975.0, + "step": 20188 + }, + { + "epoch": 2.2171095980672084, + "grad_norm": 2.290874719619751, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7373099327087402, + "num_tokens": 522460947.0, + "step": 20189 + }, + { + "epoch": 2.217219415769822, + "grad_norm": 2.1136276721954346, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.737221896648407, + "num_tokens": 522491659.0, + "step": 20190 + }, + { + "epoch": 2.217329233472436, + "grad_norm": 2.5956010818481445, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7360580563545227, + "num_tokens": 522513786.0, + "step": 20191 + }, + { + "epoch": 2.2174390511750492, + "grad_norm": 2.338894844055176, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7394101619720459, + "num_tokens": 522540420.0, + "step": 20192 + }, + { + "epoch": 2.217548868877663, + "grad_norm": 2.7169580459594727, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7566969990730286, + "num_tokens": 522560474.0, + "step": 20193 + }, + { + "epoch": 2.2176586865802768, + "grad_norm": 2.4594151973724365, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7443886995315552, + "num_tokens": 522583465.0, + "step": 20194 + }, + { + "epoch": 2.2177685042828905, + "grad_norm": 2.3223040103912354, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7351940870285034, + "num_tokens": 522609422.0, + "step": 20195 + }, + { + "epoch": 2.217878321985504, + "grad_norm": 2.317992687225342, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7267435789108276, + "num_tokens": 522634318.0, + "step": 20196 + }, + { + "epoch": 2.2179881396881176, + "grad_norm": 2.466550588607788, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.748076319694519, + "num_tokens": 522658245.0, + "step": 20197 + }, + { + "epoch": 2.2180979573907313, + "grad_norm": 2.7490384578704834, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7324404716491699, + "num_tokens": 522682109.0, + "step": 20198 + }, + { + "epoch": 2.218207775093345, + "grad_norm": 2.104702949523926, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7272839546203613, + "num_tokens": 522713748.0, + "step": 20199 + }, + { + "epoch": 2.218317592795959, + "grad_norm": 2.3166184425354004, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7048603296279907, + "num_tokens": 522742450.0, + "step": 20200 + }, + { + "epoch": 2.218427410498572, + "grad_norm": 2.2505621910095215, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7239733934402466, + "num_tokens": 522771817.0, + "step": 20201 + }, + { + "epoch": 2.218537228201186, + "grad_norm": 2.201345205307007, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7140882015228271, + "num_tokens": 522799701.0, + "step": 20202 + }, + { + "epoch": 2.2186470459037997, + "grad_norm": 2.4169278144836426, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7334131002426147, + "num_tokens": 522824493.0, + "step": 20203 + }, + { + "epoch": 2.2187568636064134, + "grad_norm": 2.5869381427764893, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7231804132461548, + "num_tokens": 522846289.0, + "step": 20204 + }, + { + "epoch": 2.218866681309027, + "grad_norm": 2.632460832595825, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7478536367416382, + "num_tokens": 522867454.0, + "step": 20205 + }, + { + "epoch": 2.2189764990116405, + "grad_norm": 2.492234945297241, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7325217127799988, + "num_tokens": 522889945.0, + "step": 20206 + }, + { + "epoch": 2.2190863167142543, + "grad_norm": 1.9551396369934082, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7356404066085815, + "num_tokens": 522923535.0, + "step": 20207 + }, + { + "epoch": 2.219196134416868, + "grad_norm": 2.59948468208313, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7352469563484192, + "num_tokens": 522945158.0, + "step": 20208 + }, + { + "epoch": 2.219305952119482, + "grad_norm": 2.106590747833252, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7315517663955688, + "num_tokens": 522975291.0, + "step": 20209 + }, + { + "epoch": 2.2194157698220955, + "grad_norm": 2.175370216369629, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7379436492919922, + "num_tokens": 523002590.0, + "step": 20210 + }, + { + "epoch": 2.219525587524709, + "grad_norm": 2.7227165699005127, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7208242416381836, + "num_tokens": 523024099.0, + "step": 20211 + }, + { + "epoch": 2.2196354052273226, + "grad_norm": 2.5267419815063477, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7313213348388672, + "num_tokens": 523046434.0, + "step": 20212 + }, + { + "epoch": 2.2197452229299364, + "grad_norm": 2.5573909282684326, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7405661344528198, + "num_tokens": 523069020.0, + "step": 20213 + }, + { + "epoch": 2.21985504063255, + "grad_norm": 2.544965982437134, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7413650155067444, + "num_tokens": 523092677.0, + "step": 20214 + }, + { + "epoch": 2.2199648583351634, + "grad_norm": 2.153959274291992, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7264081835746765, + "num_tokens": 523125967.0, + "step": 20215 + }, + { + "epoch": 2.220074676037777, + "grad_norm": 2.3165414333343506, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7362515926361084, + "num_tokens": 523152192.0, + "step": 20216 + }, + { + "epoch": 2.220184493740391, + "grad_norm": 2.4863851070404053, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7266818881034851, + "num_tokens": 523174814.0, + "step": 20217 + }, + { + "epoch": 2.2202943114430047, + "grad_norm": 2.4551331996917725, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7202000021934509, + "num_tokens": 523199753.0, + "step": 20218 + }, + { + "epoch": 2.2204041291456185, + "grad_norm": 2.2144293785095215, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7229359149932861, + "num_tokens": 523228109.0, + "step": 20219 + }, + { + "epoch": 2.220513946848232, + "grad_norm": 2.223615884780884, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7234027981758118, + "num_tokens": 523257988.0, + "step": 20220 + }, + { + "epoch": 2.2206237645508455, + "grad_norm": 2.739820957183838, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7447773814201355, + "num_tokens": 523277505.0, + "step": 20221 + }, + { + "epoch": 2.2207335822534593, + "grad_norm": 2.202134370803833, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7263689637184143, + "num_tokens": 523304940.0, + "step": 20222 + }, + { + "epoch": 2.220843399956073, + "grad_norm": 2.256685495376587, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7181397676467896, + "num_tokens": 523332632.0, + "step": 20223 + }, + { + "epoch": 2.2209532176586864, + "grad_norm": 2.411991596221924, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.703923761844635, + "num_tokens": 523361444.0, + "step": 20224 + }, + { + "epoch": 2.2210630353613, + "grad_norm": 2.6147778034210205, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7374102473258972, + "num_tokens": 523383465.0, + "step": 20225 + }, + { + "epoch": 2.221172853063914, + "grad_norm": 2.479933261871338, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7519313097000122, + "num_tokens": 523406351.0, + "step": 20226 + }, + { + "epoch": 2.2212826707665276, + "grad_norm": 2.3938353061676025, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7489397525787354, + "num_tokens": 523432018.0, + "step": 20227 + }, + { + "epoch": 2.2213924884691414, + "grad_norm": 2.216052532196045, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7134002447128296, + "num_tokens": 523463555.0, + "step": 20228 + }, + { + "epoch": 2.2215023061717547, + "grad_norm": 2.6860568523406982, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7387750148773193, + "num_tokens": 523485185.0, + "step": 20229 + }, + { + "epoch": 2.2216121238743685, + "grad_norm": 2.477855920791626, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7143246531486511, + "num_tokens": 523509089.0, + "step": 20230 + }, + { + "epoch": 2.221721941576982, + "grad_norm": 2.414605140686035, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7329020500183105, + "num_tokens": 523534084.0, + "step": 20231 + }, + { + "epoch": 2.221831759279596, + "grad_norm": 2.1595139503479004, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7192767858505249, + "num_tokens": 523564309.0, + "step": 20232 + }, + { + "epoch": 2.2219415769822097, + "grad_norm": 2.338226556777954, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7436829805374146, + "num_tokens": 523588865.0, + "step": 20233 + }, + { + "epoch": 2.222051394684823, + "grad_norm": 2.1356496810913086, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7085015773773193, + "num_tokens": 523620490.0, + "step": 20234 + }, + { + "epoch": 2.222161212387437, + "grad_norm": 2.3997087478637695, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7387887835502625, + "num_tokens": 523645323.0, + "step": 20235 + }, + { + "epoch": 2.2222710300900506, + "grad_norm": 2.483140230178833, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7218723893165588, + "num_tokens": 523668337.0, + "step": 20236 + }, + { + "epoch": 2.2223808477926643, + "grad_norm": 2.363259792327881, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.736487627029419, + "num_tokens": 523692874.0, + "step": 20237 + }, + { + "epoch": 2.222490665495278, + "grad_norm": 2.5483856201171875, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7550100088119507, + "num_tokens": 523714459.0, + "step": 20238 + }, + { + "epoch": 2.2226004831978914, + "grad_norm": 2.581010580062866, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7380056977272034, + "num_tokens": 523736494.0, + "step": 20239 + }, + { + "epoch": 2.222710300900505, + "grad_norm": 2.203733205795288, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7187427282333374, + "num_tokens": 523766414.0, + "step": 20240 + }, + { + "epoch": 2.222820118603119, + "grad_norm": 2.290440082550049, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7243351340293884, + "num_tokens": 523794278.0, + "step": 20241 + }, + { + "epoch": 2.2229299363057327, + "grad_norm": 1.9230146408081055, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7464691400527954, + "num_tokens": 523829092.0, + "step": 20242 + }, + { + "epoch": 2.223039754008346, + "grad_norm": 2.926823854446411, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7441059947013855, + "num_tokens": 523847786.0, + "step": 20243 + }, + { + "epoch": 2.2231495717109597, + "grad_norm": 2.6625306606292725, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7160578966140747, + "num_tokens": 523873224.0, + "step": 20244 + }, + { + "epoch": 2.2232593894135735, + "grad_norm": 2.4207897186279297, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.74240642786026, + "num_tokens": 523899560.0, + "step": 20245 + }, + { + "epoch": 2.2233692071161872, + "grad_norm": 2.2251222133636475, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7252548933029175, + "num_tokens": 523926135.0, + "step": 20246 + }, + { + "epoch": 2.223479024818801, + "grad_norm": 2.6673667430877686, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7383034229278564, + "num_tokens": 523947372.0, + "step": 20247 + }, + { + "epoch": 2.2235888425214143, + "grad_norm": 2.4003982543945312, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7276096343994141, + "num_tokens": 523972720.0, + "step": 20248 + }, + { + "epoch": 2.223698660224028, + "grad_norm": 2.8995845317840576, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7643567323684692, + "num_tokens": 523990408.0, + "step": 20249 + }, + { + "epoch": 2.223808477926642, + "grad_norm": 2.4092671871185303, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7358089089393616, + "num_tokens": 524018298.0, + "step": 20250 + }, + { + "epoch": 2.2239182956292556, + "grad_norm": 2.215614080429077, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7246996164321899, + "num_tokens": 524047897.0, + "step": 20251 + }, + { + "epoch": 2.224028113331869, + "grad_norm": 2.49637770652771, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7123957276344299, + "num_tokens": 524072670.0, + "step": 20252 + }, + { + "epoch": 2.2241379310344827, + "grad_norm": 2.629988670349121, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7547978162765503, + "num_tokens": 524094237.0, + "step": 20253 + }, + { + "epoch": 2.2242477487370964, + "grad_norm": 2.4988772869110107, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7151597738265991, + "num_tokens": 524119778.0, + "step": 20254 + }, + { + "epoch": 2.22435756643971, + "grad_norm": 2.2047791481018066, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7318340539932251, + "num_tokens": 524147937.0, + "step": 20255 + }, + { + "epoch": 2.224467384142324, + "grad_norm": 1.9614968299865723, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7325054407119751, + "num_tokens": 524182165.0, + "step": 20256 + }, + { + "epoch": 2.2245772018449372, + "grad_norm": 2.6996917724609375, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7577517032623291, + "num_tokens": 524201340.0, + "step": 20257 + }, + { + "epoch": 2.224687019547551, + "grad_norm": 2.4090638160705566, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.728581428527832, + "num_tokens": 524226267.0, + "step": 20258 + }, + { + "epoch": 2.2247968372501647, + "grad_norm": 2.4169304370880127, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7405152320861816, + "num_tokens": 524249983.0, + "step": 20259 + }, + { + "epoch": 2.2249066549527785, + "grad_norm": 2.228066921234131, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7091583013534546, + "num_tokens": 524277426.0, + "step": 20260 + }, + { + "epoch": 2.2250164726553923, + "grad_norm": 2.458519458770752, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7196716070175171, + "num_tokens": 524302820.0, + "step": 20261 + }, + { + "epoch": 2.2251262903580056, + "grad_norm": 2.2787344455718994, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7301042079925537, + "num_tokens": 524331358.0, + "step": 20262 + }, + { + "epoch": 2.2252361080606193, + "grad_norm": 2.27831768989563, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7264784574508667, + "num_tokens": 524358418.0, + "step": 20263 + }, + { + "epoch": 2.225345925763233, + "grad_norm": 2.592991590499878, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7642098069190979, + "num_tokens": 524379393.0, + "step": 20264 + }, + { + "epoch": 2.225455743465847, + "grad_norm": 2.2328269481658936, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7430102825164795, + "num_tokens": 524406388.0, + "step": 20265 + }, + { + "epoch": 2.22556556116846, + "grad_norm": 2.188966751098633, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7151311635971069, + "num_tokens": 524434241.0, + "step": 20266 + }, + { + "epoch": 2.225675378871074, + "grad_norm": 2.2542858123779297, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7177829146385193, + "num_tokens": 524461854.0, + "step": 20267 + }, + { + "epoch": 2.2257851965736877, + "grad_norm": 2.0432279109954834, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.693626344203949, + "num_tokens": 524496859.0, + "step": 20268 + }, + { + "epoch": 2.2258950142763014, + "grad_norm": 2.5063817501068115, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7192814350128174, + "num_tokens": 524523319.0, + "step": 20269 + }, + { + "epoch": 2.226004831978915, + "grad_norm": 2.7195188999176025, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.75494384765625, + "num_tokens": 524543772.0, + "step": 20270 + }, + { + "epoch": 2.2261146496815285, + "grad_norm": 2.309176206588745, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7163041234016418, + "num_tokens": 524570526.0, + "step": 20271 + }, + { + "epoch": 2.2262244673841423, + "grad_norm": 2.852797269821167, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7420238256454468, + "num_tokens": 524591893.0, + "step": 20272 + }, + { + "epoch": 2.226334285086756, + "grad_norm": 2.475717782974243, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7327306866645813, + "num_tokens": 524617959.0, + "step": 20273 + }, + { + "epoch": 2.2264441027893698, + "grad_norm": 2.1805577278137207, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.739698052406311, + "num_tokens": 524650018.0, + "step": 20274 + }, + { + "epoch": 2.226553920491983, + "grad_norm": 2.3143341541290283, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7301459312438965, + "num_tokens": 524677642.0, + "step": 20275 + }, + { + "epoch": 2.226663738194597, + "grad_norm": 2.6067259311676025, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7287904024124146, + "num_tokens": 524700131.0, + "step": 20276 + }, + { + "epoch": 2.2267735558972106, + "grad_norm": 2.4888038635253906, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.748276948928833, + "num_tokens": 524723625.0, + "step": 20277 + }, + { + "epoch": 2.2268833735998244, + "grad_norm": 2.3224496841430664, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7231691479682922, + "num_tokens": 524749654.0, + "step": 20278 + }, + { + "epoch": 2.226993191302438, + "grad_norm": 2.6259942054748535, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7452377676963806, + "num_tokens": 524770999.0, + "step": 20279 + }, + { + "epoch": 2.2271030090050514, + "grad_norm": 2.2129454612731934, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7490681409835815, + "num_tokens": 524798476.0, + "step": 20280 + }, + { + "epoch": 2.227212826707665, + "grad_norm": 2.186298131942749, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7192956209182739, + "num_tokens": 524830257.0, + "step": 20281 + }, + { + "epoch": 2.227322644410279, + "grad_norm": 2.1920180320739746, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7400294542312622, + "num_tokens": 524860125.0, + "step": 20282 + }, + { + "epoch": 2.2274324621128927, + "grad_norm": 2.2230660915374756, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7328294515609741, + "num_tokens": 524886906.0, + "step": 20283 + }, + { + "epoch": 2.2275422798155065, + "grad_norm": 2.362661600112915, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7297965288162231, + "num_tokens": 524913202.0, + "step": 20284 + }, + { + "epoch": 2.2276520975181198, + "grad_norm": 2.8443164825439453, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7541643977165222, + "num_tokens": 524934388.0, + "step": 20285 + }, + { + "epoch": 2.2277619152207335, + "grad_norm": 2.758334159851074, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7454829812049866, + "num_tokens": 524954701.0, + "step": 20286 + }, + { + "epoch": 2.2278717329233473, + "grad_norm": 2.2361650466918945, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7437156438827515, + "num_tokens": 524982373.0, + "step": 20287 + }, + { + "epoch": 2.227981550625961, + "grad_norm": 2.460343599319458, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7436931133270264, + "num_tokens": 525006604.0, + "step": 20288 + }, + { + "epoch": 2.228091368328575, + "grad_norm": 2.533036708831787, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7484361529350281, + "num_tokens": 525030729.0, + "step": 20289 + }, + { + "epoch": 2.228201186031188, + "grad_norm": 2.38248610496521, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7215563654899597, + "num_tokens": 525057137.0, + "step": 20290 + }, + { + "epoch": 2.228311003733802, + "grad_norm": 2.549229145050049, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7274154424667358, + "num_tokens": 525083368.0, + "step": 20291 + }, + { + "epoch": 2.2284208214364156, + "grad_norm": 2.476637363433838, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.726183295249939, + "num_tokens": 525106564.0, + "step": 20292 + }, + { + "epoch": 2.2285306391390294, + "grad_norm": 2.376657009124756, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7359119653701782, + "num_tokens": 525131346.0, + "step": 20293 + }, + { + "epoch": 2.2286404568416427, + "grad_norm": 2.4362804889678955, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7277964353561401, + "num_tokens": 525153578.0, + "step": 20294 + }, + { + "epoch": 2.2287502745442564, + "grad_norm": 2.2342588901519775, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7376549243927002, + "num_tokens": 525180802.0, + "step": 20295 + }, + { + "epoch": 2.22886009224687, + "grad_norm": 2.3253190517425537, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7137099504470825, + "num_tokens": 525207531.0, + "step": 20296 + }, + { + "epoch": 2.228969909949484, + "grad_norm": 2.166161060333252, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7419508099555969, + "num_tokens": 525235532.0, + "step": 20297 + }, + { + "epoch": 2.2290797276520977, + "grad_norm": 2.4955506324768066, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7146190404891968, + "num_tokens": 525260748.0, + "step": 20298 + }, + { + "epoch": 2.229189545354711, + "grad_norm": 2.222195625305176, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7270869016647339, + "num_tokens": 525290637.0, + "step": 20299 + }, + { + "epoch": 2.229299363057325, + "grad_norm": 2.6700439453125, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7307471036911011, + "num_tokens": 525311476.0, + "step": 20300 + }, + { + "epoch": 2.2294091807599385, + "grad_norm": 2.352018117904663, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7415968179702759, + "num_tokens": 525336307.0, + "step": 20301 + }, + { + "epoch": 2.2295189984625523, + "grad_norm": 2.428842067718506, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7314229011535645, + "num_tokens": 525360801.0, + "step": 20302 + }, + { + "epoch": 2.2296288161651656, + "grad_norm": 2.7816667556762695, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.729956328868866, + "num_tokens": 525381272.0, + "step": 20303 + }, + { + "epoch": 2.2297386338677794, + "grad_norm": 2.431508779525757, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7130941152572632, + "num_tokens": 525405143.0, + "step": 20304 + }, + { + "epoch": 2.229848451570393, + "grad_norm": 2.2972795963287354, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7232067584991455, + "num_tokens": 525432417.0, + "step": 20305 + }, + { + "epoch": 2.229958269273007, + "grad_norm": 2.3041648864746094, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7269060611724854, + "num_tokens": 525458740.0, + "step": 20306 + }, + { + "epoch": 2.2300680869756206, + "grad_norm": 2.453537940979004, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7280139923095703, + "num_tokens": 525483970.0, + "step": 20307 + }, + { + "epoch": 2.230177904678234, + "grad_norm": 2.379838228225708, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.730691134929657, + "num_tokens": 525511566.0, + "step": 20308 + }, + { + "epoch": 2.2302877223808477, + "grad_norm": 2.4921462535858154, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.750565767288208, + "num_tokens": 525534404.0, + "step": 20309 + }, + { + "epoch": 2.2303975400834615, + "grad_norm": 2.432321310043335, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7251431941986084, + "num_tokens": 525559670.0, + "step": 20310 + }, + { + "epoch": 2.2305073577860752, + "grad_norm": 2.537360668182373, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7467458248138428, + "num_tokens": 525582828.0, + "step": 20311 + }, + { + "epoch": 2.230617175488689, + "grad_norm": 2.5490951538085938, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7347114086151123, + "num_tokens": 525605437.0, + "step": 20312 + }, + { + "epoch": 2.2307269931913023, + "grad_norm": 2.4608399868011475, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7478233575820923, + "num_tokens": 525628750.0, + "step": 20313 + }, + { + "epoch": 2.230836810893916, + "grad_norm": 2.4188506603240967, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7327518463134766, + "num_tokens": 525652681.0, + "step": 20314 + }, + { + "epoch": 2.23094662859653, + "grad_norm": 2.527287721633911, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7451285719871521, + "num_tokens": 525675705.0, + "step": 20315 + }, + { + "epoch": 2.2310564462991436, + "grad_norm": 2.272932767868042, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7349855303764343, + "num_tokens": 525702444.0, + "step": 20316 + }, + { + "epoch": 2.231166264001757, + "grad_norm": 2.506747007369995, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7479936480522156, + "num_tokens": 525725438.0, + "step": 20317 + }, + { + "epoch": 2.2312760817043706, + "grad_norm": 2.0611732006073, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7328710556030273, + "num_tokens": 525755652.0, + "step": 20318 + }, + { + "epoch": 2.2313858994069844, + "grad_norm": 2.3079280853271484, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7388979196548462, + "num_tokens": 525781826.0, + "step": 20319 + }, + { + "epoch": 2.231495717109598, + "grad_norm": 1.9584033489227295, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7090047001838684, + "num_tokens": 525820208.0, + "step": 20320 + }, + { + "epoch": 2.231605534812212, + "grad_norm": 2.474453926086426, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7393131852149963, + "num_tokens": 525841822.0, + "step": 20321 + }, + { + "epoch": 2.2317153525148252, + "grad_norm": 2.131373882293701, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7066240310668945, + "num_tokens": 525873699.0, + "step": 20322 + }, + { + "epoch": 2.231825170217439, + "grad_norm": 2.674985885620117, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7191311120986938, + "num_tokens": 525896504.0, + "step": 20323 + }, + { + "epoch": 2.2319349879200527, + "grad_norm": 2.2922401428222656, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.747241735458374, + "num_tokens": 525923052.0, + "step": 20324 + }, + { + "epoch": 2.2320448056226665, + "grad_norm": 2.358928918838501, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7225923538208008, + "num_tokens": 525950202.0, + "step": 20325 + }, + { + "epoch": 2.23215462332528, + "grad_norm": 2.0396530628204346, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7026699185371399, + "num_tokens": 525983447.0, + "step": 20326 + }, + { + "epoch": 2.2322644410278936, + "grad_norm": 2.346055030822754, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7036185264587402, + "num_tokens": 526010645.0, + "step": 20327 + }, + { + "epoch": 2.2323742587305073, + "grad_norm": 2.6489646434783936, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7206933498382568, + "num_tokens": 526033189.0, + "step": 20328 + }, + { + "epoch": 2.232484076433121, + "grad_norm": 2.4700801372528076, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7254543900489807, + "num_tokens": 526057076.0, + "step": 20329 + }, + { + "epoch": 2.232593894135735, + "grad_norm": 2.1646783351898193, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7123697996139526, + "num_tokens": 526088194.0, + "step": 20330 + }, + { + "epoch": 2.232703711838348, + "grad_norm": 2.4127838611602783, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7108780145645142, + "num_tokens": 526112889.0, + "step": 20331 + }, + { + "epoch": 2.232813529540962, + "grad_norm": 2.2605323791503906, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7327408194541931, + "num_tokens": 526140702.0, + "step": 20332 + }, + { + "epoch": 2.2329233472435757, + "grad_norm": 2.0327486991882324, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.6998628377914429, + "num_tokens": 526172674.0, + "step": 20333 + }, + { + "epoch": 2.2330331649461894, + "grad_norm": 2.6311395168304443, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7508368492126465, + "num_tokens": 526194778.0, + "step": 20334 + }, + { + "epoch": 2.233142982648803, + "grad_norm": 2.392709255218506, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.727224588394165, + "num_tokens": 526221856.0, + "step": 20335 + }, + { + "epoch": 2.2332528003514165, + "grad_norm": 2.5437753200531006, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7408837080001831, + "num_tokens": 526245796.0, + "step": 20336 + }, + { + "epoch": 2.2333626180540302, + "grad_norm": 2.3586084842681885, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7432860136032104, + "num_tokens": 526269914.0, + "step": 20337 + }, + { + "epoch": 2.233472435756644, + "grad_norm": 2.485520124435425, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7643375992774963, + "num_tokens": 526292531.0, + "step": 20338 + }, + { + "epoch": 2.2335822534592578, + "grad_norm": 2.7611944675445557, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7455374598503113, + "num_tokens": 526311733.0, + "step": 20339 + }, + { + "epoch": 2.2336920711618715, + "grad_norm": 2.1713099479675293, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7571699619293213, + "num_tokens": 526338515.0, + "step": 20340 + }, + { + "epoch": 2.233801888864485, + "grad_norm": 2.1053669452667236, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7475697994232178, + "num_tokens": 526371474.0, + "step": 20341 + }, + { + "epoch": 2.2339117065670986, + "grad_norm": 2.907088041305542, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7426533699035645, + "num_tokens": 526390129.0, + "step": 20342 + }, + { + "epoch": 2.2340215242697123, + "grad_norm": 2.2816660404205322, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7362793684005737, + "num_tokens": 526417402.0, + "step": 20343 + }, + { + "epoch": 2.234131341972326, + "grad_norm": 2.420438766479492, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7597675323486328, + "num_tokens": 526440645.0, + "step": 20344 + }, + { + "epoch": 2.2342411596749394, + "grad_norm": 2.3105552196502686, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7584847211837769, + "num_tokens": 526466874.0, + "step": 20345 + }, + { + "epoch": 2.234350977377553, + "grad_norm": 2.458186149597168, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7352174520492554, + "num_tokens": 526490122.0, + "step": 20346 + }, + { + "epoch": 2.234460795080167, + "grad_norm": 2.1664936542510986, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7100741863250732, + "num_tokens": 526517745.0, + "step": 20347 + }, + { + "epoch": 2.2345706127827807, + "grad_norm": 2.5367493629455566, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7467525005340576, + "num_tokens": 526542473.0, + "step": 20348 + }, + { + "epoch": 2.2346804304853944, + "grad_norm": 2.3704822063446045, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7071176767349243, + "num_tokens": 526567896.0, + "step": 20349 + }, + { + "epoch": 2.2347902481880078, + "grad_norm": 2.7546920776367188, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7493075132369995, + "num_tokens": 526588527.0, + "step": 20350 + }, + { + "epoch": 2.2349000658906215, + "grad_norm": 2.607128143310547, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7511732578277588, + "num_tokens": 526610904.0, + "step": 20351 + }, + { + "epoch": 2.2350098835932353, + "grad_norm": 2.2436726093292236, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.736478865146637, + "num_tokens": 526638307.0, + "step": 20352 + }, + { + "epoch": 2.235119701295849, + "grad_norm": 2.6611857414245605, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7504994869232178, + "num_tokens": 526657833.0, + "step": 20353 + }, + { + "epoch": 2.2352295189984623, + "grad_norm": 2.1216518878936768, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7142513990402222, + "num_tokens": 526688820.0, + "step": 20354 + }, + { + "epoch": 2.235339336701076, + "grad_norm": 2.6158392429351807, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7329510450363159, + "num_tokens": 526712067.0, + "step": 20355 + }, + { + "epoch": 2.23544915440369, + "grad_norm": 2.567842960357666, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7423288226127625, + "num_tokens": 526734605.0, + "step": 20356 + }, + { + "epoch": 2.2355589721063036, + "grad_norm": 2.2204880714416504, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7183558344841003, + "num_tokens": 526763488.0, + "step": 20357 + }, + { + "epoch": 2.2356687898089174, + "grad_norm": 2.3762588500976562, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7324119210243225, + "num_tokens": 526788452.0, + "step": 20358 + }, + { + "epoch": 2.2357786075115307, + "grad_norm": 2.443350315093994, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7315777540206909, + "num_tokens": 526812108.0, + "step": 20359 + }, + { + "epoch": 2.2358884252141444, + "grad_norm": 2.4700233936309814, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.738825798034668, + "num_tokens": 526836705.0, + "step": 20360 + }, + { + "epoch": 2.235998242916758, + "grad_norm": 2.3778188228607178, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7529542446136475, + "num_tokens": 526860268.0, + "step": 20361 + }, + { + "epoch": 2.236108060619372, + "grad_norm": 2.216170310974121, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7299601435661316, + "num_tokens": 526888612.0, + "step": 20362 + }, + { + "epoch": 2.2362178783219857, + "grad_norm": 2.1374011039733887, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7591471672058105, + "num_tokens": 526918780.0, + "step": 20363 + }, + { + "epoch": 2.236327696024599, + "grad_norm": 2.171192169189453, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7411918640136719, + "num_tokens": 526947039.0, + "step": 20364 + }, + { + "epoch": 2.236437513727213, + "grad_norm": 2.587124824523926, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7664192914962769, + "num_tokens": 526967088.0, + "step": 20365 + }, + { + "epoch": 2.2365473314298265, + "grad_norm": 2.6568033695220947, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7319138050079346, + "num_tokens": 526988400.0, + "step": 20366 + }, + { + "epoch": 2.2366571491324403, + "grad_norm": 2.3676705360412598, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7192476987838745, + "num_tokens": 527015572.0, + "step": 20367 + }, + { + "epoch": 2.2367669668350536, + "grad_norm": 2.260812997817993, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7453374862670898, + "num_tokens": 527041790.0, + "step": 20368 + }, + { + "epoch": 2.2368767845376674, + "grad_norm": 2.443943977355957, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7352122068405151, + "num_tokens": 527066764.0, + "step": 20369 + }, + { + "epoch": 2.236986602240281, + "grad_norm": 2.450493812561035, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7341243028640747, + "num_tokens": 527092131.0, + "step": 20370 + }, + { + "epoch": 2.237096419942895, + "grad_norm": 2.494187355041504, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7186145782470703, + "num_tokens": 527115857.0, + "step": 20371 + }, + { + "epoch": 2.2372062376455086, + "grad_norm": 2.211615800857544, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7192058563232422, + "num_tokens": 527146876.0, + "step": 20372 + }, + { + "epoch": 2.237316055348122, + "grad_norm": 2.637582540512085, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7477438449859619, + "num_tokens": 527167614.0, + "step": 20373 + }, + { + "epoch": 2.2374258730507357, + "grad_norm": 2.484866142272949, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7326902151107788, + "num_tokens": 527191952.0, + "step": 20374 + }, + { + "epoch": 2.2375356907533495, + "grad_norm": 2.6253252029418945, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7583633661270142, + "num_tokens": 527211500.0, + "step": 20375 + }, + { + "epoch": 2.237645508455963, + "grad_norm": 2.430572271347046, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7404136061668396, + "num_tokens": 527236686.0, + "step": 20376 + }, + { + "epoch": 2.2377553261585765, + "grad_norm": 2.513554334640503, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7411365509033203, + "num_tokens": 527260395.0, + "step": 20377 + }, + { + "epoch": 2.2378651438611903, + "grad_norm": 2.6247687339782715, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7304785251617432, + "num_tokens": 527282151.0, + "step": 20378 + }, + { + "epoch": 2.237974961563804, + "grad_norm": 2.60610032081604, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7302224636077881, + "num_tokens": 527306096.0, + "step": 20379 + }, + { + "epoch": 2.238084779266418, + "grad_norm": 2.77776837348938, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7331662178039551, + "num_tokens": 527325235.0, + "step": 20380 + }, + { + "epoch": 2.2381945969690316, + "grad_norm": 2.5906496047973633, + "learning_rate": 1e-06, + "loss": 0.7723, + "mean_token_accuracy": 0.7537628412246704, + "num_tokens": 527347389.0, + "step": 20381 + }, + { + "epoch": 2.238304414671645, + "grad_norm": 2.4309701919555664, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7153230905532837, + "num_tokens": 527373511.0, + "step": 20382 + }, + { + "epoch": 2.2384142323742586, + "grad_norm": 2.1483590602874756, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7307368516921997, + "num_tokens": 527402166.0, + "step": 20383 + }, + { + "epoch": 2.2385240500768724, + "grad_norm": 2.7748286724090576, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7415960431098938, + "num_tokens": 527422959.0, + "step": 20384 + }, + { + "epoch": 2.238633867779486, + "grad_norm": 2.3400492668151855, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.724277913570404, + "num_tokens": 527450587.0, + "step": 20385 + }, + { + "epoch": 2.2387436854821, + "grad_norm": 2.368001937866211, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7460058927536011, + "num_tokens": 527476126.0, + "step": 20386 + }, + { + "epoch": 2.238853503184713, + "grad_norm": 2.6717774868011475, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7330896854400635, + "num_tokens": 527496509.0, + "step": 20387 + }, + { + "epoch": 2.238963320887327, + "grad_norm": 2.6092796325683594, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7431192398071289, + "num_tokens": 527518868.0, + "step": 20388 + }, + { + "epoch": 2.2390731385899407, + "grad_norm": 2.3532705307006836, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7158187031745911, + "num_tokens": 527547509.0, + "step": 20389 + }, + { + "epoch": 2.2391829562925545, + "grad_norm": 2.2348525524139404, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7268975377082825, + "num_tokens": 527579173.0, + "step": 20390 + }, + { + "epoch": 2.2392927739951682, + "grad_norm": 2.7757561206817627, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7524955868721008, + "num_tokens": 527599196.0, + "step": 20391 + }, + { + "epoch": 2.2394025916977816, + "grad_norm": 2.0907580852508545, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7362490296363831, + "num_tokens": 527629669.0, + "step": 20392 + }, + { + "epoch": 2.2395124094003953, + "grad_norm": 2.0928804874420166, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7316956520080566, + "num_tokens": 527660922.0, + "step": 20393 + }, + { + "epoch": 2.239622227103009, + "grad_norm": 2.730290412902832, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7485928535461426, + "num_tokens": 527680675.0, + "step": 20394 + }, + { + "epoch": 2.239732044805623, + "grad_norm": 2.352703332901001, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7476298809051514, + "num_tokens": 527703669.0, + "step": 20395 + }, + { + "epoch": 2.239841862508236, + "grad_norm": 2.2862861156463623, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7014567852020264, + "num_tokens": 527735082.0, + "step": 20396 + }, + { + "epoch": 2.23995168021085, + "grad_norm": 2.379901885986328, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7427464127540588, + "num_tokens": 527759437.0, + "step": 20397 + }, + { + "epoch": 2.2400614979134637, + "grad_norm": 2.202363967895508, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7029961347579956, + "num_tokens": 527789906.0, + "step": 20398 + }, + { + "epoch": 2.2401713156160774, + "grad_norm": 2.5199761390686035, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7436627149581909, + "num_tokens": 527812118.0, + "step": 20399 + }, + { + "epoch": 2.240281133318691, + "grad_norm": 2.481316089630127, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7219367027282715, + "num_tokens": 527837353.0, + "step": 20400 + }, + { + "epoch": 2.2403909510213045, + "grad_norm": 2.074148178100586, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7124879360198975, + "num_tokens": 527868683.0, + "step": 20401 + }, + { + "epoch": 2.2405007687239182, + "grad_norm": 2.370232582092285, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7418683767318726, + "num_tokens": 527892039.0, + "step": 20402 + }, + { + "epoch": 2.240610586426532, + "grad_norm": 2.3780181407928467, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7240197062492371, + "num_tokens": 527918095.0, + "step": 20403 + }, + { + "epoch": 2.2407204041291457, + "grad_norm": 2.6820852756500244, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7226977348327637, + "num_tokens": 527938217.0, + "step": 20404 + }, + { + "epoch": 2.240830221831759, + "grad_norm": 2.212287664413452, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7225512266159058, + "num_tokens": 527965025.0, + "step": 20405 + }, + { + "epoch": 2.240940039534373, + "grad_norm": 2.4453816413879395, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7125905156135559, + "num_tokens": 527989612.0, + "step": 20406 + }, + { + "epoch": 2.2410498572369866, + "grad_norm": 2.656097888946533, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7555322647094727, + "num_tokens": 528009851.0, + "step": 20407 + }, + { + "epoch": 2.2411596749396003, + "grad_norm": 1.9808481931686401, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7144076824188232, + "num_tokens": 528043087.0, + "step": 20408 + }, + { + "epoch": 2.241269492642214, + "grad_norm": 2.2288124561309814, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7096074819564819, + "num_tokens": 528072194.0, + "step": 20409 + }, + { + "epoch": 2.2413793103448274, + "grad_norm": 2.3400325775146484, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7156245708465576, + "num_tokens": 528098834.0, + "step": 20410 + }, + { + "epoch": 2.241489128047441, + "grad_norm": 2.4993412494659424, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7402788400650024, + "num_tokens": 528122187.0, + "step": 20411 + }, + { + "epoch": 2.241598945750055, + "grad_norm": 2.0777313709259033, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7083736062049866, + "num_tokens": 528154044.0, + "step": 20412 + }, + { + "epoch": 2.2417087634526687, + "grad_norm": 2.5072689056396484, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.727770209312439, + "num_tokens": 528175938.0, + "step": 20413 + }, + { + "epoch": 2.2418185811552824, + "grad_norm": 2.210986614227295, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7084686160087585, + "num_tokens": 528204258.0, + "step": 20414 + }, + { + "epoch": 2.2419283988578957, + "grad_norm": 2.0813422203063965, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.719690203666687, + "num_tokens": 528237244.0, + "step": 20415 + }, + { + "epoch": 2.2420382165605095, + "grad_norm": 2.311681032180786, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7187783122062683, + "num_tokens": 528264249.0, + "step": 20416 + }, + { + "epoch": 2.2421480342631233, + "grad_norm": 2.713559865951538, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.739071249961853, + "num_tokens": 528284839.0, + "step": 20417 + }, + { + "epoch": 2.242257851965737, + "grad_norm": 2.381674289703369, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7205348014831543, + "num_tokens": 528310191.0, + "step": 20418 + }, + { + "epoch": 2.2423676696683508, + "grad_norm": 2.3186707496643066, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7289973497390747, + "num_tokens": 528338473.0, + "step": 20419 + }, + { + "epoch": 2.242477487370964, + "grad_norm": 2.1132423877716064, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7238236665725708, + "num_tokens": 528368942.0, + "step": 20420 + }, + { + "epoch": 2.242587305073578, + "grad_norm": 2.542024850845337, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7215425968170166, + "num_tokens": 528394289.0, + "step": 20421 + }, + { + "epoch": 2.2426971227761916, + "grad_norm": 2.196967363357544, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6965226531028748, + "num_tokens": 528424723.0, + "step": 20422 + }, + { + "epoch": 2.2428069404788054, + "grad_norm": 2.332899332046509, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7060889005661011, + "num_tokens": 528451787.0, + "step": 20423 + }, + { + "epoch": 2.2429167581814187, + "grad_norm": 2.1525676250457764, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7137700915336609, + "num_tokens": 528484557.0, + "step": 20424 + }, + { + "epoch": 2.2430265758840324, + "grad_norm": 2.3701837062835693, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7175756692886353, + "num_tokens": 528511132.0, + "step": 20425 + }, + { + "epoch": 2.243136393586646, + "grad_norm": 2.239793300628662, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7303204536437988, + "num_tokens": 528539163.0, + "step": 20426 + }, + { + "epoch": 2.24324621128926, + "grad_norm": 2.243743658065796, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7394156455993652, + "num_tokens": 528567797.0, + "step": 20427 + }, + { + "epoch": 2.2433560289918737, + "grad_norm": 2.2722342014312744, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7148781418800354, + "num_tokens": 528595182.0, + "step": 20428 + }, + { + "epoch": 2.243465846694487, + "grad_norm": 2.340867042541504, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.740986168384552, + "num_tokens": 528620815.0, + "step": 20429 + }, + { + "epoch": 2.2435756643971008, + "grad_norm": 2.0786349773406982, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7180489301681519, + "num_tokens": 528653451.0, + "step": 20430 + }, + { + "epoch": 2.2436854820997145, + "grad_norm": 2.125807285308838, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7209950685501099, + "num_tokens": 528682954.0, + "step": 20431 + }, + { + "epoch": 2.2437952998023283, + "grad_norm": 2.305180549621582, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.712339460849762, + "num_tokens": 528713588.0, + "step": 20432 + }, + { + "epoch": 2.2439051175049416, + "grad_norm": 2.3209385871887207, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7450178861618042, + "num_tokens": 528738256.0, + "step": 20433 + }, + { + "epoch": 2.2440149352075554, + "grad_norm": 2.320971965789795, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.733738124370575, + "num_tokens": 528764360.0, + "step": 20434 + }, + { + "epoch": 2.244124752910169, + "grad_norm": 2.3283565044403076, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7272159457206726, + "num_tokens": 528791968.0, + "step": 20435 + }, + { + "epoch": 2.244234570612783, + "grad_norm": 2.2602012157440186, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7166018486022949, + "num_tokens": 528818306.0, + "step": 20436 + }, + { + "epoch": 2.2443443883153966, + "grad_norm": 2.414414644241333, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7439787983894348, + "num_tokens": 528844044.0, + "step": 20437 + }, + { + "epoch": 2.24445420601801, + "grad_norm": 2.3646957874298096, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.750270664691925, + "num_tokens": 528869218.0, + "step": 20438 + }, + { + "epoch": 2.2445640237206237, + "grad_norm": 2.2801015377044678, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.74153733253479, + "num_tokens": 528895694.0, + "step": 20439 + }, + { + "epoch": 2.2446738414232374, + "grad_norm": 2.425261974334717, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7141426205635071, + "num_tokens": 528924203.0, + "step": 20440 + }, + { + "epoch": 2.244783659125851, + "grad_norm": 2.3863370418548584, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7317440509796143, + "num_tokens": 528948436.0, + "step": 20441 + }, + { + "epoch": 2.244893476828465, + "grad_norm": 2.624709367752075, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7461259961128235, + "num_tokens": 528969233.0, + "step": 20442 + }, + { + "epoch": 2.2450032945310783, + "grad_norm": 2.317171812057495, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7591733932495117, + "num_tokens": 528993458.0, + "step": 20443 + }, + { + "epoch": 2.245113112233692, + "grad_norm": 2.4386913776397705, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7350785732269287, + "num_tokens": 529019352.0, + "step": 20444 + }, + { + "epoch": 2.245222929936306, + "grad_norm": 2.504779815673828, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7240978479385376, + "num_tokens": 529043241.0, + "step": 20445 + }, + { + "epoch": 2.2453327476389195, + "grad_norm": 2.3070313930511475, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7209304571151733, + "num_tokens": 529071459.0, + "step": 20446 + }, + { + "epoch": 2.245442565341533, + "grad_norm": 2.296797275543213, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7292108535766602, + "num_tokens": 529097966.0, + "step": 20447 + }, + { + "epoch": 2.2455523830441466, + "grad_norm": 2.407424211502075, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7187904119491577, + "num_tokens": 529123413.0, + "step": 20448 + }, + { + "epoch": 2.2456622007467604, + "grad_norm": 2.3611791133880615, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7098738551139832, + "num_tokens": 529151037.0, + "step": 20449 + }, + { + "epoch": 2.245772018449374, + "grad_norm": 2.5320587158203125, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7358881235122681, + "num_tokens": 529173773.0, + "step": 20450 + }, + { + "epoch": 2.245881836151988, + "grad_norm": 2.585050582885742, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.750926673412323, + "num_tokens": 529195203.0, + "step": 20451 + }, + { + "epoch": 2.245991653854601, + "grad_norm": 2.415254592895508, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7476609349250793, + "num_tokens": 529218317.0, + "step": 20452 + }, + { + "epoch": 2.246101471557215, + "grad_norm": 2.388993263244629, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.744635820388794, + "num_tokens": 529242615.0, + "step": 20453 + }, + { + "epoch": 2.2462112892598287, + "grad_norm": 2.3411293029785156, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7295305728912354, + "num_tokens": 529268984.0, + "step": 20454 + }, + { + "epoch": 2.2463211069624425, + "grad_norm": 2.3383800983428955, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7373078465461731, + "num_tokens": 529293638.0, + "step": 20455 + }, + { + "epoch": 2.246430924665056, + "grad_norm": 2.4924476146698, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7356137037277222, + "num_tokens": 529317290.0, + "step": 20456 + }, + { + "epoch": 2.2465407423676695, + "grad_norm": 2.724565029144287, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7463442087173462, + "num_tokens": 529336431.0, + "step": 20457 + }, + { + "epoch": 2.2466505600702833, + "grad_norm": 2.209069013595581, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7043311595916748, + "num_tokens": 529366266.0, + "step": 20458 + }, + { + "epoch": 2.246760377772897, + "grad_norm": 2.8383688926696777, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7527655363082886, + "num_tokens": 529383851.0, + "step": 20459 + }, + { + "epoch": 2.246870195475511, + "grad_norm": 2.3870184421539307, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7357120513916016, + "num_tokens": 529409177.0, + "step": 20460 + }, + { + "epoch": 2.246980013178124, + "grad_norm": 2.092939853668213, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7271687984466553, + "num_tokens": 529439888.0, + "step": 20461 + }, + { + "epoch": 2.247089830880738, + "grad_norm": 2.5329537391662598, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7494011521339417, + "num_tokens": 529462820.0, + "step": 20462 + }, + { + "epoch": 2.2471996485833516, + "grad_norm": 2.476649045944214, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7352493405342102, + "num_tokens": 529486913.0, + "step": 20463 + }, + { + "epoch": 2.2473094662859654, + "grad_norm": 2.2347090244293213, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7373676300048828, + "num_tokens": 529513992.0, + "step": 20464 + }, + { + "epoch": 2.247419283988579, + "grad_norm": 2.35870361328125, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7387195825576782, + "num_tokens": 529540114.0, + "step": 20465 + }, + { + "epoch": 2.2475291016911925, + "grad_norm": 2.250772714614868, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7347362637519836, + "num_tokens": 529566298.0, + "step": 20466 + }, + { + "epoch": 2.2476389193938062, + "grad_norm": 2.791297197341919, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7288001775741577, + "num_tokens": 529591700.0, + "step": 20467 + }, + { + "epoch": 2.24774873709642, + "grad_norm": 2.736764430999756, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7675503492355347, + "num_tokens": 529612780.0, + "step": 20468 + }, + { + "epoch": 2.2478585547990337, + "grad_norm": 2.1878468990325928, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7161881923675537, + "num_tokens": 529641782.0, + "step": 20469 + }, + { + "epoch": 2.2479683725016475, + "grad_norm": 2.2574374675750732, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7287211418151855, + "num_tokens": 529668437.0, + "step": 20470 + }, + { + "epoch": 2.248078190204261, + "grad_norm": 2.517133951187134, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7146667242050171, + "num_tokens": 529692229.0, + "step": 20471 + }, + { + "epoch": 2.2481880079068746, + "grad_norm": 2.083923816680908, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7373771071434021, + "num_tokens": 529722955.0, + "step": 20472 + }, + { + "epoch": 2.2482978256094883, + "grad_norm": 2.292325973510742, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7399845123291016, + "num_tokens": 529750229.0, + "step": 20473 + }, + { + "epoch": 2.248407643312102, + "grad_norm": 2.288125514984131, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7223308086395264, + "num_tokens": 529777962.0, + "step": 20474 + }, + { + "epoch": 2.2485174610147154, + "grad_norm": 2.233790397644043, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7310726642608643, + "num_tokens": 529805968.0, + "step": 20475 + }, + { + "epoch": 2.248627278717329, + "grad_norm": 2.493295431137085, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7536962628364563, + "num_tokens": 529828306.0, + "step": 20476 + }, + { + "epoch": 2.248737096419943, + "grad_norm": 2.345560312271118, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7237436771392822, + "num_tokens": 529855477.0, + "step": 20477 + }, + { + "epoch": 2.2488469141225567, + "grad_norm": 2.4896397590637207, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7486428618431091, + "num_tokens": 529879914.0, + "step": 20478 + }, + { + "epoch": 2.2489567318251704, + "grad_norm": 2.492649793624878, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7399346828460693, + "num_tokens": 529902911.0, + "step": 20479 + }, + { + "epoch": 2.2490665495277837, + "grad_norm": 2.0620198249816895, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7088236212730408, + "num_tokens": 529936411.0, + "step": 20480 + }, + { + "epoch": 2.2491763672303975, + "grad_norm": 2.588005542755127, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7371982932090759, + "num_tokens": 529957626.0, + "step": 20481 + }, + { + "epoch": 2.2492861849330112, + "grad_norm": 2.583632230758667, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7287224531173706, + "num_tokens": 529981233.0, + "step": 20482 + }, + { + "epoch": 2.249396002635625, + "grad_norm": 2.370948314666748, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7358402013778687, + "num_tokens": 530008794.0, + "step": 20483 + }, + { + "epoch": 2.2495058203382383, + "grad_norm": 2.2423460483551025, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7242701649665833, + "num_tokens": 530036435.0, + "step": 20484 + }, + { + "epoch": 2.249615638040852, + "grad_norm": 2.3009040355682373, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.736456036567688, + "num_tokens": 530060470.0, + "step": 20485 + }, + { + "epoch": 2.249725455743466, + "grad_norm": 2.2202601432800293, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7191898822784424, + "num_tokens": 530090529.0, + "step": 20486 + }, + { + "epoch": 2.2498352734460796, + "grad_norm": 2.321126699447632, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7375564575195312, + "num_tokens": 530117324.0, + "step": 20487 + }, + { + "epoch": 2.2499450911486933, + "grad_norm": 2.4449687004089355, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7367690205574036, + "num_tokens": 530139881.0, + "step": 20488 + }, + { + "epoch": 2.2500549088513067, + "grad_norm": 2.7396910190582275, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7597161531448364, + "num_tokens": 530159955.0, + "step": 20489 + }, + { + "epoch": 2.2501647265539204, + "grad_norm": 2.4367008209228516, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.727219820022583, + "num_tokens": 530183815.0, + "step": 20490 + }, + { + "epoch": 2.250274544256534, + "grad_norm": 2.375673770904541, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7342227697372437, + "num_tokens": 530209864.0, + "step": 20491 + }, + { + "epoch": 2.250384361959148, + "grad_norm": 2.4038329124450684, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7385386228561401, + "num_tokens": 530236326.0, + "step": 20492 + }, + { + "epoch": 2.2504941796617617, + "grad_norm": 2.6347434520721436, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7361313104629517, + "num_tokens": 530256498.0, + "step": 20493 + }, + { + "epoch": 2.250603997364375, + "grad_norm": 2.4046034812927246, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7313045263290405, + "num_tokens": 530280928.0, + "step": 20494 + }, + { + "epoch": 2.2507138150669888, + "grad_norm": 2.6172242164611816, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7334656119346619, + "num_tokens": 530301881.0, + "step": 20495 + }, + { + "epoch": 2.2508236327696025, + "grad_norm": 2.214240074157715, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.706058144569397, + "num_tokens": 530333726.0, + "step": 20496 + }, + { + "epoch": 2.2509334504722163, + "grad_norm": 2.3611631393432617, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7455126047134399, + "num_tokens": 530360414.0, + "step": 20497 + }, + { + "epoch": 2.25104326817483, + "grad_norm": 2.2815914154052734, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7127627730369568, + "num_tokens": 530387789.0, + "step": 20498 + }, + { + "epoch": 2.2511530858774433, + "grad_norm": 2.5794425010681152, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.726046085357666, + "num_tokens": 530410661.0, + "step": 20499 + }, + { + "epoch": 2.251262903580057, + "grad_norm": 2.2459990978240967, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7240772247314453, + "num_tokens": 530438675.0, + "step": 20500 + }, + { + "epoch": 2.251372721282671, + "grad_norm": 2.2479586601257324, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7287328839302063, + "num_tokens": 530466771.0, + "step": 20501 + }, + { + "epoch": 2.2514825389852846, + "grad_norm": 2.404188871383667, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7664662003517151, + "num_tokens": 530489353.0, + "step": 20502 + }, + { + "epoch": 2.251592356687898, + "grad_norm": 2.741887092590332, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7269980311393738, + "num_tokens": 530514382.0, + "step": 20503 + }, + { + "epoch": 2.2517021743905117, + "grad_norm": 2.3758933544158936, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7207158803939819, + "num_tokens": 530539874.0, + "step": 20504 + }, + { + "epoch": 2.2518119920931254, + "grad_norm": 2.433483123779297, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7392814755439758, + "num_tokens": 530564468.0, + "step": 20505 + }, + { + "epoch": 2.251921809795739, + "grad_norm": 2.152846097946167, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7437917590141296, + "num_tokens": 530592630.0, + "step": 20506 + }, + { + "epoch": 2.2520316274983525, + "grad_norm": 2.2922468185424805, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7200778126716614, + "num_tokens": 530621981.0, + "step": 20507 + }, + { + "epoch": 2.2521414452009663, + "grad_norm": 2.5266189575195312, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7395523190498352, + "num_tokens": 530646013.0, + "step": 20508 + }, + { + "epoch": 2.25225126290358, + "grad_norm": 2.469918727874756, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7432963848114014, + "num_tokens": 530669801.0, + "step": 20509 + }, + { + "epoch": 2.2523610806061938, + "grad_norm": 2.2822015285491943, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7659115195274353, + "num_tokens": 530695794.0, + "step": 20510 + }, + { + "epoch": 2.2524708983088075, + "grad_norm": 2.2317545413970947, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7277678847312927, + "num_tokens": 530725142.0, + "step": 20511 + }, + { + "epoch": 2.252580716011421, + "grad_norm": 2.240313768386841, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7395548224449158, + "num_tokens": 530754745.0, + "step": 20512 + }, + { + "epoch": 2.2526905337140346, + "grad_norm": 2.5001204013824463, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7302153706550598, + "num_tokens": 530778615.0, + "step": 20513 + }, + { + "epoch": 2.2528003514166484, + "grad_norm": 2.301048994064331, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7420281767845154, + "num_tokens": 530806303.0, + "step": 20514 + }, + { + "epoch": 2.252910169119262, + "grad_norm": 2.1157279014587402, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7330507040023804, + "num_tokens": 530836596.0, + "step": 20515 + }, + { + "epoch": 2.253019986821876, + "grad_norm": 2.3454551696777344, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7264378070831299, + "num_tokens": 530862211.0, + "step": 20516 + }, + { + "epoch": 2.253129804524489, + "grad_norm": 2.1728224754333496, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7091113328933716, + "num_tokens": 530892342.0, + "step": 20517 + }, + { + "epoch": 2.253239622227103, + "grad_norm": 2.143937587738037, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7017167806625366, + "num_tokens": 530924070.0, + "step": 20518 + }, + { + "epoch": 2.2533494399297167, + "grad_norm": 2.244649648666382, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7055835723876953, + "num_tokens": 530953116.0, + "step": 20519 + }, + { + "epoch": 2.2534592576323305, + "grad_norm": 2.2141635417938232, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7319177985191345, + "num_tokens": 530981887.0, + "step": 20520 + }, + { + "epoch": 2.253569075334944, + "grad_norm": 2.2174012660980225, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7339732646942139, + "num_tokens": 531009953.0, + "step": 20521 + }, + { + "epoch": 2.2536788930375575, + "grad_norm": 2.065117359161377, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7291988134384155, + "num_tokens": 531042200.0, + "step": 20522 + }, + { + "epoch": 2.2537887107401713, + "grad_norm": 2.224757194519043, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7507902979850769, + "num_tokens": 531071165.0, + "step": 20523 + }, + { + "epoch": 2.253898528442785, + "grad_norm": 2.459350347518921, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7468176484107971, + "num_tokens": 531095433.0, + "step": 20524 + }, + { + "epoch": 2.254008346145399, + "grad_norm": 2.3474948406219482, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7288814783096313, + "num_tokens": 531120481.0, + "step": 20525 + }, + { + "epoch": 2.254118163848012, + "grad_norm": 2.677156925201416, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7436052560806274, + "num_tokens": 531143677.0, + "step": 20526 + }, + { + "epoch": 2.254227981550626, + "grad_norm": 2.204273223876953, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7161155939102173, + "num_tokens": 531173405.0, + "step": 20527 + }, + { + "epoch": 2.2543377992532396, + "grad_norm": 2.3885209560394287, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7094006538391113, + "num_tokens": 531200292.0, + "step": 20528 + }, + { + "epoch": 2.2544476169558534, + "grad_norm": 2.522671699523926, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.733852207660675, + "num_tokens": 531223427.0, + "step": 20529 + }, + { + "epoch": 2.2545574346584667, + "grad_norm": 2.403425931930542, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7160168886184692, + "num_tokens": 531248597.0, + "step": 20530 + }, + { + "epoch": 2.2546672523610805, + "grad_norm": 2.5537703037261963, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7458466291427612, + "num_tokens": 531269512.0, + "step": 20531 + }, + { + "epoch": 2.254777070063694, + "grad_norm": 2.250131130218506, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7228038907051086, + "num_tokens": 531298178.0, + "step": 20532 + }, + { + "epoch": 2.254886887766308, + "grad_norm": 2.259000778198242, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7388906478881836, + "num_tokens": 531325600.0, + "step": 20533 + }, + { + "epoch": 2.2549967054689217, + "grad_norm": 2.370197057723999, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7290820479393005, + "num_tokens": 531351729.0, + "step": 20534 + }, + { + "epoch": 2.255106523171535, + "grad_norm": 2.492356538772583, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7349029183387756, + "num_tokens": 531376139.0, + "step": 20535 + }, + { + "epoch": 2.255216340874149, + "grad_norm": 2.503525972366333, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7213720083236694, + "num_tokens": 531402112.0, + "step": 20536 + }, + { + "epoch": 2.2553261585767626, + "grad_norm": 2.154252290725708, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7216008305549622, + "num_tokens": 531431035.0, + "step": 20537 + }, + { + "epoch": 2.2554359762793763, + "grad_norm": 2.209075689315796, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7232314348220825, + "num_tokens": 531459968.0, + "step": 20538 + }, + { + "epoch": 2.25554579398199, + "grad_norm": 2.5280609130859375, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.719793438911438, + "num_tokens": 531482442.0, + "step": 20539 + }, + { + "epoch": 2.2556556116846034, + "grad_norm": 2.363927125930786, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7456163763999939, + "num_tokens": 531506980.0, + "step": 20540 + }, + { + "epoch": 2.255765429387217, + "grad_norm": 2.1896393299102783, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7350866794586182, + "num_tokens": 531536863.0, + "step": 20541 + }, + { + "epoch": 2.255875247089831, + "grad_norm": 2.4272406101226807, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7415052652359009, + "num_tokens": 531561662.0, + "step": 20542 + }, + { + "epoch": 2.2559850647924446, + "grad_norm": 2.753997564315796, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7497447729110718, + "num_tokens": 531583732.0, + "step": 20543 + }, + { + "epoch": 2.2560948824950584, + "grad_norm": 2.1860289573669434, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7560092210769653, + "num_tokens": 531611244.0, + "step": 20544 + }, + { + "epoch": 2.2562047001976717, + "grad_norm": 2.954469919204712, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7638007402420044, + "num_tokens": 531629118.0, + "step": 20545 + }, + { + "epoch": 2.2563145179002855, + "grad_norm": 2.1189730167388916, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6950313448905945, + "num_tokens": 531660232.0, + "step": 20546 + }, + { + "epoch": 2.2564243356028992, + "grad_norm": 2.1591122150421143, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7400822639465332, + "num_tokens": 531691246.0, + "step": 20547 + }, + { + "epoch": 2.256534153305513, + "grad_norm": 1.9282554388046265, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7109324336051941, + "num_tokens": 531727836.0, + "step": 20548 + }, + { + "epoch": 2.2566439710081267, + "grad_norm": 2.7217602729797363, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7443307638168335, + "num_tokens": 531748360.0, + "step": 20549 + }, + { + "epoch": 2.25675378871074, + "grad_norm": 2.1000454425811768, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.6971125602722168, + "num_tokens": 531782369.0, + "step": 20550 + }, + { + "epoch": 2.256863606413354, + "grad_norm": 2.6974098682403564, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7180609703063965, + "num_tokens": 531803350.0, + "step": 20551 + }, + { + "epoch": 2.2569734241159676, + "grad_norm": 2.3570284843444824, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.726656973361969, + "num_tokens": 531830245.0, + "step": 20552 + }, + { + "epoch": 2.2570832418185813, + "grad_norm": 2.2443642616271973, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.737963080406189, + "num_tokens": 531857075.0, + "step": 20553 + }, + { + "epoch": 2.2571930595211946, + "grad_norm": 2.1024532318115234, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.721522331237793, + "num_tokens": 531887055.0, + "step": 20554 + }, + { + "epoch": 2.2573028772238084, + "grad_norm": 2.419494867324829, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.72275710105896, + "num_tokens": 531914390.0, + "step": 20555 + }, + { + "epoch": 2.257412694926422, + "grad_norm": 2.1676461696624756, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7302129864692688, + "num_tokens": 531944376.0, + "step": 20556 + }, + { + "epoch": 2.257522512629036, + "grad_norm": 2.3862264156341553, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7328758239746094, + "num_tokens": 531970531.0, + "step": 20557 + }, + { + "epoch": 2.2576323303316492, + "grad_norm": 2.3654396533966064, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7391056418418884, + "num_tokens": 531997450.0, + "step": 20558 + }, + { + "epoch": 2.257742148034263, + "grad_norm": 2.260617733001709, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7391762137413025, + "num_tokens": 532023100.0, + "step": 20559 + }, + { + "epoch": 2.2578519657368767, + "grad_norm": 2.255028486251831, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.730677604675293, + "num_tokens": 532049239.0, + "step": 20560 + }, + { + "epoch": 2.2579617834394905, + "grad_norm": 2.0394582748413086, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7263041734695435, + "num_tokens": 532083272.0, + "step": 20561 + }, + { + "epoch": 2.2580716011421043, + "grad_norm": 2.1657516956329346, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7264240980148315, + "num_tokens": 532112550.0, + "step": 20562 + }, + { + "epoch": 2.2581814188447176, + "grad_norm": 2.135448932647705, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7261394262313843, + "num_tokens": 532141991.0, + "step": 20563 + }, + { + "epoch": 2.2582912365473313, + "grad_norm": 2.1651501655578613, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7395963072776794, + "num_tokens": 532169669.0, + "step": 20564 + }, + { + "epoch": 2.258401054249945, + "grad_norm": 2.5520641803741455, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7831846475601196, + "num_tokens": 532189423.0, + "step": 20565 + }, + { + "epoch": 2.258510871952559, + "grad_norm": 2.54119873046875, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7319931983947754, + "num_tokens": 532213736.0, + "step": 20566 + }, + { + "epoch": 2.2586206896551726, + "grad_norm": 2.163421154022217, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7070995569229126, + "num_tokens": 532242845.0, + "step": 20567 + }, + { + "epoch": 2.258730507357786, + "grad_norm": 2.253319025039673, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7451995611190796, + "num_tokens": 532269830.0, + "step": 20568 + }, + { + "epoch": 2.2588403250603997, + "grad_norm": 2.414259195327759, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7457170486450195, + "num_tokens": 532294985.0, + "step": 20569 + }, + { + "epoch": 2.2589501427630134, + "grad_norm": 2.2388577461242676, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7477728128433228, + "num_tokens": 532322489.0, + "step": 20570 + }, + { + "epoch": 2.259059960465627, + "grad_norm": 2.4550533294677734, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.730634331703186, + "num_tokens": 532346418.0, + "step": 20571 + }, + { + "epoch": 2.259169778168241, + "grad_norm": 2.524864912033081, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7489690780639648, + "num_tokens": 532368992.0, + "step": 20572 + }, + { + "epoch": 2.2592795958708543, + "grad_norm": 2.4505810737609863, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7191673517227173, + "num_tokens": 532394273.0, + "step": 20573 + }, + { + "epoch": 2.259389413573468, + "grad_norm": 2.731292724609375, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7281724810600281, + "num_tokens": 532416996.0, + "step": 20574 + }, + { + "epoch": 2.2594992312760818, + "grad_norm": 2.506624221801758, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7323975563049316, + "num_tokens": 532442497.0, + "step": 20575 + }, + { + "epoch": 2.2596090489786955, + "grad_norm": 2.405024766921997, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7227948904037476, + "num_tokens": 532469051.0, + "step": 20576 + }, + { + "epoch": 2.259718866681309, + "grad_norm": 2.4240000247955322, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.731838583946228, + "num_tokens": 532495620.0, + "step": 20577 + }, + { + "epoch": 2.2598286843839226, + "grad_norm": 2.521427631378174, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7412125468254089, + "num_tokens": 532518777.0, + "step": 20578 + }, + { + "epoch": 2.2599385020865363, + "grad_norm": 2.3844449520111084, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7471460103988647, + "num_tokens": 532543082.0, + "step": 20579 + }, + { + "epoch": 2.26004831978915, + "grad_norm": 2.171966791152954, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7242813110351562, + "num_tokens": 532571941.0, + "step": 20580 + }, + { + "epoch": 2.260158137491764, + "grad_norm": 2.3445279598236084, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7386102676391602, + "num_tokens": 532596864.0, + "step": 20581 + }, + { + "epoch": 2.260267955194377, + "grad_norm": 2.2794744968414307, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7159661054611206, + "num_tokens": 532624412.0, + "step": 20582 + }, + { + "epoch": 2.260377772896991, + "grad_norm": 2.6319992542266846, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7542397975921631, + "num_tokens": 532645362.0, + "step": 20583 + }, + { + "epoch": 2.2604875905996047, + "grad_norm": 2.844933271408081, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7355693578720093, + "num_tokens": 532665259.0, + "step": 20584 + }, + { + "epoch": 2.2605974083022184, + "grad_norm": 2.458855628967285, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7174347639083862, + "num_tokens": 532688492.0, + "step": 20585 + }, + { + "epoch": 2.2607072260048318, + "grad_norm": 2.5137031078338623, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.741806149482727, + "num_tokens": 532712808.0, + "step": 20586 + }, + { + "epoch": 2.2608170437074455, + "grad_norm": 2.1628997325897217, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7417740225791931, + "num_tokens": 532741960.0, + "step": 20587 + }, + { + "epoch": 2.2609268614100593, + "grad_norm": 2.047372579574585, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7275511026382446, + "num_tokens": 532775010.0, + "step": 20588 + }, + { + "epoch": 2.261036679112673, + "grad_norm": 2.354440212249756, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.731762707233429, + "num_tokens": 532799610.0, + "step": 20589 + }, + { + "epoch": 2.261146496815287, + "grad_norm": 2.5531065464019775, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7405549883842468, + "num_tokens": 532822318.0, + "step": 20590 + }, + { + "epoch": 2.2612563145179, + "grad_norm": 2.358659505844116, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7445966601371765, + "num_tokens": 532848644.0, + "step": 20591 + }, + { + "epoch": 2.261366132220514, + "grad_norm": 2.2893688678741455, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7205365896224976, + "num_tokens": 532874857.0, + "step": 20592 + }, + { + "epoch": 2.2614759499231276, + "grad_norm": 2.3396458625793457, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7353376150131226, + "num_tokens": 532901140.0, + "step": 20593 + }, + { + "epoch": 2.2615857676257414, + "grad_norm": 2.099322557449341, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6993703842163086, + "num_tokens": 532933633.0, + "step": 20594 + }, + { + "epoch": 2.261695585328355, + "grad_norm": 2.3801839351654053, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7204862833023071, + "num_tokens": 532960356.0, + "step": 20595 + }, + { + "epoch": 2.2618054030309684, + "grad_norm": 2.642918586730957, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.723991870880127, + "num_tokens": 532982102.0, + "step": 20596 + }, + { + "epoch": 2.261915220733582, + "grad_norm": 2.3814704418182373, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7379876375198364, + "num_tokens": 533008657.0, + "step": 20597 + }, + { + "epoch": 2.262025038436196, + "grad_norm": 2.005451202392578, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.715142011642456, + "num_tokens": 533040050.0, + "step": 20598 + }, + { + "epoch": 2.2621348561388097, + "grad_norm": 2.3783652782440186, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.71881502866745, + "num_tokens": 533065654.0, + "step": 20599 + }, + { + "epoch": 2.2622446738414235, + "grad_norm": 2.444890022277832, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7489843368530273, + "num_tokens": 533089382.0, + "step": 20600 + }, + { + "epoch": 2.262354491544037, + "grad_norm": 2.187849283218384, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.6986271739006042, + "num_tokens": 533119669.0, + "step": 20601 + }, + { + "epoch": 2.2624643092466505, + "grad_norm": 2.7338316440582275, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.774753987789154, + "num_tokens": 533138406.0, + "step": 20602 + }, + { + "epoch": 2.2625741269492643, + "grad_norm": 2.235844373703003, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7388913631439209, + "num_tokens": 533165108.0, + "step": 20603 + }, + { + "epoch": 2.262683944651878, + "grad_norm": 2.172085762023926, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7313092350959778, + "num_tokens": 533196382.0, + "step": 20604 + }, + { + "epoch": 2.2627937623544914, + "grad_norm": 2.3698232173919678, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.729607105255127, + "num_tokens": 533221981.0, + "step": 20605 + }, + { + "epoch": 2.262903580057105, + "grad_norm": 2.4232542514801025, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.717445969581604, + "num_tokens": 533248532.0, + "step": 20606 + }, + { + "epoch": 2.263013397759719, + "grad_norm": 2.165724277496338, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7212219834327698, + "num_tokens": 533276848.0, + "step": 20607 + }, + { + "epoch": 2.2631232154623326, + "grad_norm": 2.1227874755859375, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.704728901386261, + "num_tokens": 533309991.0, + "step": 20608 + }, + { + "epoch": 2.263233033164946, + "grad_norm": 2.3626389503479004, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7427456378936768, + "num_tokens": 533337717.0, + "step": 20609 + }, + { + "epoch": 2.2633428508675597, + "grad_norm": 2.4334359169006348, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7253894209861755, + "num_tokens": 533361471.0, + "step": 20610 + }, + { + "epoch": 2.2634526685701735, + "grad_norm": 2.2717854976654053, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7346705198287964, + "num_tokens": 533390149.0, + "step": 20611 + }, + { + "epoch": 2.263562486272787, + "grad_norm": 2.455716371536255, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7622716426849365, + "num_tokens": 533412367.0, + "step": 20612 + }, + { + "epoch": 2.263672303975401, + "grad_norm": 2.3837573528289795, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7333012819290161, + "num_tokens": 533439118.0, + "step": 20613 + }, + { + "epoch": 2.2637821216780143, + "grad_norm": 2.4677817821502686, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7174943685531616, + "num_tokens": 533465411.0, + "step": 20614 + }, + { + "epoch": 2.263891939380628, + "grad_norm": 2.4594500064849854, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7271661758422852, + "num_tokens": 533491447.0, + "step": 20615 + }, + { + "epoch": 2.264001757083242, + "grad_norm": 2.6117382049560547, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7427964210510254, + "num_tokens": 533513943.0, + "step": 20616 + }, + { + "epoch": 2.2641115747858556, + "grad_norm": 2.418538808822632, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.712095320224762, + "num_tokens": 533541238.0, + "step": 20617 + }, + { + "epoch": 2.2642213924884693, + "grad_norm": 2.4222357273101807, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7428323030471802, + "num_tokens": 533565011.0, + "step": 20618 + }, + { + "epoch": 2.2643312101910826, + "grad_norm": 2.042229652404785, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7396670579910278, + "num_tokens": 533598006.0, + "step": 20619 + }, + { + "epoch": 2.2644410278936964, + "grad_norm": 2.4540677070617676, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7091983556747437, + "num_tokens": 533623268.0, + "step": 20620 + }, + { + "epoch": 2.26455084559631, + "grad_norm": 2.2036373615264893, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7311264872550964, + "num_tokens": 533652105.0, + "step": 20621 + }, + { + "epoch": 2.264660663298924, + "grad_norm": 2.8400461673736572, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7387689352035522, + "num_tokens": 533670980.0, + "step": 20622 + }, + { + "epoch": 2.2647704810015377, + "grad_norm": 2.462557077407837, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7268094420433044, + "num_tokens": 533697461.0, + "step": 20623 + }, + { + "epoch": 2.264880298704151, + "grad_norm": 2.288888931274414, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7289975881576538, + "num_tokens": 533723770.0, + "step": 20624 + }, + { + "epoch": 2.2649901164067647, + "grad_norm": 2.117687702178955, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.741106390953064, + "num_tokens": 533752403.0, + "step": 20625 + }, + { + "epoch": 2.2650999341093785, + "grad_norm": 2.387402296066284, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7358884811401367, + "num_tokens": 533777072.0, + "step": 20626 + }, + { + "epoch": 2.2652097518119922, + "grad_norm": 2.6010076999664307, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7451827526092529, + "num_tokens": 533800909.0, + "step": 20627 + }, + { + "epoch": 2.265319569514606, + "grad_norm": 2.4192516803741455, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7580690979957581, + "num_tokens": 533825942.0, + "step": 20628 + }, + { + "epoch": 2.2654293872172193, + "grad_norm": 2.295032262802124, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7464863061904907, + "num_tokens": 533851246.0, + "step": 20629 + }, + { + "epoch": 2.265539204919833, + "grad_norm": 2.729153871536255, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.748680830001831, + "num_tokens": 533871288.0, + "step": 20630 + }, + { + "epoch": 2.265649022622447, + "grad_norm": 2.3409922122955322, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7167520523071289, + "num_tokens": 533897718.0, + "step": 20631 + }, + { + "epoch": 2.2657588403250606, + "grad_norm": 2.4680559635162354, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7440218925476074, + "num_tokens": 533919924.0, + "step": 20632 + }, + { + "epoch": 2.265868658027674, + "grad_norm": 2.344811201095581, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7245806455612183, + "num_tokens": 533945249.0, + "step": 20633 + }, + { + "epoch": 2.2659784757302877, + "grad_norm": 2.281780242919922, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7101224064826965, + "num_tokens": 533973516.0, + "step": 20634 + }, + { + "epoch": 2.2660882934329014, + "grad_norm": 2.4756245613098145, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7498360276222229, + "num_tokens": 533997573.0, + "step": 20635 + }, + { + "epoch": 2.266198111135515, + "grad_norm": 2.8202764987945557, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7595536708831787, + "num_tokens": 534016343.0, + "step": 20636 + }, + { + "epoch": 2.2663079288381285, + "grad_norm": 2.143078565597534, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7247123718261719, + "num_tokens": 534046554.0, + "step": 20637 + }, + { + "epoch": 2.2664177465407422, + "grad_norm": 2.5358080863952637, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7398627996444702, + "num_tokens": 534070506.0, + "step": 20638 + }, + { + "epoch": 2.266527564243356, + "grad_norm": 2.2421672344207764, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7163307666778564, + "num_tokens": 534098458.0, + "step": 20639 + }, + { + "epoch": 2.2666373819459698, + "grad_norm": 2.714301824569702, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7470472455024719, + "num_tokens": 534119676.0, + "step": 20640 + }, + { + "epoch": 2.2667471996485835, + "grad_norm": 2.2915420532226562, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7577473521232605, + "num_tokens": 534147125.0, + "step": 20641 + }, + { + "epoch": 2.266857017351197, + "grad_norm": 2.2471113204956055, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.6953814029693604, + "num_tokens": 534175831.0, + "step": 20642 + }, + { + "epoch": 2.2669668350538106, + "grad_norm": 2.2830910682678223, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7475835084915161, + "num_tokens": 534202981.0, + "step": 20643 + }, + { + "epoch": 2.2670766527564243, + "grad_norm": 2.3313512802124023, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7150905728340149, + "num_tokens": 534232202.0, + "step": 20644 + }, + { + "epoch": 2.267186470459038, + "grad_norm": 2.518319845199585, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7546876668930054, + "num_tokens": 534256887.0, + "step": 20645 + }, + { + "epoch": 2.267296288161652, + "grad_norm": 2.28053879737854, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7455703020095825, + "num_tokens": 534283370.0, + "step": 20646 + }, + { + "epoch": 2.267406105864265, + "grad_norm": 2.574803113937378, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7511714100837708, + "num_tokens": 534305855.0, + "step": 20647 + }, + { + "epoch": 2.267515923566879, + "grad_norm": 2.926863193511963, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7334333658218384, + "num_tokens": 534324343.0, + "step": 20648 + }, + { + "epoch": 2.2676257412694927, + "grad_norm": 2.39985990524292, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7448036670684814, + "num_tokens": 534349733.0, + "step": 20649 + }, + { + "epoch": 2.2677355589721064, + "grad_norm": 2.4323105812072754, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7200517654418945, + "num_tokens": 534376612.0, + "step": 20650 + }, + { + "epoch": 2.26784537667472, + "grad_norm": 2.194772243499756, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7361640930175781, + "num_tokens": 534405067.0, + "step": 20651 + }, + { + "epoch": 2.2679551943773335, + "grad_norm": 2.1572601795196533, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7039345502853394, + "num_tokens": 534434465.0, + "step": 20652 + }, + { + "epoch": 2.2680650120799473, + "grad_norm": 2.5781326293945312, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7261461019515991, + "num_tokens": 534457313.0, + "step": 20653 + }, + { + "epoch": 2.268174829782561, + "grad_norm": 2.294978618621826, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7410756945610046, + "num_tokens": 534483692.0, + "step": 20654 + }, + { + "epoch": 2.2682846474851748, + "grad_norm": 2.456374406814575, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7284218668937683, + "num_tokens": 534506119.0, + "step": 20655 + }, + { + "epoch": 2.268394465187788, + "grad_norm": 2.1720244884490967, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7079953551292419, + "num_tokens": 534539490.0, + "step": 20656 + }, + { + "epoch": 2.268504282890402, + "grad_norm": 2.2795794010162354, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.759151816368103, + "num_tokens": 534566135.0, + "step": 20657 + }, + { + "epoch": 2.2686141005930156, + "grad_norm": 2.4463577270507812, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7459337115287781, + "num_tokens": 534591151.0, + "step": 20658 + }, + { + "epoch": 2.2687239182956294, + "grad_norm": 2.175039529800415, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7284670472145081, + "num_tokens": 534620416.0, + "step": 20659 + }, + { + "epoch": 2.2688337359982427, + "grad_norm": 2.5025739669799805, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7439343929290771, + "num_tokens": 534643272.0, + "step": 20660 + }, + { + "epoch": 2.2689435537008564, + "grad_norm": 2.1985292434692383, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.717069685459137, + "num_tokens": 534674244.0, + "step": 20661 + }, + { + "epoch": 2.26905337140347, + "grad_norm": 2.1643776893615723, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7453769445419312, + "num_tokens": 534702621.0, + "step": 20662 + }, + { + "epoch": 2.269163189106084, + "grad_norm": 2.616391181945801, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7412598729133606, + "num_tokens": 534724554.0, + "step": 20663 + }, + { + "epoch": 2.2692730068086977, + "grad_norm": 2.317690372467041, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7145368456840515, + "num_tokens": 534754776.0, + "step": 20664 + }, + { + "epoch": 2.269382824511311, + "grad_norm": 2.405184507369995, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7322914600372314, + "num_tokens": 534781638.0, + "step": 20665 + }, + { + "epoch": 2.2694926422139248, + "grad_norm": 2.7372875213623047, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7481238842010498, + "num_tokens": 534800711.0, + "step": 20666 + }, + { + "epoch": 2.2696024599165385, + "grad_norm": 2.1921606063842773, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7060458660125732, + "num_tokens": 534833402.0, + "step": 20667 + }, + { + "epoch": 2.2697122776191523, + "grad_norm": 2.381141424179077, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7467300891876221, + "num_tokens": 534858057.0, + "step": 20668 + }, + { + "epoch": 2.269822095321766, + "grad_norm": 2.234501838684082, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7109061479568481, + "num_tokens": 534884852.0, + "step": 20669 + }, + { + "epoch": 2.2699319130243794, + "grad_norm": 2.610405445098877, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7383667230606079, + "num_tokens": 534906785.0, + "step": 20670 + }, + { + "epoch": 2.270041730726993, + "grad_norm": 2.199817657470703, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7280185222625732, + "num_tokens": 534939871.0, + "step": 20671 + }, + { + "epoch": 2.270151548429607, + "grad_norm": 2.8787245750427246, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7591683268547058, + "num_tokens": 534957889.0, + "step": 20672 + }, + { + "epoch": 2.2702613661322206, + "grad_norm": 2.264540910720825, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7136899828910828, + "num_tokens": 534988918.0, + "step": 20673 + }, + { + "epoch": 2.2703711838348344, + "grad_norm": 2.403563976287842, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7211405038833618, + "num_tokens": 535016145.0, + "step": 20674 + }, + { + "epoch": 2.2704810015374477, + "grad_norm": 2.5110886096954346, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7475056648254395, + "num_tokens": 535040159.0, + "step": 20675 + }, + { + "epoch": 2.2705908192400615, + "grad_norm": 2.1359639167785645, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7476957440376282, + "num_tokens": 535068420.0, + "step": 20676 + }, + { + "epoch": 2.270700636942675, + "grad_norm": 2.851583957672119, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7321658134460449, + "num_tokens": 535086914.0, + "step": 20677 + }, + { + "epoch": 2.270810454645289, + "grad_norm": 2.543384313583374, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7475212216377258, + "num_tokens": 535107860.0, + "step": 20678 + }, + { + "epoch": 2.2709202723479027, + "grad_norm": 2.3583624362945557, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7385408878326416, + "num_tokens": 535134240.0, + "step": 20679 + }, + { + "epoch": 2.271030090050516, + "grad_norm": 2.634599447250366, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7170263528823853, + "num_tokens": 535159061.0, + "step": 20680 + }, + { + "epoch": 2.27113990775313, + "grad_norm": 2.3461720943450928, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7399359941482544, + "num_tokens": 535183824.0, + "step": 20681 + }, + { + "epoch": 2.2712497254557436, + "grad_norm": 2.4511454105377197, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7284194827079773, + "num_tokens": 535209253.0, + "step": 20682 + }, + { + "epoch": 2.2713595431583573, + "grad_norm": 2.599834680557251, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7230372428894043, + "num_tokens": 535232310.0, + "step": 20683 + }, + { + "epoch": 2.2714693608609706, + "grad_norm": 2.466242790222168, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7392558455467224, + "num_tokens": 535254604.0, + "step": 20684 + }, + { + "epoch": 2.2715791785635844, + "grad_norm": 2.070403575897217, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.717740535736084, + "num_tokens": 535288578.0, + "step": 20685 + }, + { + "epoch": 2.271688996266198, + "grad_norm": 2.4725685119628906, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7247285842895508, + "num_tokens": 535313776.0, + "step": 20686 + }, + { + "epoch": 2.271798813968812, + "grad_norm": 2.145451307296753, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.728588342666626, + "num_tokens": 535343650.0, + "step": 20687 + }, + { + "epoch": 2.271908631671425, + "grad_norm": 2.7491965293884277, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.748703122138977, + "num_tokens": 535363387.0, + "step": 20688 + }, + { + "epoch": 2.272018449374039, + "grad_norm": 2.5638692378997803, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7578730583190918, + "num_tokens": 535385358.0, + "step": 20689 + }, + { + "epoch": 2.2721282670766527, + "grad_norm": 2.300560474395752, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.699561357498169, + "num_tokens": 535416079.0, + "step": 20690 + }, + { + "epoch": 2.2722380847792665, + "grad_norm": 2.4038407802581787, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7402207851409912, + "num_tokens": 535440675.0, + "step": 20691 + }, + { + "epoch": 2.2723479024818802, + "grad_norm": 2.23752760887146, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7208104133605957, + "num_tokens": 535469567.0, + "step": 20692 + }, + { + "epoch": 2.2724577201844935, + "grad_norm": 2.0743062496185303, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.706331729888916, + "num_tokens": 535503782.0, + "step": 20693 + }, + { + "epoch": 2.2725675378871073, + "grad_norm": 2.324738025665283, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7493858933448792, + "num_tokens": 535529052.0, + "step": 20694 + }, + { + "epoch": 2.272677355589721, + "grad_norm": 2.7724950313568115, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7518731951713562, + "num_tokens": 535550344.0, + "step": 20695 + }, + { + "epoch": 2.272787173292335, + "grad_norm": 2.5275585651397705, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7298557162284851, + "num_tokens": 535574011.0, + "step": 20696 + }, + { + "epoch": 2.2728969909949486, + "grad_norm": 2.283020257949829, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.723076343536377, + "num_tokens": 535604376.0, + "step": 20697 + }, + { + "epoch": 2.273006808697562, + "grad_norm": 2.2659876346588135, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7273083925247192, + "num_tokens": 535633623.0, + "step": 20698 + }, + { + "epoch": 2.2731166264001756, + "grad_norm": 2.4622135162353516, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7238322496414185, + "num_tokens": 535658667.0, + "step": 20699 + }, + { + "epoch": 2.2732264441027894, + "grad_norm": 2.4189796447753906, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.761621356010437, + "num_tokens": 535682680.0, + "step": 20700 + }, + { + "epoch": 2.273336261805403, + "grad_norm": 2.560525417327881, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7533892393112183, + "num_tokens": 535704776.0, + "step": 20701 + }, + { + "epoch": 2.273446079508017, + "grad_norm": 2.4962310791015625, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.727886974811554, + "num_tokens": 535731321.0, + "step": 20702 + }, + { + "epoch": 2.2735558972106302, + "grad_norm": 2.823080062866211, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.75339674949646, + "num_tokens": 535749357.0, + "step": 20703 + }, + { + "epoch": 2.273665714913244, + "grad_norm": 2.214038372039795, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7462801933288574, + "num_tokens": 535776414.0, + "step": 20704 + }, + { + "epoch": 2.2737755326158577, + "grad_norm": 2.522343873977661, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7112013697624207, + "num_tokens": 535800583.0, + "step": 20705 + }, + { + "epoch": 2.2738853503184715, + "grad_norm": 2.2633657455444336, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7271014451980591, + "num_tokens": 535827948.0, + "step": 20706 + }, + { + "epoch": 2.273995168021085, + "grad_norm": 2.2766499519348145, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7507786750793457, + "num_tokens": 535855953.0, + "step": 20707 + }, + { + "epoch": 2.2741049857236986, + "grad_norm": 2.5418078899383545, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.75167316198349, + "num_tokens": 535878268.0, + "step": 20708 + }, + { + "epoch": 2.2742148034263123, + "grad_norm": 2.257417678833008, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7311452031135559, + "num_tokens": 535906332.0, + "step": 20709 + }, + { + "epoch": 2.274324621128926, + "grad_norm": 2.6078872680664062, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7300083041191101, + "num_tokens": 535930170.0, + "step": 20710 + }, + { + "epoch": 2.2744344388315394, + "grad_norm": 2.4151437282562256, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7326269149780273, + "num_tokens": 535956811.0, + "step": 20711 + }, + { + "epoch": 2.274544256534153, + "grad_norm": 2.186870813369751, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7322909832000732, + "num_tokens": 535987062.0, + "step": 20712 + }, + { + "epoch": 2.274654074236767, + "grad_norm": 2.238162040710449, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7473361492156982, + "num_tokens": 536016727.0, + "step": 20713 + }, + { + "epoch": 2.2747638919393807, + "grad_norm": 2.4838569164276123, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7281967997550964, + "num_tokens": 536040280.0, + "step": 20714 + }, + { + "epoch": 2.2748737096419944, + "grad_norm": 2.3831779956817627, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7194465398788452, + "num_tokens": 536067056.0, + "step": 20715 + }, + { + "epoch": 2.2749835273446077, + "grad_norm": 2.5682213306427, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7142002582550049, + "num_tokens": 536090216.0, + "step": 20716 + }, + { + "epoch": 2.2750933450472215, + "grad_norm": 2.508967161178589, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7438992261886597, + "num_tokens": 536113676.0, + "step": 20717 + }, + { + "epoch": 2.2752031627498353, + "grad_norm": 2.3385121822357178, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7453426122665405, + "num_tokens": 536140983.0, + "step": 20718 + }, + { + "epoch": 2.275312980452449, + "grad_norm": 2.518538236618042, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7111225724220276, + "num_tokens": 536164566.0, + "step": 20719 + }, + { + "epoch": 2.2754227981550628, + "grad_norm": 2.365644693374634, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7449028491973877, + "num_tokens": 536189222.0, + "step": 20720 + }, + { + "epoch": 2.275532615857676, + "grad_norm": 2.7673306465148926, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7457417249679565, + "num_tokens": 536207842.0, + "step": 20721 + }, + { + "epoch": 2.27564243356029, + "grad_norm": 2.2471792697906494, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7355218529701233, + "num_tokens": 536238776.0, + "step": 20722 + }, + { + "epoch": 2.2757522512629036, + "grad_norm": 2.690990924835205, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7381447553634644, + "num_tokens": 536258384.0, + "step": 20723 + }, + { + "epoch": 2.2758620689655173, + "grad_norm": 2.1957225799560547, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7222133874893188, + "num_tokens": 536284965.0, + "step": 20724 + }, + { + "epoch": 2.275971886668131, + "grad_norm": 2.2665112018585205, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7344251871109009, + "num_tokens": 536313428.0, + "step": 20725 + }, + { + "epoch": 2.2760817043707444, + "grad_norm": 2.4829227924346924, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.718595027923584, + "num_tokens": 536339517.0, + "step": 20726 + }, + { + "epoch": 2.276191522073358, + "grad_norm": 2.258147954940796, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7342942357063293, + "num_tokens": 536365847.0, + "step": 20727 + }, + { + "epoch": 2.276301339775972, + "grad_norm": 2.2763707637786865, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.709083080291748, + "num_tokens": 536393986.0, + "step": 20728 + }, + { + "epoch": 2.2764111574785857, + "grad_norm": 2.279193639755249, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7070920467376709, + "num_tokens": 536422355.0, + "step": 20729 + }, + { + "epoch": 2.2765209751811994, + "grad_norm": 2.480397939682007, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7269392013549805, + "num_tokens": 536447075.0, + "step": 20730 + }, + { + "epoch": 2.2766307928838128, + "grad_norm": 2.173231363296509, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.6975399255752563, + "num_tokens": 536478409.0, + "step": 20731 + }, + { + "epoch": 2.2767406105864265, + "grad_norm": 2.372354745864868, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7189432382583618, + "num_tokens": 536505080.0, + "step": 20732 + }, + { + "epoch": 2.2768504282890403, + "grad_norm": 2.379075050354004, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7526060938835144, + "num_tokens": 536529626.0, + "step": 20733 + }, + { + "epoch": 2.276960245991654, + "grad_norm": 2.2159531116485596, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7458903789520264, + "num_tokens": 536556253.0, + "step": 20734 + }, + { + "epoch": 2.2770700636942673, + "grad_norm": 2.5913360118865967, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7584189772605896, + "num_tokens": 536578185.0, + "step": 20735 + }, + { + "epoch": 2.277179881396881, + "grad_norm": 2.6026558876037598, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7257967591285706, + "num_tokens": 536601047.0, + "step": 20736 + }, + { + "epoch": 2.277289699099495, + "grad_norm": 2.193692922592163, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7405945062637329, + "num_tokens": 536631078.0, + "step": 20737 + }, + { + "epoch": 2.2773995168021086, + "grad_norm": 2.8494510650634766, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7322394847869873, + "num_tokens": 536649282.0, + "step": 20738 + }, + { + "epoch": 2.277509334504722, + "grad_norm": 2.342289686203003, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7190631628036499, + "num_tokens": 536676114.0, + "step": 20739 + }, + { + "epoch": 2.2776191522073357, + "grad_norm": 2.3999404907226562, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7426620721817017, + "num_tokens": 536701464.0, + "step": 20740 + }, + { + "epoch": 2.2777289699099494, + "grad_norm": 2.0645558834075928, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7191148996353149, + "num_tokens": 536733933.0, + "step": 20741 + }, + { + "epoch": 2.277838787612563, + "grad_norm": 2.098226308822632, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7423195242881775, + "num_tokens": 536764702.0, + "step": 20742 + }, + { + "epoch": 2.277948605315177, + "grad_norm": 2.3694369792938232, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7163382172584534, + "num_tokens": 536789400.0, + "step": 20743 + }, + { + "epoch": 2.2780584230177903, + "grad_norm": 2.603126049041748, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7576045989990234, + "num_tokens": 536810565.0, + "step": 20744 + }, + { + "epoch": 2.278168240720404, + "grad_norm": 2.4019713401794434, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.731985867023468, + "num_tokens": 536835174.0, + "step": 20745 + }, + { + "epoch": 2.278278058423018, + "grad_norm": 2.416579246520996, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7294187545776367, + "num_tokens": 536860085.0, + "step": 20746 + }, + { + "epoch": 2.2783878761256315, + "grad_norm": 2.5483949184417725, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7335418462753296, + "num_tokens": 536883060.0, + "step": 20747 + }, + { + "epoch": 2.2784976938282453, + "grad_norm": 2.1712942123413086, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7247780561447144, + "num_tokens": 536912050.0, + "step": 20748 + }, + { + "epoch": 2.2786075115308586, + "grad_norm": 2.5367202758789062, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7385045289993286, + "num_tokens": 536934709.0, + "step": 20749 + }, + { + "epoch": 2.2787173292334724, + "grad_norm": 2.1841533184051514, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7398327589035034, + "num_tokens": 536963123.0, + "step": 20750 + }, + { + "epoch": 2.278827146936086, + "grad_norm": 2.1038641929626465, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7196164131164551, + "num_tokens": 536994318.0, + "step": 20751 + }, + { + "epoch": 2.2789369646387, + "grad_norm": 2.8721230030059814, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7404063940048218, + "num_tokens": 537013036.0, + "step": 20752 + }, + { + "epoch": 2.2790467823413136, + "grad_norm": 2.2819931507110596, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7068902254104614, + "num_tokens": 537042769.0, + "step": 20753 + }, + { + "epoch": 2.279156600043927, + "grad_norm": 2.294463872909546, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7380516529083252, + "num_tokens": 537070224.0, + "step": 20754 + }, + { + "epoch": 2.2792664177465407, + "grad_norm": 2.212850570678711, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7033546566963196, + "num_tokens": 537100398.0, + "step": 20755 + }, + { + "epoch": 2.2793762354491545, + "grad_norm": 2.3184380531311035, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7418048977851868, + "num_tokens": 537124562.0, + "step": 20756 + }, + { + "epoch": 2.279486053151768, + "grad_norm": 2.5641205310821533, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7202709913253784, + "num_tokens": 537148606.0, + "step": 20757 + }, + { + "epoch": 2.2795958708543815, + "grad_norm": 2.227856397628784, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7218377590179443, + "num_tokens": 537177375.0, + "step": 20758 + }, + { + "epoch": 2.2797056885569953, + "grad_norm": 2.537525177001953, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.752494752407074, + "num_tokens": 537198936.0, + "step": 20759 + }, + { + "epoch": 2.279815506259609, + "grad_norm": 2.496823787689209, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7519331574440002, + "num_tokens": 537222002.0, + "step": 20760 + }, + { + "epoch": 2.279925323962223, + "grad_norm": 2.2097771167755127, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7249680757522583, + "num_tokens": 537252011.0, + "step": 20761 + }, + { + "epoch": 2.2800351416648366, + "grad_norm": 2.4453372955322266, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7439601421356201, + "num_tokens": 537275960.0, + "step": 20762 + }, + { + "epoch": 2.28014495936745, + "grad_norm": 2.055288791656494, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7339397668838501, + "num_tokens": 537306553.0, + "step": 20763 + }, + { + "epoch": 2.2802547770700636, + "grad_norm": 2.535029172897339, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7450205087661743, + "num_tokens": 537329446.0, + "step": 20764 + }, + { + "epoch": 2.2803645947726774, + "grad_norm": 2.713726758956909, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7375698685646057, + "num_tokens": 537352940.0, + "step": 20765 + }, + { + "epoch": 2.280474412475291, + "grad_norm": 2.3203320503234863, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7408734560012817, + "num_tokens": 537381267.0, + "step": 20766 + }, + { + "epoch": 2.2805842301779045, + "grad_norm": 2.385171890258789, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7345052361488342, + "num_tokens": 537407946.0, + "step": 20767 + }, + { + "epoch": 2.280694047880518, + "grad_norm": 2.7267181873321533, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.74907386302948, + "num_tokens": 537426873.0, + "step": 20768 + }, + { + "epoch": 2.280803865583132, + "grad_norm": 2.351335048675537, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.730392575263977, + "num_tokens": 537453441.0, + "step": 20769 + }, + { + "epoch": 2.2809136832857457, + "grad_norm": 2.4506735801696777, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7309895753860474, + "num_tokens": 537477762.0, + "step": 20770 + }, + { + "epoch": 2.2810235009883595, + "grad_norm": 2.3337695598602295, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7237227559089661, + "num_tokens": 537504972.0, + "step": 20771 + }, + { + "epoch": 2.281133318690973, + "grad_norm": 2.069676399230957, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7038992643356323, + "num_tokens": 537538606.0, + "step": 20772 + }, + { + "epoch": 2.2812431363935866, + "grad_norm": 2.0940725803375244, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7188916802406311, + "num_tokens": 537570654.0, + "step": 20773 + }, + { + "epoch": 2.2813529540962003, + "grad_norm": 2.342791795730591, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7268303632736206, + "num_tokens": 537597100.0, + "step": 20774 + }, + { + "epoch": 2.281462771798814, + "grad_norm": 2.388676881790161, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7380920648574829, + "num_tokens": 537621795.0, + "step": 20775 + }, + { + "epoch": 2.281572589501428, + "grad_norm": 2.4922053813934326, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.735236406326294, + "num_tokens": 537645966.0, + "step": 20776 + }, + { + "epoch": 2.281682407204041, + "grad_norm": 2.232987880706787, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7163185477256775, + "num_tokens": 537675897.0, + "step": 20777 + }, + { + "epoch": 2.281792224906655, + "grad_norm": 2.3734519481658936, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7333820462226868, + "num_tokens": 537703630.0, + "step": 20778 + }, + { + "epoch": 2.2819020426092687, + "grad_norm": 2.280398368835449, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7033647894859314, + "num_tokens": 537731596.0, + "step": 20779 + }, + { + "epoch": 2.2820118603118824, + "grad_norm": 2.2672667503356934, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7266652584075928, + "num_tokens": 537759309.0, + "step": 20780 + }, + { + "epoch": 2.282121678014496, + "grad_norm": 2.598261833190918, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7366947531700134, + "num_tokens": 537779453.0, + "step": 20781 + }, + { + "epoch": 2.2822314957171095, + "grad_norm": 2.4243781566619873, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7459220886230469, + "num_tokens": 537804432.0, + "step": 20782 + }, + { + "epoch": 2.2823413134197232, + "grad_norm": 2.112330436706543, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7362769842147827, + "num_tokens": 537833522.0, + "step": 20783 + }, + { + "epoch": 2.282451131122337, + "grad_norm": 2.2532777786254883, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7281834483146667, + "num_tokens": 537861995.0, + "step": 20784 + }, + { + "epoch": 2.2825609488249508, + "grad_norm": 2.7005274295806885, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7499825954437256, + "num_tokens": 537883858.0, + "step": 20785 + }, + { + "epoch": 2.282670766527564, + "grad_norm": 2.3136298656463623, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7126851081848145, + "num_tokens": 537910691.0, + "step": 20786 + }, + { + "epoch": 2.282780584230178, + "grad_norm": 2.7334201335906982, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.730249285697937, + "num_tokens": 537932131.0, + "step": 20787 + }, + { + "epoch": 2.2828904019327916, + "grad_norm": 2.3908309936523438, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7125231027603149, + "num_tokens": 537959397.0, + "step": 20788 + }, + { + "epoch": 2.2830002196354053, + "grad_norm": 2.754211902618408, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7217971682548523, + "num_tokens": 537980465.0, + "step": 20789 + }, + { + "epoch": 2.2831100373380186, + "grad_norm": 2.1135716438293457, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.719019889831543, + "num_tokens": 538012996.0, + "step": 20790 + }, + { + "epoch": 2.2832198550406324, + "grad_norm": 2.0966479778289795, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7108259201049805, + "num_tokens": 538042730.0, + "step": 20791 + }, + { + "epoch": 2.283329672743246, + "grad_norm": 2.7572059631347656, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7369862794876099, + "num_tokens": 538064188.0, + "step": 20792 + }, + { + "epoch": 2.28343949044586, + "grad_norm": 2.5073907375335693, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7483457326889038, + "num_tokens": 538089270.0, + "step": 20793 + }, + { + "epoch": 2.2835493081484737, + "grad_norm": 2.2463135719299316, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7266250848770142, + "num_tokens": 538117966.0, + "step": 20794 + }, + { + "epoch": 2.283659125851087, + "grad_norm": 2.2045116424560547, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7445759773254395, + "num_tokens": 538146195.0, + "step": 20795 + }, + { + "epoch": 2.2837689435537007, + "grad_norm": 2.3441832065582275, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7342331409454346, + "num_tokens": 538171123.0, + "step": 20796 + }, + { + "epoch": 2.2838787612563145, + "grad_norm": 2.444070816040039, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7143486738204956, + "num_tokens": 538195696.0, + "step": 20797 + }, + { + "epoch": 2.2839885789589283, + "grad_norm": 2.1687583923339844, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7284883856773376, + "num_tokens": 538224843.0, + "step": 20798 + }, + { + "epoch": 2.284098396661542, + "grad_norm": 2.802313804626465, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7431433200836182, + "num_tokens": 538243669.0, + "step": 20799 + }, + { + "epoch": 2.2842082143641553, + "grad_norm": 2.2369372844696045, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7374379634857178, + "num_tokens": 538271782.0, + "step": 20800 + }, + { + "epoch": 2.284318032066769, + "grad_norm": 2.407233476638794, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7286208868026733, + "num_tokens": 538295796.0, + "step": 20801 + }, + { + "epoch": 2.284427849769383, + "grad_norm": 2.558239698410034, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7212005853652954, + "num_tokens": 538319690.0, + "step": 20802 + }, + { + "epoch": 2.2845376674719966, + "grad_norm": 2.3930349349975586, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7254878282546997, + "num_tokens": 538345376.0, + "step": 20803 + }, + { + "epoch": 2.2846474851746104, + "grad_norm": 2.392204523086548, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.740035355091095, + "num_tokens": 538368600.0, + "step": 20804 + }, + { + "epoch": 2.2847573028772237, + "grad_norm": 2.247365951538086, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7341741323471069, + "num_tokens": 538396941.0, + "step": 20805 + }, + { + "epoch": 2.2848671205798374, + "grad_norm": 2.540623664855957, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7064356207847595, + "num_tokens": 538423386.0, + "step": 20806 + }, + { + "epoch": 2.284976938282451, + "grad_norm": 2.3941004276275635, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7238646745681763, + "num_tokens": 538449073.0, + "step": 20807 + }, + { + "epoch": 2.285086755985065, + "grad_norm": 2.138495445251465, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7285590171813965, + "num_tokens": 538481914.0, + "step": 20808 + }, + { + "epoch": 2.2851965736876787, + "grad_norm": 2.6246509552001953, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7513911724090576, + "num_tokens": 538502236.0, + "step": 20809 + }, + { + "epoch": 2.285306391390292, + "grad_norm": 2.7152364253997803, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7323489189147949, + "num_tokens": 538522773.0, + "step": 20810 + }, + { + "epoch": 2.2854162090929058, + "grad_norm": 2.478163003921509, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7399981021881104, + "num_tokens": 538547524.0, + "step": 20811 + }, + { + "epoch": 2.2855260267955195, + "grad_norm": 2.5917203426361084, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7483956217765808, + "num_tokens": 538569996.0, + "step": 20812 + }, + { + "epoch": 2.2856358444981333, + "grad_norm": 2.231374502182007, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7270443439483643, + "num_tokens": 538597151.0, + "step": 20813 + }, + { + "epoch": 2.2857456622007466, + "grad_norm": 2.1565980911254883, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7105324268341064, + "num_tokens": 538627382.0, + "step": 20814 + }, + { + "epoch": 2.2858554799033604, + "grad_norm": 2.4543557167053223, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7328078746795654, + "num_tokens": 538652426.0, + "step": 20815 + }, + { + "epoch": 2.285965297605974, + "grad_norm": 2.2596747875213623, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6908231973648071, + "num_tokens": 538681280.0, + "step": 20816 + }, + { + "epoch": 2.286075115308588, + "grad_norm": 2.214067220687866, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7223875522613525, + "num_tokens": 538709258.0, + "step": 20817 + }, + { + "epoch": 2.286184933011201, + "grad_norm": 2.4868338108062744, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7323579788208008, + "num_tokens": 538734645.0, + "step": 20818 + }, + { + "epoch": 2.286294750713815, + "grad_norm": 2.398771286010742, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7438822388648987, + "num_tokens": 538759285.0, + "step": 20819 + }, + { + "epoch": 2.2864045684164287, + "grad_norm": 2.1263785362243652, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7033442258834839, + "num_tokens": 538789988.0, + "step": 20820 + }, + { + "epoch": 2.2865143861190425, + "grad_norm": 2.490612030029297, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7403239011764526, + "num_tokens": 538813693.0, + "step": 20821 + }, + { + "epoch": 2.286624203821656, + "grad_norm": 2.288386344909668, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.745082676410675, + "num_tokens": 538839601.0, + "step": 20822 + }, + { + "epoch": 2.2867340215242695, + "grad_norm": 2.063135862350464, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7290233373641968, + "num_tokens": 538869111.0, + "step": 20823 + }, + { + "epoch": 2.2868438392268833, + "grad_norm": 2.4193432331085205, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7287896275520325, + "num_tokens": 538895327.0, + "step": 20824 + }, + { + "epoch": 2.286953656929497, + "grad_norm": 2.606353521347046, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7140687704086304, + "num_tokens": 538917551.0, + "step": 20825 + }, + { + "epoch": 2.287063474632111, + "grad_norm": 2.1353671550750732, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7169234752655029, + "num_tokens": 538948810.0, + "step": 20826 + }, + { + "epoch": 2.2871732923347246, + "grad_norm": 2.302619695663452, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7062546610832214, + "num_tokens": 538977071.0, + "step": 20827 + }, + { + "epoch": 2.287283110037338, + "grad_norm": 2.242027997970581, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7434391975402832, + "num_tokens": 539005221.0, + "step": 20828 + }, + { + "epoch": 2.2873929277399516, + "grad_norm": 2.0958006381988525, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.720923662185669, + "num_tokens": 539037391.0, + "step": 20829 + }, + { + "epoch": 2.2875027454425654, + "grad_norm": 3.085298776626587, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7413824796676636, + "num_tokens": 539054853.0, + "step": 20830 + }, + { + "epoch": 2.287612563145179, + "grad_norm": 2.567779541015625, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7512733340263367, + "num_tokens": 539076794.0, + "step": 20831 + }, + { + "epoch": 2.287722380847793, + "grad_norm": 2.4368979930877686, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7643607258796692, + "num_tokens": 539099218.0, + "step": 20832 + }, + { + "epoch": 2.287832198550406, + "grad_norm": 2.311429262161255, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7296514511108398, + "num_tokens": 539124379.0, + "step": 20833 + }, + { + "epoch": 2.28794201625302, + "grad_norm": 2.3075571060180664, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.727242648601532, + "num_tokens": 539150985.0, + "step": 20834 + }, + { + "epoch": 2.2880518339556337, + "grad_norm": 2.6508686542510986, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7467304468154907, + "num_tokens": 539171635.0, + "step": 20835 + }, + { + "epoch": 2.2881616516582475, + "grad_norm": 2.4655518531799316, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7418428063392639, + "num_tokens": 539195741.0, + "step": 20836 + }, + { + "epoch": 2.288271469360861, + "grad_norm": 2.1831018924713135, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.737975001335144, + "num_tokens": 539224786.0, + "step": 20837 + }, + { + "epoch": 2.2883812870634745, + "grad_norm": 2.4995226860046387, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7146097421646118, + "num_tokens": 539249631.0, + "step": 20838 + }, + { + "epoch": 2.2884911047660883, + "grad_norm": 2.220402717590332, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7288559675216675, + "num_tokens": 539277615.0, + "step": 20839 + }, + { + "epoch": 2.288600922468702, + "grad_norm": 2.5030486583709717, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7335114479064941, + "num_tokens": 539300415.0, + "step": 20840 + }, + { + "epoch": 2.2887107401713154, + "grad_norm": 2.392010450363159, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7449557781219482, + "num_tokens": 539324690.0, + "step": 20841 + }, + { + "epoch": 2.288820557873929, + "grad_norm": 2.2081799507141113, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7425243854522705, + "num_tokens": 539353522.0, + "step": 20842 + }, + { + "epoch": 2.288930375576543, + "grad_norm": 2.309359312057495, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7342958450317383, + "num_tokens": 539380717.0, + "step": 20843 + }, + { + "epoch": 2.2890401932791566, + "grad_norm": 2.4702320098876953, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7069077491760254, + "num_tokens": 539406095.0, + "step": 20844 + }, + { + "epoch": 2.2891500109817704, + "grad_norm": 2.64054799079895, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.741530179977417, + "num_tokens": 539428055.0, + "step": 20845 + }, + { + "epoch": 2.2892598286843837, + "grad_norm": 2.355923652648926, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7456081509590149, + "num_tokens": 539453392.0, + "step": 20846 + }, + { + "epoch": 2.2893696463869975, + "grad_norm": 2.3413281440734863, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7358572483062744, + "num_tokens": 539478657.0, + "step": 20847 + }, + { + "epoch": 2.2894794640896112, + "grad_norm": 2.320070743560791, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7369223833084106, + "num_tokens": 539503239.0, + "step": 20848 + }, + { + "epoch": 2.289589281792225, + "grad_norm": 2.254673480987549, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7515827417373657, + "num_tokens": 539529013.0, + "step": 20849 + }, + { + "epoch": 2.2896990994948387, + "grad_norm": 2.385815143585205, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7473113536834717, + "num_tokens": 539552767.0, + "step": 20850 + }, + { + "epoch": 2.289808917197452, + "grad_norm": 2.2944233417510986, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7555892467498779, + "num_tokens": 539579488.0, + "step": 20851 + }, + { + "epoch": 2.289918734900066, + "grad_norm": 2.301896333694458, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7224438190460205, + "num_tokens": 539606583.0, + "step": 20852 + }, + { + "epoch": 2.2900285526026796, + "grad_norm": 2.4654061794281006, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7181903719902039, + "num_tokens": 539631989.0, + "step": 20853 + }, + { + "epoch": 2.2901383703052933, + "grad_norm": 2.5604188442230225, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7282135486602783, + "num_tokens": 539656172.0, + "step": 20854 + }, + { + "epoch": 2.290248188007907, + "grad_norm": 2.303480863571167, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7246354818344116, + "num_tokens": 539684394.0, + "step": 20855 + }, + { + "epoch": 2.2903580057105204, + "grad_norm": 2.264524221420288, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7077795267105103, + "num_tokens": 539712695.0, + "step": 20856 + }, + { + "epoch": 2.290467823413134, + "grad_norm": 2.2395410537719727, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.714652955532074, + "num_tokens": 539741458.0, + "step": 20857 + }, + { + "epoch": 2.290577641115748, + "grad_norm": 2.2746052742004395, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7272493839263916, + "num_tokens": 539770286.0, + "step": 20858 + }, + { + "epoch": 2.2906874588183617, + "grad_norm": 2.415210723876953, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7181429862976074, + "num_tokens": 539797632.0, + "step": 20859 + }, + { + "epoch": 2.2907972765209754, + "grad_norm": 2.024397850036621, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7169032692909241, + "num_tokens": 539831806.0, + "step": 20860 + }, + { + "epoch": 2.2909070942235887, + "grad_norm": 2.157904863357544, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.726787805557251, + "num_tokens": 539859437.0, + "step": 20861 + }, + { + "epoch": 2.2910169119262025, + "grad_norm": 2.6818530559539795, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7218754291534424, + "num_tokens": 539880540.0, + "step": 20862 + }, + { + "epoch": 2.2911267296288162, + "grad_norm": 2.8102996349334717, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7521511316299438, + "num_tokens": 539899973.0, + "step": 20863 + }, + { + "epoch": 2.29123654733143, + "grad_norm": 1.9834709167480469, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7216953039169312, + "num_tokens": 539934351.0, + "step": 20864 + }, + { + "epoch": 2.2913463650340433, + "grad_norm": 2.58313250541687, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7442545890808105, + "num_tokens": 539956011.0, + "step": 20865 + }, + { + "epoch": 2.291456182736657, + "grad_norm": 2.6255171298980713, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7437587976455688, + "num_tokens": 539976916.0, + "step": 20866 + }, + { + "epoch": 2.291566000439271, + "grad_norm": 2.3203914165496826, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.717674732208252, + "num_tokens": 540002841.0, + "step": 20867 + }, + { + "epoch": 2.2916758181418846, + "grad_norm": 2.3034932613372803, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7285211682319641, + "num_tokens": 540028328.0, + "step": 20868 + }, + { + "epoch": 2.291785635844498, + "grad_norm": 2.272015333175659, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7271971702575684, + "num_tokens": 540054231.0, + "step": 20869 + }, + { + "epoch": 2.2918954535471117, + "grad_norm": 2.733713388442993, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7360438704490662, + "num_tokens": 540072985.0, + "step": 20870 + }, + { + "epoch": 2.2920052712497254, + "grad_norm": 2.5751802921295166, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7449134588241577, + "num_tokens": 540093628.0, + "step": 20871 + }, + { + "epoch": 2.292115088952339, + "grad_norm": 2.592726469039917, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.734259307384491, + "num_tokens": 540115163.0, + "step": 20872 + }, + { + "epoch": 2.292224906654953, + "grad_norm": 2.5718252658843994, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.749954104423523, + "num_tokens": 540137017.0, + "step": 20873 + }, + { + "epoch": 2.2923347243575662, + "grad_norm": 2.651522397994995, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7318865060806274, + "num_tokens": 540158223.0, + "step": 20874 + }, + { + "epoch": 2.29244454206018, + "grad_norm": 2.3500306606292725, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7152390480041504, + "num_tokens": 540184447.0, + "step": 20875 + }, + { + "epoch": 2.2925543597627938, + "grad_norm": 2.1947507858276367, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7196252346038818, + "num_tokens": 540214098.0, + "step": 20876 + }, + { + "epoch": 2.2926641774654075, + "grad_norm": 2.4195144176483154, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7674691081047058, + "num_tokens": 540236665.0, + "step": 20877 + }, + { + "epoch": 2.2927739951680213, + "grad_norm": 2.3767316341400146, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7465437054634094, + "num_tokens": 540262127.0, + "step": 20878 + }, + { + "epoch": 2.2928838128706346, + "grad_norm": 2.310546398162842, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7357178330421448, + "num_tokens": 540286459.0, + "step": 20879 + }, + { + "epoch": 2.2929936305732483, + "grad_norm": 2.652299404144287, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7246128916740417, + "num_tokens": 540308637.0, + "step": 20880 + }, + { + "epoch": 2.293103448275862, + "grad_norm": 2.4641690254211426, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7529534101486206, + "num_tokens": 540330801.0, + "step": 20881 + }, + { + "epoch": 2.293213265978476, + "grad_norm": 2.721000909805298, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.761325478553772, + "num_tokens": 540350290.0, + "step": 20882 + }, + { + "epoch": 2.2933230836810896, + "grad_norm": 2.3454830646514893, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7402642369270325, + "num_tokens": 540375822.0, + "step": 20883 + }, + { + "epoch": 2.293432901383703, + "grad_norm": 2.6414034366607666, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7276932001113892, + "num_tokens": 540400247.0, + "step": 20884 + }, + { + "epoch": 2.2935427190863167, + "grad_norm": 2.750246047973633, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7621890902519226, + "num_tokens": 540419036.0, + "step": 20885 + }, + { + "epoch": 2.2936525367889304, + "grad_norm": 1.9902420043945312, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.6998408436775208, + "num_tokens": 540454269.0, + "step": 20886 + }, + { + "epoch": 2.293762354491544, + "grad_norm": 2.4073069095611572, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7412710785865784, + "num_tokens": 540478961.0, + "step": 20887 + }, + { + "epoch": 2.2938721721941575, + "grad_norm": 2.3081350326538086, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7463478446006775, + "num_tokens": 540505132.0, + "step": 20888 + }, + { + "epoch": 2.2939819898967713, + "grad_norm": 2.8983027935028076, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7381582260131836, + "num_tokens": 540524365.0, + "step": 20889 + }, + { + "epoch": 2.294091807599385, + "grad_norm": 2.3953745365142822, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7227821350097656, + "num_tokens": 540551399.0, + "step": 20890 + }, + { + "epoch": 2.294201625301999, + "grad_norm": 2.3758981227874756, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7468153834342957, + "num_tokens": 540575806.0, + "step": 20891 + }, + { + "epoch": 2.294311443004612, + "grad_norm": 2.3914954662323, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7296110391616821, + "num_tokens": 540599400.0, + "step": 20892 + }, + { + "epoch": 2.294421260707226, + "grad_norm": 2.1412625312805176, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7319251894950867, + "num_tokens": 540629135.0, + "step": 20893 + }, + { + "epoch": 2.2945310784098396, + "grad_norm": 2.1920247077941895, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7001656293869019, + "num_tokens": 540659440.0, + "step": 20894 + }, + { + "epoch": 2.2946408961124534, + "grad_norm": 2.7280025482177734, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7477669715881348, + "num_tokens": 540680501.0, + "step": 20895 + }, + { + "epoch": 2.294750713815067, + "grad_norm": 2.452157974243164, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7360681891441345, + "num_tokens": 540703039.0, + "step": 20896 + }, + { + "epoch": 2.2948605315176804, + "grad_norm": 2.251058340072632, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.719480037689209, + "num_tokens": 540732665.0, + "step": 20897 + }, + { + "epoch": 2.294970349220294, + "grad_norm": 2.356752872467041, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7303768396377563, + "num_tokens": 540759246.0, + "step": 20898 + }, + { + "epoch": 2.295080166922908, + "grad_norm": 2.171750783920288, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7145766019821167, + "num_tokens": 540788127.0, + "step": 20899 + }, + { + "epoch": 2.2951899846255217, + "grad_norm": 2.898716926574707, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7495628595352173, + "num_tokens": 540806931.0, + "step": 20900 + }, + { + "epoch": 2.2952998023281355, + "grad_norm": 2.8658246994018555, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7362655401229858, + "num_tokens": 540826294.0, + "step": 20901 + }, + { + "epoch": 2.2954096200307488, + "grad_norm": 2.3732075691223145, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7265647053718567, + "num_tokens": 540851938.0, + "step": 20902 + }, + { + "epoch": 2.2955194377333625, + "grad_norm": 2.247471809387207, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7128211855888367, + "num_tokens": 540879803.0, + "step": 20903 + }, + { + "epoch": 2.2956292554359763, + "grad_norm": 2.334097146987915, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7114787697792053, + "num_tokens": 540908534.0, + "step": 20904 + }, + { + "epoch": 2.29573907313859, + "grad_norm": 2.6235809326171875, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7232801914215088, + "num_tokens": 540930373.0, + "step": 20905 + }, + { + "epoch": 2.295848890841204, + "grad_norm": 2.387053966522217, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7411350607872009, + "num_tokens": 540954331.0, + "step": 20906 + }, + { + "epoch": 2.295958708543817, + "grad_norm": 2.5830392837524414, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7464943528175354, + "num_tokens": 540975078.0, + "step": 20907 + }, + { + "epoch": 2.296068526246431, + "grad_norm": 2.4026105403900146, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7211054563522339, + "num_tokens": 541000601.0, + "step": 20908 + }, + { + "epoch": 2.2961783439490446, + "grad_norm": 2.334468364715576, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7187513113021851, + "num_tokens": 541028698.0, + "step": 20909 + }, + { + "epoch": 2.2962881616516584, + "grad_norm": 2.133362293243408, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7081218361854553, + "num_tokens": 541060964.0, + "step": 20910 + }, + { + "epoch": 2.296397979354272, + "grad_norm": 2.472048044204712, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7275717854499817, + "num_tokens": 541083273.0, + "step": 20911 + }, + { + "epoch": 2.2965077970568855, + "grad_norm": 2.531454563140869, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.706523060798645, + "num_tokens": 541107607.0, + "step": 20912 + }, + { + "epoch": 2.296617614759499, + "grad_norm": 2.392474889755249, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7325979471206665, + "num_tokens": 541131056.0, + "step": 20913 + }, + { + "epoch": 2.296727432462113, + "grad_norm": 2.222341775894165, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7204608917236328, + "num_tokens": 541158773.0, + "step": 20914 + }, + { + "epoch": 2.2968372501647267, + "grad_norm": 2.4723219871520996, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7394685745239258, + "num_tokens": 541181050.0, + "step": 20915 + }, + { + "epoch": 2.29694706786734, + "grad_norm": 2.3136985301971436, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7286958694458008, + "num_tokens": 541210007.0, + "step": 20916 + }, + { + "epoch": 2.297056885569954, + "grad_norm": 2.2915737628936768, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7213212251663208, + "num_tokens": 541235748.0, + "step": 20917 + }, + { + "epoch": 2.2971667032725676, + "grad_norm": 2.172337293624878, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7257786989212036, + "num_tokens": 541267118.0, + "step": 20918 + }, + { + "epoch": 2.2972765209751813, + "grad_norm": 2.463937520980835, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.737257182598114, + "num_tokens": 541290251.0, + "step": 20919 + }, + { + "epoch": 2.2973863386777946, + "grad_norm": 2.078963279724121, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7109517455101013, + "num_tokens": 541321159.0, + "step": 20920 + }, + { + "epoch": 2.2974961563804084, + "grad_norm": 2.236107110977173, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7220832109451294, + "num_tokens": 541348450.0, + "step": 20921 + }, + { + "epoch": 2.297605974083022, + "grad_norm": 2.3965091705322266, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7441663146018982, + "num_tokens": 541372135.0, + "step": 20922 + }, + { + "epoch": 2.297715791785636, + "grad_norm": 2.1953723430633545, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7229890823364258, + "num_tokens": 541401150.0, + "step": 20923 + }, + { + "epoch": 2.2978256094882497, + "grad_norm": 2.18056583404541, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7160767316818237, + "num_tokens": 541435158.0, + "step": 20924 + }, + { + "epoch": 2.297935427190863, + "grad_norm": 2.9266843795776367, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7360407114028931, + "num_tokens": 541453374.0, + "step": 20925 + }, + { + "epoch": 2.2980452448934767, + "grad_norm": 2.4082822799682617, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7843325138092041, + "num_tokens": 541475660.0, + "step": 20926 + }, + { + "epoch": 2.2981550625960905, + "grad_norm": 2.239264488220215, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7532118558883667, + "num_tokens": 541503522.0, + "step": 20927 + }, + { + "epoch": 2.2982648802987042, + "grad_norm": 2.4198832511901855, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7391211986541748, + "num_tokens": 541528360.0, + "step": 20928 + }, + { + "epoch": 2.298374698001318, + "grad_norm": 2.4951136112213135, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7211774587631226, + "num_tokens": 541554598.0, + "step": 20929 + }, + { + "epoch": 2.2984845157039313, + "grad_norm": 2.49980092048645, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7565503716468811, + "num_tokens": 541577840.0, + "step": 20930 + }, + { + "epoch": 2.298594333406545, + "grad_norm": 2.2541346549987793, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.709131121635437, + "num_tokens": 541608864.0, + "step": 20931 + }, + { + "epoch": 2.298704151109159, + "grad_norm": 2.2797107696533203, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7035846710205078, + "num_tokens": 541636803.0, + "step": 20932 + }, + { + "epoch": 2.2988139688117726, + "grad_norm": 2.2429206371307373, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7292384505271912, + "num_tokens": 541664820.0, + "step": 20933 + }, + { + "epoch": 2.2989237865143863, + "grad_norm": 2.4780220985412598, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7250649929046631, + "num_tokens": 541690577.0, + "step": 20934 + }, + { + "epoch": 2.2990336042169996, + "grad_norm": 2.2932517528533936, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7305896282196045, + "num_tokens": 541716375.0, + "step": 20935 + }, + { + "epoch": 2.2991434219196134, + "grad_norm": 2.3931403160095215, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7449580430984497, + "num_tokens": 541742331.0, + "step": 20936 + }, + { + "epoch": 2.299253239622227, + "grad_norm": 2.362433433532715, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7380726337432861, + "num_tokens": 541767319.0, + "step": 20937 + }, + { + "epoch": 2.299363057324841, + "grad_norm": 2.269822120666504, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7082937359809875, + "num_tokens": 541795125.0, + "step": 20938 + }, + { + "epoch": 2.2994728750274542, + "grad_norm": 2.0304388999938965, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7151225805282593, + "num_tokens": 541830697.0, + "step": 20939 + }, + { + "epoch": 2.299582692730068, + "grad_norm": 2.502312660217285, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.741436243057251, + "num_tokens": 541856540.0, + "step": 20940 + }, + { + "epoch": 2.2996925104326817, + "grad_norm": 2.756699323654175, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7521274089813232, + "num_tokens": 541876615.0, + "step": 20941 + }, + { + "epoch": 2.2998023281352955, + "grad_norm": 2.203498601913452, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7246660590171814, + "num_tokens": 541903735.0, + "step": 20942 + }, + { + "epoch": 2.2999121458379093, + "grad_norm": 2.0948824882507324, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7126684188842773, + "num_tokens": 541935807.0, + "step": 20943 + }, + { + "epoch": 2.3000219635405226, + "grad_norm": 2.3441579341888428, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7411406636238098, + "num_tokens": 541962688.0, + "step": 20944 + }, + { + "epoch": 2.3001317812431363, + "grad_norm": 2.2977941036224365, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7202808856964111, + "num_tokens": 541988938.0, + "step": 20945 + }, + { + "epoch": 2.30024159894575, + "grad_norm": 2.6336731910705566, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7619233727455139, + "num_tokens": 542010466.0, + "step": 20946 + }, + { + "epoch": 2.300351416648364, + "grad_norm": 2.391300678253174, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7363080382347107, + "num_tokens": 542034293.0, + "step": 20947 + }, + { + "epoch": 2.300461234350977, + "grad_norm": 2.466999053955078, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7134764194488525, + "num_tokens": 542059149.0, + "step": 20948 + }, + { + "epoch": 2.300571052053591, + "grad_norm": 2.054272174835205, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7288128137588501, + "num_tokens": 542089035.0, + "step": 20949 + }, + { + "epoch": 2.3006808697562047, + "grad_norm": 2.4087486267089844, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.700093150138855, + "num_tokens": 542116054.0, + "step": 20950 + }, + { + "epoch": 2.3007906874588184, + "grad_norm": 2.3057780265808105, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7205958366394043, + "num_tokens": 542145231.0, + "step": 20951 + }, + { + "epoch": 2.300900505161432, + "grad_norm": 2.2298500537872314, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7255450487136841, + "num_tokens": 542173544.0, + "step": 20952 + }, + { + "epoch": 2.3010103228640455, + "grad_norm": 2.177602529525757, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7103042602539062, + "num_tokens": 542202205.0, + "step": 20953 + }, + { + "epoch": 2.3011201405666593, + "grad_norm": 2.374821424484253, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7242798805236816, + "num_tokens": 542227842.0, + "step": 20954 + }, + { + "epoch": 2.301229958269273, + "grad_norm": 2.1304705142974854, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7261556386947632, + "num_tokens": 542255069.0, + "step": 20955 + }, + { + "epoch": 2.3013397759718868, + "grad_norm": 2.290138006210327, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7299607396125793, + "num_tokens": 542282401.0, + "step": 20956 + }, + { + "epoch": 2.3014495936745005, + "grad_norm": 2.314204692840576, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7561993598937988, + "num_tokens": 542308418.0, + "step": 20957 + }, + { + "epoch": 2.301559411377114, + "grad_norm": 2.38730525970459, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7277867794036865, + "num_tokens": 542333647.0, + "step": 20958 + }, + { + "epoch": 2.3016692290797276, + "grad_norm": 2.292060613632202, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7335633039474487, + "num_tokens": 542359218.0, + "step": 20959 + }, + { + "epoch": 2.3017790467823414, + "grad_norm": 2.2004828453063965, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7408815622329712, + "num_tokens": 542388516.0, + "step": 20960 + }, + { + "epoch": 2.301888864484955, + "grad_norm": 2.248384475708008, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7178696393966675, + "num_tokens": 542417076.0, + "step": 20961 + }, + { + "epoch": 2.301998682187569, + "grad_norm": 2.580571413040161, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7323200702667236, + "num_tokens": 542439792.0, + "step": 20962 + }, + { + "epoch": 2.302108499890182, + "grad_norm": 2.1899924278259277, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7250678539276123, + "num_tokens": 542469589.0, + "step": 20963 + }, + { + "epoch": 2.302218317592796, + "grad_norm": 2.4047651290893555, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7234730124473572, + "num_tokens": 542494322.0, + "step": 20964 + }, + { + "epoch": 2.3023281352954097, + "grad_norm": 2.349989175796509, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7507137060165405, + "num_tokens": 542519882.0, + "step": 20965 + }, + { + "epoch": 2.3024379529980235, + "grad_norm": 2.279463529586792, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7234652042388916, + "num_tokens": 542548456.0, + "step": 20966 + }, + { + "epoch": 2.3025477707006368, + "grad_norm": 2.5106518268585205, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7299816608428955, + "num_tokens": 542573067.0, + "step": 20967 + }, + { + "epoch": 2.3026575884032505, + "grad_norm": 2.233703136444092, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7217221856117249, + "num_tokens": 542602039.0, + "step": 20968 + }, + { + "epoch": 2.3027674061058643, + "grad_norm": 2.737356185913086, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7506940960884094, + "num_tokens": 542622463.0, + "step": 20969 + }, + { + "epoch": 2.302877223808478, + "grad_norm": 2.2716004848480225, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7095047235488892, + "num_tokens": 542649997.0, + "step": 20970 + }, + { + "epoch": 2.3029870415110913, + "grad_norm": 2.184265375137329, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7395750284194946, + "num_tokens": 542679946.0, + "step": 20971 + }, + { + "epoch": 2.303096859213705, + "grad_norm": 2.477651357650757, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.736629068851471, + "num_tokens": 542704940.0, + "step": 20972 + }, + { + "epoch": 2.303206676916319, + "grad_norm": 2.1436054706573486, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7255839705467224, + "num_tokens": 542735386.0, + "step": 20973 + }, + { + "epoch": 2.3033164946189326, + "grad_norm": 2.3825185298919678, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7399620413780212, + "num_tokens": 542760679.0, + "step": 20974 + }, + { + "epoch": 2.3034263123215464, + "grad_norm": 2.0918586254119873, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7425345778465271, + "num_tokens": 542789594.0, + "step": 20975 + }, + { + "epoch": 2.3035361300241597, + "grad_norm": 2.1899383068084717, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7208098769187927, + "num_tokens": 542817666.0, + "step": 20976 + }, + { + "epoch": 2.3036459477267734, + "grad_norm": 2.39998459815979, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.750212550163269, + "num_tokens": 542841983.0, + "step": 20977 + }, + { + "epoch": 2.303755765429387, + "grad_norm": 2.3044660091400146, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7336781024932861, + "num_tokens": 542869857.0, + "step": 20978 + }, + { + "epoch": 2.303865583132001, + "grad_norm": 2.1635568141937256, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7250799536705017, + "num_tokens": 542900501.0, + "step": 20979 + }, + { + "epoch": 2.3039754008346147, + "grad_norm": 2.709221124649048, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7351548671722412, + "num_tokens": 542922771.0, + "step": 20980 + }, + { + "epoch": 2.304085218537228, + "grad_norm": 2.516535997390747, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7531690001487732, + "num_tokens": 542945980.0, + "step": 20981 + }, + { + "epoch": 2.304195036239842, + "grad_norm": 2.582082509994507, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7203088402748108, + "num_tokens": 542969733.0, + "step": 20982 + }, + { + "epoch": 2.3043048539424555, + "grad_norm": 2.3081605434417725, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7446043491363525, + "num_tokens": 542997008.0, + "step": 20983 + }, + { + "epoch": 2.3044146716450693, + "grad_norm": 2.414834499359131, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7367841005325317, + "num_tokens": 543021143.0, + "step": 20984 + }, + { + "epoch": 2.304524489347683, + "grad_norm": 2.3718342781066895, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7447595596313477, + "num_tokens": 543045313.0, + "step": 20985 + }, + { + "epoch": 2.3046343070502964, + "grad_norm": 2.507791042327881, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7421333193778992, + "num_tokens": 543068448.0, + "step": 20986 + }, + { + "epoch": 2.30474412475291, + "grad_norm": 2.1343188285827637, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7228442430496216, + "num_tokens": 543099282.0, + "step": 20987 + }, + { + "epoch": 2.304853942455524, + "grad_norm": 2.6529836654663086, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7334237098693848, + "num_tokens": 543120078.0, + "step": 20988 + }, + { + "epoch": 2.3049637601581376, + "grad_norm": 2.4732248783111572, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7345424294471741, + "num_tokens": 543143251.0, + "step": 20989 + }, + { + "epoch": 2.3050735778607514, + "grad_norm": 2.392997980117798, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7208917737007141, + "num_tokens": 543167052.0, + "step": 20990 + }, + { + "epoch": 2.3051833955633647, + "grad_norm": 2.669041156768799, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7294890284538269, + "num_tokens": 543188857.0, + "step": 20991 + }, + { + "epoch": 2.3052932132659785, + "grad_norm": 2.421597719192505, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7348021864891052, + "num_tokens": 543215243.0, + "step": 20992 + }, + { + "epoch": 2.3054030309685922, + "grad_norm": 2.5306196212768555, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7257500290870667, + "num_tokens": 543239720.0, + "step": 20993 + }, + { + "epoch": 2.305512848671206, + "grad_norm": 2.643941879272461, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.738689661026001, + "num_tokens": 543261966.0, + "step": 20994 + }, + { + "epoch": 2.3056226663738193, + "grad_norm": 2.6889283657073975, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7485253810882568, + "num_tokens": 543281693.0, + "step": 20995 + }, + { + "epoch": 2.305732484076433, + "grad_norm": 2.538571834564209, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7441837787628174, + "num_tokens": 543303138.0, + "step": 20996 + }, + { + "epoch": 2.305842301779047, + "grad_norm": 2.463860034942627, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.707355797290802, + "num_tokens": 543330442.0, + "step": 20997 + }, + { + "epoch": 2.3059521194816606, + "grad_norm": 2.1550827026367188, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7168638706207275, + "num_tokens": 543362534.0, + "step": 20998 + }, + { + "epoch": 2.306061937184274, + "grad_norm": 2.66491961479187, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.739636242389679, + "num_tokens": 543382404.0, + "step": 20999 + }, + { + "epoch": 2.3061717548868876, + "grad_norm": 2.4478814601898193, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7206850051879883, + "num_tokens": 543407543.0, + "step": 21000 + }, + { + "epoch": 2.3062815725895014, + "grad_norm": 2.515833616256714, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7450510263442993, + "num_tokens": 543430168.0, + "step": 21001 + }, + { + "epoch": 2.306391390292115, + "grad_norm": 2.2826709747314453, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7231091260910034, + "num_tokens": 543458488.0, + "step": 21002 + }, + { + "epoch": 2.306501207994729, + "grad_norm": 2.6042869091033936, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7425789833068848, + "num_tokens": 543481233.0, + "step": 21003 + }, + { + "epoch": 2.306611025697342, + "grad_norm": 2.4869916439056396, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.729162871837616, + "num_tokens": 543504692.0, + "step": 21004 + }, + { + "epoch": 2.306720843399956, + "grad_norm": 2.6085922718048096, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7403228878974915, + "num_tokens": 543526621.0, + "step": 21005 + }, + { + "epoch": 2.3068306611025697, + "grad_norm": 2.158050775527954, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7280597686767578, + "num_tokens": 543554414.0, + "step": 21006 + }, + { + "epoch": 2.3069404788051835, + "grad_norm": 2.1009578704833984, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7072091698646545, + "num_tokens": 543587486.0, + "step": 21007 + }, + { + "epoch": 2.3070502965077972, + "grad_norm": 2.265033721923828, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7191617488861084, + "num_tokens": 543615678.0, + "step": 21008 + }, + { + "epoch": 2.3071601142104106, + "grad_norm": 2.476562023162842, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7539446353912354, + "num_tokens": 543639927.0, + "step": 21009 + }, + { + "epoch": 2.3072699319130243, + "grad_norm": 2.303438425064087, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7148799896240234, + "num_tokens": 543668013.0, + "step": 21010 + }, + { + "epoch": 2.307379749615638, + "grad_norm": 2.5024898052215576, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.75018310546875, + "num_tokens": 543692129.0, + "step": 21011 + }, + { + "epoch": 2.307489567318252, + "grad_norm": 2.538705348968506, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7439658641815186, + "num_tokens": 543714848.0, + "step": 21012 + }, + { + "epoch": 2.3075993850208656, + "grad_norm": 2.452432632446289, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7292049527168274, + "num_tokens": 543739756.0, + "step": 21013 + }, + { + "epoch": 2.307709202723479, + "grad_norm": 2.530123710632324, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7284339666366577, + "num_tokens": 543761497.0, + "step": 21014 + }, + { + "epoch": 2.3078190204260927, + "grad_norm": 2.6887829303741455, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7336094379425049, + "num_tokens": 543783793.0, + "step": 21015 + }, + { + "epoch": 2.3079288381287064, + "grad_norm": 2.4015355110168457, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.722162127494812, + "num_tokens": 543809473.0, + "step": 21016 + }, + { + "epoch": 2.30803865583132, + "grad_norm": 2.479933977127075, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7365956902503967, + "num_tokens": 543832119.0, + "step": 21017 + }, + { + "epoch": 2.3081484735339335, + "grad_norm": 2.4280149936676025, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7290350794792175, + "num_tokens": 543858050.0, + "step": 21018 + }, + { + "epoch": 2.3082582912365472, + "grad_norm": 2.418370485305786, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7277465462684631, + "num_tokens": 543883102.0, + "step": 21019 + }, + { + "epoch": 2.308368108939161, + "grad_norm": 2.6307785511016846, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7536057233810425, + "num_tokens": 543904191.0, + "step": 21020 + }, + { + "epoch": 2.3084779266417748, + "grad_norm": 2.6497631072998047, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7380766272544861, + "num_tokens": 543925282.0, + "step": 21021 + }, + { + "epoch": 2.308587744344388, + "grad_norm": 2.6166679859161377, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7283073663711548, + "num_tokens": 543948199.0, + "step": 21022 + }, + { + "epoch": 2.308697562047002, + "grad_norm": 2.056979179382324, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7503312826156616, + "num_tokens": 543978822.0, + "step": 21023 + }, + { + "epoch": 2.3088073797496156, + "grad_norm": 2.2018256187438965, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7101050615310669, + "num_tokens": 544009634.0, + "step": 21024 + }, + { + "epoch": 2.3089171974522293, + "grad_norm": 1.9762436151504517, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7156969308853149, + "num_tokens": 544045825.0, + "step": 21025 + }, + { + "epoch": 2.309027015154843, + "grad_norm": 2.59287166595459, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.739810049533844, + "num_tokens": 544066655.0, + "step": 21026 + }, + { + "epoch": 2.3091368328574564, + "grad_norm": 2.5345168113708496, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7236540913581848, + "num_tokens": 544088602.0, + "step": 21027 + }, + { + "epoch": 2.30924665056007, + "grad_norm": 2.439284324645996, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.743438184261322, + "num_tokens": 544111969.0, + "step": 21028 + }, + { + "epoch": 2.309356468262684, + "grad_norm": 2.3576905727386475, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7519360184669495, + "num_tokens": 544136419.0, + "step": 21029 + }, + { + "epoch": 2.3094662859652977, + "grad_norm": 2.1026053428649902, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7233849763870239, + "num_tokens": 544167652.0, + "step": 21030 + }, + { + "epoch": 2.3095761036679114, + "grad_norm": 2.140875816345215, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7252568006515503, + "num_tokens": 544196676.0, + "step": 21031 + }, + { + "epoch": 2.3096859213705248, + "grad_norm": 2.404794931411743, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7399862408638, + "num_tokens": 544222160.0, + "step": 21032 + }, + { + "epoch": 2.3097957390731385, + "grad_norm": 2.319347381591797, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7177361845970154, + "num_tokens": 544250547.0, + "step": 21033 + }, + { + "epoch": 2.3099055567757523, + "grad_norm": 2.5032410621643066, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7164479494094849, + "num_tokens": 544275216.0, + "step": 21034 + }, + { + "epoch": 2.310015374478366, + "grad_norm": 2.237621545791626, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7296109795570374, + "num_tokens": 544302957.0, + "step": 21035 + }, + { + "epoch": 2.31012519218098, + "grad_norm": 2.423236608505249, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7224931716918945, + "num_tokens": 544330432.0, + "step": 21036 + }, + { + "epoch": 2.310235009883593, + "grad_norm": 2.4610202312469482, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7315233945846558, + "num_tokens": 544358067.0, + "step": 21037 + }, + { + "epoch": 2.310344827586207, + "grad_norm": 2.700582504272461, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7882809042930603, + "num_tokens": 544378540.0, + "step": 21038 + }, + { + "epoch": 2.3104546452888206, + "grad_norm": 2.4479589462280273, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7611749172210693, + "num_tokens": 544402832.0, + "step": 21039 + }, + { + "epoch": 2.3105644629914344, + "grad_norm": 2.521440267562866, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7586470246315002, + "num_tokens": 544424797.0, + "step": 21040 + }, + { + "epoch": 2.310674280694048, + "grad_norm": 2.1394002437591553, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7231807708740234, + "num_tokens": 544457150.0, + "step": 21041 + }, + { + "epoch": 2.3107840983966614, + "grad_norm": 2.099048614501953, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7296279668807983, + "num_tokens": 544488923.0, + "step": 21042 + }, + { + "epoch": 2.310893916099275, + "grad_norm": 2.512436866760254, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7372978925704956, + "num_tokens": 544518724.0, + "step": 21043 + }, + { + "epoch": 2.311003733801889, + "grad_norm": 2.000035047531128, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7358570098876953, + "num_tokens": 544551721.0, + "step": 21044 + }, + { + "epoch": 2.3111135515045027, + "grad_norm": 2.538426399230957, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7184776067733765, + "num_tokens": 544575492.0, + "step": 21045 + }, + { + "epoch": 2.311223369207116, + "grad_norm": 2.2301132678985596, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7251046299934387, + "num_tokens": 544602952.0, + "step": 21046 + }, + { + "epoch": 2.3113331869097298, + "grad_norm": 2.4164557456970215, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.74552321434021, + "num_tokens": 544628679.0, + "step": 21047 + }, + { + "epoch": 2.3114430046123435, + "grad_norm": 2.3833582401275635, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7333064675331116, + "num_tokens": 544655406.0, + "step": 21048 + }, + { + "epoch": 2.3115528223149573, + "grad_norm": 2.5524919033050537, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7207252979278564, + "num_tokens": 544676038.0, + "step": 21049 + }, + { + "epoch": 2.3116626400175706, + "grad_norm": 2.6155519485473633, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7531970739364624, + "num_tokens": 544696723.0, + "step": 21050 + }, + { + "epoch": 2.3117724577201844, + "grad_norm": 2.4195353984832764, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7410845160484314, + "num_tokens": 544720954.0, + "step": 21051 + }, + { + "epoch": 2.311882275422798, + "grad_norm": 2.300564765930176, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7286388874053955, + "num_tokens": 544748396.0, + "step": 21052 + }, + { + "epoch": 2.311992093125412, + "grad_norm": 2.4347307682037354, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7380712032318115, + "num_tokens": 544771867.0, + "step": 21053 + }, + { + "epoch": 2.3121019108280256, + "grad_norm": 2.440671920776367, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7491896152496338, + "num_tokens": 544794970.0, + "step": 21054 + }, + { + "epoch": 2.312211728530639, + "grad_norm": 2.367384195327759, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7450300455093384, + "num_tokens": 544821841.0, + "step": 21055 + }, + { + "epoch": 2.3123215462332527, + "grad_norm": 2.318225145339966, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7275649309158325, + "num_tokens": 544849110.0, + "step": 21056 + }, + { + "epoch": 2.3124313639358665, + "grad_norm": 2.3769969940185547, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7254514694213867, + "num_tokens": 544876829.0, + "step": 21057 + }, + { + "epoch": 2.31254118163848, + "grad_norm": 2.2219600677490234, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7482433319091797, + "num_tokens": 544904983.0, + "step": 21058 + }, + { + "epoch": 2.312650999341094, + "grad_norm": 2.6851606369018555, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7175449132919312, + "num_tokens": 544926799.0, + "step": 21059 + }, + { + "epoch": 2.3127608170437073, + "grad_norm": 2.2308712005615234, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7036047577857971, + "num_tokens": 544956576.0, + "step": 21060 + }, + { + "epoch": 2.312870634746321, + "grad_norm": 2.6391451358795166, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7488920092582703, + "num_tokens": 544977466.0, + "step": 21061 + }, + { + "epoch": 2.312980452448935, + "grad_norm": 2.2490592002868652, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7144957780838013, + "num_tokens": 545006051.0, + "step": 21062 + }, + { + "epoch": 2.3130902701515486, + "grad_norm": 2.349003791809082, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7190304398536682, + "num_tokens": 545032919.0, + "step": 21063 + }, + { + "epoch": 2.3132000878541623, + "grad_norm": 2.2905194759368896, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7293791770935059, + "num_tokens": 545058743.0, + "step": 21064 + }, + { + "epoch": 2.3133099055567756, + "grad_norm": 2.514833688735962, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7492936253547668, + "num_tokens": 545080941.0, + "step": 21065 + }, + { + "epoch": 2.3134197232593894, + "grad_norm": 2.2369558811187744, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7186551690101624, + "num_tokens": 545111743.0, + "step": 21066 + }, + { + "epoch": 2.313529540962003, + "grad_norm": 2.413532257080078, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7193230390548706, + "num_tokens": 545137372.0, + "step": 21067 + }, + { + "epoch": 2.313639358664617, + "grad_norm": 2.4919495582580566, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7249249219894409, + "num_tokens": 545161184.0, + "step": 21068 + }, + { + "epoch": 2.31374917636723, + "grad_norm": 2.4075005054473877, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7647103071212769, + "num_tokens": 545184304.0, + "step": 21069 + }, + { + "epoch": 2.313858994069844, + "grad_norm": 2.316155195236206, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7233304977416992, + "num_tokens": 545210656.0, + "step": 21070 + }, + { + "epoch": 2.3139688117724577, + "grad_norm": 2.437439441680908, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7582641839981079, + "num_tokens": 545235013.0, + "step": 21071 + }, + { + "epoch": 2.3140786294750715, + "grad_norm": 2.2983922958374023, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7295751571655273, + "num_tokens": 545261614.0, + "step": 21072 + }, + { + "epoch": 2.314188447177685, + "grad_norm": 2.2715330123901367, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7340105772018433, + "num_tokens": 545288530.0, + "step": 21073 + }, + { + "epoch": 2.3142982648802986, + "grad_norm": 2.271726369857788, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7171844244003296, + "num_tokens": 545316653.0, + "step": 21074 + }, + { + "epoch": 2.3144080825829123, + "grad_norm": 2.3439927101135254, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7343400716781616, + "num_tokens": 545344921.0, + "step": 21075 + }, + { + "epoch": 2.314517900285526, + "grad_norm": 2.131141424179077, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7078909277915955, + "num_tokens": 545375523.0, + "step": 21076 + }, + { + "epoch": 2.31462771798814, + "grad_norm": 2.0282669067382812, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7309837341308594, + "num_tokens": 545407587.0, + "step": 21077 + }, + { + "epoch": 2.314737535690753, + "grad_norm": 2.454251289367676, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7454771995544434, + "num_tokens": 545432350.0, + "step": 21078 + }, + { + "epoch": 2.314847353393367, + "grad_norm": 2.2850229740142822, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7473958730697632, + "num_tokens": 545460827.0, + "step": 21079 + }, + { + "epoch": 2.3149571710959806, + "grad_norm": 2.4702839851379395, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.737967848777771, + "num_tokens": 545484607.0, + "step": 21080 + }, + { + "epoch": 2.3150669887985944, + "grad_norm": 2.3726367950439453, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7261770963668823, + "num_tokens": 545508854.0, + "step": 21081 + }, + { + "epoch": 2.315176806501208, + "grad_norm": 2.3258168697357178, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7024053335189819, + "num_tokens": 545536439.0, + "step": 21082 + }, + { + "epoch": 2.3152866242038215, + "grad_norm": 2.3612749576568604, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7279860377311707, + "num_tokens": 545562043.0, + "step": 21083 + }, + { + "epoch": 2.3153964419064352, + "grad_norm": 2.2197678089141846, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7334327697753906, + "num_tokens": 545589760.0, + "step": 21084 + }, + { + "epoch": 2.315506259609049, + "grad_norm": 2.323105812072754, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7409254312515259, + "num_tokens": 545614440.0, + "step": 21085 + }, + { + "epoch": 2.3156160773116627, + "grad_norm": 2.318659543991089, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.735689103603363, + "num_tokens": 545639962.0, + "step": 21086 + }, + { + "epoch": 2.3157258950142765, + "grad_norm": 2.621258497238159, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7177999019622803, + "num_tokens": 545663324.0, + "step": 21087 + }, + { + "epoch": 2.31583571271689, + "grad_norm": 2.6206343173980713, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7524303197860718, + "num_tokens": 545685668.0, + "step": 21088 + }, + { + "epoch": 2.3159455304195036, + "grad_norm": 2.4621431827545166, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.729576051235199, + "num_tokens": 545709631.0, + "step": 21089 + }, + { + "epoch": 2.3160553481221173, + "grad_norm": 2.5257999897003174, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7483399510383606, + "num_tokens": 545733222.0, + "step": 21090 + }, + { + "epoch": 2.316165165824731, + "grad_norm": 2.329219102859497, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7192851305007935, + "num_tokens": 545759942.0, + "step": 21091 + }, + { + "epoch": 2.316274983527345, + "grad_norm": 2.300773859024048, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7477700710296631, + "num_tokens": 545785642.0, + "step": 21092 + }, + { + "epoch": 2.316384801229958, + "grad_norm": 2.25591778755188, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7376701831817627, + "num_tokens": 545814865.0, + "step": 21093 + }, + { + "epoch": 2.316494618932572, + "grad_norm": 2.5049219131469727, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7234516739845276, + "num_tokens": 545838748.0, + "step": 21094 + }, + { + "epoch": 2.3166044366351857, + "grad_norm": 1.9585773944854736, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7167502641677856, + "num_tokens": 545876189.0, + "step": 21095 + }, + { + "epoch": 2.3167142543377994, + "grad_norm": 2.23040771484375, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7154837846755981, + "num_tokens": 545906308.0, + "step": 21096 + }, + { + "epoch": 2.3168240720404127, + "grad_norm": 2.3612191677093506, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7363050580024719, + "num_tokens": 545931183.0, + "step": 21097 + }, + { + "epoch": 2.3169338897430265, + "grad_norm": 2.2320914268493652, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7512953281402588, + "num_tokens": 545958608.0, + "step": 21098 + }, + { + "epoch": 2.3170437074456403, + "grad_norm": 2.165925979614258, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7241559624671936, + "num_tokens": 545988236.0, + "step": 21099 + }, + { + "epoch": 2.317153525148254, + "grad_norm": 2.227839946746826, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7268519997596741, + "num_tokens": 546016863.0, + "step": 21100 + }, + { + "epoch": 2.3172633428508673, + "grad_norm": 2.604463815689087, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7596118450164795, + "num_tokens": 546038858.0, + "step": 21101 + }, + { + "epoch": 2.317373160553481, + "grad_norm": 2.3729488849639893, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.743681788444519, + "num_tokens": 546064374.0, + "step": 21102 + }, + { + "epoch": 2.317482978256095, + "grad_norm": 2.314871072769165, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7406742572784424, + "num_tokens": 546091560.0, + "step": 21103 + }, + { + "epoch": 2.3175927959587086, + "grad_norm": 2.2363650798797607, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7348541021347046, + "num_tokens": 546121413.0, + "step": 21104 + }, + { + "epoch": 2.3177026136613224, + "grad_norm": 2.357802629470825, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7083714008331299, + "num_tokens": 546146179.0, + "step": 21105 + }, + { + "epoch": 2.3178124313639357, + "grad_norm": 2.480262517929077, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7512061595916748, + "num_tokens": 546169669.0, + "step": 21106 + }, + { + "epoch": 2.3179222490665494, + "grad_norm": 2.27030348777771, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7094501256942749, + "num_tokens": 546196400.0, + "step": 21107 + }, + { + "epoch": 2.318032066769163, + "grad_norm": 2.6319291591644287, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7361434102058411, + "num_tokens": 546219794.0, + "step": 21108 + }, + { + "epoch": 2.318141884471777, + "grad_norm": 2.086709976196289, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7126368284225464, + "num_tokens": 546252721.0, + "step": 21109 + }, + { + "epoch": 2.3182517021743907, + "grad_norm": 2.441509485244751, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7619050741195679, + "num_tokens": 546275903.0, + "step": 21110 + }, + { + "epoch": 2.318361519877004, + "grad_norm": 2.8633317947387695, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.743903636932373, + "num_tokens": 546296258.0, + "step": 21111 + }, + { + "epoch": 2.3184713375796178, + "grad_norm": 2.4498276710510254, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7260869741439819, + "num_tokens": 546320414.0, + "step": 21112 + }, + { + "epoch": 2.3185811552822315, + "grad_norm": 2.481907844543457, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7547228336334229, + "num_tokens": 546341684.0, + "step": 21113 + }, + { + "epoch": 2.3186909729848453, + "grad_norm": 2.5341796875, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7592546939849854, + "num_tokens": 546362874.0, + "step": 21114 + }, + { + "epoch": 2.318800790687459, + "grad_norm": 2.383572816848755, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7200415134429932, + "num_tokens": 546390052.0, + "step": 21115 + }, + { + "epoch": 2.3189106083900723, + "grad_norm": 2.1778292655944824, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7225297093391418, + "num_tokens": 546419092.0, + "step": 21116 + }, + { + "epoch": 2.319020426092686, + "grad_norm": 2.5699820518493652, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7334538698196411, + "num_tokens": 546441330.0, + "step": 21117 + }, + { + "epoch": 2.3191302437953, + "grad_norm": 2.6296675205230713, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7353078722953796, + "num_tokens": 546462439.0, + "step": 21118 + }, + { + "epoch": 2.3192400614979136, + "grad_norm": 2.336331367492676, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7594794034957886, + "num_tokens": 546487977.0, + "step": 21119 + }, + { + "epoch": 2.319349879200527, + "grad_norm": 2.986459970474243, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7518949508666992, + "num_tokens": 546505219.0, + "step": 21120 + }, + { + "epoch": 2.3194596969031407, + "grad_norm": 2.155224561691284, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7337826490402222, + "num_tokens": 546533972.0, + "step": 21121 + }, + { + "epoch": 2.3195695146057544, + "grad_norm": 2.438715934753418, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7181888818740845, + "num_tokens": 546559537.0, + "step": 21122 + }, + { + "epoch": 2.319679332308368, + "grad_norm": 2.4923458099365234, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7148549556732178, + "num_tokens": 546583319.0, + "step": 21123 + }, + { + "epoch": 2.319789150010982, + "grad_norm": 2.2664031982421875, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7322218418121338, + "num_tokens": 546609774.0, + "step": 21124 + }, + { + "epoch": 2.3198989677135953, + "grad_norm": 2.2096362113952637, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7280189990997314, + "num_tokens": 546637470.0, + "step": 21125 + }, + { + "epoch": 2.320008785416209, + "grad_norm": 2.3321292400360107, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7466727495193481, + "num_tokens": 546663033.0, + "step": 21126 + }, + { + "epoch": 2.320118603118823, + "grad_norm": 2.6302783489227295, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6944732666015625, + "num_tokens": 546686081.0, + "step": 21127 + }, + { + "epoch": 2.3202284208214365, + "grad_norm": 2.383803606033325, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7466222047805786, + "num_tokens": 546711171.0, + "step": 21128 + }, + { + "epoch": 2.32033823852405, + "grad_norm": 2.6096668243408203, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7145520448684692, + "num_tokens": 546733502.0, + "step": 21129 + }, + { + "epoch": 2.3204480562266636, + "grad_norm": 2.23757266998291, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7197774648666382, + "num_tokens": 546762358.0, + "step": 21130 + }, + { + "epoch": 2.3205578739292774, + "grad_norm": 2.445067882537842, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7309485673904419, + "num_tokens": 546787678.0, + "step": 21131 + }, + { + "epoch": 2.320667691631891, + "grad_norm": 2.125840425491333, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7243990302085876, + "num_tokens": 546818176.0, + "step": 21132 + }, + { + "epoch": 2.320777509334505, + "grad_norm": 2.473752737045288, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.736443042755127, + "num_tokens": 546842222.0, + "step": 21133 + }, + { + "epoch": 2.320887327037118, + "grad_norm": 2.43831729888916, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7325807213783264, + "num_tokens": 546866707.0, + "step": 21134 + }, + { + "epoch": 2.320997144739732, + "grad_norm": 2.4137825965881348, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7434864044189453, + "num_tokens": 546891770.0, + "step": 21135 + }, + { + "epoch": 2.3211069624423457, + "grad_norm": 2.6536340713500977, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7269161343574524, + "num_tokens": 546915119.0, + "step": 21136 + }, + { + "epoch": 2.3212167801449595, + "grad_norm": 2.2009012699127197, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7157468199729919, + "num_tokens": 546946013.0, + "step": 21137 + }, + { + "epoch": 2.3213265978475732, + "grad_norm": 2.4482672214508057, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7468072175979614, + "num_tokens": 546971462.0, + "step": 21138 + }, + { + "epoch": 2.3214364155501865, + "grad_norm": 2.2516160011291504, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7266839742660522, + "num_tokens": 547000981.0, + "step": 21139 + }, + { + "epoch": 2.3215462332528003, + "grad_norm": 2.5334506034851074, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7453880310058594, + "num_tokens": 547025222.0, + "step": 21140 + }, + { + "epoch": 2.321656050955414, + "grad_norm": 2.525282144546509, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7317991256713867, + "num_tokens": 547049261.0, + "step": 21141 + }, + { + "epoch": 2.321765868658028, + "grad_norm": 2.4628777503967285, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7463738918304443, + "num_tokens": 547071238.0, + "step": 21142 + }, + { + "epoch": 2.3218756863606416, + "grad_norm": 2.1508169174194336, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7188855409622192, + "num_tokens": 547102348.0, + "step": 21143 + }, + { + "epoch": 2.321985504063255, + "grad_norm": 2.2988924980163574, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7223731279373169, + "num_tokens": 547130826.0, + "step": 21144 + }, + { + "epoch": 2.3220953217658686, + "grad_norm": 2.711420774459839, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7392807006835938, + "num_tokens": 547152390.0, + "step": 21145 + }, + { + "epoch": 2.3222051394684824, + "grad_norm": 2.3909695148468018, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7268549799919128, + "num_tokens": 547176962.0, + "step": 21146 + }, + { + "epoch": 2.322314957171096, + "grad_norm": 2.329643726348877, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7324786186218262, + "num_tokens": 547201818.0, + "step": 21147 + }, + { + "epoch": 2.3224247748737095, + "grad_norm": 2.315535545349121, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7404054999351501, + "num_tokens": 547227966.0, + "step": 21148 + }, + { + "epoch": 2.322534592576323, + "grad_norm": 2.321928024291992, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7465195655822754, + "num_tokens": 547254343.0, + "step": 21149 + }, + { + "epoch": 2.322644410278937, + "grad_norm": 2.3602230548858643, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7385671138763428, + "num_tokens": 547281730.0, + "step": 21150 + }, + { + "epoch": 2.3227542279815507, + "grad_norm": 2.1360981464385986, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7243950366973877, + "num_tokens": 547310001.0, + "step": 21151 + }, + { + "epoch": 2.322864045684164, + "grad_norm": 2.715266466140747, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7492697834968567, + "num_tokens": 547331369.0, + "step": 21152 + }, + { + "epoch": 2.322973863386778, + "grad_norm": 2.320543050765991, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7313200235366821, + "num_tokens": 547357263.0, + "step": 21153 + }, + { + "epoch": 2.3230836810893916, + "grad_norm": 2.2316668033599854, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7119562029838562, + "num_tokens": 547386568.0, + "step": 21154 + }, + { + "epoch": 2.3231934987920053, + "grad_norm": 1.9564522504806519, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6853357553482056, + "num_tokens": 547422711.0, + "step": 21155 + }, + { + "epoch": 2.323303316494619, + "grad_norm": 2.3440983295440674, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7166591882705688, + "num_tokens": 547448999.0, + "step": 21156 + }, + { + "epoch": 2.3234131341972324, + "grad_norm": 2.3944170475006104, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7355089783668518, + "num_tokens": 547473272.0, + "step": 21157 + }, + { + "epoch": 2.323522951899846, + "grad_norm": 2.789005994796753, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7380310297012329, + "num_tokens": 547492875.0, + "step": 21158 + }, + { + "epoch": 2.32363276960246, + "grad_norm": 2.102994441986084, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7187249064445496, + "num_tokens": 547526202.0, + "step": 21159 + }, + { + "epoch": 2.3237425873050737, + "grad_norm": 2.4484055042266846, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7317252159118652, + "num_tokens": 547551740.0, + "step": 21160 + }, + { + "epoch": 2.3238524050076874, + "grad_norm": 2.5342085361480713, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.721320390701294, + "num_tokens": 547576042.0, + "step": 21161 + }, + { + "epoch": 2.3239622227103007, + "grad_norm": 2.207761526107788, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7170525789260864, + "num_tokens": 547605542.0, + "step": 21162 + }, + { + "epoch": 2.3240720404129145, + "grad_norm": 2.079789400100708, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7054417133331299, + "num_tokens": 547636849.0, + "step": 21163 + }, + { + "epoch": 2.3241818581155282, + "grad_norm": 2.527674674987793, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.719219982624054, + "num_tokens": 547661953.0, + "step": 21164 + }, + { + "epoch": 2.324291675818142, + "grad_norm": 2.4410598278045654, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7443366050720215, + "num_tokens": 547685654.0, + "step": 21165 + }, + { + "epoch": 2.3244014935207558, + "grad_norm": 2.342738628387451, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7167609333992004, + "num_tokens": 547712398.0, + "step": 21166 + }, + { + "epoch": 2.324511311223369, + "grad_norm": 2.37009596824646, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7339599132537842, + "num_tokens": 547737811.0, + "step": 21167 + }, + { + "epoch": 2.324621128925983, + "grad_norm": 2.48097825050354, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7305774688720703, + "num_tokens": 547762265.0, + "step": 21168 + }, + { + "epoch": 2.3247309466285966, + "grad_norm": 2.6325907707214355, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7437902092933655, + "num_tokens": 547783120.0, + "step": 21169 + }, + { + "epoch": 2.3248407643312103, + "grad_norm": 2.10386323928833, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7133601903915405, + "num_tokens": 547814700.0, + "step": 21170 + }, + { + "epoch": 2.324950582033824, + "grad_norm": 2.354511022567749, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7129232883453369, + "num_tokens": 547842383.0, + "step": 21171 + }, + { + "epoch": 2.3250603997364374, + "grad_norm": 2.3998215198516846, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7218193411827087, + "num_tokens": 547868389.0, + "step": 21172 + }, + { + "epoch": 2.325170217439051, + "grad_norm": 2.227851152420044, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7203853130340576, + "num_tokens": 547898639.0, + "step": 21173 + }, + { + "epoch": 2.325280035141665, + "grad_norm": 2.4203734397888184, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7362408638000488, + "num_tokens": 547925825.0, + "step": 21174 + }, + { + "epoch": 2.3253898528442787, + "grad_norm": 2.5239381790161133, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.73444664478302, + "num_tokens": 547949645.0, + "step": 21175 + }, + { + "epoch": 2.325499670546892, + "grad_norm": 2.595749855041504, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7187097072601318, + "num_tokens": 547974075.0, + "step": 21176 + }, + { + "epoch": 2.3256094882495058, + "grad_norm": 2.3965704441070557, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7312152981758118, + "num_tokens": 548000066.0, + "step": 21177 + }, + { + "epoch": 2.3257193059521195, + "grad_norm": 2.217794418334961, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7111226320266724, + "num_tokens": 548028726.0, + "step": 21178 + }, + { + "epoch": 2.3258291236547333, + "grad_norm": 2.4722211360931396, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7625064849853516, + "num_tokens": 548050967.0, + "step": 21179 + }, + { + "epoch": 2.3259389413573466, + "grad_norm": 2.4583170413970947, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7294070720672607, + "num_tokens": 548075738.0, + "step": 21180 + }, + { + "epoch": 2.3260487590599603, + "grad_norm": 2.5556414127349854, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7677086591720581, + "num_tokens": 548097259.0, + "step": 21181 + }, + { + "epoch": 2.326158576762574, + "grad_norm": 2.173823595046997, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7159132957458496, + "num_tokens": 548128313.0, + "step": 21182 + }, + { + "epoch": 2.326268394465188, + "grad_norm": 2.376391887664795, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7249538898468018, + "num_tokens": 548155143.0, + "step": 21183 + }, + { + "epoch": 2.3263782121678016, + "grad_norm": 2.713754892349243, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7379322052001953, + "num_tokens": 548176848.0, + "step": 21184 + }, + { + "epoch": 2.326488029870415, + "grad_norm": 2.8155081272125244, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7444943189620972, + "num_tokens": 548196503.0, + "step": 21185 + }, + { + "epoch": 2.3265978475730287, + "grad_norm": 2.8314433097839355, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.743465781211853, + "num_tokens": 548214433.0, + "step": 21186 + }, + { + "epoch": 2.3267076652756424, + "grad_norm": 2.422245502471924, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7137220501899719, + "num_tokens": 548242208.0, + "step": 21187 + }, + { + "epoch": 2.326817482978256, + "grad_norm": 2.385094165802002, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.731849730014801, + "num_tokens": 548267871.0, + "step": 21188 + }, + { + "epoch": 2.32692730068087, + "grad_norm": 2.897484540939331, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7385731935501099, + "num_tokens": 548285796.0, + "step": 21189 + }, + { + "epoch": 2.3270371183834833, + "grad_norm": 2.4025022983551025, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7242119312286377, + "num_tokens": 548314029.0, + "step": 21190 + }, + { + "epoch": 2.327146936086097, + "grad_norm": 2.4171340465545654, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7595044374465942, + "num_tokens": 548339195.0, + "step": 21191 + }, + { + "epoch": 2.3272567537887108, + "grad_norm": 2.4361748695373535, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7280585169792175, + "num_tokens": 548363915.0, + "step": 21192 + }, + { + "epoch": 2.3273665714913245, + "grad_norm": 2.256391763687134, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7258495688438416, + "num_tokens": 548392975.0, + "step": 21193 + }, + { + "epoch": 2.3274763891939383, + "grad_norm": 2.1158668994903564, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7058942317962646, + "num_tokens": 548423278.0, + "step": 21194 + }, + { + "epoch": 2.3275862068965516, + "grad_norm": 2.360140562057495, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7356735467910767, + "num_tokens": 548451510.0, + "step": 21195 + }, + { + "epoch": 2.3276960245991654, + "grad_norm": 2.420300245285034, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7392635345458984, + "num_tokens": 548477218.0, + "step": 21196 + }, + { + "epoch": 2.327805842301779, + "grad_norm": 2.3276023864746094, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7491560578346252, + "num_tokens": 548503026.0, + "step": 21197 + }, + { + "epoch": 2.327915660004393, + "grad_norm": 2.617384433746338, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7394512295722961, + "num_tokens": 548525537.0, + "step": 21198 + }, + { + "epoch": 2.328025477707006, + "grad_norm": 2.2834322452545166, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7170155048370361, + "num_tokens": 548554097.0, + "step": 21199 + }, + { + "epoch": 2.32813529540962, + "grad_norm": 2.471895694732666, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7350630760192871, + "num_tokens": 548578518.0, + "step": 21200 + }, + { + "epoch": 2.3282451131122337, + "grad_norm": 2.363285541534424, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7349820137023926, + "num_tokens": 548603344.0, + "step": 21201 + }, + { + "epoch": 2.3283549308148475, + "grad_norm": 2.2222611904144287, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7470158338546753, + "num_tokens": 548627826.0, + "step": 21202 + }, + { + "epoch": 2.3284647485174608, + "grad_norm": 2.3865890502929688, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7410979270935059, + "num_tokens": 548652117.0, + "step": 21203 + }, + { + "epoch": 2.3285745662200745, + "grad_norm": 2.4092907905578613, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7496732473373413, + "num_tokens": 548677237.0, + "step": 21204 + }, + { + "epoch": 2.3286843839226883, + "grad_norm": 2.0948092937469482, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7264410853385925, + "num_tokens": 548709854.0, + "step": 21205 + }, + { + "epoch": 2.328794201625302, + "grad_norm": 2.357545852661133, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7219853401184082, + "num_tokens": 548736894.0, + "step": 21206 + }, + { + "epoch": 2.328904019327916, + "grad_norm": 2.5036654472351074, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7405552864074707, + "num_tokens": 548758761.0, + "step": 21207 + }, + { + "epoch": 2.329013837030529, + "grad_norm": 2.4437971115112305, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7245926856994629, + "num_tokens": 548783443.0, + "step": 21208 + }, + { + "epoch": 2.329123654733143, + "grad_norm": 2.379857301712036, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7182071805000305, + "num_tokens": 548809750.0, + "step": 21209 + }, + { + "epoch": 2.3292334724357566, + "grad_norm": 2.119959592819214, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7274311780929565, + "num_tokens": 548839854.0, + "step": 21210 + }, + { + "epoch": 2.3293432901383704, + "grad_norm": 2.380763053894043, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7548998594284058, + "num_tokens": 548864134.0, + "step": 21211 + }, + { + "epoch": 2.329453107840984, + "grad_norm": 2.461061477661133, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7053847312927246, + "num_tokens": 548893718.0, + "step": 21212 + }, + { + "epoch": 2.3295629255435975, + "grad_norm": 2.373319625854492, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7194143533706665, + "num_tokens": 548920479.0, + "step": 21213 + }, + { + "epoch": 2.329672743246211, + "grad_norm": 2.598881244659424, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7271969318389893, + "num_tokens": 548944177.0, + "step": 21214 + }, + { + "epoch": 2.329782560948825, + "grad_norm": 2.536548137664795, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7168968915939331, + "num_tokens": 548969981.0, + "step": 21215 + }, + { + "epoch": 2.3298923786514387, + "grad_norm": 2.669398546218872, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.720137357711792, + "num_tokens": 548991299.0, + "step": 21216 + }, + { + "epoch": 2.3300021963540525, + "grad_norm": 2.57491397857666, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7357766628265381, + "num_tokens": 549013991.0, + "step": 21217 + }, + { + "epoch": 2.330112014056666, + "grad_norm": 2.3580892086029053, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7620318531990051, + "num_tokens": 549038032.0, + "step": 21218 + }, + { + "epoch": 2.3302218317592795, + "grad_norm": 2.35404372215271, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7075309753417969, + "num_tokens": 549064922.0, + "step": 21219 + }, + { + "epoch": 2.3303316494618933, + "grad_norm": 2.0406486988067627, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.697770357131958, + "num_tokens": 549100841.0, + "step": 21220 + }, + { + "epoch": 2.330441467164507, + "grad_norm": 2.5146195888519287, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7182995676994324, + "num_tokens": 549125312.0, + "step": 21221 + }, + { + "epoch": 2.330551284867121, + "grad_norm": 2.794321298599243, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7400898337364197, + "num_tokens": 549144977.0, + "step": 21222 + }, + { + "epoch": 2.330661102569734, + "grad_norm": 2.123598575592041, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.6979982852935791, + "num_tokens": 549178168.0, + "step": 21223 + }, + { + "epoch": 2.330770920272348, + "grad_norm": 2.4151086807250977, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7374986410140991, + "num_tokens": 549203052.0, + "step": 21224 + }, + { + "epoch": 2.3308807379749616, + "grad_norm": 2.4460384845733643, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7598536014556885, + "num_tokens": 549226087.0, + "step": 21225 + }, + { + "epoch": 2.3309905556775754, + "grad_norm": 2.3760342597961426, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7551523447036743, + "num_tokens": 549251485.0, + "step": 21226 + }, + { + "epoch": 2.3311003733801887, + "grad_norm": 2.2588422298431396, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7186081409454346, + "num_tokens": 549280757.0, + "step": 21227 + }, + { + "epoch": 2.3312101910828025, + "grad_norm": 2.099224328994751, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7302483320236206, + "num_tokens": 549311095.0, + "step": 21228 + }, + { + "epoch": 2.3313200087854162, + "grad_norm": 2.428365707397461, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7359414100646973, + "num_tokens": 549336923.0, + "step": 21229 + }, + { + "epoch": 2.33142982648803, + "grad_norm": 2.2204782962799072, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7161632180213928, + "num_tokens": 549367414.0, + "step": 21230 + }, + { + "epoch": 2.3315396441906433, + "grad_norm": 2.587254047393799, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7418258190155029, + "num_tokens": 549390980.0, + "step": 21231 + }, + { + "epoch": 2.331649461893257, + "grad_norm": 2.4386484622955322, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.738128662109375, + "num_tokens": 549415582.0, + "step": 21232 + }, + { + "epoch": 2.331759279595871, + "grad_norm": 2.3016672134399414, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7076509594917297, + "num_tokens": 549442023.0, + "step": 21233 + }, + { + "epoch": 2.3318690972984846, + "grad_norm": 2.250802993774414, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7441407442092896, + "num_tokens": 549468923.0, + "step": 21234 + }, + { + "epoch": 2.3319789150010983, + "grad_norm": 2.624516010284424, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7171444892883301, + "num_tokens": 549493366.0, + "step": 21235 + }, + { + "epoch": 2.3320887327037116, + "grad_norm": 2.2013015747070312, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7236452102661133, + "num_tokens": 549521336.0, + "step": 21236 + }, + { + "epoch": 2.3321985504063254, + "grad_norm": 2.3813507556915283, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7146252393722534, + "num_tokens": 549547172.0, + "step": 21237 + }, + { + "epoch": 2.332308368108939, + "grad_norm": 2.8141262531280518, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7538593411445618, + "num_tokens": 549566702.0, + "step": 21238 + }, + { + "epoch": 2.332418185811553, + "grad_norm": 2.366584300994873, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7497293949127197, + "num_tokens": 549591628.0, + "step": 21239 + }, + { + "epoch": 2.3325280035141667, + "grad_norm": 2.2665657997131348, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.718777596950531, + "num_tokens": 549618880.0, + "step": 21240 + }, + { + "epoch": 2.33263782121678, + "grad_norm": 1.9457454681396484, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7275190353393555, + "num_tokens": 549652810.0, + "step": 21241 + }, + { + "epoch": 2.3327476389193937, + "grad_norm": 2.7757840156555176, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7719415426254272, + "num_tokens": 549671444.0, + "step": 21242 + }, + { + "epoch": 2.3328574566220075, + "grad_norm": 2.8015100955963135, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7530261278152466, + "num_tokens": 549691769.0, + "step": 21243 + }, + { + "epoch": 2.3329672743246213, + "grad_norm": 2.5228147506713867, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7311753034591675, + "num_tokens": 549715034.0, + "step": 21244 + }, + { + "epoch": 2.333077092027235, + "grad_norm": 2.273203134536743, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7354779243469238, + "num_tokens": 549740360.0, + "step": 21245 + }, + { + "epoch": 2.3331869097298483, + "grad_norm": 2.3101699352264404, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7575552463531494, + "num_tokens": 549766285.0, + "step": 21246 + }, + { + "epoch": 2.333296727432462, + "grad_norm": 2.140162467956543, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7094868421554565, + "num_tokens": 549796952.0, + "step": 21247 + }, + { + "epoch": 2.333406545135076, + "grad_norm": 2.691042184829712, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7515789270401001, + "num_tokens": 549816546.0, + "step": 21248 + }, + { + "epoch": 2.3335163628376896, + "grad_norm": 2.378105640411377, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7362651824951172, + "num_tokens": 549840980.0, + "step": 21249 + }, + { + "epoch": 2.333626180540303, + "grad_norm": 2.5062978267669678, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7099494934082031, + "num_tokens": 549865581.0, + "step": 21250 + }, + { + "epoch": 2.3337359982429167, + "grad_norm": 2.3903021812438965, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7291476726531982, + "num_tokens": 549893505.0, + "step": 21251 + }, + { + "epoch": 2.3338458159455304, + "grad_norm": 2.7048423290252686, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7399290800094604, + "num_tokens": 549913166.0, + "step": 21252 + }, + { + "epoch": 2.333955633648144, + "grad_norm": 2.542994976043701, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7312242984771729, + "num_tokens": 549936725.0, + "step": 21253 + }, + { + "epoch": 2.334065451350758, + "grad_norm": 2.230754852294922, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7356340289115906, + "num_tokens": 549965484.0, + "step": 21254 + }, + { + "epoch": 2.3341752690533712, + "grad_norm": 2.2840611934661865, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7337278127670288, + "num_tokens": 549990882.0, + "step": 21255 + }, + { + "epoch": 2.334285086755985, + "grad_norm": 2.3047380447387695, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7085496783256531, + "num_tokens": 550019734.0, + "step": 21256 + }, + { + "epoch": 2.3343949044585988, + "grad_norm": 2.6087393760681152, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7509050965309143, + "num_tokens": 550040980.0, + "step": 21257 + }, + { + "epoch": 2.3345047221612125, + "grad_norm": 2.3160340785980225, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7198976278305054, + "num_tokens": 550065783.0, + "step": 21258 + }, + { + "epoch": 2.334614539863826, + "grad_norm": 2.1656360626220703, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7490150332450867, + "num_tokens": 550094010.0, + "step": 21259 + }, + { + "epoch": 2.3347243575664396, + "grad_norm": 2.740640640258789, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7249720096588135, + "num_tokens": 550114344.0, + "step": 21260 + }, + { + "epoch": 2.3348341752690533, + "grad_norm": 2.43467378616333, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7267752885818481, + "num_tokens": 550142710.0, + "step": 21261 + }, + { + "epoch": 2.334943992971667, + "grad_norm": 2.3042662143707275, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7410716414451599, + "num_tokens": 550168823.0, + "step": 21262 + }, + { + "epoch": 2.335053810674281, + "grad_norm": 2.359999656677246, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7331767082214355, + "num_tokens": 550193547.0, + "step": 21263 + }, + { + "epoch": 2.335163628376894, + "grad_norm": 2.3236663341522217, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7345709204673767, + "num_tokens": 550219685.0, + "step": 21264 + }, + { + "epoch": 2.335273446079508, + "grad_norm": 2.190474510192871, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7153623104095459, + "num_tokens": 550248353.0, + "step": 21265 + }, + { + "epoch": 2.3353832637821217, + "grad_norm": 2.166952610015869, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7405297756195068, + "num_tokens": 550278798.0, + "step": 21266 + }, + { + "epoch": 2.3354930814847354, + "grad_norm": 2.3706109523773193, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7182092666625977, + "num_tokens": 550306111.0, + "step": 21267 + }, + { + "epoch": 2.335602899187349, + "grad_norm": 2.2408626079559326, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7427153587341309, + "num_tokens": 550333461.0, + "step": 21268 + }, + { + "epoch": 2.3357127168899625, + "grad_norm": 2.5890910625457764, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7783346772193909, + "num_tokens": 550354430.0, + "step": 21269 + }, + { + "epoch": 2.3358225345925763, + "grad_norm": 2.512535333633423, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7293033003807068, + "num_tokens": 550379375.0, + "step": 21270 + }, + { + "epoch": 2.33593235229519, + "grad_norm": 2.1547458171844482, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.718510627746582, + "num_tokens": 550408939.0, + "step": 21271 + }, + { + "epoch": 2.336042169997804, + "grad_norm": 2.6359076499938965, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7365773916244507, + "num_tokens": 550430170.0, + "step": 21272 + }, + { + "epoch": 2.3361519877004175, + "grad_norm": 2.522858142852783, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.755603551864624, + "num_tokens": 550453836.0, + "step": 21273 + }, + { + "epoch": 2.336261805403031, + "grad_norm": 2.0727691650390625, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7214375138282776, + "num_tokens": 550485340.0, + "step": 21274 + }, + { + "epoch": 2.3363716231056446, + "grad_norm": 2.077634811401367, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7049343585968018, + "num_tokens": 550520121.0, + "step": 21275 + }, + { + "epoch": 2.3364814408082584, + "grad_norm": 2.5412850379943848, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7384362816810608, + "num_tokens": 550545685.0, + "step": 21276 + }, + { + "epoch": 2.336591258510872, + "grad_norm": 2.197395086288452, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6960382461547852, + "num_tokens": 550576946.0, + "step": 21277 + }, + { + "epoch": 2.3367010762134854, + "grad_norm": 2.5517890453338623, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7504196166992188, + "num_tokens": 550598771.0, + "step": 21278 + }, + { + "epoch": 2.336810893916099, + "grad_norm": 2.3912525177001953, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.723118007183075, + "num_tokens": 550625923.0, + "step": 21279 + }, + { + "epoch": 2.336920711618713, + "grad_norm": 2.2197117805480957, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7218976020812988, + "num_tokens": 550654256.0, + "step": 21280 + }, + { + "epoch": 2.3370305293213267, + "grad_norm": 2.1425399780273438, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7171944379806519, + "num_tokens": 550684929.0, + "step": 21281 + }, + { + "epoch": 2.33714034702394, + "grad_norm": 2.3231418132781982, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7215496301651001, + "num_tokens": 550712460.0, + "step": 21282 + }, + { + "epoch": 2.337250164726554, + "grad_norm": 2.232501268386841, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7317453622817993, + "num_tokens": 550740829.0, + "step": 21283 + }, + { + "epoch": 2.3373599824291675, + "grad_norm": 2.373352289199829, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7316170930862427, + "num_tokens": 550767310.0, + "step": 21284 + }, + { + "epoch": 2.3374698001317813, + "grad_norm": 2.233424425125122, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7437254190444946, + "num_tokens": 550793297.0, + "step": 21285 + }, + { + "epoch": 2.337579617834395, + "grad_norm": 2.509803056716919, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7407995462417603, + "num_tokens": 550814235.0, + "step": 21286 + }, + { + "epoch": 2.3376894355370084, + "grad_norm": 2.421121120452881, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7536380290985107, + "num_tokens": 550838545.0, + "step": 21287 + }, + { + "epoch": 2.337799253239622, + "grad_norm": 2.57399845123291, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7302955389022827, + "num_tokens": 550861019.0, + "step": 21288 + }, + { + "epoch": 2.337909070942236, + "grad_norm": 2.465043544769287, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7324731349945068, + "num_tokens": 550884329.0, + "step": 21289 + }, + { + "epoch": 2.3380188886448496, + "grad_norm": 2.156090259552002, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7189037799835205, + "num_tokens": 550914916.0, + "step": 21290 + }, + { + "epoch": 2.3381287063474634, + "grad_norm": 2.675718069076538, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7437614798545837, + "num_tokens": 550937653.0, + "step": 21291 + }, + { + "epoch": 2.3382385240500767, + "grad_norm": 2.318772077560425, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7211464047431946, + "num_tokens": 550964748.0, + "step": 21292 + }, + { + "epoch": 2.3383483417526905, + "grad_norm": 2.4897286891937256, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7340220212936401, + "num_tokens": 550988746.0, + "step": 21293 + }, + { + "epoch": 2.338458159455304, + "grad_norm": 2.346256732940674, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7440769076347351, + "num_tokens": 551012690.0, + "step": 21294 + }, + { + "epoch": 2.338567977157918, + "grad_norm": 2.111077308654785, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7381390333175659, + "num_tokens": 551046199.0, + "step": 21295 + }, + { + "epoch": 2.3386777948605317, + "grad_norm": 2.299982786178589, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7250896692276001, + "num_tokens": 551072392.0, + "step": 21296 + }, + { + "epoch": 2.338787612563145, + "grad_norm": 2.491975784301758, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.719304084777832, + "num_tokens": 551096181.0, + "step": 21297 + }, + { + "epoch": 2.338897430265759, + "grad_norm": 2.3204970359802246, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6978317499160767, + "num_tokens": 551126784.0, + "step": 21298 + }, + { + "epoch": 2.3390072479683726, + "grad_norm": 2.189173460006714, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7182080149650574, + "num_tokens": 551157975.0, + "step": 21299 + }, + { + "epoch": 2.3391170656709863, + "grad_norm": 2.1999356746673584, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.714693009853363, + "num_tokens": 551188296.0, + "step": 21300 + }, + { + "epoch": 2.3392268833735996, + "grad_norm": 2.5286080837249756, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7493290901184082, + "num_tokens": 551209808.0, + "step": 21301 + }, + { + "epoch": 2.3393367010762134, + "grad_norm": 2.1950247287750244, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7291884422302246, + "num_tokens": 551237982.0, + "step": 21302 + }, + { + "epoch": 2.339446518778827, + "grad_norm": 2.3556573390960693, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7359738349914551, + "num_tokens": 551263064.0, + "step": 21303 + }, + { + "epoch": 2.339556336481441, + "grad_norm": 2.3666107654571533, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7408596873283386, + "num_tokens": 551290330.0, + "step": 21304 + }, + { + "epoch": 2.3396661541840547, + "grad_norm": 2.1261239051818848, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7374657392501831, + "num_tokens": 551324588.0, + "step": 21305 + }, + { + "epoch": 2.339775971886668, + "grad_norm": 2.561915636062622, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7279241681098938, + "num_tokens": 551346696.0, + "step": 21306 + }, + { + "epoch": 2.3398857895892817, + "grad_norm": 2.322784423828125, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7076767086982727, + "num_tokens": 551374637.0, + "step": 21307 + }, + { + "epoch": 2.3399956072918955, + "grad_norm": 2.244115114212036, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7267187237739563, + "num_tokens": 551402293.0, + "step": 21308 + }, + { + "epoch": 2.3401054249945092, + "grad_norm": 2.284867763519287, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7188611030578613, + "num_tokens": 551430867.0, + "step": 21309 + }, + { + "epoch": 2.3402152426971226, + "grad_norm": 2.535836696624756, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7138112187385559, + "num_tokens": 551456102.0, + "step": 21310 + }, + { + "epoch": 2.3403250603997363, + "grad_norm": 2.5347015857696533, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7286708354949951, + "num_tokens": 551479576.0, + "step": 21311 + }, + { + "epoch": 2.34043487810235, + "grad_norm": 2.1447370052337646, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7638531923294067, + "num_tokens": 551508248.0, + "step": 21312 + }, + { + "epoch": 2.340544695804964, + "grad_norm": 2.630697250366211, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7272821068763733, + "num_tokens": 551529376.0, + "step": 21313 + }, + { + "epoch": 2.3406545135075776, + "grad_norm": 2.322535276412964, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7397459745407104, + "num_tokens": 551554610.0, + "step": 21314 + }, + { + "epoch": 2.340764331210191, + "grad_norm": 2.6533780097961426, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7325626611709595, + "num_tokens": 551579653.0, + "step": 21315 + }, + { + "epoch": 2.3408741489128047, + "grad_norm": 2.400653839111328, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7385461330413818, + "num_tokens": 551603025.0, + "step": 21316 + }, + { + "epoch": 2.3409839666154184, + "grad_norm": 2.3081986904144287, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7272424697875977, + "num_tokens": 551631007.0, + "step": 21317 + }, + { + "epoch": 2.341093784318032, + "grad_norm": 2.2124955654144287, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.6970789432525635, + "num_tokens": 551662782.0, + "step": 21318 + }, + { + "epoch": 2.341203602020646, + "grad_norm": 2.7561981678009033, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7323787212371826, + "num_tokens": 551683837.0, + "step": 21319 + }, + { + "epoch": 2.3413134197232592, + "grad_norm": 2.5417118072509766, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7581642866134644, + "num_tokens": 551705496.0, + "step": 21320 + }, + { + "epoch": 2.341423237425873, + "grad_norm": 2.6259307861328125, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7260435819625854, + "num_tokens": 551728394.0, + "step": 21321 + }, + { + "epoch": 2.3415330551284868, + "grad_norm": 2.665820837020874, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7466173768043518, + "num_tokens": 551748347.0, + "step": 21322 + }, + { + "epoch": 2.3416428728311005, + "grad_norm": 2.475827693939209, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7422679662704468, + "num_tokens": 551771301.0, + "step": 21323 + }, + { + "epoch": 2.3417526905337143, + "grad_norm": 2.6006953716278076, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7424362897872925, + "num_tokens": 551791860.0, + "step": 21324 + }, + { + "epoch": 2.3418625082363276, + "grad_norm": 2.243839740753174, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7317131757736206, + "num_tokens": 551820556.0, + "step": 21325 + }, + { + "epoch": 2.3419723259389413, + "grad_norm": 2.4094693660736084, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7409577369689941, + "num_tokens": 551845083.0, + "step": 21326 + }, + { + "epoch": 2.342082143641555, + "grad_norm": 2.582575559616089, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7576123476028442, + "num_tokens": 551866820.0, + "step": 21327 + }, + { + "epoch": 2.342191961344169, + "grad_norm": 2.4245448112487793, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7389183044433594, + "num_tokens": 551890731.0, + "step": 21328 + }, + { + "epoch": 2.342301779046782, + "grad_norm": 2.218261480331421, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7215088605880737, + "num_tokens": 551919352.0, + "step": 21329 + }, + { + "epoch": 2.342411596749396, + "grad_norm": 2.40828537940979, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7223340272903442, + "num_tokens": 551946592.0, + "step": 21330 + }, + { + "epoch": 2.3425214144520097, + "grad_norm": 2.406053304672241, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.721792995929718, + "num_tokens": 551972924.0, + "step": 21331 + }, + { + "epoch": 2.3426312321546234, + "grad_norm": 2.558176040649414, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7554237842559814, + "num_tokens": 551994666.0, + "step": 21332 + }, + { + "epoch": 2.3427410498572367, + "grad_norm": 2.385021448135376, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7171253561973572, + "num_tokens": 552021422.0, + "step": 21333 + }, + { + "epoch": 2.3428508675598505, + "grad_norm": 2.247032880783081, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7268191576004028, + "num_tokens": 552051876.0, + "step": 21334 + }, + { + "epoch": 2.3429606852624643, + "grad_norm": 2.5610551834106445, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7433657050132751, + "num_tokens": 552074281.0, + "step": 21335 + }, + { + "epoch": 2.343070502965078, + "grad_norm": 2.3169944286346436, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7343578338623047, + "num_tokens": 552101939.0, + "step": 21336 + }, + { + "epoch": 2.3431803206676918, + "grad_norm": 2.1681668758392334, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7359999418258667, + "num_tokens": 552133240.0, + "step": 21337 + }, + { + "epoch": 2.343290138370305, + "grad_norm": 2.2279043197631836, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7453044652938843, + "num_tokens": 552160652.0, + "step": 21338 + }, + { + "epoch": 2.343399956072919, + "grad_norm": 2.311227798461914, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7238712906837463, + "num_tokens": 552188326.0, + "step": 21339 + }, + { + "epoch": 2.3435097737755326, + "grad_norm": 2.4901845455169678, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7508950233459473, + "num_tokens": 552211141.0, + "step": 21340 + }, + { + "epoch": 2.3436195914781464, + "grad_norm": 2.551302194595337, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7317551374435425, + "num_tokens": 552235120.0, + "step": 21341 + }, + { + "epoch": 2.34372940918076, + "grad_norm": 2.5040786266326904, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7470476627349854, + "num_tokens": 552257858.0, + "step": 21342 + }, + { + "epoch": 2.3438392268833734, + "grad_norm": 2.250861644744873, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7386336326599121, + "num_tokens": 552286546.0, + "step": 21343 + }, + { + "epoch": 2.343949044585987, + "grad_norm": 2.3607616424560547, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7541694641113281, + "num_tokens": 552308667.0, + "step": 21344 + }, + { + "epoch": 2.344058862288601, + "grad_norm": 2.1448447704315186, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6919804811477661, + "num_tokens": 552342909.0, + "step": 21345 + }, + { + "epoch": 2.3441686799912147, + "grad_norm": 2.272368907928467, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7529087662696838, + "num_tokens": 552368696.0, + "step": 21346 + }, + { + "epoch": 2.3442784976938285, + "grad_norm": 2.3057737350463867, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7068898677825928, + "num_tokens": 552395639.0, + "step": 21347 + }, + { + "epoch": 2.3443883153964418, + "grad_norm": 2.506389617919922, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7196357250213623, + "num_tokens": 552420544.0, + "step": 21348 + }, + { + "epoch": 2.3444981330990555, + "grad_norm": 2.7968170642852783, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7447300553321838, + "num_tokens": 552439038.0, + "step": 21349 + }, + { + "epoch": 2.3446079508016693, + "grad_norm": 2.3813059329986572, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7352343797683716, + "num_tokens": 552464579.0, + "step": 21350 + }, + { + "epoch": 2.344717768504283, + "grad_norm": 2.15629243850708, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7250458598136902, + "num_tokens": 552495383.0, + "step": 21351 + }, + { + "epoch": 2.344827586206897, + "grad_norm": 2.3000895977020264, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7214611768722534, + "num_tokens": 552520115.0, + "step": 21352 + }, + { + "epoch": 2.34493740390951, + "grad_norm": 2.648453950881958, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7360063791275024, + "num_tokens": 552541490.0, + "step": 21353 + }, + { + "epoch": 2.345047221612124, + "grad_norm": 2.1386377811431885, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7260626554489136, + "num_tokens": 552571728.0, + "step": 21354 + }, + { + "epoch": 2.3451570393147376, + "grad_norm": 2.183096170425415, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.733786940574646, + "num_tokens": 552599496.0, + "step": 21355 + }, + { + "epoch": 2.3452668570173514, + "grad_norm": 2.3352324962615967, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7198638916015625, + "num_tokens": 552627444.0, + "step": 21356 + }, + { + "epoch": 2.3453766747199647, + "grad_norm": 2.4740378856658936, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7328193187713623, + "num_tokens": 552651759.0, + "step": 21357 + }, + { + "epoch": 2.3454864924225785, + "grad_norm": 2.083915948867798, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7347803115844727, + "num_tokens": 552684943.0, + "step": 21358 + }, + { + "epoch": 2.345596310125192, + "grad_norm": 2.485152244567871, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7304224371910095, + "num_tokens": 552708085.0, + "step": 21359 + }, + { + "epoch": 2.345706127827806, + "grad_norm": 2.2046937942504883, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7152079343795776, + "num_tokens": 552738251.0, + "step": 21360 + }, + { + "epoch": 2.3458159455304193, + "grad_norm": 2.1631813049316406, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7380557656288147, + "num_tokens": 552770263.0, + "step": 21361 + }, + { + "epoch": 2.345925763233033, + "grad_norm": 2.171894073486328, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.714984118938446, + "num_tokens": 552799588.0, + "step": 21362 + }, + { + "epoch": 2.346035580935647, + "grad_norm": 2.275388240814209, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.728771448135376, + "num_tokens": 552825885.0, + "step": 21363 + }, + { + "epoch": 2.3461453986382605, + "grad_norm": 2.6122868061065674, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7228856682777405, + "num_tokens": 552849550.0, + "step": 21364 + }, + { + "epoch": 2.3462552163408743, + "grad_norm": 2.793401002883911, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7490202784538269, + "num_tokens": 552870271.0, + "step": 21365 + }, + { + "epoch": 2.3463650340434876, + "grad_norm": 2.1626288890838623, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7442018985748291, + "num_tokens": 552897487.0, + "step": 21366 + }, + { + "epoch": 2.3464748517461014, + "grad_norm": 2.190596103668213, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7153875827789307, + "num_tokens": 552931433.0, + "step": 21367 + }, + { + "epoch": 2.346584669448715, + "grad_norm": 2.611086845397949, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7361874580383301, + "num_tokens": 552952389.0, + "step": 21368 + }, + { + "epoch": 2.346694487151329, + "grad_norm": 2.2685000896453857, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7291287183761597, + "num_tokens": 552980525.0, + "step": 21369 + }, + { + "epoch": 2.3468043048539426, + "grad_norm": 2.2638823986053467, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7285903692245483, + "num_tokens": 553006900.0, + "step": 21370 + }, + { + "epoch": 2.346914122556556, + "grad_norm": 2.3148369789123535, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7244850397109985, + "num_tokens": 553032119.0, + "step": 21371 + }, + { + "epoch": 2.3470239402591697, + "grad_norm": 2.7351796627044678, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7558009028434753, + "num_tokens": 553051757.0, + "step": 21372 + }, + { + "epoch": 2.3471337579617835, + "grad_norm": 3.0050501823425293, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7559692859649658, + "num_tokens": 553068911.0, + "step": 21373 + }, + { + "epoch": 2.3472435756643972, + "grad_norm": 2.407999277114868, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7424125075340271, + "num_tokens": 553091522.0, + "step": 21374 + }, + { + "epoch": 2.347353393367011, + "grad_norm": 2.6374635696411133, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7260677218437195, + "num_tokens": 553114261.0, + "step": 21375 + }, + { + "epoch": 2.3474632110696243, + "grad_norm": 2.3874142169952393, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7235031723976135, + "num_tokens": 553138696.0, + "step": 21376 + }, + { + "epoch": 2.347573028772238, + "grad_norm": 2.5357086658477783, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7118555903434753, + "num_tokens": 553164347.0, + "step": 21377 + }, + { + "epoch": 2.347682846474852, + "grad_norm": 2.0845835208892822, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7068578004837036, + "num_tokens": 553196295.0, + "step": 21378 + }, + { + "epoch": 2.3477926641774656, + "grad_norm": 2.2808284759521484, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7210357785224915, + "num_tokens": 553224688.0, + "step": 21379 + }, + { + "epoch": 2.347902481880079, + "grad_norm": 2.48807692527771, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7357831001281738, + "num_tokens": 553248386.0, + "step": 21380 + }, + { + "epoch": 2.3480122995826926, + "grad_norm": 2.368565320968628, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7197545766830444, + "num_tokens": 553275638.0, + "step": 21381 + }, + { + "epoch": 2.3481221172853064, + "grad_norm": 2.679395914077759, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7604019641876221, + "num_tokens": 553294839.0, + "step": 21382 + }, + { + "epoch": 2.34823193498792, + "grad_norm": 2.5955793857574463, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.75297611951828, + "num_tokens": 553316714.0, + "step": 21383 + }, + { + "epoch": 2.3483417526905335, + "grad_norm": 2.1575632095336914, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.721266508102417, + "num_tokens": 553346341.0, + "step": 21384 + }, + { + "epoch": 2.3484515703931472, + "grad_norm": 2.521329879760742, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7248635292053223, + "num_tokens": 553368820.0, + "step": 21385 + }, + { + "epoch": 2.348561388095761, + "grad_norm": 2.565485715866089, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7528849840164185, + "num_tokens": 553390215.0, + "step": 21386 + }, + { + "epoch": 2.3486712057983747, + "grad_norm": 2.3989474773406982, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7265114784240723, + "num_tokens": 553416577.0, + "step": 21387 + }, + { + "epoch": 2.3487810235009885, + "grad_norm": 2.2591657638549805, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7442554831504822, + "num_tokens": 553443738.0, + "step": 21388 + }, + { + "epoch": 2.348890841203602, + "grad_norm": 2.603693723678589, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7231768369674683, + "num_tokens": 553465023.0, + "step": 21389 + }, + { + "epoch": 2.3490006589062156, + "grad_norm": 2.5318071842193604, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7121611833572388, + "num_tokens": 553489967.0, + "step": 21390 + }, + { + "epoch": 2.3491104766088293, + "grad_norm": 2.318480968475342, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7445933818817139, + "num_tokens": 553516728.0, + "step": 21391 + }, + { + "epoch": 2.349220294311443, + "grad_norm": 2.499516725540161, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.743804931640625, + "num_tokens": 553538784.0, + "step": 21392 + }, + { + "epoch": 2.349330112014057, + "grad_norm": 2.2128779888153076, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.694063127040863, + "num_tokens": 553568410.0, + "step": 21393 + }, + { + "epoch": 2.34943992971667, + "grad_norm": 2.202979326248169, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7491766214370728, + "num_tokens": 553597423.0, + "step": 21394 + }, + { + "epoch": 2.349549747419284, + "grad_norm": 2.5809922218322754, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7333956956863403, + "num_tokens": 553619223.0, + "step": 21395 + }, + { + "epoch": 2.3496595651218977, + "grad_norm": 2.1051058769226074, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.718761682510376, + "num_tokens": 553648604.0, + "step": 21396 + }, + { + "epoch": 2.3497693828245114, + "grad_norm": 2.774688243865967, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7430893778800964, + "num_tokens": 553668209.0, + "step": 21397 + }, + { + "epoch": 2.349879200527125, + "grad_norm": 2.3384480476379395, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.733177661895752, + "num_tokens": 553693394.0, + "step": 21398 + }, + { + "epoch": 2.3499890182297385, + "grad_norm": 2.644714832305908, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7349607348442078, + "num_tokens": 553714622.0, + "step": 21399 + }, + { + "epoch": 2.3500988359323522, + "grad_norm": 2.389704465866089, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7541955709457397, + "num_tokens": 553738769.0, + "step": 21400 + }, + { + "epoch": 2.350208653634966, + "grad_norm": 2.541757345199585, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7069628834724426, + "num_tokens": 553762917.0, + "step": 21401 + }, + { + "epoch": 2.3503184713375798, + "grad_norm": 1.997869849205017, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7237038612365723, + "num_tokens": 553798724.0, + "step": 21402 + }, + { + "epoch": 2.3504282890401935, + "grad_norm": 2.7579164505004883, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7372250556945801, + "num_tokens": 553819757.0, + "step": 21403 + }, + { + "epoch": 2.350538106742807, + "grad_norm": 2.3612189292907715, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7341295480728149, + "num_tokens": 553846996.0, + "step": 21404 + }, + { + "epoch": 2.3506479244454206, + "grad_norm": 2.4896507263183594, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7220703959465027, + "num_tokens": 553871149.0, + "step": 21405 + }, + { + "epoch": 2.3507577421480343, + "grad_norm": 2.139817237854004, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7189415693283081, + "num_tokens": 553904582.0, + "step": 21406 + }, + { + "epoch": 2.350867559850648, + "grad_norm": 2.6944689750671387, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7474178075790405, + "num_tokens": 553925932.0, + "step": 21407 + }, + { + "epoch": 2.3509773775532614, + "grad_norm": 2.43839693069458, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7503650784492493, + "num_tokens": 553950760.0, + "step": 21408 + }, + { + "epoch": 2.351087195255875, + "grad_norm": 2.202552080154419, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7125678062438965, + "num_tokens": 553980795.0, + "step": 21409 + }, + { + "epoch": 2.351197012958489, + "grad_norm": 2.280898094177246, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7558529376983643, + "num_tokens": 554007473.0, + "step": 21410 + }, + { + "epoch": 2.3513068306611027, + "grad_norm": 2.1619811058044434, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7159102559089661, + "num_tokens": 554036706.0, + "step": 21411 + }, + { + "epoch": 2.351416648363716, + "grad_norm": 2.788471221923828, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7665873169898987, + "num_tokens": 554055256.0, + "step": 21412 + }, + { + "epoch": 2.3515264660663298, + "grad_norm": 2.4632315635681152, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7328851222991943, + "num_tokens": 554081341.0, + "step": 21413 + }, + { + "epoch": 2.3516362837689435, + "grad_norm": 2.1060726642608643, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7287525534629822, + "num_tokens": 554110544.0, + "step": 21414 + }, + { + "epoch": 2.3517461014715573, + "grad_norm": 2.3000917434692383, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7305547595024109, + "num_tokens": 554136803.0, + "step": 21415 + }, + { + "epoch": 2.351855919174171, + "grad_norm": 2.1654906272888184, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7244359254837036, + "num_tokens": 554164323.0, + "step": 21416 + }, + { + "epoch": 2.3519657368767843, + "grad_norm": 2.2901864051818848, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7450342774391174, + "num_tokens": 554188916.0, + "step": 21417 + }, + { + "epoch": 2.352075554579398, + "grad_norm": 2.5682454109191895, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7369042634963989, + "num_tokens": 554210815.0, + "step": 21418 + }, + { + "epoch": 2.352185372282012, + "grad_norm": 2.591773271560669, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.738702654838562, + "num_tokens": 554233278.0, + "step": 21419 + }, + { + "epoch": 2.3522951899846256, + "grad_norm": 2.537863254547119, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.726763904094696, + "num_tokens": 554256319.0, + "step": 21420 + }, + { + "epoch": 2.3524050076872394, + "grad_norm": 2.786038875579834, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7268617153167725, + "num_tokens": 554275103.0, + "step": 21421 + }, + { + "epoch": 2.3525148253898527, + "grad_norm": 2.334908962249756, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7219877243041992, + "num_tokens": 554302972.0, + "step": 21422 + }, + { + "epoch": 2.3526246430924664, + "grad_norm": 2.442857027053833, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7319698333740234, + "num_tokens": 554326597.0, + "step": 21423 + }, + { + "epoch": 2.35273446079508, + "grad_norm": 2.430835008621216, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7130154371261597, + "num_tokens": 554350438.0, + "step": 21424 + }, + { + "epoch": 2.352844278497694, + "grad_norm": 2.675147533416748, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7322691082954407, + "num_tokens": 554372166.0, + "step": 21425 + }, + { + "epoch": 2.3529540962003077, + "grad_norm": 2.526197910308838, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7422933578491211, + "num_tokens": 554394105.0, + "step": 21426 + }, + { + "epoch": 2.353063913902921, + "grad_norm": 2.416102170944214, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.722728431224823, + "num_tokens": 554419285.0, + "step": 21427 + }, + { + "epoch": 2.353173731605535, + "grad_norm": 2.5096802711486816, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7407402992248535, + "num_tokens": 554442757.0, + "step": 21428 + }, + { + "epoch": 2.3532835493081485, + "grad_norm": 2.4106056690216064, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7137156128883362, + "num_tokens": 554468867.0, + "step": 21429 + }, + { + "epoch": 2.3533933670107623, + "grad_norm": 2.2093498706817627, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7265329360961914, + "num_tokens": 554499879.0, + "step": 21430 + }, + { + "epoch": 2.3535031847133756, + "grad_norm": 2.4639062881469727, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7357986569404602, + "num_tokens": 554524143.0, + "step": 21431 + }, + { + "epoch": 2.3536130024159894, + "grad_norm": 2.5877909660339355, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7380411624908447, + "num_tokens": 554546452.0, + "step": 21432 + }, + { + "epoch": 2.353722820118603, + "grad_norm": 2.3799760341644287, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7413957118988037, + "num_tokens": 554571337.0, + "step": 21433 + }, + { + "epoch": 2.353832637821217, + "grad_norm": 2.527708053588867, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7328400611877441, + "num_tokens": 554595132.0, + "step": 21434 + }, + { + "epoch": 2.3539424555238306, + "grad_norm": 2.415257215499878, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7251198291778564, + "num_tokens": 554622249.0, + "step": 21435 + }, + { + "epoch": 2.354052273226444, + "grad_norm": 2.06396484375, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7179054021835327, + "num_tokens": 554654011.0, + "step": 21436 + }, + { + "epoch": 2.3541620909290577, + "grad_norm": 2.3383514881134033, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7319526672363281, + "num_tokens": 554681248.0, + "step": 21437 + }, + { + "epoch": 2.3542719086316715, + "grad_norm": 2.373300313949585, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7438949942588806, + "num_tokens": 554704839.0, + "step": 21438 + }, + { + "epoch": 2.354381726334285, + "grad_norm": 2.056499481201172, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7265157699584961, + "num_tokens": 554736566.0, + "step": 21439 + }, + { + "epoch": 2.3544915440368985, + "grad_norm": 2.5355913639068604, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7331432700157166, + "num_tokens": 554759453.0, + "step": 21440 + }, + { + "epoch": 2.3546013617395123, + "grad_norm": 2.3711609840393066, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7298447489738464, + "num_tokens": 554785923.0, + "step": 21441 + }, + { + "epoch": 2.354711179442126, + "grad_norm": 2.628937005996704, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7242188453674316, + "num_tokens": 554807141.0, + "step": 21442 + }, + { + "epoch": 2.35482099714474, + "grad_norm": 2.3793416023254395, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7674164772033691, + "num_tokens": 554830120.0, + "step": 21443 + }, + { + "epoch": 2.3549308148473536, + "grad_norm": 2.351555824279785, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7336606383323669, + "num_tokens": 554854882.0, + "step": 21444 + }, + { + "epoch": 2.355040632549967, + "grad_norm": 2.5274834632873535, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7014434933662415, + "num_tokens": 554879697.0, + "step": 21445 + }, + { + "epoch": 2.3551504502525806, + "grad_norm": 2.474590301513672, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7551871538162231, + "num_tokens": 554901742.0, + "step": 21446 + }, + { + "epoch": 2.3552602679551944, + "grad_norm": 2.380124807357788, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7274613976478577, + "num_tokens": 554927477.0, + "step": 21447 + }, + { + "epoch": 2.355370085657808, + "grad_norm": 2.8565897941589355, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7668713331222534, + "num_tokens": 554945121.0, + "step": 21448 + }, + { + "epoch": 2.355479903360422, + "grad_norm": 2.448366165161133, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7235978245735168, + "num_tokens": 554971482.0, + "step": 21449 + }, + { + "epoch": 2.355589721063035, + "grad_norm": 2.737361192703247, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7266291379928589, + "num_tokens": 554991702.0, + "step": 21450 + }, + { + "epoch": 2.355699538765649, + "grad_norm": 2.5263240337371826, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7496242523193359, + "num_tokens": 555012775.0, + "step": 21451 + }, + { + "epoch": 2.3558093564682627, + "grad_norm": 2.3190789222717285, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7267023324966431, + "num_tokens": 555042151.0, + "step": 21452 + }, + { + "epoch": 2.3559191741708765, + "grad_norm": 2.2110965251922607, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7314596772193909, + "num_tokens": 555069862.0, + "step": 21453 + }, + { + "epoch": 2.3560289918734902, + "grad_norm": 2.593459129333496, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7493661046028137, + "num_tokens": 555092922.0, + "step": 21454 + }, + { + "epoch": 2.3561388095761036, + "grad_norm": 2.2567262649536133, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7410078048706055, + "num_tokens": 555119169.0, + "step": 21455 + }, + { + "epoch": 2.3562486272787173, + "grad_norm": 2.3080849647521973, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7288486957550049, + "num_tokens": 555145956.0, + "step": 21456 + }, + { + "epoch": 2.356358444981331, + "grad_norm": 2.2393672466278076, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7187109589576721, + "num_tokens": 555170733.0, + "step": 21457 + }, + { + "epoch": 2.356468262683945, + "grad_norm": 2.5887205600738525, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.7569135427474976, + "num_tokens": 555190957.0, + "step": 21458 + }, + { + "epoch": 2.356578080386558, + "grad_norm": 2.7504446506500244, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7337218523025513, + "num_tokens": 555211908.0, + "step": 21459 + }, + { + "epoch": 2.356687898089172, + "grad_norm": 2.4559195041656494, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7277474403381348, + "num_tokens": 555236327.0, + "step": 21460 + }, + { + "epoch": 2.3567977157917857, + "grad_norm": 2.557098388671875, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.737910807132721, + "num_tokens": 555260806.0, + "step": 21461 + }, + { + "epoch": 2.3569075334943994, + "grad_norm": 2.2460038661956787, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7256903648376465, + "num_tokens": 555290160.0, + "step": 21462 + }, + { + "epoch": 2.3570173511970127, + "grad_norm": 2.7116284370422363, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.739080548286438, + "num_tokens": 555310957.0, + "step": 21463 + }, + { + "epoch": 2.3571271688996265, + "grad_norm": 2.47953724861145, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7269521355628967, + "num_tokens": 555335161.0, + "step": 21464 + }, + { + "epoch": 2.3572369866022402, + "grad_norm": 2.262861728668213, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7225804328918457, + "num_tokens": 555362095.0, + "step": 21465 + }, + { + "epoch": 2.357346804304854, + "grad_norm": 2.5037176609039307, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7467436194419861, + "num_tokens": 555384986.0, + "step": 21466 + }, + { + "epoch": 2.3574566220074678, + "grad_norm": 2.477186441421509, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7395920753479004, + "num_tokens": 555410007.0, + "step": 21467 + }, + { + "epoch": 2.357566439710081, + "grad_norm": 2.1805903911590576, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7229360342025757, + "num_tokens": 555441338.0, + "step": 21468 + }, + { + "epoch": 2.357676257412695, + "grad_norm": 2.255643844604492, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7162103652954102, + "num_tokens": 555470582.0, + "step": 21469 + }, + { + "epoch": 2.3577860751153086, + "grad_norm": 2.0743396282196045, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7320612668991089, + "num_tokens": 555503886.0, + "step": 21470 + }, + { + "epoch": 2.3578958928179223, + "grad_norm": 2.463256597518921, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7362213134765625, + "num_tokens": 555528617.0, + "step": 21471 + }, + { + "epoch": 2.358005710520536, + "grad_norm": 2.33406138420105, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.740838885307312, + "num_tokens": 555555610.0, + "step": 21472 + }, + { + "epoch": 2.3581155282231494, + "grad_norm": 2.3262388706207275, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7056305408477783, + "num_tokens": 555581989.0, + "step": 21473 + }, + { + "epoch": 2.358225345925763, + "grad_norm": 2.7703938484191895, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7335516810417175, + "num_tokens": 555604084.0, + "step": 21474 + }, + { + "epoch": 2.358335163628377, + "grad_norm": 2.240844488143921, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.744530200958252, + "num_tokens": 555631310.0, + "step": 21475 + }, + { + "epoch": 2.3584449813309907, + "grad_norm": 2.1904728412628174, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7214824557304382, + "num_tokens": 555661471.0, + "step": 21476 + }, + { + "epoch": 2.3585547990336044, + "grad_norm": 2.274742603302002, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7529574632644653, + "num_tokens": 555686611.0, + "step": 21477 + }, + { + "epoch": 2.3586646167362177, + "grad_norm": 2.6267733573913574, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7449674606323242, + "num_tokens": 555706852.0, + "step": 21478 + }, + { + "epoch": 2.3587744344388315, + "grad_norm": 2.3412997722625732, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7561908960342407, + "num_tokens": 555731573.0, + "step": 21479 + }, + { + "epoch": 2.3588842521414453, + "grad_norm": 2.3804073333740234, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7550097107887268, + "num_tokens": 555754399.0, + "step": 21480 + }, + { + "epoch": 2.358994069844059, + "grad_norm": 2.4376001358032227, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7194056510925293, + "num_tokens": 555779355.0, + "step": 21481 + }, + { + "epoch": 2.3591038875466728, + "grad_norm": 2.4886038303375244, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7446846961975098, + "num_tokens": 555801736.0, + "step": 21482 + }, + { + "epoch": 2.359213705249286, + "grad_norm": 2.425741672515869, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7283806800842285, + "num_tokens": 555828488.0, + "step": 21483 + }, + { + "epoch": 2.3593235229519, + "grad_norm": 2.4358694553375244, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7559728026390076, + "num_tokens": 555853675.0, + "step": 21484 + }, + { + "epoch": 2.3594333406545136, + "grad_norm": 2.472339630126953, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7422692179679871, + "num_tokens": 555877499.0, + "step": 21485 + }, + { + "epoch": 2.3595431583571274, + "grad_norm": 2.2346949577331543, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7277818918228149, + "num_tokens": 555904816.0, + "step": 21486 + }, + { + "epoch": 2.3596529760597407, + "grad_norm": 2.398836374282837, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7428464293479919, + "num_tokens": 555930791.0, + "step": 21487 + }, + { + "epoch": 2.3597627937623544, + "grad_norm": 2.498136043548584, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7231357097625732, + "num_tokens": 555954763.0, + "step": 21488 + }, + { + "epoch": 2.359872611464968, + "grad_norm": 2.461318254470825, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7589025497436523, + "num_tokens": 555978907.0, + "step": 21489 + }, + { + "epoch": 2.359982429167582, + "grad_norm": 2.3516790866851807, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.72782301902771, + "num_tokens": 556004729.0, + "step": 21490 + }, + { + "epoch": 2.3600922468701953, + "grad_norm": 2.339859962463379, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7233636975288391, + "num_tokens": 556031476.0, + "step": 21491 + }, + { + "epoch": 2.360202064572809, + "grad_norm": 2.5117361545562744, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7331054210662842, + "num_tokens": 556054552.0, + "step": 21492 + }, + { + "epoch": 2.3603118822754228, + "grad_norm": 2.596947431564331, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7155441045761108, + "num_tokens": 556079224.0, + "step": 21493 + }, + { + "epoch": 2.3604216999780365, + "grad_norm": 2.6918489933013916, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7451398968696594, + "num_tokens": 556099862.0, + "step": 21494 + }, + { + "epoch": 2.3605315176806503, + "grad_norm": 2.0628199577331543, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7113946676254272, + "num_tokens": 556131139.0, + "step": 21495 + }, + { + "epoch": 2.3606413353832636, + "grad_norm": 2.4360384941101074, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7440071702003479, + "num_tokens": 556154231.0, + "step": 21496 + }, + { + "epoch": 2.3607511530858774, + "grad_norm": 2.1747610569000244, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7254797220230103, + "num_tokens": 556185699.0, + "step": 21497 + }, + { + "epoch": 2.360860970788491, + "grad_norm": 2.1893088817596436, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7413190603256226, + "num_tokens": 556213295.0, + "step": 21498 + }, + { + "epoch": 2.360970788491105, + "grad_norm": 2.4714560508728027, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7191442251205444, + "num_tokens": 556236909.0, + "step": 21499 + }, + { + "epoch": 2.3610806061937186, + "grad_norm": 2.580195665359497, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7391822934150696, + "num_tokens": 556260911.0, + "step": 21500 + }, + { + "epoch": 2.361190423896332, + "grad_norm": 2.5542829036712646, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7487450838088989, + "num_tokens": 556284155.0, + "step": 21501 + }, + { + "epoch": 2.3613002415989457, + "grad_norm": 2.3798739910125732, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7244316339492798, + "num_tokens": 556311007.0, + "step": 21502 + }, + { + "epoch": 2.3614100593015594, + "grad_norm": 2.8961427211761475, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7753976583480835, + "num_tokens": 556327874.0, + "step": 21503 + }, + { + "epoch": 2.361519877004173, + "grad_norm": 2.2951674461364746, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7359557747840881, + "num_tokens": 556354663.0, + "step": 21504 + }, + { + "epoch": 2.361629694706787, + "grad_norm": 2.407357931137085, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.761703372001648, + "num_tokens": 556380131.0, + "step": 21505 + }, + { + "epoch": 2.3617395124094003, + "grad_norm": 2.065464735031128, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7340248823165894, + "num_tokens": 556412535.0, + "step": 21506 + }, + { + "epoch": 2.361849330112014, + "grad_norm": 2.298027753829956, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7269037961959839, + "num_tokens": 556438374.0, + "step": 21507 + }, + { + "epoch": 2.361959147814628, + "grad_norm": 2.517481565475464, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7303540110588074, + "num_tokens": 556461616.0, + "step": 21508 + }, + { + "epoch": 2.3620689655172415, + "grad_norm": 2.4440927505493164, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7416858077049255, + "num_tokens": 556483612.0, + "step": 21509 + }, + { + "epoch": 2.362178783219855, + "grad_norm": 2.366903305053711, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7445951104164124, + "num_tokens": 556508354.0, + "step": 21510 + }, + { + "epoch": 2.3622886009224686, + "grad_norm": 2.4070558547973633, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7361162900924683, + "num_tokens": 556533953.0, + "step": 21511 + }, + { + "epoch": 2.3623984186250824, + "grad_norm": 2.401697874069214, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7299295663833618, + "num_tokens": 556559964.0, + "step": 21512 + }, + { + "epoch": 2.362508236327696, + "grad_norm": 2.1949777603149414, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7205536961555481, + "num_tokens": 556591345.0, + "step": 21513 + }, + { + "epoch": 2.3626180540303094, + "grad_norm": 2.85119366645813, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.735785961151123, + "num_tokens": 556610964.0, + "step": 21514 + }, + { + "epoch": 2.362727871732923, + "grad_norm": 2.3898773193359375, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7514420747756958, + "num_tokens": 556635293.0, + "step": 21515 + }, + { + "epoch": 2.362837689435537, + "grad_norm": 2.2525594234466553, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7124372720718384, + "num_tokens": 556663735.0, + "step": 21516 + }, + { + "epoch": 2.3629475071381507, + "grad_norm": 2.68007755279541, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7339930534362793, + "num_tokens": 556686187.0, + "step": 21517 + }, + { + "epoch": 2.3630573248407645, + "grad_norm": 2.3597404956817627, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7255455255508423, + "num_tokens": 556714669.0, + "step": 21518 + }, + { + "epoch": 2.363167142543378, + "grad_norm": 2.2762081623077393, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7273424863815308, + "num_tokens": 556741696.0, + "step": 21519 + }, + { + "epoch": 2.3632769602459915, + "grad_norm": 2.4787039756774902, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7388726472854614, + "num_tokens": 556763765.0, + "step": 21520 + }, + { + "epoch": 2.3633867779486053, + "grad_norm": 2.3086111545562744, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7247961759567261, + "num_tokens": 556790514.0, + "step": 21521 + }, + { + "epoch": 2.363496595651219, + "grad_norm": 2.534656047821045, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7317296862602234, + "num_tokens": 556813619.0, + "step": 21522 + }, + { + "epoch": 2.363606413353833, + "grad_norm": 2.2892298698425293, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7452051639556885, + "num_tokens": 556841250.0, + "step": 21523 + }, + { + "epoch": 2.363716231056446, + "grad_norm": 2.5263888835906982, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7514404654502869, + "num_tokens": 556863528.0, + "step": 21524 + }, + { + "epoch": 2.36382604875906, + "grad_norm": 2.212123394012451, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7267321944236755, + "num_tokens": 556892962.0, + "step": 21525 + }, + { + "epoch": 2.3639358664616736, + "grad_norm": 2.2690131664276123, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7236998677253723, + "num_tokens": 556920673.0, + "step": 21526 + }, + { + "epoch": 2.3640456841642874, + "grad_norm": 2.596574306488037, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.758247971534729, + "num_tokens": 556943449.0, + "step": 21527 + }, + { + "epoch": 2.364155501866901, + "grad_norm": 2.214000701904297, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7185421586036682, + "num_tokens": 556972204.0, + "step": 21528 + }, + { + "epoch": 2.3642653195695145, + "grad_norm": 2.2602131366729736, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7060441374778748, + "num_tokens": 557002786.0, + "step": 21529 + }, + { + "epoch": 2.3643751372721282, + "grad_norm": 2.561884641647339, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7086361646652222, + "num_tokens": 557028432.0, + "step": 21530 + }, + { + "epoch": 2.364484954974742, + "grad_norm": 2.5059924125671387, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7115153670310974, + "num_tokens": 557053635.0, + "step": 21531 + }, + { + "epoch": 2.3645947726773557, + "grad_norm": 2.3549015522003174, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7380908727645874, + "num_tokens": 557080238.0, + "step": 21532 + }, + { + "epoch": 2.3647045903799695, + "grad_norm": 2.4770398139953613, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7395671606063843, + "num_tokens": 557102780.0, + "step": 21533 + }, + { + "epoch": 2.364814408082583, + "grad_norm": 2.299309253692627, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7225500345230103, + "num_tokens": 557131178.0, + "step": 21534 + }, + { + "epoch": 2.3649242257851966, + "grad_norm": 2.588495969772339, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7425265908241272, + "num_tokens": 557153394.0, + "step": 21535 + }, + { + "epoch": 2.3650340434878103, + "grad_norm": 2.276618480682373, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7309032678604126, + "num_tokens": 557180490.0, + "step": 21536 + }, + { + "epoch": 2.365143861190424, + "grad_norm": 2.101564407348633, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7296880483627319, + "num_tokens": 557213714.0, + "step": 21537 + }, + { + "epoch": 2.3652536788930374, + "grad_norm": 2.390375852584839, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7398364543914795, + "num_tokens": 557238581.0, + "step": 21538 + }, + { + "epoch": 2.365363496595651, + "grad_norm": 2.3667638301849365, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7186824083328247, + "num_tokens": 557264846.0, + "step": 21539 + }, + { + "epoch": 2.365473314298265, + "grad_norm": 2.402412176132202, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7264150977134705, + "num_tokens": 557291147.0, + "step": 21540 + }, + { + "epoch": 2.3655831320008787, + "grad_norm": 2.568354368209839, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7352166175842285, + "num_tokens": 557312834.0, + "step": 21541 + }, + { + "epoch": 2.365692949703492, + "grad_norm": 2.4249911308288574, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7465444803237915, + "num_tokens": 557339964.0, + "step": 21542 + }, + { + "epoch": 2.3658027674061057, + "grad_norm": 2.3460357189178467, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7434156537055969, + "num_tokens": 557366144.0, + "step": 21543 + }, + { + "epoch": 2.3659125851087195, + "grad_norm": 2.3607585430145264, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7231107950210571, + "num_tokens": 557391270.0, + "step": 21544 + }, + { + "epoch": 2.3660224028113332, + "grad_norm": 2.300837278366089, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7307955622673035, + "num_tokens": 557417639.0, + "step": 21545 + }, + { + "epoch": 2.366132220513947, + "grad_norm": 2.45511794090271, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7365598678588867, + "num_tokens": 557444990.0, + "step": 21546 + }, + { + "epoch": 2.3662420382165603, + "grad_norm": 2.623220920562744, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7057123184204102, + "num_tokens": 557467022.0, + "step": 21547 + }, + { + "epoch": 2.366351855919174, + "grad_norm": 2.30363130569458, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7221486568450928, + "num_tokens": 557495147.0, + "step": 21548 + }, + { + "epoch": 2.366461673621788, + "grad_norm": 2.4679172039031982, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7404788136482239, + "num_tokens": 557519017.0, + "step": 21549 + }, + { + "epoch": 2.3665714913244016, + "grad_norm": 2.387479782104492, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7668302059173584, + "num_tokens": 557543080.0, + "step": 21550 + }, + { + "epoch": 2.3666813090270153, + "grad_norm": 2.343519926071167, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7314125895500183, + "num_tokens": 557569638.0, + "step": 21551 + }, + { + "epoch": 2.3667911267296287, + "grad_norm": 2.1890227794647217, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.743934154510498, + "num_tokens": 557598536.0, + "step": 21552 + }, + { + "epoch": 2.3669009444322424, + "grad_norm": 2.070659637451172, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7200374603271484, + "num_tokens": 557634436.0, + "step": 21553 + }, + { + "epoch": 2.367010762134856, + "grad_norm": 2.800022602081299, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7525877356529236, + "num_tokens": 557654703.0, + "step": 21554 + }, + { + "epoch": 2.36712057983747, + "grad_norm": 2.3110525608062744, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7261127233505249, + "num_tokens": 557683104.0, + "step": 21555 + }, + { + "epoch": 2.3672303975400837, + "grad_norm": 2.2060723304748535, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7154415845870972, + "num_tokens": 557715197.0, + "step": 21556 + }, + { + "epoch": 2.367340215242697, + "grad_norm": 2.2662153244018555, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7321634292602539, + "num_tokens": 557742148.0, + "step": 21557 + }, + { + "epoch": 2.3674500329453108, + "grad_norm": 2.193190813064575, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7253391742706299, + "num_tokens": 557769927.0, + "step": 21558 + }, + { + "epoch": 2.3675598506479245, + "grad_norm": 2.51208758354187, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7303742170333862, + "num_tokens": 557793071.0, + "step": 21559 + }, + { + "epoch": 2.3676696683505383, + "grad_norm": 2.1435325145721436, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7488346099853516, + "num_tokens": 557822869.0, + "step": 21560 + }, + { + "epoch": 2.3677794860531516, + "grad_norm": 2.259763717651367, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7318320274353027, + "num_tokens": 557852209.0, + "step": 21561 + }, + { + "epoch": 2.3678893037557653, + "grad_norm": 2.4174728393554688, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7239799499511719, + "num_tokens": 557876654.0, + "step": 21562 + }, + { + "epoch": 2.367999121458379, + "grad_norm": 2.1077349185943604, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7412148714065552, + "num_tokens": 557906940.0, + "step": 21563 + }, + { + "epoch": 2.368108939160993, + "grad_norm": 2.1697232723236084, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7240811586380005, + "num_tokens": 557938472.0, + "step": 21564 + }, + { + "epoch": 2.368218756863606, + "grad_norm": 2.618079900741577, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7270771861076355, + "num_tokens": 557960421.0, + "step": 21565 + }, + { + "epoch": 2.36832857456622, + "grad_norm": 2.6840615272521973, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.742655873298645, + "num_tokens": 557981998.0, + "step": 21566 + }, + { + "epoch": 2.3684383922688337, + "grad_norm": 2.522843360900879, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7390599846839905, + "num_tokens": 558004848.0, + "step": 21567 + }, + { + "epoch": 2.3685482099714474, + "grad_norm": 2.520381450653076, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7415125966072083, + "num_tokens": 558028237.0, + "step": 21568 + }, + { + "epoch": 2.368658027674061, + "grad_norm": 2.2023918628692627, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7246014475822449, + "num_tokens": 558058594.0, + "step": 21569 + }, + { + "epoch": 2.3687678453766745, + "grad_norm": 2.3033347129821777, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7540983557701111, + "num_tokens": 558084242.0, + "step": 21570 + }, + { + "epoch": 2.3688776630792883, + "grad_norm": 2.1215381622314453, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7012530565261841, + "num_tokens": 558115959.0, + "step": 21571 + }, + { + "epoch": 2.368987480781902, + "grad_norm": 2.3394360542297363, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7218514680862427, + "num_tokens": 558141050.0, + "step": 21572 + }, + { + "epoch": 2.369097298484516, + "grad_norm": 2.1194632053375244, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6931822299957275, + "num_tokens": 558173423.0, + "step": 21573 + }, + { + "epoch": 2.3692071161871295, + "grad_norm": 2.3782687187194824, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7360621690750122, + "num_tokens": 558197401.0, + "step": 21574 + }, + { + "epoch": 2.369316933889743, + "grad_norm": 2.616105318069458, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7404006123542786, + "num_tokens": 558220872.0, + "step": 21575 + }, + { + "epoch": 2.3694267515923566, + "grad_norm": 2.4111437797546387, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7318177223205566, + "num_tokens": 558245340.0, + "step": 21576 + }, + { + "epoch": 2.3695365692949704, + "grad_norm": 2.3617324829101562, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7525095343589783, + "num_tokens": 558269052.0, + "step": 21577 + }, + { + "epoch": 2.369646386997584, + "grad_norm": 2.2455897331237793, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7384119033813477, + "num_tokens": 558296135.0, + "step": 21578 + }, + { + "epoch": 2.369756204700198, + "grad_norm": 2.5655312538146973, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7246509790420532, + "num_tokens": 558319289.0, + "step": 21579 + }, + { + "epoch": 2.369866022402811, + "grad_norm": 2.5333781242370605, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7436246871948242, + "num_tokens": 558341681.0, + "step": 21580 + }, + { + "epoch": 2.369975840105425, + "grad_norm": 2.242297410964966, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7095115184783936, + "num_tokens": 558369106.0, + "step": 21581 + }, + { + "epoch": 2.3700856578080387, + "grad_norm": 2.207592487335205, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7330787181854248, + "num_tokens": 558398945.0, + "step": 21582 + }, + { + "epoch": 2.3701954755106525, + "grad_norm": 2.7548024654388428, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7419383525848389, + "num_tokens": 558422580.0, + "step": 21583 + }, + { + "epoch": 2.370305293213266, + "grad_norm": 2.2318763732910156, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7382527589797974, + "num_tokens": 558450847.0, + "step": 21584 + }, + { + "epoch": 2.3704151109158795, + "grad_norm": 2.3329975605010986, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7329447865486145, + "num_tokens": 558475667.0, + "step": 21585 + }, + { + "epoch": 2.3705249286184933, + "grad_norm": 2.2117037773132324, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7432499527931213, + "num_tokens": 558504062.0, + "step": 21586 + }, + { + "epoch": 2.370634746321107, + "grad_norm": 2.5231921672821045, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7398595809936523, + "num_tokens": 558527171.0, + "step": 21587 + }, + { + "epoch": 2.370744564023721, + "grad_norm": 2.790045976638794, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7438811659812927, + "num_tokens": 558546295.0, + "step": 21588 + }, + { + "epoch": 2.370854381726334, + "grad_norm": 2.52402400970459, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.721906304359436, + "num_tokens": 558568236.0, + "step": 21589 + }, + { + "epoch": 2.370964199428948, + "grad_norm": 2.189028739929199, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7054115533828735, + "num_tokens": 558598369.0, + "step": 21590 + }, + { + "epoch": 2.3710740171315616, + "grad_norm": 2.6326258182525635, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7607419490814209, + "num_tokens": 558620381.0, + "step": 21591 + }, + { + "epoch": 2.3711838348341754, + "grad_norm": 2.6414694786071777, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.740432620048523, + "num_tokens": 558642173.0, + "step": 21592 + }, + { + "epoch": 2.3712936525367887, + "grad_norm": 2.431577205657959, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7257153987884521, + "num_tokens": 558669174.0, + "step": 21593 + }, + { + "epoch": 2.3714034702394025, + "grad_norm": 2.541870355606079, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7376049757003784, + "num_tokens": 558691184.0, + "step": 21594 + }, + { + "epoch": 2.371513287942016, + "grad_norm": 2.9817755222320557, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7580649256706238, + "num_tokens": 558707598.0, + "step": 21595 + }, + { + "epoch": 2.37162310564463, + "grad_norm": 2.025191307067871, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6930348873138428, + "num_tokens": 558744396.0, + "step": 21596 + }, + { + "epoch": 2.3717329233472437, + "grad_norm": 2.4539763927459717, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7438129186630249, + "num_tokens": 558769673.0, + "step": 21597 + }, + { + "epoch": 2.371842741049857, + "grad_norm": 2.471271514892578, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7325968146324158, + "num_tokens": 558795749.0, + "step": 21598 + }, + { + "epoch": 2.371952558752471, + "grad_norm": 2.128890037536621, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7464860081672668, + "num_tokens": 558826193.0, + "step": 21599 + }, + { + "epoch": 2.3720623764550846, + "grad_norm": 2.482607126235962, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7491371631622314, + "num_tokens": 558850118.0, + "step": 21600 + }, + { + "epoch": 2.3721721941576983, + "grad_norm": 2.698915958404541, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7315708994865417, + "num_tokens": 558870899.0, + "step": 21601 + }, + { + "epoch": 2.372282011860312, + "grad_norm": 2.2915101051330566, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7231963276863098, + "num_tokens": 558897874.0, + "step": 21602 + }, + { + "epoch": 2.3723918295629254, + "grad_norm": 2.2608869075775146, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7222409844398499, + "num_tokens": 558925023.0, + "step": 21603 + }, + { + "epoch": 2.372501647265539, + "grad_norm": 2.431506633758545, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7352253198623657, + "num_tokens": 558950567.0, + "step": 21604 + }, + { + "epoch": 2.372611464968153, + "grad_norm": 2.3778114318847656, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7432104349136353, + "num_tokens": 558974871.0, + "step": 21605 + }, + { + "epoch": 2.3727212826707667, + "grad_norm": 2.1761884689331055, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7207398414611816, + "num_tokens": 559005912.0, + "step": 21606 + }, + { + "epoch": 2.3728311003733804, + "grad_norm": 2.283231735229492, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7274838089942932, + "num_tokens": 559032318.0, + "step": 21607 + }, + { + "epoch": 2.3729409180759937, + "grad_norm": 2.871634006500244, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7512105703353882, + "num_tokens": 559050923.0, + "step": 21608 + }, + { + "epoch": 2.3730507357786075, + "grad_norm": 2.037304162979126, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7516050338745117, + "num_tokens": 559081779.0, + "step": 21609 + }, + { + "epoch": 2.3731605534812212, + "grad_norm": 2.7407093048095703, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7208091020584106, + "num_tokens": 559101853.0, + "step": 21610 + }, + { + "epoch": 2.373270371183835, + "grad_norm": 2.6552979946136475, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.714919924736023, + "num_tokens": 559125924.0, + "step": 21611 + }, + { + "epoch": 2.3733801888864483, + "grad_norm": 2.222496747970581, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7423380613327026, + "num_tokens": 559155653.0, + "step": 21612 + }, + { + "epoch": 2.373490006589062, + "grad_norm": 2.670348644256592, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7340299487113953, + "num_tokens": 559179068.0, + "step": 21613 + }, + { + "epoch": 2.373599824291676, + "grad_norm": 2.2662055492401123, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7229191064834595, + "num_tokens": 559206875.0, + "step": 21614 + }, + { + "epoch": 2.3737096419942896, + "grad_norm": 2.1857595443725586, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7287712693214417, + "num_tokens": 559235853.0, + "step": 21615 + }, + { + "epoch": 2.3738194596969033, + "grad_norm": 2.7285706996917725, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7608951926231384, + "num_tokens": 559256191.0, + "step": 21616 + }, + { + "epoch": 2.3739292773995166, + "grad_norm": 2.6132290363311768, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.716490626335144, + "num_tokens": 559278183.0, + "step": 21617 + }, + { + "epoch": 2.3740390951021304, + "grad_norm": 2.6540915966033936, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7399775385856628, + "num_tokens": 559299964.0, + "step": 21618 + }, + { + "epoch": 2.374148912804744, + "grad_norm": 2.3372879028320312, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7219746112823486, + "num_tokens": 559324445.0, + "step": 21619 + }, + { + "epoch": 2.374258730507358, + "grad_norm": 2.389194965362549, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.727229118347168, + "num_tokens": 559349323.0, + "step": 21620 + }, + { + "epoch": 2.3743685482099712, + "grad_norm": 2.2418625354766846, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7256275415420532, + "num_tokens": 559378716.0, + "step": 21621 + }, + { + "epoch": 2.374478365912585, + "grad_norm": 2.017103910446167, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7094076871871948, + "num_tokens": 559412640.0, + "step": 21622 + }, + { + "epoch": 2.3745881836151987, + "grad_norm": 2.764765977859497, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7463037967681885, + "num_tokens": 559431472.0, + "step": 21623 + }, + { + "epoch": 2.3746980013178125, + "grad_norm": 2.5652520656585693, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7453823685646057, + "num_tokens": 559453465.0, + "step": 21624 + }, + { + "epoch": 2.3748078190204263, + "grad_norm": 2.4429385662078857, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7160930037498474, + "num_tokens": 559479660.0, + "step": 21625 + }, + { + "epoch": 2.3749176367230396, + "grad_norm": 2.1363489627838135, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7445461750030518, + "num_tokens": 559509927.0, + "step": 21626 + }, + { + "epoch": 2.3750274544256533, + "grad_norm": 2.384537696838379, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7273319959640503, + "num_tokens": 559535535.0, + "step": 21627 + }, + { + "epoch": 2.375137272128267, + "grad_norm": 2.2780814170837402, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7392480373382568, + "num_tokens": 559563278.0, + "step": 21628 + }, + { + "epoch": 2.375247089830881, + "grad_norm": 2.2802717685699463, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7298564910888672, + "num_tokens": 559588563.0, + "step": 21629 + }, + { + "epoch": 2.3753569075334946, + "grad_norm": 2.2510204315185547, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7123053669929504, + "num_tokens": 559619051.0, + "step": 21630 + }, + { + "epoch": 2.375466725236108, + "grad_norm": 2.6190366744995117, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7567743062973022, + "num_tokens": 559639925.0, + "step": 21631 + }, + { + "epoch": 2.3755765429387217, + "grad_norm": 2.1522562503814697, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7516219019889832, + "num_tokens": 559671210.0, + "step": 21632 + }, + { + "epoch": 2.3756863606413354, + "grad_norm": 2.592740774154663, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7301493883132935, + "num_tokens": 559694236.0, + "step": 21633 + }, + { + "epoch": 2.375796178343949, + "grad_norm": 2.75488543510437, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7223268747329712, + "num_tokens": 559714548.0, + "step": 21634 + }, + { + "epoch": 2.375905996046563, + "grad_norm": 2.2113382816314697, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7189983129501343, + "num_tokens": 559744762.0, + "step": 21635 + }, + { + "epoch": 2.3760158137491763, + "grad_norm": 2.3708295822143555, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7322875261306763, + "num_tokens": 559770635.0, + "step": 21636 + }, + { + "epoch": 2.37612563145179, + "grad_norm": 2.759573459625244, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7346901893615723, + "num_tokens": 559791610.0, + "step": 21637 + }, + { + "epoch": 2.3762354491544038, + "grad_norm": 2.5287351608276367, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.731105387210846, + "num_tokens": 559814186.0, + "step": 21638 + }, + { + "epoch": 2.3763452668570175, + "grad_norm": 2.609957695007324, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7233357429504395, + "num_tokens": 559835190.0, + "step": 21639 + }, + { + "epoch": 2.376455084559631, + "grad_norm": 2.0975794792175293, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7205013632774353, + "num_tokens": 559866682.0, + "step": 21640 + }, + { + "epoch": 2.3765649022622446, + "grad_norm": 2.232700824737549, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.724079966545105, + "num_tokens": 559895962.0, + "step": 21641 + }, + { + "epoch": 2.3766747199648584, + "grad_norm": 2.5115580558776855, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7535616159439087, + "num_tokens": 559918730.0, + "step": 21642 + }, + { + "epoch": 2.376784537667472, + "grad_norm": 2.460907220840454, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7284133434295654, + "num_tokens": 559942407.0, + "step": 21643 + }, + { + "epoch": 2.3768943553700854, + "grad_norm": 2.420717716217041, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7119272947311401, + "num_tokens": 559969006.0, + "step": 21644 + }, + { + "epoch": 2.377004173072699, + "grad_norm": 2.490133047103882, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7314898371696472, + "num_tokens": 559992947.0, + "step": 21645 + }, + { + "epoch": 2.377113990775313, + "grad_norm": 2.341695547103882, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7079479694366455, + "num_tokens": 560020395.0, + "step": 21646 + }, + { + "epoch": 2.3772238084779267, + "grad_norm": 2.2233216762542725, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7377241253852844, + "num_tokens": 560048467.0, + "step": 21647 + }, + { + "epoch": 2.3773336261805404, + "grad_norm": 2.6166889667510986, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7348992824554443, + "num_tokens": 560070449.0, + "step": 21648 + }, + { + "epoch": 2.3774434438831538, + "grad_norm": 2.0382094383239746, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7069814801216125, + "num_tokens": 560102206.0, + "step": 21649 + }, + { + "epoch": 2.3775532615857675, + "grad_norm": 2.4494006633758545, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7289028763771057, + "num_tokens": 560124865.0, + "step": 21650 + }, + { + "epoch": 2.3776630792883813, + "grad_norm": 2.4137566089630127, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7429268956184387, + "num_tokens": 560146976.0, + "step": 21651 + }, + { + "epoch": 2.377772896990995, + "grad_norm": 2.621161699295044, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7467231750488281, + "num_tokens": 560166722.0, + "step": 21652 + }, + { + "epoch": 2.377882714693609, + "grad_norm": 2.5721468925476074, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7386074066162109, + "num_tokens": 560189172.0, + "step": 21653 + }, + { + "epoch": 2.377992532396222, + "grad_norm": 2.4351565837860107, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7211698293685913, + "num_tokens": 560214066.0, + "step": 21654 + }, + { + "epoch": 2.378102350098836, + "grad_norm": 2.0739119052886963, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7384821176528931, + "num_tokens": 560246232.0, + "step": 21655 + }, + { + "epoch": 2.3782121678014496, + "grad_norm": 2.363821268081665, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7343015074729919, + "num_tokens": 560272804.0, + "step": 21656 + }, + { + "epoch": 2.3783219855040634, + "grad_norm": 2.4881083965301514, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7362803220748901, + "num_tokens": 560296297.0, + "step": 21657 + }, + { + "epoch": 2.378431803206677, + "grad_norm": 2.3603103160858154, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7179461717605591, + "num_tokens": 560319923.0, + "step": 21658 + }, + { + "epoch": 2.3785416209092904, + "grad_norm": 2.43997859954834, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7256891131401062, + "num_tokens": 560345057.0, + "step": 21659 + }, + { + "epoch": 2.378651438611904, + "grad_norm": 2.7021687030792236, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7345013618469238, + "num_tokens": 560365551.0, + "step": 21660 + }, + { + "epoch": 2.378761256314518, + "grad_norm": 2.347102403640747, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7565139532089233, + "num_tokens": 560390724.0, + "step": 21661 + }, + { + "epoch": 2.3788710740171317, + "grad_norm": 2.6246893405914307, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7211623191833496, + "num_tokens": 560412951.0, + "step": 21662 + }, + { + "epoch": 2.3789808917197455, + "grad_norm": 2.4616787433624268, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7314965128898621, + "num_tokens": 560437061.0, + "step": 21663 + }, + { + "epoch": 2.379090709422359, + "grad_norm": 2.4392318725585938, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7574591636657715, + "num_tokens": 560459323.0, + "step": 21664 + }, + { + "epoch": 2.3792005271249725, + "grad_norm": 2.7028684616088867, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7236851453781128, + "num_tokens": 560480023.0, + "step": 21665 + }, + { + "epoch": 2.3793103448275863, + "grad_norm": 2.488166093826294, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7193976044654846, + "num_tokens": 560507057.0, + "step": 21666 + }, + { + "epoch": 2.3794201625302, + "grad_norm": 2.3947482109069824, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7217425107955933, + "num_tokens": 560531806.0, + "step": 21667 + }, + { + "epoch": 2.3795299802328134, + "grad_norm": 2.036409854888916, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.735008955001831, + "num_tokens": 560562967.0, + "step": 21668 + }, + { + "epoch": 2.379639797935427, + "grad_norm": 2.497096061706543, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7453969120979309, + "num_tokens": 560587734.0, + "step": 21669 + }, + { + "epoch": 2.379749615638041, + "grad_norm": 2.5345168113708496, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7245504856109619, + "num_tokens": 560612771.0, + "step": 21670 + }, + { + "epoch": 2.3798594333406546, + "grad_norm": 2.809537172317505, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.785591721534729, + "num_tokens": 560631954.0, + "step": 21671 + }, + { + "epoch": 2.379969251043268, + "grad_norm": 2.088275194168091, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7002977132797241, + "num_tokens": 560664443.0, + "step": 21672 + }, + { + "epoch": 2.3800790687458817, + "grad_norm": 2.4386560916900635, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7307556867599487, + "num_tokens": 560689391.0, + "step": 21673 + }, + { + "epoch": 2.3801888864484955, + "grad_norm": 2.218426465988159, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7238095998764038, + "num_tokens": 560717244.0, + "step": 21674 + }, + { + "epoch": 2.3802987041511092, + "grad_norm": 2.4897851943969727, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7467143535614014, + "num_tokens": 560738910.0, + "step": 21675 + }, + { + "epoch": 2.380408521853723, + "grad_norm": 2.1092238426208496, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7285528182983398, + "num_tokens": 560769723.0, + "step": 21676 + }, + { + "epoch": 2.3805183395563363, + "grad_norm": 1.8746190071105957, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7197228670120239, + "num_tokens": 560806610.0, + "step": 21677 + }, + { + "epoch": 2.38062815725895, + "grad_norm": 2.5491819381713867, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7274118661880493, + "num_tokens": 560829502.0, + "step": 21678 + }, + { + "epoch": 2.380737974961564, + "grad_norm": 2.2864930629730225, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7559351325035095, + "num_tokens": 560855250.0, + "step": 21679 + }, + { + "epoch": 2.3808477926641776, + "grad_norm": 2.6511764526367188, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7503489255905151, + "num_tokens": 560876357.0, + "step": 21680 + }, + { + "epoch": 2.3809576103667913, + "grad_norm": 2.4392194747924805, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.715208113193512, + "num_tokens": 560902717.0, + "step": 21681 + }, + { + "epoch": 2.3810674280694046, + "grad_norm": 2.2117865085601807, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7423055768013, + "num_tokens": 560930540.0, + "step": 21682 + }, + { + "epoch": 2.3811772457720184, + "grad_norm": 2.4184064865112305, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7441515326499939, + "num_tokens": 560954786.0, + "step": 21683 + }, + { + "epoch": 2.381287063474632, + "grad_norm": 2.2920172214508057, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7317950129508972, + "num_tokens": 560981734.0, + "step": 21684 + }, + { + "epoch": 2.381396881177246, + "grad_norm": 2.4361064434051514, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7398122549057007, + "num_tokens": 561004489.0, + "step": 21685 + }, + { + "epoch": 2.3815066988798597, + "grad_norm": 2.6657519340515137, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7485238909721375, + "num_tokens": 561024652.0, + "step": 21686 + }, + { + "epoch": 2.381616516582473, + "grad_norm": 2.40596866607666, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7375778555870056, + "num_tokens": 561049401.0, + "step": 21687 + }, + { + "epoch": 2.3817263342850867, + "grad_norm": 2.637418031692505, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7458838820457458, + "num_tokens": 561071127.0, + "step": 21688 + }, + { + "epoch": 2.3818361519877005, + "grad_norm": 2.4053704738616943, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7127553224563599, + "num_tokens": 561095712.0, + "step": 21689 + }, + { + "epoch": 2.3819459696903142, + "grad_norm": 2.5423855781555176, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7416698336601257, + "num_tokens": 561118285.0, + "step": 21690 + }, + { + "epoch": 2.3820557873929276, + "grad_norm": 2.2232866287231445, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7130032777786255, + "num_tokens": 561147901.0, + "step": 21691 + }, + { + "epoch": 2.3821656050955413, + "grad_norm": 2.1709704399108887, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7260030508041382, + "num_tokens": 561176893.0, + "step": 21692 + }, + { + "epoch": 2.382275422798155, + "grad_norm": 2.2063851356506348, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6908845901489258, + "num_tokens": 561207332.0, + "step": 21693 + }, + { + "epoch": 2.382385240500769, + "grad_norm": 2.609206438064575, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7414082288742065, + "num_tokens": 561228511.0, + "step": 21694 + }, + { + "epoch": 2.382495058203382, + "grad_norm": 2.2743890285491943, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7239727973937988, + "num_tokens": 561256955.0, + "step": 21695 + }, + { + "epoch": 2.382604875905996, + "grad_norm": 2.5420188903808594, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7378026247024536, + "num_tokens": 561279513.0, + "step": 21696 + }, + { + "epoch": 2.3827146936086097, + "grad_norm": 2.2593507766723633, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7228761911392212, + "num_tokens": 561307164.0, + "step": 21697 + }, + { + "epoch": 2.3828245113112234, + "grad_norm": 2.141054153442383, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6946418881416321, + "num_tokens": 561337569.0, + "step": 21698 + }, + { + "epoch": 2.382934329013837, + "grad_norm": 2.2316181659698486, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7220642566680908, + "num_tokens": 561367097.0, + "step": 21699 + }, + { + "epoch": 2.3830441467164505, + "grad_norm": 2.2634570598602295, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7227297425270081, + "num_tokens": 561394499.0, + "step": 21700 + }, + { + "epoch": 2.3831539644190642, + "grad_norm": 2.2767961025238037, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7302416563034058, + "num_tokens": 561422943.0, + "step": 21701 + }, + { + "epoch": 2.383263782121678, + "grad_norm": 2.4461841583251953, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7437598705291748, + "num_tokens": 561446362.0, + "step": 21702 + }, + { + "epoch": 2.3833735998242918, + "grad_norm": 2.44494366645813, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7200316190719604, + "num_tokens": 561471673.0, + "step": 21703 + }, + { + "epoch": 2.3834834175269055, + "grad_norm": 2.814641237258911, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7370376586914062, + "num_tokens": 561491792.0, + "step": 21704 + }, + { + "epoch": 2.383593235229519, + "grad_norm": 2.1381747722625732, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.70937180519104, + "num_tokens": 561523490.0, + "step": 21705 + }, + { + "epoch": 2.3837030529321326, + "grad_norm": 2.520498275756836, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7317997217178345, + "num_tokens": 561546471.0, + "step": 21706 + }, + { + "epoch": 2.3838128706347463, + "grad_norm": 2.5319948196411133, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7521514892578125, + "num_tokens": 561568689.0, + "step": 21707 + }, + { + "epoch": 2.38392268833736, + "grad_norm": 2.36100172996521, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7248989343643188, + "num_tokens": 561595060.0, + "step": 21708 + }, + { + "epoch": 2.384032506039974, + "grad_norm": 2.385462522506714, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7466428875923157, + "num_tokens": 561619795.0, + "step": 21709 + }, + { + "epoch": 2.384142323742587, + "grad_norm": 3.148545742034912, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.746491551399231, + "num_tokens": 561636235.0, + "step": 21710 + }, + { + "epoch": 2.384252141445201, + "grad_norm": 2.0989160537719727, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7510801553726196, + "num_tokens": 561665324.0, + "step": 21711 + }, + { + "epoch": 2.3843619591478147, + "grad_norm": 2.5515732765197754, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7129251956939697, + "num_tokens": 561686989.0, + "step": 21712 + }, + { + "epoch": 2.3844717768504284, + "grad_norm": 2.3776004314422607, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7212606072425842, + "num_tokens": 561713475.0, + "step": 21713 + }, + { + "epoch": 2.384581594553042, + "grad_norm": 2.2165589332580566, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7312970161437988, + "num_tokens": 561740807.0, + "step": 21714 + }, + { + "epoch": 2.3846914122556555, + "grad_norm": 2.499152898788452, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.74528568983078, + "num_tokens": 561764139.0, + "step": 21715 + }, + { + "epoch": 2.3848012299582693, + "grad_norm": 2.2072055339813232, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7478779554367065, + "num_tokens": 561791526.0, + "step": 21716 + }, + { + "epoch": 2.384911047660883, + "grad_norm": 2.669032335281372, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7343190908432007, + "num_tokens": 561813069.0, + "step": 21717 + }, + { + "epoch": 2.385020865363497, + "grad_norm": 2.2639851570129395, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7421153783798218, + "num_tokens": 561838720.0, + "step": 21718 + }, + { + "epoch": 2.38513068306611, + "grad_norm": 2.4589505195617676, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7399811744689941, + "num_tokens": 561864126.0, + "step": 21719 + }, + { + "epoch": 2.385240500768724, + "grad_norm": 2.4085960388183594, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7338027954101562, + "num_tokens": 561889744.0, + "step": 21720 + }, + { + "epoch": 2.3853503184713376, + "grad_norm": 2.3269612789154053, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7305208444595337, + "num_tokens": 561914889.0, + "step": 21721 + }, + { + "epoch": 2.3854601361739514, + "grad_norm": 2.4767675399780273, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7309970855712891, + "num_tokens": 561939297.0, + "step": 21722 + }, + { + "epoch": 2.3855699538765647, + "grad_norm": 2.3556063175201416, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7326348423957825, + "num_tokens": 561965447.0, + "step": 21723 + }, + { + "epoch": 2.3856797715791784, + "grad_norm": 2.290980815887451, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7018533945083618, + "num_tokens": 561994561.0, + "step": 21724 + }, + { + "epoch": 2.385789589281792, + "grad_norm": 1.9718732833862305, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7465853691101074, + "num_tokens": 562028541.0, + "step": 21725 + }, + { + "epoch": 2.385899406984406, + "grad_norm": 2.166949510574341, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.73316490650177, + "num_tokens": 562059217.0, + "step": 21726 + }, + { + "epoch": 2.3860092246870197, + "grad_norm": 2.294999122619629, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.740129828453064, + "num_tokens": 562085241.0, + "step": 21727 + }, + { + "epoch": 2.386119042389633, + "grad_norm": 2.4933533668518066, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7471365928649902, + "num_tokens": 562109433.0, + "step": 21728 + }, + { + "epoch": 2.3862288600922468, + "grad_norm": 2.7512123584747314, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.736896276473999, + "num_tokens": 562130464.0, + "step": 21729 + }, + { + "epoch": 2.3863386777948605, + "grad_norm": 2.5151989459991455, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7443167567253113, + "num_tokens": 562154334.0, + "step": 21730 + }, + { + "epoch": 2.3864484954974743, + "grad_norm": 2.3328232765197754, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7517554759979248, + "num_tokens": 562179353.0, + "step": 21731 + }, + { + "epoch": 2.386558313200088, + "grad_norm": 2.233255624771118, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7334825992584229, + "num_tokens": 562209174.0, + "step": 21732 + }, + { + "epoch": 2.3866681309027014, + "grad_norm": 2.306701183319092, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7456409931182861, + "num_tokens": 562238300.0, + "step": 21733 + }, + { + "epoch": 2.386777948605315, + "grad_norm": 2.2605607509613037, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7196673154830933, + "num_tokens": 562267018.0, + "step": 21734 + }, + { + "epoch": 2.386887766307929, + "grad_norm": 2.1164004802703857, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7571619749069214, + "num_tokens": 562296651.0, + "step": 21735 + }, + { + "epoch": 2.3869975840105426, + "grad_norm": 2.5073611736297607, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7222047448158264, + "num_tokens": 562322156.0, + "step": 21736 + }, + { + "epoch": 2.3871074017131564, + "grad_norm": 2.522434711456299, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.73704993724823, + "num_tokens": 562344786.0, + "step": 21737 + }, + { + "epoch": 2.3872172194157697, + "grad_norm": 2.483252763748169, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7326020002365112, + "num_tokens": 562368680.0, + "step": 21738 + }, + { + "epoch": 2.3873270371183835, + "grad_norm": 2.2419631481170654, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7107340097427368, + "num_tokens": 562396086.0, + "step": 21739 + }, + { + "epoch": 2.387436854820997, + "grad_norm": 2.2730462551116943, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7315011620521545, + "num_tokens": 562424322.0, + "step": 21740 + }, + { + "epoch": 2.387546672523611, + "grad_norm": 2.415587902069092, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.707237720489502, + "num_tokens": 562449818.0, + "step": 21741 + }, + { + "epoch": 2.3876564902262243, + "grad_norm": 2.2114996910095215, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7298884391784668, + "num_tokens": 562479687.0, + "step": 21742 + }, + { + "epoch": 2.387766307928838, + "grad_norm": 2.6155943870544434, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7534749507904053, + "num_tokens": 562500708.0, + "step": 21743 + }, + { + "epoch": 2.387876125631452, + "grad_norm": 2.570891857147217, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7173417806625366, + "num_tokens": 562524041.0, + "step": 21744 + }, + { + "epoch": 2.3879859433340656, + "grad_norm": 2.4463982582092285, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7272783517837524, + "num_tokens": 562551157.0, + "step": 21745 + }, + { + "epoch": 2.388095761036679, + "grad_norm": 2.0646047592163086, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7390461564064026, + "num_tokens": 562583952.0, + "step": 21746 + }, + { + "epoch": 2.3882055787392926, + "grad_norm": 2.5871565341949463, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7408381104469299, + "num_tokens": 562606927.0, + "step": 21747 + }, + { + "epoch": 2.3883153964419064, + "grad_norm": 2.741312265396118, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7306461334228516, + "num_tokens": 562627104.0, + "step": 21748 + }, + { + "epoch": 2.38842521414452, + "grad_norm": 2.595900058746338, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7265352010726929, + "num_tokens": 562649867.0, + "step": 21749 + }, + { + "epoch": 2.388535031847134, + "grad_norm": 2.622375965118408, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7155443429946899, + "num_tokens": 562672739.0, + "step": 21750 + }, + { + "epoch": 2.388644849549747, + "grad_norm": 2.4047951698303223, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7157059907913208, + "num_tokens": 562699666.0, + "step": 21751 + }, + { + "epoch": 2.388754667252361, + "grad_norm": 2.633334159851074, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7481004595756531, + "num_tokens": 562719402.0, + "step": 21752 + }, + { + "epoch": 2.3888644849549747, + "grad_norm": 2.5627262592315674, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7255162000656128, + "num_tokens": 562743460.0, + "step": 21753 + }, + { + "epoch": 2.3889743026575885, + "grad_norm": 2.565746307373047, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7299742698669434, + "num_tokens": 562766855.0, + "step": 21754 + }, + { + "epoch": 2.3890841203602022, + "grad_norm": 2.537208318710327, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7373640537261963, + "num_tokens": 562789578.0, + "step": 21755 + }, + { + "epoch": 2.3891939380628155, + "grad_norm": 2.254467487335205, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7400170564651489, + "num_tokens": 562816256.0, + "step": 21756 + }, + { + "epoch": 2.3893037557654293, + "grad_norm": 2.458949327468872, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.6970689296722412, + "num_tokens": 562842188.0, + "step": 21757 + }, + { + "epoch": 2.389413573468043, + "grad_norm": 2.2683451175689697, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7352097034454346, + "num_tokens": 562870493.0, + "step": 21758 + }, + { + "epoch": 2.389523391170657, + "grad_norm": 2.3464815616607666, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7339342832565308, + "num_tokens": 562896126.0, + "step": 21759 + }, + { + "epoch": 2.3896332088732706, + "grad_norm": 2.7109410762786865, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7475470304489136, + "num_tokens": 562916736.0, + "step": 21760 + }, + { + "epoch": 2.389743026575884, + "grad_norm": 2.074049949645996, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7208749055862427, + "num_tokens": 562948370.0, + "step": 21761 + }, + { + "epoch": 2.3898528442784976, + "grad_norm": 2.2298548221588135, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7467571496963501, + "num_tokens": 562974389.0, + "step": 21762 + }, + { + "epoch": 2.3899626619811114, + "grad_norm": 2.2279551029205322, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7326170802116394, + "num_tokens": 563001419.0, + "step": 21763 + }, + { + "epoch": 2.390072479683725, + "grad_norm": 2.398219108581543, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7125420570373535, + "num_tokens": 563028210.0, + "step": 21764 + }, + { + "epoch": 2.390182297386339, + "grad_norm": 2.312933921813965, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7425587177276611, + "num_tokens": 563052099.0, + "step": 21765 + }, + { + "epoch": 2.3902921150889522, + "grad_norm": 2.6740663051605225, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7208821177482605, + "num_tokens": 563074501.0, + "step": 21766 + }, + { + "epoch": 2.390401932791566, + "grad_norm": 2.122044324874878, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7295207381248474, + "num_tokens": 563105438.0, + "step": 21767 + }, + { + "epoch": 2.3905117504941797, + "grad_norm": 2.498098373413086, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7340373396873474, + "num_tokens": 563127829.0, + "step": 21768 + }, + { + "epoch": 2.3906215681967935, + "grad_norm": 2.544996738433838, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.735938549041748, + "num_tokens": 563150649.0, + "step": 21769 + }, + { + "epoch": 2.390731385899407, + "grad_norm": 2.405601978302002, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.723415732383728, + "num_tokens": 563174548.0, + "step": 21770 + }, + { + "epoch": 2.3908412036020206, + "grad_norm": 1.966421127319336, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7104994058609009, + "num_tokens": 563211815.0, + "step": 21771 + }, + { + "epoch": 2.3909510213046343, + "grad_norm": 2.2039449214935303, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7166933417320251, + "num_tokens": 563240838.0, + "step": 21772 + }, + { + "epoch": 2.391060839007248, + "grad_norm": 2.367669105529785, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7337319254875183, + "num_tokens": 563266701.0, + "step": 21773 + }, + { + "epoch": 2.3911706567098614, + "grad_norm": 2.61376953125, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7549564242362976, + "num_tokens": 563290876.0, + "step": 21774 + }, + { + "epoch": 2.391280474412475, + "grad_norm": 2.716198205947876, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7496510744094849, + "num_tokens": 563310841.0, + "step": 21775 + }, + { + "epoch": 2.391390292115089, + "grad_norm": 2.4442849159240723, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7500284910202026, + "num_tokens": 563335458.0, + "step": 21776 + }, + { + "epoch": 2.3915001098177027, + "grad_norm": 2.2374954223632812, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7546788454055786, + "num_tokens": 563362466.0, + "step": 21777 + }, + { + "epoch": 2.3916099275203164, + "grad_norm": 2.17387056350708, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7329725027084351, + "num_tokens": 563391871.0, + "step": 21778 + }, + { + "epoch": 2.3917197452229297, + "grad_norm": 2.4505088329315186, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7246853113174438, + "num_tokens": 563417280.0, + "step": 21779 + }, + { + "epoch": 2.3918295629255435, + "grad_norm": 2.187863826751709, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7148084044456482, + "num_tokens": 563446860.0, + "step": 21780 + }, + { + "epoch": 2.3919393806281573, + "grad_norm": 2.195772886276245, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7286748886108398, + "num_tokens": 563476694.0, + "step": 21781 + }, + { + "epoch": 2.392049198330771, + "grad_norm": 2.4137842655181885, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7493702173233032, + "num_tokens": 563501967.0, + "step": 21782 + }, + { + "epoch": 2.3921590160333848, + "grad_norm": 2.5595297813415527, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.723095178604126, + "num_tokens": 563533006.0, + "step": 21783 + }, + { + "epoch": 2.392268833735998, + "grad_norm": 2.469496726989746, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7550622224807739, + "num_tokens": 563556085.0, + "step": 21784 + }, + { + "epoch": 2.392378651438612, + "grad_norm": 2.2749857902526855, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.734510600566864, + "num_tokens": 563583818.0, + "step": 21785 + }, + { + "epoch": 2.3924884691412256, + "grad_norm": 2.266782522201538, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7209016680717468, + "num_tokens": 563611700.0, + "step": 21786 + }, + { + "epoch": 2.3925982868438394, + "grad_norm": 2.640892744064331, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.727198600769043, + "num_tokens": 563635091.0, + "step": 21787 + }, + { + "epoch": 2.392708104546453, + "grad_norm": 2.2493104934692383, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7335718870162964, + "num_tokens": 563664243.0, + "step": 21788 + }, + { + "epoch": 2.3928179222490664, + "grad_norm": 2.6145377159118652, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7414634227752686, + "num_tokens": 563687430.0, + "step": 21789 + }, + { + "epoch": 2.39292773995168, + "grad_norm": 2.617840051651001, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7263810634613037, + "num_tokens": 563716757.0, + "step": 21790 + }, + { + "epoch": 2.393037557654294, + "grad_norm": 2.495675563812256, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7465869784355164, + "num_tokens": 563739681.0, + "step": 21791 + }, + { + "epoch": 2.3931473753569077, + "grad_norm": 2.408221483230591, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7423148155212402, + "num_tokens": 563763013.0, + "step": 21792 + }, + { + "epoch": 2.393257193059521, + "grad_norm": 2.3757729530334473, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7099194526672363, + "num_tokens": 563789203.0, + "step": 21793 + }, + { + "epoch": 2.3933670107621348, + "grad_norm": 2.4818859100341797, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.736649751663208, + "num_tokens": 563813546.0, + "step": 21794 + }, + { + "epoch": 2.3934768284647485, + "grad_norm": 2.432532787322998, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7220954895019531, + "num_tokens": 563840301.0, + "step": 21795 + }, + { + "epoch": 2.3935866461673623, + "grad_norm": 2.3602020740509033, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7248826622962952, + "num_tokens": 563867125.0, + "step": 21796 + }, + { + "epoch": 2.393696463869976, + "grad_norm": 2.6140873432159424, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.73384690284729, + "num_tokens": 563888995.0, + "step": 21797 + }, + { + "epoch": 2.3938062815725893, + "grad_norm": 2.282118082046509, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7374440431594849, + "num_tokens": 563914370.0, + "step": 21798 + }, + { + "epoch": 2.393916099275203, + "grad_norm": 2.320384979248047, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7416710257530212, + "num_tokens": 563941515.0, + "step": 21799 + }, + { + "epoch": 2.394025916977817, + "grad_norm": 2.320833921432495, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7502487301826477, + "num_tokens": 563967176.0, + "step": 21800 + }, + { + "epoch": 2.3941357346804306, + "grad_norm": 2.546057939529419, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7473787069320679, + "num_tokens": 563989351.0, + "step": 21801 + }, + { + "epoch": 2.394245552383044, + "grad_norm": 2.248866081237793, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7383171319961548, + "num_tokens": 564015982.0, + "step": 21802 + }, + { + "epoch": 2.3943553700856577, + "grad_norm": 2.2214460372924805, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7407734394073486, + "num_tokens": 564043171.0, + "step": 21803 + }, + { + "epoch": 2.3944651877882714, + "grad_norm": 2.145922899246216, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7197743654251099, + "num_tokens": 564073133.0, + "step": 21804 + }, + { + "epoch": 2.394575005490885, + "grad_norm": 2.312551975250244, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7638070583343506, + "num_tokens": 564099305.0, + "step": 21805 + }, + { + "epoch": 2.394684823193499, + "grad_norm": 2.2521231174468994, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7465216517448425, + "num_tokens": 564124842.0, + "step": 21806 + }, + { + "epoch": 2.3947946408961123, + "grad_norm": 2.1444270610809326, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7107678651809692, + "num_tokens": 564158542.0, + "step": 21807 + }, + { + "epoch": 2.394904458598726, + "grad_norm": 2.318535089492798, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7287114262580872, + "num_tokens": 564185746.0, + "step": 21808 + }, + { + "epoch": 2.39501427630134, + "grad_norm": 2.4619178771972656, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7387380599975586, + "num_tokens": 564208325.0, + "step": 21809 + }, + { + "epoch": 2.3951240940039535, + "grad_norm": 2.19262433052063, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7287888526916504, + "num_tokens": 564237384.0, + "step": 21810 + }, + { + "epoch": 2.3952339117065673, + "grad_norm": 2.504084348678589, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.719129204750061, + "num_tokens": 564263098.0, + "step": 21811 + }, + { + "epoch": 2.3953437294091806, + "grad_norm": 2.4405603408813477, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7148374319076538, + "num_tokens": 564287499.0, + "step": 21812 + }, + { + "epoch": 2.3954535471117944, + "grad_norm": 2.725539445877075, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7387009263038635, + "num_tokens": 564307598.0, + "step": 21813 + }, + { + "epoch": 2.395563364814408, + "grad_norm": 2.2135255336761475, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.711540937423706, + "num_tokens": 564338375.0, + "step": 21814 + }, + { + "epoch": 2.395673182517022, + "grad_norm": 2.428960084915161, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7349971532821655, + "num_tokens": 564361749.0, + "step": 21815 + }, + { + "epoch": 2.3957830002196356, + "grad_norm": 2.339035987854004, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7275770306587219, + "num_tokens": 564387269.0, + "step": 21816 + }, + { + "epoch": 2.395892817922249, + "grad_norm": 2.358370542526245, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7201963663101196, + "num_tokens": 564411993.0, + "step": 21817 + }, + { + "epoch": 2.3960026356248627, + "grad_norm": 2.2768101692199707, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7405252456665039, + "num_tokens": 564437303.0, + "step": 21818 + }, + { + "epoch": 2.3961124533274765, + "grad_norm": 2.343285083770752, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7349376678466797, + "num_tokens": 564467197.0, + "step": 21819 + }, + { + "epoch": 2.3962222710300902, + "grad_norm": 2.0617003440856934, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7230865359306335, + "num_tokens": 564501039.0, + "step": 21820 + }, + { + "epoch": 2.3963320887327035, + "grad_norm": 2.2070648670196533, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7239946722984314, + "num_tokens": 564529656.0, + "step": 21821 + }, + { + "epoch": 2.3964419064353173, + "grad_norm": 2.0646657943725586, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7341932058334351, + "num_tokens": 564560411.0, + "step": 21822 + }, + { + "epoch": 2.396551724137931, + "grad_norm": 2.1440024375915527, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7351475954055786, + "num_tokens": 564589332.0, + "step": 21823 + }, + { + "epoch": 2.396661541840545, + "grad_norm": 2.1893656253814697, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7429803013801575, + "num_tokens": 564617369.0, + "step": 21824 + }, + { + "epoch": 2.396771359543158, + "grad_norm": 2.586366891860962, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7255972623825073, + "num_tokens": 564640388.0, + "step": 21825 + }, + { + "epoch": 2.396881177245772, + "grad_norm": 2.5087788105010986, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7138023972511292, + "num_tokens": 564664886.0, + "step": 21826 + }, + { + "epoch": 2.3969909949483856, + "grad_norm": 2.3098888397216797, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7190111875534058, + "num_tokens": 564693525.0, + "step": 21827 + }, + { + "epoch": 2.3971008126509994, + "grad_norm": 2.36800217628479, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7241872549057007, + "num_tokens": 564721609.0, + "step": 21828 + }, + { + "epoch": 2.397210630353613, + "grad_norm": 2.3687734603881836, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7321544289588928, + "num_tokens": 564747590.0, + "step": 21829 + }, + { + "epoch": 2.3973204480562265, + "grad_norm": 2.090765953063965, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7038004398345947, + "num_tokens": 564778965.0, + "step": 21830 + }, + { + "epoch": 2.39743026575884, + "grad_norm": 2.5006840229034424, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7405371069908142, + "num_tokens": 564800943.0, + "step": 21831 + }, + { + "epoch": 2.397540083461454, + "grad_norm": 2.391288995742798, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7105860710144043, + "num_tokens": 564826490.0, + "step": 21832 + }, + { + "epoch": 2.3976499011640677, + "grad_norm": 2.437145948410034, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7180522084236145, + "num_tokens": 564852336.0, + "step": 21833 + }, + { + "epoch": 2.3977597188666815, + "grad_norm": 2.4223172664642334, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7409319281578064, + "num_tokens": 564876595.0, + "step": 21834 + }, + { + "epoch": 2.397869536569295, + "grad_norm": 2.315413236618042, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7288085222244263, + "num_tokens": 564906207.0, + "step": 21835 + }, + { + "epoch": 2.3979793542719086, + "grad_norm": 2.7508440017700195, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.742235541343689, + "num_tokens": 564925607.0, + "step": 21836 + }, + { + "epoch": 2.3980891719745223, + "grad_norm": 2.4629619121551514, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7463374137878418, + "num_tokens": 564948981.0, + "step": 21837 + }, + { + "epoch": 2.398198989677136, + "grad_norm": 2.121546745300293, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.719921350479126, + "num_tokens": 564980853.0, + "step": 21838 + }, + { + "epoch": 2.39830880737975, + "grad_norm": 2.215409755706787, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7112483978271484, + "num_tokens": 565010521.0, + "step": 21839 + }, + { + "epoch": 2.398418625082363, + "grad_norm": 2.719395637512207, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7554014325141907, + "num_tokens": 565030037.0, + "step": 21840 + }, + { + "epoch": 2.398528442784977, + "grad_norm": 2.4249024391174316, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7245721817016602, + "num_tokens": 565056268.0, + "step": 21841 + }, + { + "epoch": 2.3986382604875907, + "grad_norm": 2.347452163696289, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7168195247650146, + "num_tokens": 565082378.0, + "step": 21842 + }, + { + "epoch": 2.3987480781902044, + "grad_norm": 2.908554792404175, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7477423548698425, + "num_tokens": 565100237.0, + "step": 21843 + }, + { + "epoch": 2.398857895892818, + "grad_norm": 2.3482515811920166, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7175467610359192, + "num_tokens": 565125361.0, + "step": 21844 + }, + { + "epoch": 2.3989677135954315, + "grad_norm": 2.608665943145752, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7590245008468628, + "num_tokens": 565146362.0, + "step": 21845 + }, + { + "epoch": 2.3990775312980452, + "grad_norm": 2.5244925022125244, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7514845132827759, + "num_tokens": 565168609.0, + "step": 21846 + }, + { + "epoch": 2.399187349000659, + "grad_norm": 2.4820618629455566, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7216532230377197, + "num_tokens": 565192456.0, + "step": 21847 + }, + { + "epoch": 2.3992971667032728, + "grad_norm": 2.2601869106292725, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.718667209148407, + "num_tokens": 565221728.0, + "step": 21848 + }, + { + "epoch": 2.399406984405886, + "grad_norm": 2.12941312789917, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7377699613571167, + "num_tokens": 565253334.0, + "step": 21849 + }, + { + "epoch": 2.3995168021085, + "grad_norm": 2.3384602069854736, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7188064455986023, + "num_tokens": 565280534.0, + "step": 21850 + }, + { + "epoch": 2.3996266198111136, + "grad_norm": 2.3279576301574707, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7309035062789917, + "num_tokens": 565307008.0, + "step": 21851 + }, + { + "epoch": 2.3997364375137273, + "grad_norm": 2.5225486755371094, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7404775619506836, + "num_tokens": 565329287.0, + "step": 21852 + }, + { + "epoch": 2.3998462552163407, + "grad_norm": 2.1489756107330322, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7435578107833862, + "num_tokens": 565358958.0, + "step": 21853 + }, + { + "epoch": 2.3999560729189544, + "grad_norm": 2.265549659729004, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7479680776596069, + "num_tokens": 565388515.0, + "step": 21854 + }, + { + "epoch": 2.400065890621568, + "grad_norm": 2.5833189487457275, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7320554256439209, + "num_tokens": 565411671.0, + "step": 21855 + }, + { + "epoch": 2.400175708324182, + "grad_norm": 2.4765055179595947, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7456151247024536, + "num_tokens": 565434986.0, + "step": 21856 + }, + { + "epoch": 2.4002855260267957, + "grad_norm": 2.324414014816284, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7277148962020874, + "num_tokens": 565460261.0, + "step": 21857 + }, + { + "epoch": 2.400395343729409, + "grad_norm": 2.2263193130493164, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7187917828559875, + "num_tokens": 565489292.0, + "step": 21858 + }, + { + "epoch": 2.4005051614320227, + "grad_norm": 2.278164863586426, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7580989599227905, + "num_tokens": 565512923.0, + "step": 21859 + }, + { + "epoch": 2.4006149791346365, + "grad_norm": 2.3038644790649414, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.6993759870529175, + "num_tokens": 565542048.0, + "step": 21860 + }, + { + "epoch": 2.4007247968372503, + "grad_norm": 2.226473808288574, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.715930700302124, + "num_tokens": 565571858.0, + "step": 21861 + }, + { + "epoch": 2.400834614539864, + "grad_norm": 2.4145936965942383, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7205682396888733, + "num_tokens": 565597292.0, + "step": 21862 + }, + { + "epoch": 2.4009444322424773, + "grad_norm": 2.2849605083465576, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7421349883079529, + "num_tokens": 565622850.0, + "step": 21863 + }, + { + "epoch": 2.401054249945091, + "grad_norm": 2.531900405883789, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7245035171508789, + "num_tokens": 565645293.0, + "step": 21864 + }, + { + "epoch": 2.401164067647705, + "grad_norm": 2.3573386669158936, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.770565390586853, + "num_tokens": 565673669.0, + "step": 21865 + }, + { + "epoch": 2.4012738853503186, + "grad_norm": 2.182812452316284, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7101386785507202, + "num_tokens": 565703567.0, + "step": 21866 + }, + { + "epoch": 2.4013837030529324, + "grad_norm": 2.3654685020446777, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7183654308319092, + "num_tokens": 565728942.0, + "step": 21867 + }, + { + "epoch": 2.4014935207555457, + "grad_norm": 2.083207845687866, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7305626273155212, + "num_tokens": 565760420.0, + "step": 21868 + }, + { + "epoch": 2.4016033384581594, + "grad_norm": 2.5173611640930176, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7452927827835083, + "num_tokens": 565782604.0, + "step": 21869 + }, + { + "epoch": 2.401713156160773, + "grad_norm": 2.6461291313171387, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7210335731506348, + "num_tokens": 565803210.0, + "step": 21870 + }, + { + "epoch": 2.401822973863387, + "grad_norm": 2.2516026496887207, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7296751737594604, + "num_tokens": 565830695.0, + "step": 21871 + }, + { + "epoch": 2.4019327915660003, + "grad_norm": 2.4034998416900635, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7308282852172852, + "num_tokens": 565855945.0, + "step": 21872 + }, + { + "epoch": 2.402042609268614, + "grad_norm": 2.760467052459717, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7428404092788696, + "num_tokens": 565875323.0, + "step": 21873 + }, + { + "epoch": 2.4021524269712278, + "grad_norm": 2.3009612560272217, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7062551379203796, + "num_tokens": 565904005.0, + "step": 21874 + }, + { + "epoch": 2.4022622446738415, + "grad_norm": 2.1462080478668213, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7197614908218384, + "num_tokens": 565933325.0, + "step": 21875 + }, + { + "epoch": 2.402372062376455, + "grad_norm": 2.8297295570373535, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7311235666275024, + "num_tokens": 565952176.0, + "step": 21876 + }, + { + "epoch": 2.4024818800790686, + "grad_norm": 2.4819271564483643, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.746645450592041, + "num_tokens": 565975579.0, + "step": 21877 + }, + { + "epoch": 2.4025916977816824, + "grad_norm": 2.1991870403289795, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7169855237007141, + "num_tokens": 566003823.0, + "step": 21878 + }, + { + "epoch": 2.402701515484296, + "grad_norm": 2.342067241668701, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7383164167404175, + "num_tokens": 566031231.0, + "step": 21879 + }, + { + "epoch": 2.40281133318691, + "grad_norm": 2.5971298217773438, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7365812063217163, + "num_tokens": 566053359.0, + "step": 21880 + }, + { + "epoch": 2.402921150889523, + "grad_norm": 2.3055167198181152, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7373688817024231, + "num_tokens": 566081184.0, + "step": 21881 + }, + { + "epoch": 2.403030968592137, + "grad_norm": 2.424222946166992, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.721450924873352, + "num_tokens": 566107003.0, + "step": 21882 + }, + { + "epoch": 2.4031407862947507, + "grad_norm": 2.39379620552063, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7226020097732544, + "num_tokens": 566131996.0, + "step": 21883 + }, + { + "epoch": 2.4032506039973645, + "grad_norm": 2.303729772567749, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7414172291755676, + "num_tokens": 566156993.0, + "step": 21884 + }, + { + "epoch": 2.403360421699978, + "grad_norm": 2.381498098373413, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7167075872421265, + "num_tokens": 566181627.0, + "step": 21885 + }, + { + "epoch": 2.4034702394025915, + "grad_norm": 2.4478869438171387, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7291555404663086, + "num_tokens": 566207643.0, + "step": 21886 + }, + { + "epoch": 2.4035800571052053, + "grad_norm": 2.4074175357818604, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7284836173057556, + "num_tokens": 566232623.0, + "step": 21887 + }, + { + "epoch": 2.403689874807819, + "grad_norm": 2.3955841064453125, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7235651016235352, + "num_tokens": 566259132.0, + "step": 21888 + }, + { + "epoch": 2.403799692510433, + "grad_norm": 2.5038671493530273, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7515802383422852, + "num_tokens": 566283076.0, + "step": 21889 + }, + { + "epoch": 2.4039095102130466, + "grad_norm": 2.453366994857788, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7426168322563171, + "num_tokens": 566307400.0, + "step": 21890 + }, + { + "epoch": 2.40401932791566, + "grad_norm": 2.872708797454834, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.738927960395813, + "num_tokens": 566326041.0, + "step": 21891 + }, + { + "epoch": 2.4041291456182736, + "grad_norm": 2.241558074951172, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7325187921524048, + "num_tokens": 566355322.0, + "step": 21892 + }, + { + "epoch": 2.4042389633208874, + "grad_norm": 2.312100410461426, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7320905923843384, + "num_tokens": 566381855.0, + "step": 21893 + }, + { + "epoch": 2.404348781023501, + "grad_norm": 2.484710216522217, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7467703223228455, + "num_tokens": 566405157.0, + "step": 21894 + }, + { + "epoch": 2.404458598726115, + "grad_norm": 2.1895148754119873, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7553056478500366, + "num_tokens": 566432532.0, + "step": 21895 + }, + { + "epoch": 2.404568416428728, + "grad_norm": 2.8219797611236572, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7446395754814148, + "num_tokens": 566452057.0, + "step": 21896 + }, + { + "epoch": 2.404678234131342, + "grad_norm": 2.694108486175537, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.743830680847168, + "num_tokens": 566472312.0, + "step": 21897 + }, + { + "epoch": 2.4047880518339557, + "grad_norm": 2.4716975688934326, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7396259903907776, + "num_tokens": 566498063.0, + "step": 21898 + }, + { + "epoch": 2.4048978695365695, + "grad_norm": 2.5453550815582275, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7349613308906555, + "num_tokens": 566520570.0, + "step": 21899 + }, + { + "epoch": 2.405007687239183, + "grad_norm": 2.1757967472076416, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7287523746490479, + "num_tokens": 566547892.0, + "step": 21900 + }, + { + "epoch": 2.4051175049417965, + "grad_norm": 2.5165417194366455, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7412183880805969, + "num_tokens": 566569457.0, + "step": 21901 + }, + { + "epoch": 2.4052273226444103, + "grad_norm": 2.092860221862793, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7247858047485352, + "num_tokens": 566601355.0, + "step": 21902 + }, + { + "epoch": 2.405337140347024, + "grad_norm": 2.5477383136749268, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7450196743011475, + "num_tokens": 566622944.0, + "step": 21903 + }, + { + "epoch": 2.4054469580496374, + "grad_norm": 2.3479065895080566, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7168484926223755, + "num_tokens": 566649726.0, + "step": 21904 + }, + { + "epoch": 2.405556775752251, + "grad_norm": 2.4389684200286865, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7147225141525269, + "num_tokens": 566676527.0, + "step": 21905 + }, + { + "epoch": 2.405666593454865, + "grad_norm": 2.199084758758545, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7352210879325867, + "num_tokens": 566703545.0, + "step": 21906 + }, + { + "epoch": 2.4057764111574786, + "grad_norm": 2.282325267791748, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7505820393562317, + "num_tokens": 566730234.0, + "step": 21907 + }, + { + "epoch": 2.4058862288600924, + "grad_norm": 2.292994976043701, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7311737537384033, + "num_tokens": 566755985.0, + "step": 21908 + }, + { + "epoch": 2.4059960465627057, + "grad_norm": 2.4117138385772705, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7283099889755249, + "num_tokens": 566780924.0, + "step": 21909 + }, + { + "epoch": 2.4061058642653195, + "grad_norm": 2.4984378814697266, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7446539402008057, + "num_tokens": 566805480.0, + "step": 21910 + }, + { + "epoch": 2.4062156819679332, + "grad_norm": 2.4816880226135254, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6971420645713806, + "num_tokens": 566832750.0, + "step": 21911 + }, + { + "epoch": 2.406325499670547, + "grad_norm": 2.4778873920440674, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7395963072776794, + "num_tokens": 566855160.0, + "step": 21912 + }, + { + "epoch": 2.4064353173731607, + "grad_norm": 2.668149471282959, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.741661787033081, + "num_tokens": 566876672.0, + "step": 21913 + }, + { + "epoch": 2.406545135075774, + "grad_norm": 2.253021478652954, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7150238752365112, + "num_tokens": 566906330.0, + "step": 21914 + }, + { + "epoch": 2.406654952778388, + "grad_norm": 2.4705207347869873, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7192809581756592, + "num_tokens": 566930922.0, + "step": 21915 + }, + { + "epoch": 2.4067647704810016, + "grad_norm": 2.239046096801758, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.7465050220489502, + "num_tokens": 566957596.0, + "step": 21916 + }, + { + "epoch": 2.4068745881836153, + "grad_norm": 2.384676694869995, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7292783260345459, + "num_tokens": 566982957.0, + "step": 21917 + }, + { + "epoch": 2.406984405886229, + "grad_norm": 2.4401307106018066, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7312394380569458, + "num_tokens": 567006482.0, + "step": 21918 + }, + { + "epoch": 2.4070942235888424, + "grad_norm": 2.3395979404449463, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7386409044265747, + "num_tokens": 567032029.0, + "step": 21919 + }, + { + "epoch": 2.407204041291456, + "grad_norm": 2.437213182449341, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7346854209899902, + "num_tokens": 567057629.0, + "step": 21920 + }, + { + "epoch": 2.40731385899407, + "grad_norm": 2.321598529815674, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7249277234077454, + "num_tokens": 567085534.0, + "step": 21921 + }, + { + "epoch": 2.4074236766966837, + "grad_norm": 2.107764720916748, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7183487415313721, + "num_tokens": 567118239.0, + "step": 21922 + }, + { + "epoch": 2.407533494399297, + "grad_norm": 2.5336530208587646, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7450251579284668, + "num_tokens": 567141682.0, + "step": 21923 + }, + { + "epoch": 2.4076433121019107, + "grad_norm": 2.0951321125030518, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7305905818939209, + "num_tokens": 567174548.0, + "step": 21924 + }, + { + "epoch": 2.4077531298045245, + "grad_norm": 2.485870838165283, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7448019981384277, + "num_tokens": 567197996.0, + "step": 21925 + }, + { + "epoch": 2.4078629475071383, + "grad_norm": 2.0337071418762207, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7231431603431702, + "num_tokens": 567231224.0, + "step": 21926 + }, + { + "epoch": 2.4079727652097516, + "grad_norm": 2.400362253189087, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7254740595817566, + "num_tokens": 567257919.0, + "step": 21927 + }, + { + "epoch": 2.4080825829123653, + "grad_norm": 2.5107452869415283, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7252311706542969, + "num_tokens": 567283142.0, + "step": 21928 + }, + { + "epoch": 2.408192400614979, + "grad_norm": 2.3309333324432373, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7174624800682068, + "num_tokens": 567308445.0, + "step": 21929 + }, + { + "epoch": 2.408302218317593, + "grad_norm": 2.1697466373443604, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7284340262413025, + "num_tokens": 567337026.0, + "step": 21930 + }, + { + "epoch": 2.4084120360202066, + "grad_norm": 2.456975221633911, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7115234136581421, + "num_tokens": 567362751.0, + "step": 21931 + }, + { + "epoch": 2.40852185372282, + "grad_norm": 2.241217851638794, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7365390062332153, + "num_tokens": 567389259.0, + "step": 21932 + }, + { + "epoch": 2.4086316714254337, + "grad_norm": 2.407168388366699, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7338464260101318, + "num_tokens": 567413614.0, + "step": 21933 + }, + { + "epoch": 2.4087414891280474, + "grad_norm": 1.957326889038086, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7353760004043579, + "num_tokens": 567449268.0, + "step": 21934 + }, + { + "epoch": 2.408851306830661, + "grad_norm": 2.6600327491760254, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7521319389343262, + "num_tokens": 567469816.0, + "step": 21935 + }, + { + "epoch": 2.408961124533275, + "grad_norm": 2.4464845657348633, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7247704267501831, + "num_tokens": 567495710.0, + "step": 21936 + }, + { + "epoch": 2.4090709422358882, + "grad_norm": 2.568413734436035, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7444781064987183, + "num_tokens": 567518594.0, + "step": 21937 + }, + { + "epoch": 2.409180759938502, + "grad_norm": 2.3423783779144287, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.734774112701416, + "num_tokens": 567545070.0, + "step": 21938 + }, + { + "epoch": 2.4092905776411158, + "grad_norm": 2.2607319355010986, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7399903535842896, + "num_tokens": 567572442.0, + "step": 21939 + }, + { + "epoch": 2.4094003953437295, + "grad_norm": 2.496032238006592, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7233591079711914, + "num_tokens": 567597535.0, + "step": 21940 + }, + { + "epoch": 2.4095102130463433, + "grad_norm": 2.3464572429656982, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7063164710998535, + "num_tokens": 567625135.0, + "step": 21941 + }, + { + "epoch": 2.4096200307489566, + "grad_norm": 2.5207009315490723, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7492479085922241, + "num_tokens": 567647853.0, + "step": 21942 + }, + { + "epoch": 2.4097298484515703, + "grad_norm": 2.2224457263946533, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7201162576675415, + "num_tokens": 567676905.0, + "step": 21943 + }, + { + "epoch": 2.409839666154184, + "grad_norm": 2.561386823654175, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7529809474945068, + "num_tokens": 567698642.0, + "step": 21944 + }, + { + "epoch": 2.409949483856798, + "grad_norm": 2.172895908355713, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7375718951225281, + "num_tokens": 567729030.0, + "step": 21945 + }, + { + "epoch": 2.4100593015594116, + "grad_norm": 2.6947951316833496, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7542747855186462, + "num_tokens": 567747945.0, + "step": 21946 + }, + { + "epoch": 2.410169119262025, + "grad_norm": 2.363497734069824, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7290692329406738, + "num_tokens": 567773337.0, + "step": 21947 + }, + { + "epoch": 2.4102789369646387, + "grad_norm": 2.474191665649414, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7148447632789612, + "num_tokens": 567798204.0, + "step": 21948 + }, + { + "epoch": 2.4103887546672524, + "grad_norm": 2.438795804977417, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7233407497406006, + "num_tokens": 567824427.0, + "step": 21949 + }, + { + "epoch": 2.410498572369866, + "grad_norm": 2.184488296508789, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7505850791931152, + "num_tokens": 567853385.0, + "step": 21950 + }, + { + "epoch": 2.4106083900724795, + "grad_norm": 2.692392349243164, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7374106049537659, + "num_tokens": 567874455.0, + "step": 21951 + }, + { + "epoch": 2.4107182077750933, + "grad_norm": 2.143416404724121, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7319288849830627, + "num_tokens": 567904011.0, + "step": 21952 + }, + { + "epoch": 2.410828025477707, + "grad_norm": 2.383890390396118, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7147915959358215, + "num_tokens": 567929546.0, + "step": 21953 + }, + { + "epoch": 2.410937843180321, + "grad_norm": 2.323831796646118, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7671738862991333, + "num_tokens": 567955058.0, + "step": 21954 + }, + { + "epoch": 2.411047660882934, + "grad_norm": 2.318331003189087, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7444326281547546, + "num_tokens": 567982081.0, + "step": 21955 + }, + { + "epoch": 2.411157478585548, + "grad_norm": 2.415058135986328, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7352752685546875, + "num_tokens": 568007640.0, + "step": 21956 + }, + { + "epoch": 2.4112672962881616, + "grad_norm": 2.501805305480957, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7279825210571289, + "num_tokens": 568031630.0, + "step": 21957 + }, + { + "epoch": 2.4113771139907754, + "grad_norm": 2.3922269344329834, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7344454526901245, + "num_tokens": 568054579.0, + "step": 21958 + }, + { + "epoch": 2.411486931693389, + "grad_norm": 2.5028865337371826, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7489380240440369, + "num_tokens": 568078160.0, + "step": 21959 + }, + { + "epoch": 2.4115967493960024, + "grad_norm": 2.3814311027526855, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7252102494239807, + "num_tokens": 568102673.0, + "step": 21960 + }, + { + "epoch": 2.411706567098616, + "grad_norm": 2.25256085395813, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7334919571876526, + "num_tokens": 568129707.0, + "step": 21961 + }, + { + "epoch": 2.41181638480123, + "grad_norm": 2.131166696548462, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7346771955490112, + "num_tokens": 568159400.0, + "step": 21962 + }, + { + "epoch": 2.4119262025038437, + "grad_norm": 2.3469269275665283, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7244940400123596, + "num_tokens": 568184122.0, + "step": 21963 + }, + { + "epoch": 2.4120360202064575, + "grad_norm": 2.4563446044921875, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.719855546951294, + "num_tokens": 568207931.0, + "step": 21964 + }, + { + "epoch": 2.412145837909071, + "grad_norm": 2.09486722946167, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7313690185546875, + "num_tokens": 568239815.0, + "step": 21965 + }, + { + "epoch": 2.4122556556116845, + "grad_norm": 1.956827163696289, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7149733304977417, + "num_tokens": 568277237.0, + "step": 21966 + }, + { + "epoch": 2.4123654733142983, + "grad_norm": 2.093250274658203, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7202661037445068, + "num_tokens": 568311920.0, + "step": 21967 + }, + { + "epoch": 2.412475291016912, + "grad_norm": 2.6789658069610596, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7730123996734619, + "num_tokens": 568330760.0, + "step": 21968 + }, + { + "epoch": 2.412585108719526, + "grad_norm": 2.486851692199707, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7186006307601929, + "num_tokens": 568355507.0, + "step": 21969 + }, + { + "epoch": 2.412694926422139, + "grad_norm": 2.3168227672576904, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7211132049560547, + "num_tokens": 568382024.0, + "step": 21970 + }, + { + "epoch": 2.412804744124753, + "grad_norm": 2.5925791263580322, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7293684482574463, + "num_tokens": 568405147.0, + "step": 21971 + }, + { + "epoch": 2.4129145618273666, + "grad_norm": 2.2454891204833984, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7257179617881775, + "num_tokens": 568433274.0, + "step": 21972 + }, + { + "epoch": 2.4130243795299804, + "grad_norm": 2.481708526611328, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.734133243560791, + "num_tokens": 568457178.0, + "step": 21973 + }, + { + "epoch": 2.4131341972325937, + "grad_norm": 2.026099920272827, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7459262609481812, + "num_tokens": 568487821.0, + "step": 21974 + }, + { + "epoch": 2.4132440149352075, + "grad_norm": 2.57352876663208, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7277479767799377, + "num_tokens": 568511891.0, + "step": 21975 + }, + { + "epoch": 2.413353832637821, + "grad_norm": 2.614530324935913, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7325525283813477, + "num_tokens": 568533065.0, + "step": 21976 + }, + { + "epoch": 2.413463650340435, + "grad_norm": 2.2215700149536133, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7099473476409912, + "num_tokens": 568562977.0, + "step": 21977 + }, + { + "epoch": 2.4135734680430487, + "grad_norm": 2.268310308456421, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7179083824157715, + "num_tokens": 568588649.0, + "step": 21978 + }, + { + "epoch": 2.413683285745662, + "grad_norm": 2.7651870250701904, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7326722145080566, + "num_tokens": 568609408.0, + "step": 21979 + }, + { + "epoch": 2.413793103448276, + "grad_norm": 2.2877001762390137, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.713631272315979, + "num_tokens": 568635159.0, + "step": 21980 + }, + { + "epoch": 2.4139029211508896, + "grad_norm": 2.3395910263061523, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7080524563789368, + "num_tokens": 568662805.0, + "step": 21981 + }, + { + "epoch": 2.4140127388535033, + "grad_norm": 2.5667803287506104, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7413366436958313, + "num_tokens": 568686176.0, + "step": 21982 + }, + { + "epoch": 2.4141225565561166, + "grad_norm": 2.7456820011138916, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7213308215141296, + "num_tokens": 568706853.0, + "step": 21983 + }, + { + "epoch": 2.4142323742587304, + "grad_norm": 2.190152406692505, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7178137302398682, + "num_tokens": 568736656.0, + "step": 21984 + }, + { + "epoch": 2.414342191961344, + "grad_norm": 2.525902032852173, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7607583999633789, + "num_tokens": 568758166.0, + "step": 21985 + }, + { + "epoch": 2.414452009663958, + "grad_norm": 2.569650173187256, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7434275150299072, + "num_tokens": 568778611.0, + "step": 21986 + }, + { + "epoch": 2.4145618273665717, + "grad_norm": 2.7248282432556152, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7390799522399902, + "num_tokens": 568799349.0, + "step": 21987 + }, + { + "epoch": 2.414671645069185, + "grad_norm": 2.1166186332702637, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7132994532585144, + "num_tokens": 568830941.0, + "step": 21988 + }, + { + "epoch": 2.4147814627717987, + "grad_norm": 2.0613977909088135, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7144191861152649, + "num_tokens": 568861571.0, + "step": 21989 + }, + { + "epoch": 2.4148912804744125, + "grad_norm": 2.2319560050964355, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7396864891052246, + "num_tokens": 568889511.0, + "step": 21990 + }, + { + "epoch": 2.4150010981770262, + "grad_norm": 2.2308483123779297, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7452627420425415, + "num_tokens": 568916393.0, + "step": 21991 + }, + { + "epoch": 2.41511091587964, + "grad_norm": 2.2529356479644775, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7229552268981934, + "num_tokens": 568943616.0, + "step": 21992 + }, + { + "epoch": 2.4152207335822533, + "grad_norm": 2.5731778144836426, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7439581155776978, + "num_tokens": 568966047.0, + "step": 21993 + }, + { + "epoch": 2.415330551284867, + "grad_norm": 2.5577402114868164, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7580254673957825, + "num_tokens": 568987775.0, + "step": 21994 + }, + { + "epoch": 2.415440368987481, + "grad_norm": 2.123502492904663, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7281743884086609, + "num_tokens": 569017181.0, + "step": 21995 + }, + { + "epoch": 2.4155501866900946, + "grad_norm": 2.430839776992798, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7470011711120605, + "num_tokens": 569043843.0, + "step": 21996 + }, + { + "epoch": 2.4156600043927083, + "grad_norm": 2.5150930881500244, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7238001227378845, + "num_tokens": 569070037.0, + "step": 21997 + }, + { + "epoch": 2.4157698220953217, + "grad_norm": 2.716109275817871, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7429933547973633, + "num_tokens": 569090111.0, + "step": 21998 + }, + { + "epoch": 2.4158796397979354, + "grad_norm": 3.242414951324463, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7144293189048767, + "num_tokens": 569108485.0, + "step": 21999 + }, + { + "epoch": 2.415989457500549, + "grad_norm": 2.4424922466278076, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.714834451675415, + "num_tokens": 569134939.0, + "step": 22000 + }, + { + "epoch": 2.416099275203163, + "grad_norm": 2.296459197998047, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7378209829330444, + "num_tokens": 569163020.0, + "step": 22001 + }, + { + "epoch": 2.4162090929057762, + "grad_norm": 2.4729108810424805, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7295191287994385, + "num_tokens": 569188176.0, + "step": 22002 + }, + { + "epoch": 2.41631891060839, + "grad_norm": 2.174898147583008, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7421913146972656, + "num_tokens": 569216999.0, + "step": 22003 + }, + { + "epoch": 2.4164287283110037, + "grad_norm": 2.0775508880615234, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.733102560043335, + "num_tokens": 569247007.0, + "step": 22004 + }, + { + "epoch": 2.4165385460136175, + "grad_norm": 2.8574273586273193, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7475990056991577, + "num_tokens": 569264811.0, + "step": 22005 + }, + { + "epoch": 2.416648363716231, + "grad_norm": 2.5589065551757812, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7248384952545166, + "num_tokens": 569289177.0, + "step": 22006 + }, + { + "epoch": 2.4167581814188446, + "grad_norm": 2.0891966819763184, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7178478240966797, + "num_tokens": 569320467.0, + "step": 22007 + }, + { + "epoch": 2.4168679991214583, + "grad_norm": 2.6083576679229736, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7295575737953186, + "num_tokens": 569344687.0, + "step": 22008 + }, + { + "epoch": 2.416977816824072, + "grad_norm": 2.043917655944824, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7256132960319519, + "num_tokens": 569376539.0, + "step": 22009 + }, + { + "epoch": 2.417087634526686, + "grad_norm": 2.437662124633789, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7505110502243042, + "num_tokens": 569399611.0, + "step": 22010 + }, + { + "epoch": 2.417197452229299, + "grad_norm": 2.4853322505950928, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7100156545639038, + "num_tokens": 569425410.0, + "step": 22011 + }, + { + "epoch": 2.417307269931913, + "grad_norm": 2.55161714553833, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.738822877407074, + "num_tokens": 569447821.0, + "step": 22012 + }, + { + "epoch": 2.4174170876345267, + "grad_norm": 2.596787691116333, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7230483293533325, + "num_tokens": 569471760.0, + "step": 22013 + }, + { + "epoch": 2.4175269053371404, + "grad_norm": 2.435993194580078, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7339321374893188, + "num_tokens": 569495548.0, + "step": 22014 + }, + { + "epoch": 2.417636723039754, + "grad_norm": 2.452096462249756, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7283459901809692, + "num_tokens": 569518643.0, + "step": 22015 + }, + { + "epoch": 2.4177465407423675, + "grad_norm": 2.7823352813720703, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7380048632621765, + "num_tokens": 569538571.0, + "step": 22016 + }, + { + "epoch": 2.4178563584449813, + "grad_norm": 2.1144726276397705, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7216373682022095, + "num_tokens": 569569174.0, + "step": 22017 + }, + { + "epoch": 2.417966176147595, + "grad_norm": 2.3151116371154785, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7150393724441528, + "num_tokens": 569595496.0, + "step": 22018 + }, + { + "epoch": 2.4180759938502088, + "grad_norm": 2.0506086349487305, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7123104333877563, + "num_tokens": 569629327.0, + "step": 22019 + }, + { + "epoch": 2.4181858115528225, + "grad_norm": 2.2935092449188232, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7558740377426147, + "num_tokens": 569654413.0, + "step": 22020 + }, + { + "epoch": 2.418295629255436, + "grad_norm": 2.575670003890991, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7544485926628113, + "num_tokens": 569676924.0, + "step": 22021 + }, + { + "epoch": 2.4184054469580496, + "grad_norm": 2.2529001235961914, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7314959168434143, + "num_tokens": 569706225.0, + "step": 22022 + }, + { + "epoch": 2.4185152646606634, + "grad_norm": 2.411301612854004, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7380038499832153, + "num_tokens": 569729987.0, + "step": 22023 + }, + { + "epoch": 2.418625082363277, + "grad_norm": 2.6302649974823, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7185085415840149, + "num_tokens": 569751708.0, + "step": 22024 + }, + { + "epoch": 2.418734900065891, + "grad_norm": 2.588378667831421, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7325904369354248, + "num_tokens": 569775611.0, + "step": 22025 + }, + { + "epoch": 2.418844717768504, + "grad_norm": 2.0281147956848145, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6951415538787842, + "num_tokens": 569811089.0, + "step": 22026 + }, + { + "epoch": 2.418954535471118, + "grad_norm": 2.3308510780334473, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7267038822174072, + "num_tokens": 569836572.0, + "step": 22027 + }, + { + "epoch": 2.4190643531737317, + "grad_norm": 2.800283432006836, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7463915348052979, + "num_tokens": 569856953.0, + "step": 22028 + }, + { + "epoch": 2.4191741708763455, + "grad_norm": 2.3104023933410645, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7489182353019714, + "num_tokens": 569883376.0, + "step": 22029 + }, + { + "epoch": 2.4192839885789588, + "grad_norm": 2.502854347229004, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7283545136451721, + "num_tokens": 569907999.0, + "step": 22030 + }, + { + "epoch": 2.4193938062815725, + "grad_norm": 2.1974267959594727, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.727831244468689, + "num_tokens": 569938111.0, + "step": 22031 + }, + { + "epoch": 2.4195036239841863, + "grad_norm": 2.2583274841308594, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7161600589752197, + "num_tokens": 569966184.0, + "step": 22032 + }, + { + "epoch": 2.4196134416868, + "grad_norm": 2.8124866485595703, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7674459218978882, + "num_tokens": 569985005.0, + "step": 22033 + }, + { + "epoch": 2.4197232593894134, + "grad_norm": 2.4055392742156982, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7241709232330322, + "num_tokens": 570010300.0, + "step": 22034 + }, + { + "epoch": 2.419833077092027, + "grad_norm": 2.4982845783233643, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7421892285346985, + "num_tokens": 570032860.0, + "step": 22035 + }, + { + "epoch": 2.419942894794641, + "grad_norm": 2.2699992656707764, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7484760880470276, + "num_tokens": 570061059.0, + "step": 22036 + }, + { + "epoch": 2.4200527124972546, + "grad_norm": 2.2920305728912354, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7401142120361328, + "num_tokens": 570087969.0, + "step": 22037 + }, + { + "epoch": 2.4201625301998684, + "grad_norm": 2.2882165908813477, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7091065645217896, + "num_tokens": 570115084.0, + "step": 22038 + }, + { + "epoch": 2.4202723479024817, + "grad_norm": 2.073974847793579, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7283521294593811, + "num_tokens": 570143869.0, + "step": 22039 + }, + { + "epoch": 2.4203821656050954, + "grad_norm": 2.131091356277466, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7283947467803955, + "num_tokens": 570173988.0, + "step": 22040 + }, + { + "epoch": 2.420491983307709, + "grad_norm": 2.272312879562378, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7151024341583252, + "num_tokens": 570202500.0, + "step": 22041 + }, + { + "epoch": 2.420601801010323, + "grad_norm": 2.6109585762023926, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7306909561157227, + "num_tokens": 570226143.0, + "step": 22042 + }, + { + "epoch": 2.4207116187129367, + "grad_norm": 2.605095386505127, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7430475354194641, + "num_tokens": 570246966.0, + "step": 22043 + }, + { + "epoch": 2.42082143641555, + "grad_norm": 2.1527748107910156, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7499860525131226, + "num_tokens": 570276122.0, + "step": 22044 + }, + { + "epoch": 2.420931254118164, + "grad_norm": 2.4491848945617676, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7272772789001465, + "num_tokens": 570299470.0, + "step": 22045 + }, + { + "epoch": 2.4210410718207775, + "grad_norm": 2.328695297241211, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7327529788017273, + "num_tokens": 570326235.0, + "step": 22046 + }, + { + "epoch": 2.4211508895233913, + "grad_norm": 2.6348989009857178, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7460403442382812, + "num_tokens": 570346722.0, + "step": 22047 + }, + { + "epoch": 2.421260707226005, + "grad_norm": 2.7143986225128174, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7195507287979126, + "num_tokens": 570368090.0, + "step": 22048 + }, + { + "epoch": 2.4213705249286184, + "grad_norm": 2.6022069454193115, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7383151054382324, + "num_tokens": 570391613.0, + "step": 22049 + }, + { + "epoch": 2.421480342631232, + "grad_norm": 2.631817579269409, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7490040063858032, + "num_tokens": 570411032.0, + "step": 22050 + }, + { + "epoch": 2.421590160333846, + "grad_norm": 2.429112434387207, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7238688468933105, + "num_tokens": 570435697.0, + "step": 22051 + }, + { + "epoch": 2.4216999780364596, + "grad_norm": 2.6388320922851562, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7248986959457397, + "num_tokens": 570457632.0, + "step": 22052 + }, + { + "epoch": 2.421809795739073, + "grad_norm": 2.3828418254852295, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7515464425086975, + "num_tokens": 570482577.0, + "step": 22053 + }, + { + "epoch": 2.4219196134416867, + "grad_norm": 2.1618638038635254, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7194236516952515, + "num_tokens": 570515113.0, + "step": 22054 + }, + { + "epoch": 2.4220294311443005, + "grad_norm": 2.1485910415649414, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7221322059631348, + "num_tokens": 570546724.0, + "step": 22055 + }, + { + "epoch": 2.4221392488469142, + "grad_norm": 2.3144538402557373, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7359762191772461, + "num_tokens": 570572622.0, + "step": 22056 + }, + { + "epoch": 2.4222490665495275, + "grad_norm": 2.7076470851898193, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7548590898513794, + "num_tokens": 570592893.0, + "step": 22057 + }, + { + "epoch": 2.4223588842521413, + "grad_norm": 2.373910427093506, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7258003354072571, + "num_tokens": 570617906.0, + "step": 22058 + }, + { + "epoch": 2.422468701954755, + "grad_norm": 2.625164031982422, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.730594277381897, + "num_tokens": 570640269.0, + "step": 22059 + }, + { + "epoch": 2.422578519657369, + "grad_norm": 2.7866106033325195, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7471771240234375, + "num_tokens": 570661526.0, + "step": 22060 + }, + { + "epoch": 2.4226883373599826, + "grad_norm": 2.146594762802124, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7180845141410828, + "num_tokens": 570692755.0, + "step": 22061 + }, + { + "epoch": 2.422798155062596, + "grad_norm": 2.6106958389282227, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7369125485420227, + "num_tokens": 570715163.0, + "step": 22062 + }, + { + "epoch": 2.4229079727652096, + "grad_norm": 2.1153690814971924, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7129831314086914, + "num_tokens": 570746436.0, + "step": 22063 + }, + { + "epoch": 2.4230177904678234, + "grad_norm": 2.676332712173462, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7202054262161255, + "num_tokens": 570768185.0, + "step": 22064 + }, + { + "epoch": 2.423127608170437, + "grad_norm": 2.4752724170684814, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7333977222442627, + "num_tokens": 570793782.0, + "step": 22065 + }, + { + "epoch": 2.423237425873051, + "grad_norm": 2.3324201107025146, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7492339611053467, + "num_tokens": 570817523.0, + "step": 22066 + }, + { + "epoch": 2.4233472435756642, + "grad_norm": 2.20011305809021, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7262632250785828, + "num_tokens": 570847349.0, + "step": 22067 + }, + { + "epoch": 2.423457061278278, + "grad_norm": 2.1903445720672607, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7448480129241943, + "num_tokens": 570874899.0, + "step": 22068 + }, + { + "epoch": 2.4235668789808917, + "grad_norm": 2.36425518989563, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.775914192199707, + "num_tokens": 570898312.0, + "step": 22069 + }, + { + "epoch": 2.4236766966835055, + "grad_norm": 2.4220986366271973, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7292507886886597, + "num_tokens": 570923526.0, + "step": 22070 + }, + { + "epoch": 2.4237865143861193, + "grad_norm": 2.104583501815796, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7348562479019165, + "num_tokens": 570955857.0, + "step": 22071 + }, + { + "epoch": 2.4238963320887326, + "grad_norm": 2.234748363494873, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7476479411125183, + "num_tokens": 570982408.0, + "step": 22072 + }, + { + "epoch": 2.4240061497913463, + "grad_norm": 2.3131072521209717, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7273200154304504, + "num_tokens": 571012542.0, + "step": 22073 + }, + { + "epoch": 2.42411596749396, + "grad_norm": 2.3653862476348877, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7155616283416748, + "num_tokens": 571038782.0, + "step": 22074 + }, + { + "epoch": 2.424225785196574, + "grad_norm": 2.0092451572418213, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7489306926727295, + "num_tokens": 571071042.0, + "step": 22075 + }, + { + "epoch": 2.4243356028991876, + "grad_norm": 2.3421952724456787, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7286705374717712, + "num_tokens": 571097481.0, + "step": 22076 + }, + { + "epoch": 2.424445420601801, + "grad_norm": 2.366445779800415, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7342256307601929, + "num_tokens": 571123289.0, + "step": 22077 + }, + { + "epoch": 2.4245552383044147, + "grad_norm": 2.3240413665771484, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7339143753051758, + "num_tokens": 571150458.0, + "step": 22078 + }, + { + "epoch": 2.4246650560070284, + "grad_norm": 2.4503018856048584, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7552961707115173, + "num_tokens": 571174379.0, + "step": 22079 + }, + { + "epoch": 2.424774873709642, + "grad_norm": 2.2136740684509277, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7340701222419739, + "num_tokens": 571203623.0, + "step": 22080 + }, + { + "epoch": 2.4248846914122555, + "grad_norm": 2.102848529815674, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7085787057876587, + "num_tokens": 571236688.0, + "step": 22081 + }, + { + "epoch": 2.4249945091148692, + "grad_norm": 2.101872205734253, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7213656902313232, + "num_tokens": 571266124.0, + "step": 22082 + }, + { + "epoch": 2.425104326817483, + "grad_norm": 2.503331422805786, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7194260954856873, + "num_tokens": 571289727.0, + "step": 22083 + }, + { + "epoch": 2.4252141445200968, + "grad_norm": 2.2027180194854736, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7335947155952454, + "num_tokens": 571317444.0, + "step": 22084 + }, + { + "epoch": 2.42532396222271, + "grad_norm": 2.3148577213287354, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7222899198532104, + "num_tokens": 571346632.0, + "step": 22085 + }, + { + "epoch": 2.425433779925324, + "grad_norm": 2.352400541305542, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7248401641845703, + "num_tokens": 571374381.0, + "step": 22086 + }, + { + "epoch": 2.4255435976279376, + "grad_norm": 2.2560653686523438, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.722793698310852, + "num_tokens": 571400742.0, + "step": 22087 + }, + { + "epoch": 2.4256534153305513, + "grad_norm": 2.3321335315704346, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7497228384017944, + "num_tokens": 571427455.0, + "step": 22088 + }, + { + "epoch": 2.425763233033165, + "grad_norm": 2.3299450874328613, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7195014357566833, + "num_tokens": 571452419.0, + "step": 22089 + }, + { + "epoch": 2.4258730507357784, + "grad_norm": 2.294732093811035, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7311044931411743, + "num_tokens": 571479041.0, + "step": 22090 + }, + { + "epoch": 2.425982868438392, + "grad_norm": 2.4851512908935547, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7450799345970154, + "num_tokens": 571501627.0, + "step": 22091 + }, + { + "epoch": 2.426092686141006, + "grad_norm": 2.3469784259796143, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7279396057128906, + "num_tokens": 571527922.0, + "step": 22092 + }, + { + "epoch": 2.4262025038436197, + "grad_norm": 2.4790878295898438, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7538823485374451, + "num_tokens": 571552399.0, + "step": 22093 + }, + { + "epoch": 2.4263123215462334, + "grad_norm": 2.411389112472534, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7353879809379578, + "num_tokens": 571577848.0, + "step": 22094 + }, + { + "epoch": 2.4264221392488468, + "grad_norm": 2.2197113037109375, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7206973433494568, + "num_tokens": 571607660.0, + "step": 22095 + }, + { + "epoch": 2.4265319569514605, + "grad_norm": 2.3563344478607178, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7383515238761902, + "num_tokens": 571633207.0, + "step": 22096 + }, + { + "epoch": 2.4266417746540743, + "grad_norm": 2.2423665523529053, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7197322249412537, + "num_tokens": 571660053.0, + "step": 22097 + }, + { + "epoch": 2.426751592356688, + "grad_norm": 2.1908042430877686, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.733871340751648, + "num_tokens": 571689648.0, + "step": 22098 + }, + { + "epoch": 2.426861410059302, + "grad_norm": 2.2733964920043945, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7084816694259644, + "num_tokens": 571718070.0, + "step": 22099 + }, + { + "epoch": 2.426971227761915, + "grad_norm": 1.953088402748108, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7292807698249817, + "num_tokens": 571755031.0, + "step": 22100 + }, + { + "epoch": 2.427081045464529, + "grad_norm": 2.4078617095947266, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7329777479171753, + "num_tokens": 571779782.0, + "step": 22101 + }, + { + "epoch": 2.4271908631671426, + "grad_norm": 2.6561477184295654, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7496169805526733, + "num_tokens": 571800827.0, + "step": 22102 + }, + { + "epoch": 2.4273006808697564, + "grad_norm": 2.424527406692505, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7258093357086182, + "num_tokens": 571823948.0, + "step": 22103 + }, + { + "epoch": 2.4274104985723697, + "grad_norm": 2.7723147869110107, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.747734785079956, + "num_tokens": 571844500.0, + "step": 22104 + }, + { + "epoch": 2.4275203162749834, + "grad_norm": 2.32167911529541, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.725545346736908, + "num_tokens": 571871878.0, + "step": 22105 + }, + { + "epoch": 2.427630133977597, + "grad_norm": 2.276865243911743, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7180672287940979, + "num_tokens": 571899440.0, + "step": 22106 + }, + { + "epoch": 2.427739951680211, + "grad_norm": 2.253052234649658, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7339917421340942, + "num_tokens": 571930948.0, + "step": 22107 + }, + { + "epoch": 2.4278497693828243, + "grad_norm": 2.1216914653778076, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.722801923751831, + "num_tokens": 571960027.0, + "step": 22108 + }, + { + "epoch": 2.427959587085438, + "grad_norm": 2.297698497772217, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.6990099549293518, + "num_tokens": 571989873.0, + "step": 22109 + }, + { + "epoch": 2.4280694047880518, + "grad_norm": 2.3431954383850098, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.726014256477356, + "num_tokens": 572017896.0, + "step": 22110 + }, + { + "epoch": 2.4281792224906655, + "grad_norm": 2.535998582839966, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7324328422546387, + "num_tokens": 572042412.0, + "step": 22111 + }, + { + "epoch": 2.4282890401932793, + "grad_norm": 2.295806884765625, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7453151345252991, + "num_tokens": 572068675.0, + "step": 22112 + }, + { + "epoch": 2.4283988578958926, + "grad_norm": 2.4712295532226562, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.731581449508667, + "num_tokens": 572092858.0, + "step": 22113 + }, + { + "epoch": 2.4285086755985064, + "grad_norm": 2.3093578815460205, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7379173040390015, + "num_tokens": 572122554.0, + "step": 22114 + }, + { + "epoch": 2.42861849330112, + "grad_norm": 2.5652503967285156, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7237979173660278, + "num_tokens": 572145919.0, + "step": 22115 + }, + { + "epoch": 2.428728311003734, + "grad_norm": 2.1933205127716064, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7218389511108398, + "num_tokens": 572174593.0, + "step": 22116 + }, + { + "epoch": 2.4288381287063476, + "grad_norm": 2.7465336322784424, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7393358945846558, + "num_tokens": 572194188.0, + "step": 22117 + }, + { + "epoch": 2.428947946408961, + "grad_norm": 2.769045829772949, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7382010817527771, + "num_tokens": 572213760.0, + "step": 22118 + }, + { + "epoch": 2.4290577641115747, + "grad_norm": 2.313551902770996, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7222487926483154, + "num_tokens": 572240846.0, + "step": 22119 + }, + { + "epoch": 2.4291675818141885, + "grad_norm": 2.8020942211151123, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7370567917823792, + "num_tokens": 572260410.0, + "step": 22120 + }, + { + "epoch": 2.429277399516802, + "grad_norm": 2.311575174331665, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7153308391571045, + "num_tokens": 572288503.0, + "step": 22121 + }, + { + "epoch": 2.429387217219416, + "grad_norm": 2.322565793991089, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7243255376815796, + "num_tokens": 572314762.0, + "step": 22122 + }, + { + "epoch": 2.4294970349220293, + "grad_norm": 2.4493043422698975, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7300738096237183, + "num_tokens": 572338449.0, + "step": 22123 + }, + { + "epoch": 2.429606852624643, + "grad_norm": 2.1610846519470215, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7275413274765015, + "num_tokens": 572371814.0, + "step": 22124 + }, + { + "epoch": 2.429716670327257, + "grad_norm": 2.3777239322662354, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7173112034797668, + "num_tokens": 572401330.0, + "step": 22125 + }, + { + "epoch": 2.4298264880298706, + "grad_norm": 2.1013970375061035, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7430077791213989, + "num_tokens": 572431993.0, + "step": 22126 + }, + { + "epoch": 2.4299363057324843, + "grad_norm": 2.1500356197357178, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7503365278244019, + "num_tokens": 572461968.0, + "step": 22127 + }, + { + "epoch": 2.4300461234350976, + "grad_norm": 2.397834539413452, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7124191522598267, + "num_tokens": 572487597.0, + "step": 22128 + }, + { + "epoch": 2.4301559411377114, + "grad_norm": 2.277564764022827, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7397608757019043, + "num_tokens": 572513688.0, + "step": 22129 + }, + { + "epoch": 2.430265758840325, + "grad_norm": 2.4415109157562256, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7266681790351868, + "num_tokens": 572537892.0, + "step": 22130 + }, + { + "epoch": 2.430375576542939, + "grad_norm": 2.480966567993164, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7412769794464111, + "num_tokens": 572562504.0, + "step": 22131 + }, + { + "epoch": 2.430485394245552, + "grad_norm": 2.5125515460968018, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7659450769424438, + "num_tokens": 572585240.0, + "step": 22132 + }, + { + "epoch": 2.430595211948166, + "grad_norm": 2.611142873764038, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7280063629150391, + "num_tokens": 572608644.0, + "step": 22133 + }, + { + "epoch": 2.4307050296507797, + "grad_norm": 2.379652261734009, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7479256391525269, + "num_tokens": 572634180.0, + "step": 22134 + }, + { + "epoch": 2.4308148473533935, + "grad_norm": 2.7186124324798584, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7342019081115723, + "num_tokens": 572655918.0, + "step": 22135 + }, + { + "epoch": 2.430924665056007, + "grad_norm": 2.24196195602417, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7406972646713257, + "num_tokens": 572683866.0, + "step": 22136 + }, + { + "epoch": 2.4310344827586206, + "grad_norm": 2.575446605682373, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7489757537841797, + "num_tokens": 572706081.0, + "step": 22137 + }, + { + "epoch": 2.4311443004612343, + "grad_norm": 2.520090341567993, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7484723329544067, + "num_tokens": 572729135.0, + "step": 22138 + }, + { + "epoch": 2.431254118163848, + "grad_norm": 2.5359585285186768, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7631245255470276, + "num_tokens": 572752702.0, + "step": 22139 + }, + { + "epoch": 2.431363935866462, + "grad_norm": 2.3273937702178955, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7423343658447266, + "num_tokens": 572778951.0, + "step": 22140 + }, + { + "epoch": 2.431473753569075, + "grad_norm": 2.818624496459961, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7713096737861633, + "num_tokens": 572797376.0, + "step": 22141 + }, + { + "epoch": 2.431583571271689, + "grad_norm": 2.419086456298828, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7269480228424072, + "num_tokens": 572821225.0, + "step": 22142 + }, + { + "epoch": 2.4316933889743026, + "grad_norm": 2.161531925201416, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7457875609397888, + "num_tokens": 572851640.0, + "step": 22143 + }, + { + "epoch": 2.4318032066769164, + "grad_norm": 2.3907833099365234, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7224284410476685, + "num_tokens": 572876784.0, + "step": 22144 + }, + { + "epoch": 2.43191302437953, + "grad_norm": 2.653287172317505, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7377103567123413, + "num_tokens": 572901633.0, + "step": 22145 + }, + { + "epoch": 2.4320228420821435, + "grad_norm": 2.425204277038574, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7237350940704346, + "num_tokens": 572925554.0, + "step": 22146 + }, + { + "epoch": 2.4321326597847572, + "grad_norm": 2.7484049797058105, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7515463829040527, + "num_tokens": 572945689.0, + "step": 22147 + }, + { + "epoch": 2.432242477487371, + "grad_norm": 2.418848752975464, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7340266704559326, + "num_tokens": 572969929.0, + "step": 22148 + }, + { + "epoch": 2.4323522951899847, + "grad_norm": 2.4018948078155518, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7477088570594788, + "num_tokens": 572996094.0, + "step": 22149 + }, + { + "epoch": 2.4324621128925985, + "grad_norm": 3.0005359649658203, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7459884881973267, + "num_tokens": 573013678.0, + "step": 22150 + }, + { + "epoch": 2.432571930595212, + "grad_norm": 2.274526357650757, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7222759127616882, + "num_tokens": 573044076.0, + "step": 22151 + }, + { + "epoch": 2.4326817482978256, + "grad_norm": 2.2846717834472656, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7170758247375488, + "num_tokens": 573075911.0, + "step": 22152 + }, + { + "epoch": 2.4327915660004393, + "grad_norm": 2.2769036293029785, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7213096618652344, + "num_tokens": 573104044.0, + "step": 22153 + }, + { + "epoch": 2.432901383703053, + "grad_norm": 2.497821092605591, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7269518375396729, + "num_tokens": 573128751.0, + "step": 22154 + }, + { + "epoch": 2.4330112014056664, + "grad_norm": 2.0930371284484863, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7535057067871094, + "num_tokens": 573161004.0, + "step": 22155 + }, + { + "epoch": 2.43312101910828, + "grad_norm": 2.510713577270508, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.746295690536499, + "num_tokens": 573183890.0, + "step": 22156 + }, + { + "epoch": 2.433230836810894, + "grad_norm": 2.120821952819824, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7087975740432739, + "num_tokens": 573214899.0, + "step": 22157 + }, + { + "epoch": 2.4333406545135077, + "grad_norm": 2.217834711074829, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7253280878067017, + "num_tokens": 573244349.0, + "step": 22158 + }, + { + "epoch": 2.4334504722161214, + "grad_norm": 2.567577838897705, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7347859144210815, + "num_tokens": 573266804.0, + "step": 22159 + }, + { + "epoch": 2.4335602899187347, + "grad_norm": 2.8017072677612305, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7377520203590393, + "num_tokens": 573285840.0, + "step": 22160 + }, + { + "epoch": 2.4336701076213485, + "grad_norm": 2.1921441555023193, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7246692776679993, + "num_tokens": 573313675.0, + "step": 22161 + }, + { + "epoch": 2.4337799253239623, + "grad_norm": 2.2603752613067627, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7308189868927002, + "num_tokens": 573341634.0, + "step": 22162 + }, + { + "epoch": 2.433889743026576, + "grad_norm": 2.106044054031372, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7138321995735168, + "num_tokens": 573372386.0, + "step": 22163 + }, + { + "epoch": 2.4339995607291893, + "grad_norm": 2.592024803161621, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7475167512893677, + "num_tokens": 573393531.0, + "step": 22164 + }, + { + "epoch": 2.434109378431803, + "grad_norm": 2.367203712463379, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7396698594093323, + "num_tokens": 573421354.0, + "step": 22165 + }, + { + "epoch": 2.434219196134417, + "grad_norm": 2.1472604274749756, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7101552486419678, + "num_tokens": 573450837.0, + "step": 22166 + }, + { + "epoch": 2.4343290138370306, + "grad_norm": 2.6268014907836914, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7414750456809998, + "num_tokens": 573472620.0, + "step": 22167 + }, + { + "epoch": 2.4344388315396444, + "grad_norm": 2.3706722259521484, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7156583070755005, + "num_tokens": 573499618.0, + "step": 22168 + }, + { + "epoch": 2.4345486492422577, + "grad_norm": 2.4349007606506348, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7401371598243713, + "num_tokens": 573525270.0, + "step": 22169 + }, + { + "epoch": 2.4346584669448714, + "grad_norm": 2.1647706031799316, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.6951167583465576, + "num_tokens": 573558375.0, + "step": 22170 + }, + { + "epoch": 2.434768284647485, + "grad_norm": 2.544060707092285, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7303698062896729, + "num_tokens": 573580819.0, + "step": 22171 + }, + { + "epoch": 2.434878102350099, + "grad_norm": 2.6024837493896484, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7191899418830872, + "num_tokens": 573603032.0, + "step": 22172 + }, + { + "epoch": 2.4349879200527127, + "grad_norm": 2.5475542545318604, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7523863315582275, + "num_tokens": 573624413.0, + "step": 22173 + }, + { + "epoch": 2.435097737755326, + "grad_norm": 2.4587254524230957, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7383389472961426, + "num_tokens": 573647899.0, + "step": 22174 + }, + { + "epoch": 2.4352075554579398, + "grad_norm": 2.64941668510437, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7081829309463501, + "num_tokens": 573672654.0, + "step": 22175 + }, + { + "epoch": 2.4353173731605535, + "grad_norm": 2.7049243450164795, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7650290727615356, + "num_tokens": 573692131.0, + "step": 22176 + }, + { + "epoch": 2.4354271908631673, + "grad_norm": 2.9678738117218018, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7289602756500244, + "num_tokens": 573712666.0, + "step": 22177 + }, + { + "epoch": 2.435537008565781, + "grad_norm": 2.7006001472473145, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7547916769981384, + "num_tokens": 573732616.0, + "step": 22178 + }, + { + "epoch": 2.4356468262683943, + "grad_norm": 2.441366195678711, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7318633794784546, + "num_tokens": 573755985.0, + "step": 22179 + }, + { + "epoch": 2.435756643971008, + "grad_norm": 2.410370111465454, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7134566307067871, + "num_tokens": 573781635.0, + "step": 22180 + }, + { + "epoch": 2.435866461673622, + "grad_norm": 2.536325693130493, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.756930947303772, + "num_tokens": 573803827.0, + "step": 22181 + }, + { + "epoch": 2.4359762793762356, + "grad_norm": 2.453066110610962, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7475597858428955, + "num_tokens": 573828707.0, + "step": 22182 + }, + { + "epoch": 2.436086097078849, + "grad_norm": 2.0480220317840576, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7510502934455872, + "num_tokens": 573858685.0, + "step": 22183 + }, + { + "epoch": 2.4361959147814627, + "grad_norm": 2.3176913261413574, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7296121120452881, + "num_tokens": 573885476.0, + "step": 22184 + }, + { + "epoch": 2.4363057324840764, + "grad_norm": 2.4872660636901855, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7308957576751709, + "num_tokens": 573910411.0, + "step": 22185 + }, + { + "epoch": 2.43641555018669, + "grad_norm": 2.144840955734253, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7400230169296265, + "num_tokens": 573940680.0, + "step": 22186 + }, + { + "epoch": 2.4365253678893035, + "grad_norm": 1.9977976083755493, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.744966447353363, + "num_tokens": 573974782.0, + "step": 22187 + }, + { + "epoch": 2.4366351855919173, + "grad_norm": 2.3197503089904785, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.702215850353241, + "num_tokens": 574001647.0, + "step": 22188 + }, + { + "epoch": 2.436745003294531, + "grad_norm": 2.3814237117767334, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7267400026321411, + "num_tokens": 574027374.0, + "step": 22189 + }, + { + "epoch": 2.436854820997145, + "grad_norm": 2.784771680831909, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7394900321960449, + "num_tokens": 574047848.0, + "step": 22190 + }, + { + "epoch": 2.4369646386997585, + "grad_norm": 2.443019151687622, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7444291114807129, + "num_tokens": 574071840.0, + "step": 22191 + }, + { + "epoch": 2.437074456402372, + "grad_norm": 2.4721715450286865, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7234032154083252, + "num_tokens": 574096632.0, + "step": 22192 + }, + { + "epoch": 2.4371842741049856, + "grad_norm": 2.443178415298462, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7258671522140503, + "num_tokens": 574121632.0, + "step": 22193 + }, + { + "epoch": 2.4372940918075994, + "grad_norm": 2.590690851211548, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7408225536346436, + "num_tokens": 574143569.0, + "step": 22194 + }, + { + "epoch": 2.437403909510213, + "grad_norm": 2.6629655361175537, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7454941868782043, + "num_tokens": 574165403.0, + "step": 22195 + }, + { + "epoch": 2.437513727212827, + "grad_norm": 2.255446195602417, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7221238613128662, + "num_tokens": 574192232.0, + "step": 22196 + }, + { + "epoch": 2.43762354491544, + "grad_norm": 2.2955281734466553, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7394384145736694, + "num_tokens": 574218583.0, + "step": 22197 + }, + { + "epoch": 2.437733362618054, + "grad_norm": 2.2490310668945312, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7334674000740051, + "num_tokens": 574247284.0, + "step": 22198 + }, + { + "epoch": 2.4378431803206677, + "grad_norm": 2.339019775390625, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7316527366638184, + "num_tokens": 574272439.0, + "step": 22199 + }, + { + "epoch": 2.4379529980232815, + "grad_norm": 2.323798418045044, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7399500608444214, + "num_tokens": 574300708.0, + "step": 22200 + }, + { + "epoch": 2.4380628157258952, + "grad_norm": 2.185731887817383, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.727211594581604, + "num_tokens": 574329878.0, + "step": 22201 + }, + { + "epoch": 2.4381726334285085, + "grad_norm": 2.054666757583618, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7281053066253662, + "num_tokens": 574362905.0, + "step": 22202 + }, + { + "epoch": 2.4382824511311223, + "grad_norm": 2.309483051300049, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.699430525302887, + "num_tokens": 574389659.0, + "step": 22203 + }, + { + "epoch": 2.438392268833736, + "grad_norm": 2.679684638977051, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7243027687072754, + "num_tokens": 574415611.0, + "step": 22204 + }, + { + "epoch": 2.43850208653635, + "grad_norm": 2.1999080181121826, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7159062623977661, + "num_tokens": 574444974.0, + "step": 22205 + }, + { + "epoch": 2.4386119042389636, + "grad_norm": 2.498224973678589, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7473419904708862, + "num_tokens": 574467654.0, + "step": 22206 + }, + { + "epoch": 2.438721721941577, + "grad_norm": 2.886202096939087, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.751704752445221, + "num_tokens": 574487929.0, + "step": 22207 + }, + { + "epoch": 2.4388315396441906, + "grad_norm": 2.6622607707977295, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7346667051315308, + "num_tokens": 574510732.0, + "step": 22208 + }, + { + "epoch": 2.4389413573468044, + "grad_norm": 2.4193880558013916, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7263416051864624, + "num_tokens": 574536138.0, + "step": 22209 + }, + { + "epoch": 2.439051175049418, + "grad_norm": 2.222919464111328, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7425766587257385, + "num_tokens": 574563499.0, + "step": 22210 + }, + { + "epoch": 2.4391609927520315, + "grad_norm": 2.350360631942749, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.719027042388916, + "num_tokens": 574589829.0, + "step": 22211 + }, + { + "epoch": 2.439270810454645, + "grad_norm": 2.659691095352173, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7367125749588013, + "num_tokens": 574611065.0, + "step": 22212 + }, + { + "epoch": 2.439380628157259, + "grad_norm": 2.318967819213867, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7141964435577393, + "num_tokens": 574637669.0, + "step": 22213 + }, + { + "epoch": 2.4394904458598727, + "grad_norm": 2.8790860176086426, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7568070292472839, + "num_tokens": 574656869.0, + "step": 22214 + }, + { + "epoch": 2.439600263562486, + "grad_norm": 2.2932498455047607, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7314035892486572, + "num_tokens": 574684604.0, + "step": 22215 + }, + { + "epoch": 2.4397100812651, + "grad_norm": 2.513990640640259, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7152706384658813, + "num_tokens": 574708139.0, + "step": 22216 + }, + { + "epoch": 2.4398198989677136, + "grad_norm": 2.4621024131774902, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.764478325843811, + "num_tokens": 574732374.0, + "step": 22217 + }, + { + "epoch": 2.4399297166703273, + "grad_norm": 2.267604112625122, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7067400217056274, + "num_tokens": 574760582.0, + "step": 22218 + }, + { + "epoch": 2.440039534372941, + "grad_norm": 2.398681163787842, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.735990047454834, + "num_tokens": 574786011.0, + "step": 22219 + }, + { + "epoch": 2.4401493520755544, + "grad_norm": 2.3815228939056396, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7287976741790771, + "num_tokens": 574812478.0, + "step": 22220 + }, + { + "epoch": 2.440259169778168, + "grad_norm": 2.270631790161133, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7383703589439392, + "num_tokens": 574838868.0, + "step": 22221 + }, + { + "epoch": 2.440368987480782, + "grad_norm": 2.220794677734375, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7126864194869995, + "num_tokens": 574867871.0, + "step": 22222 + }, + { + "epoch": 2.4404788051833957, + "grad_norm": 2.4445078372955322, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7318611145019531, + "num_tokens": 574890716.0, + "step": 22223 + }, + { + "epoch": 2.4405886228860094, + "grad_norm": 2.3135106563568115, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7309792041778564, + "num_tokens": 574916637.0, + "step": 22224 + }, + { + "epoch": 2.4406984405886227, + "grad_norm": 2.3962113857269287, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7345725297927856, + "num_tokens": 574943381.0, + "step": 22225 + }, + { + "epoch": 2.4408082582912365, + "grad_norm": 2.218527317047119, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7263977527618408, + "num_tokens": 574972351.0, + "step": 22226 + }, + { + "epoch": 2.4409180759938502, + "grad_norm": 2.1922528743743896, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7398272156715393, + "num_tokens": 575000134.0, + "step": 22227 + }, + { + "epoch": 2.441027893696464, + "grad_norm": 2.027284622192383, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.731717050075531, + "num_tokens": 575035356.0, + "step": 22228 + }, + { + "epoch": 2.4411377113990778, + "grad_norm": 2.669517755508423, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7359327077865601, + "num_tokens": 575057505.0, + "step": 22229 + }, + { + "epoch": 2.441247529101691, + "grad_norm": 2.379188060760498, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7444248199462891, + "num_tokens": 575081014.0, + "step": 22230 + }, + { + "epoch": 2.441357346804305, + "grad_norm": 2.016327381134033, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.701432466506958, + "num_tokens": 575115581.0, + "step": 22231 + }, + { + "epoch": 2.4414671645069186, + "grad_norm": 2.4750187397003174, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7227129936218262, + "num_tokens": 575139436.0, + "step": 22232 + }, + { + "epoch": 2.4415769822095323, + "grad_norm": 2.169126272201538, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7110050916671753, + "num_tokens": 575171593.0, + "step": 22233 + }, + { + "epoch": 2.4416867999121457, + "grad_norm": 2.378970146179199, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7439099550247192, + "num_tokens": 575197302.0, + "step": 22234 + }, + { + "epoch": 2.4417966176147594, + "grad_norm": 2.2894744873046875, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.726370096206665, + "num_tokens": 575225667.0, + "step": 22235 + }, + { + "epoch": 2.441906435317373, + "grad_norm": 2.380068778991699, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7323974370956421, + "num_tokens": 575250392.0, + "step": 22236 + }, + { + "epoch": 2.442016253019987, + "grad_norm": 2.666851758956909, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7313427925109863, + "num_tokens": 575270867.0, + "step": 22237 + }, + { + "epoch": 2.4421260707226002, + "grad_norm": 2.029378890991211, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.714554488658905, + "num_tokens": 575303943.0, + "step": 22238 + }, + { + "epoch": 2.442235888425214, + "grad_norm": 2.596102237701416, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7340006232261658, + "num_tokens": 575325586.0, + "step": 22239 + }, + { + "epoch": 2.4423457061278278, + "grad_norm": 2.7949321269989014, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7397211194038391, + "num_tokens": 575345010.0, + "step": 22240 + }, + { + "epoch": 2.4424555238304415, + "grad_norm": 2.319211483001709, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7063758969306946, + "num_tokens": 575371827.0, + "step": 22241 + }, + { + "epoch": 2.4425653415330553, + "grad_norm": 2.3854517936706543, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7440369725227356, + "num_tokens": 575397851.0, + "step": 22242 + }, + { + "epoch": 2.4426751592356686, + "grad_norm": 2.5267186164855957, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7285572290420532, + "num_tokens": 575420803.0, + "step": 22243 + }, + { + "epoch": 2.4427849769382823, + "grad_norm": 2.66182017326355, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7479403614997864, + "num_tokens": 575441226.0, + "step": 22244 + }, + { + "epoch": 2.442894794640896, + "grad_norm": 2.336315631866455, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7312135100364685, + "num_tokens": 575467323.0, + "step": 22245 + }, + { + "epoch": 2.44300461234351, + "grad_norm": 2.4902539253234863, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7312263250350952, + "num_tokens": 575492186.0, + "step": 22246 + }, + { + "epoch": 2.4431144300461236, + "grad_norm": 2.737107992172241, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7383712530136108, + "num_tokens": 575512859.0, + "step": 22247 + }, + { + "epoch": 2.443224247748737, + "grad_norm": 2.5108695030212402, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7434470653533936, + "num_tokens": 575534950.0, + "step": 22248 + }, + { + "epoch": 2.4433340654513507, + "grad_norm": 2.401576280593872, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7465814352035522, + "num_tokens": 575560010.0, + "step": 22249 + }, + { + "epoch": 2.4434438831539644, + "grad_norm": 2.3742473125457764, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7138411998748779, + "num_tokens": 575587780.0, + "step": 22250 + }, + { + "epoch": 2.443553700856578, + "grad_norm": 2.4285435676574707, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7226780652999878, + "num_tokens": 575613729.0, + "step": 22251 + }, + { + "epoch": 2.443663518559192, + "grad_norm": 2.3796679973602295, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.6962124705314636, + "num_tokens": 575641897.0, + "step": 22252 + }, + { + "epoch": 2.4437733362618053, + "grad_norm": 2.5280344486236572, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7391765117645264, + "num_tokens": 575663939.0, + "step": 22253 + }, + { + "epoch": 2.443883153964419, + "grad_norm": 2.3734614849090576, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7247700691223145, + "num_tokens": 575693398.0, + "step": 22254 + }, + { + "epoch": 2.4439929716670328, + "grad_norm": 2.4923059940338135, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.735382080078125, + "num_tokens": 575716003.0, + "step": 22255 + }, + { + "epoch": 2.4441027893696465, + "grad_norm": 2.575924873352051, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7340394258499146, + "num_tokens": 575737645.0, + "step": 22256 + }, + { + "epoch": 2.4442126070722603, + "grad_norm": 2.2983269691467285, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7358369827270508, + "num_tokens": 575764482.0, + "step": 22257 + }, + { + "epoch": 2.4443224247748736, + "grad_norm": 2.6044094562530518, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7215867042541504, + "num_tokens": 575785339.0, + "step": 22258 + }, + { + "epoch": 2.4444322424774874, + "grad_norm": 2.40312123298645, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7607988715171814, + "num_tokens": 575810353.0, + "step": 22259 + }, + { + "epoch": 2.444542060180101, + "grad_norm": 2.4193313121795654, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7269145846366882, + "num_tokens": 575837686.0, + "step": 22260 + }, + { + "epoch": 2.444651877882715, + "grad_norm": 2.0763609409332275, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.707877516746521, + "num_tokens": 575873131.0, + "step": 22261 + }, + { + "epoch": 2.444761695585328, + "grad_norm": 2.1725995540618896, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.71155846118927, + "num_tokens": 575901282.0, + "step": 22262 + }, + { + "epoch": 2.444871513287942, + "grad_norm": 2.646799325942993, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7344423532485962, + "num_tokens": 575923327.0, + "step": 22263 + }, + { + "epoch": 2.4449813309905557, + "grad_norm": 2.123849630355835, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7193390130996704, + "num_tokens": 575955089.0, + "step": 22264 + }, + { + "epoch": 2.4450911486931695, + "grad_norm": 2.3505451679229736, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7101468443870544, + "num_tokens": 575982716.0, + "step": 22265 + }, + { + "epoch": 2.4452009663957828, + "grad_norm": 2.253329038619995, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7067748308181763, + "num_tokens": 576012552.0, + "step": 22266 + }, + { + "epoch": 2.4453107840983965, + "grad_norm": 2.6478681564331055, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7416466474533081, + "num_tokens": 576033064.0, + "step": 22267 + }, + { + "epoch": 2.4454206018010103, + "grad_norm": 2.107499837875366, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7144216895103455, + "num_tokens": 576063360.0, + "step": 22268 + }, + { + "epoch": 2.445530419503624, + "grad_norm": 2.2894840240478516, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7483617663383484, + "num_tokens": 576088138.0, + "step": 22269 + }, + { + "epoch": 2.445640237206238, + "grad_norm": 2.0492615699768066, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7067630887031555, + "num_tokens": 576120804.0, + "step": 22270 + }, + { + "epoch": 2.445750054908851, + "grad_norm": 2.2858781814575195, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7339188456535339, + "num_tokens": 576146306.0, + "step": 22271 + }, + { + "epoch": 2.445859872611465, + "grad_norm": 2.550217628479004, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7245228290557861, + "num_tokens": 576171014.0, + "step": 22272 + }, + { + "epoch": 2.4459696903140786, + "grad_norm": 2.3643782138824463, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7221512794494629, + "num_tokens": 576197866.0, + "step": 22273 + }, + { + "epoch": 2.4460795080166924, + "grad_norm": 2.3833987712860107, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.6999739408493042, + "num_tokens": 576225642.0, + "step": 22274 + }, + { + "epoch": 2.446189325719306, + "grad_norm": 2.2237234115600586, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7341861128807068, + "num_tokens": 576253010.0, + "step": 22275 + }, + { + "epoch": 2.4462991434219195, + "grad_norm": 2.3031809329986572, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7248402833938599, + "num_tokens": 576279767.0, + "step": 22276 + }, + { + "epoch": 2.446408961124533, + "grad_norm": 2.2150182723999023, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.711097240447998, + "num_tokens": 576310196.0, + "step": 22277 + }, + { + "epoch": 2.446518778827147, + "grad_norm": 2.39481520652771, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7069363594055176, + "num_tokens": 576336773.0, + "step": 22278 + }, + { + "epoch": 2.4466285965297607, + "grad_norm": 2.638521671295166, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7353576421737671, + "num_tokens": 576358569.0, + "step": 22279 + }, + { + "epoch": 2.4467384142323745, + "grad_norm": 2.305748701095581, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7361516356468201, + "num_tokens": 576386669.0, + "step": 22280 + }, + { + "epoch": 2.446848231934988, + "grad_norm": 2.4930717945098877, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7417935132980347, + "num_tokens": 576409761.0, + "step": 22281 + }, + { + "epoch": 2.4469580496376016, + "grad_norm": 2.653855085372925, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.741755485534668, + "num_tokens": 576432019.0, + "step": 22282 + }, + { + "epoch": 2.4470678673402153, + "grad_norm": 2.3777146339416504, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7326592206954956, + "num_tokens": 576459040.0, + "step": 22283 + }, + { + "epoch": 2.447177685042829, + "grad_norm": 2.3511409759521484, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7507919073104858, + "num_tokens": 576485630.0, + "step": 22284 + }, + { + "epoch": 2.4472875027454424, + "grad_norm": 2.402376890182495, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7444509267807007, + "num_tokens": 576510185.0, + "step": 22285 + }, + { + "epoch": 2.447397320448056, + "grad_norm": 2.5653629302978516, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7309747338294983, + "num_tokens": 576533097.0, + "step": 22286 + }, + { + "epoch": 2.44750713815067, + "grad_norm": 2.8776791095733643, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7356081008911133, + "num_tokens": 576552424.0, + "step": 22287 + }, + { + "epoch": 2.4476169558532836, + "grad_norm": 2.608556032180786, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7517634630203247, + "num_tokens": 576572851.0, + "step": 22288 + }, + { + "epoch": 2.447726773555897, + "grad_norm": 2.2622549533843994, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7390366792678833, + "num_tokens": 576599655.0, + "step": 22289 + }, + { + "epoch": 2.4478365912585107, + "grad_norm": 2.1125845909118652, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7305985689163208, + "num_tokens": 576628544.0, + "step": 22290 + }, + { + "epoch": 2.4479464089611245, + "grad_norm": 2.1772289276123047, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7432712912559509, + "num_tokens": 576657072.0, + "step": 22291 + }, + { + "epoch": 2.4480562266637382, + "grad_norm": 2.3161256313323975, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.727968156337738, + "num_tokens": 576683784.0, + "step": 22292 + }, + { + "epoch": 2.448166044366352, + "grad_norm": 2.511789321899414, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7447052001953125, + "num_tokens": 576706149.0, + "step": 22293 + }, + { + "epoch": 2.4482758620689653, + "grad_norm": 2.5420167446136475, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7291363477706909, + "num_tokens": 576727558.0, + "step": 22294 + }, + { + "epoch": 2.448385679771579, + "grad_norm": 2.235936164855957, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.733843207359314, + "num_tokens": 576754030.0, + "step": 22295 + }, + { + "epoch": 2.448495497474193, + "grad_norm": 2.27219295501709, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7312665581703186, + "num_tokens": 576782912.0, + "step": 22296 + }, + { + "epoch": 2.4486053151768066, + "grad_norm": 2.3655738830566406, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7342234253883362, + "num_tokens": 576807119.0, + "step": 22297 + }, + { + "epoch": 2.4487151328794203, + "grad_norm": 2.4807682037353516, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7289830446243286, + "num_tokens": 576830037.0, + "step": 22298 + }, + { + "epoch": 2.4488249505820336, + "grad_norm": 2.541827440261841, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7689832448959351, + "num_tokens": 576851476.0, + "step": 22299 + }, + { + "epoch": 2.4489347682846474, + "grad_norm": 2.5562891960144043, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7189016342163086, + "num_tokens": 576875256.0, + "step": 22300 + }, + { + "epoch": 2.449044585987261, + "grad_norm": 2.1709301471710205, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7281458377838135, + "num_tokens": 576905309.0, + "step": 22301 + }, + { + "epoch": 2.449154403689875, + "grad_norm": 2.3912925720214844, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7259564399719238, + "num_tokens": 576930590.0, + "step": 22302 + }, + { + "epoch": 2.4492642213924887, + "grad_norm": 2.651667594909668, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7355338931083679, + "num_tokens": 576952543.0, + "step": 22303 + }, + { + "epoch": 2.449374039095102, + "grad_norm": 2.377901315689087, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7220439314842224, + "num_tokens": 576978843.0, + "step": 22304 + }, + { + "epoch": 2.4494838567977157, + "grad_norm": 2.480168581008911, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7243866324424744, + "num_tokens": 577004525.0, + "step": 22305 + }, + { + "epoch": 2.4495936745003295, + "grad_norm": 2.422776460647583, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7341516613960266, + "num_tokens": 577028989.0, + "step": 22306 + }, + { + "epoch": 2.4497034922029433, + "grad_norm": 2.619992733001709, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7502306699752808, + "num_tokens": 577049797.0, + "step": 22307 + }, + { + "epoch": 2.449813309905557, + "grad_norm": 2.591189384460449, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.7357655763626099, + "num_tokens": 577072989.0, + "step": 22308 + }, + { + "epoch": 2.4499231276081703, + "grad_norm": 2.377903461456299, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7312703132629395, + "num_tokens": 577097721.0, + "step": 22309 + }, + { + "epoch": 2.450032945310784, + "grad_norm": 2.206617593765259, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7092982530593872, + "num_tokens": 577125235.0, + "step": 22310 + }, + { + "epoch": 2.450142763013398, + "grad_norm": 2.8449289798736572, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7492246031761169, + "num_tokens": 577146621.0, + "step": 22311 + }, + { + "epoch": 2.4502525807160116, + "grad_norm": 2.8138070106506348, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7533230781555176, + "num_tokens": 577165943.0, + "step": 22312 + }, + { + "epoch": 2.450362398418625, + "grad_norm": 2.6256327629089355, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7215588092803955, + "num_tokens": 577187725.0, + "step": 22313 + }, + { + "epoch": 2.4504722161212387, + "grad_norm": 2.3433947563171387, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7406859993934631, + "num_tokens": 577214425.0, + "step": 22314 + }, + { + "epoch": 2.4505820338238524, + "grad_norm": 2.3125951290130615, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7365594506263733, + "num_tokens": 577241342.0, + "step": 22315 + }, + { + "epoch": 2.450691851526466, + "grad_norm": 2.4306910037994385, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7445025444030762, + "num_tokens": 577264693.0, + "step": 22316 + }, + { + "epoch": 2.4508016692290795, + "grad_norm": 2.2375495433807373, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7389562129974365, + "num_tokens": 577293109.0, + "step": 22317 + }, + { + "epoch": 2.4509114869316933, + "grad_norm": 2.254917860031128, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7336502075195312, + "num_tokens": 577322748.0, + "step": 22318 + }, + { + "epoch": 2.451021304634307, + "grad_norm": 2.434474229812622, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.73222815990448, + "num_tokens": 577346093.0, + "step": 22319 + }, + { + "epoch": 2.4511311223369208, + "grad_norm": 2.1481213569641113, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7335531115531921, + "num_tokens": 577376360.0, + "step": 22320 + }, + { + "epoch": 2.4512409400395345, + "grad_norm": 2.369680166244507, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7239781022071838, + "num_tokens": 577406450.0, + "step": 22321 + }, + { + "epoch": 2.451350757742148, + "grad_norm": 2.639597177505493, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7195463180541992, + "num_tokens": 577428068.0, + "step": 22322 + }, + { + "epoch": 2.4514605754447616, + "grad_norm": 2.4795496463775635, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7412564754486084, + "num_tokens": 577451745.0, + "step": 22323 + }, + { + "epoch": 2.4515703931473753, + "grad_norm": 2.6600117683410645, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7143383622169495, + "num_tokens": 577473353.0, + "step": 22324 + }, + { + "epoch": 2.451680210849989, + "grad_norm": 1.932033658027649, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7252170443534851, + "num_tokens": 577511202.0, + "step": 22325 + }, + { + "epoch": 2.451790028552603, + "grad_norm": 2.210029125213623, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7429040670394897, + "num_tokens": 577540374.0, + "step": 22326 + }, + { + "epoch": 2.451899846255216, + "grad_norm": 2.1481540203094482, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7079421877861023, + "num_tokens": 577571983.0, + "step": 22327 + }, + { + "epoch": 2.45200966395783, + "grad_norm": 2.550030469894409, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7227851152420044, + "num_tokens": 577596579.0, + "step": 22328 + }, + { + "epoch": 2.4521194816604437, + "grad_norm": 2.2128660678863525, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7527853846549988, + "num_tokens": 577626159.0, + "step": 22329 + }, + { + "epoch": 2.4522292993630574, + "grad_norm": 2.25869083404541, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7210677266120911, + "num_tokens": 577653838.0, + "step": 22330 + }, + { + "epoch": 2.452339117065671, + "grad_norm": 2.238637685775757, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7265727519989014, + "num_tokens": 577681468.0, + "step": 22331 + }, + { + "epoch": 2.4524489347682845, + "grad_norm": 2.2035675048828125, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7278046607971191, + "num_tokens": 577711263.0, + "step": 22332 + }, + { + "epoch": 2.4525587524708983, + "grad_norm": 2.465306043624878, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7322673797607422, + "num_tokens": 577734729.0, + "step": 22333 + }, + { + "epoch": 2.452668570173512, + "grad_norm": 2.6745615005493164, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7415509223937988, + "num_tokens": 577756749.0, + "step": 22334 + }, + { + "epoch": 2.452778387876126, + "grad_norm": 2.180168867111206, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7282154560089111, + "num_tokens": 577785506.0, + "step": 22335 + }, + { + "epoch": 2.452888205578739, + "grad_norm": 2.464463472366333, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7319376468658447, + "num_tokens": 577808889.0, + "step": 22336 + }, + { + "epoch": 2.452998023281353, + "grad_norm": 2.296626567840576, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7182484865188599, + "num_tokens": 577837719.0, + "step": 22337 + }, + { + "epoch": 2.4531078409839666, + "grad_norm": 2.3683769702911377, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7359668016433716, + "num_tokens": 577863447.0, + "step": 22338 + }, + { + "epoch": 2.4532176586865804, + "grad_norm": 2.7284560203552246, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7293787002563477, + "num_tokens": 577883916.0, + "step": 22339 + }, + { + "epoch": 2.453327476389194, + "grad_norm": 2.380232810974121, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.712935209274292, + "num_tokens": 577910642.0, + "step": 22340 + }, + { + "epoch": 2.4534372940918074, + "grad_norm": 2.3866100311279297, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7574458122253418, + "num_tokens": 577934895.0, + "step": 22341 + }, + { + "epoch": 2.453547111794421, + "grad_norm": 2.421597957611084, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7288224101066589, + "num_tokens": 577960405.0, + "step": 22342 + }, + { + "epoch": 2.453656929497035, + "grad_norm": 2.1298916339874268, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6956120133399963, + "num_tokens": 577991669.0, + "step": 22343 + }, + { + "epoch": 2.4537667471996487, + "grad_norm": 2.4987952709198, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7552627325057983, + "num_tokens": 578012901.0, + "step": 22344 + }, + { + "epoch": 2.453876564902262, + "grad_norm": 2.643433094024658, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7486827373504639, + "num_tokens": 578033330.0, + "step": 22345 + }, + { + "epoch": 2.453986382604876, + "grad_norm": 2.8174216747283936, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7495847344398499, + "num_tokens": 578053690.0, + "step": 22346 + }, + { + "epoch": 2.4540962003074895, + "grad_norm": 2.2230324745178223, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7297725677490234, + "num_tokens": 578084004.0, + "step": 22347 + }, + { + "epoch": 2.4542060180101033, + "grad_norm": 2.310107946395874, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7399780750274658, + "num_tokens": 578109929.0, + "step": 22348 + }, + { + "epoch": 2.454315835712717, + "grad_norm": 2.0976927280426025, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7236942648887634, + "num_tokens": 578143144.0, + "step": 22349 + }, + { + "epoch": 2.4544256534153304, + "grad_norm": 2.1299214363098145, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7204369902610779, + "num_tokens": 578172629.0, + "step": 22350 + }, + { + "epoch": 2.454535471117944, + "grad_norm": 2.407961845397949, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7354669570922852, + "num_tokens": 578196976.0, + "step": 22351 + }, + { + "epoch": 2.454645288820558, + "grad_norm": 2.3180315494537354, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7164756655693054, + "num_tokens": 578226395.0, + "step": 22352 + }, + { + "epoch": 2.4547551065231716, + "grad_norm": 2.3243465423583984, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7243932485580444, + "num_tokens": 578254367.0, + "step": 22353 + }, + { + "epoch": 2.4548649242257854, + "grad_norm": 2.369732141494751, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7323795557022095, + "num_tokens": 578279572.0, + "step": 22354 + }, + { + "epoch": 2.4549747419283987, + "grad_norm": 2.5410499572753906, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7755373120307922, + "num_tokens": 578302458.0, + "step": 22355 + }, + { + "epoch": 2.4550845596310125, + "grad_norm": 2.1260986328125, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.713706374168396, + "num_tokens": 578333129.0, + "step": 22356 + }, + { + "epoch": 2.455194377333626, + "grad_norm": 2.4506266117095947, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.744626522064209, + "num_tokens": 578357560.0, + "step": 22357 + }, + { + "epoch": 2.45530419503624, + "grad_norm": 2.196054220199585, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7442931532859802, + "num_tokens": 578384080.0, + "step": 22358 + }, + { + "epoch": 2.4554140127388537, + "grad_norm": 2.242464065551758, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7160981893539429, + "num_tokens": 578413855.0, + "step": 22359 + }, + { + "epoch": 2.455523830441467, + "grad_norm": 2.4466841220855713, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7304626703262329, + "num_tokens": 578441063.0, + "step": 22360 + }, + { + "epoch": 2.455633648144081, + "grad_norm": 2.2735841274261475, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7162952423095703, + "num_tokens": 578469577.0, + "step": 22361 + }, + { + "epoch": 2.4557434658466946, + "grad_norm": 2.4190123081207275, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7478891015052795, + "num_tokens": 578492684.0, + "step": 22362 + }, + { + "epoch": 2.4558532835493083, + "grad_norm": 2.4262452125549316, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7290267944335938, + "num_tokens": 578516977.0, + "step": 22363 + }, + { + "epoch": 2.4559631012519216, + "grad_norm": 2.808093309402466, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7158519625663757, + "num_tokens": 578542266.0, + "step": 22364 + }, + { + "epoch": 2.4560729189545354, + "grad_norm": 2.53684401512146, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7357719540596008, + "num_tokens": 578564821.0, + "step": 22365 + }, + { + "epoch": 2.456182736657149, + "grad_norm": 2.7417166233062744, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7372879981994629, + "num_tokens": 578586215.0, + "step": 22366 + }, + { + "epoch": 2.456292554359763, + "grad_norm": 2.3927500247955322, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7251816391944885, + "num_tokens": 578612609.0, + "step": 22367 + }, + { + "epoch": 2.456402372062376, + "grad_norm": 2.465620994567871, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7097201943397522, + "num_tokens": 578637678.0, + "step": 22368 + }, + { + "epoch": 2.45651218976499, + "grad_norm": 2.0903542041778564, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.715726375579834, + "num_tokens": 578669710.0, + "step": 22369 + }, + { + "epoch": 2.4566220074676037, + "grad_norm": 2.473914384841919, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7487358450889587, + "num_tokens": 578693789.0, + "step": 22370 + }, + { + "epoch": 2.4567318251702175, + "grad_norm": 2.198662519454956, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7282785177230835, + "num_tokens": 578724447.0, + "step": 22371 + }, + { + "epoch": 2.4568416428728312, + "grad_norm": 2.3275949954986572, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7242304086685181, + "num_tokens": 578750989.0, + "step": 22372 + }, + { + "epoch": 2.4569514605754446, + "grad_norm": 2.7076945304870605, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7446004152297974, + "num_tokens": 578772778.0, + "step": 22373 + }, + { + "epoch": 2.4570612782780583, + "grad_norm": 2.262084722518921, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.722947895526886, + "num_tokens": 578800895.0, + "step": 22374 + }, + { + "epoch": 2.457171095980672, + "grad_norm": 2.173598289489746, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7542120218276978, + "num_tokens": 578829230.0, + "step": 22375 + }, + { + "epoch": 2.457280913683286, + "grad_norm": 2.207218885421753, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7159697413444519, + "num_tokens": 578860437.0, + "step": 22376 + }, + { + "epoch": 2.4573907313858996, + "grad_norm": 2.3134517669677734, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7074767351150513, + "num_tokens": 578887887.0, + "step": 22377 + }, + { + "epoch": 2.457500549088513, + "grad_norm": 2.2562570571899414, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7200781106948853, + "num_tokens": 578916142.0, + "step": 22378 + }, + { + "epoch": 2.4576103667911267, + "grad_norm": 2.316889762878418, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7329987287521362, + "num_tokens": 578943239.0, + "step": 22379 + }, + { + "epoch": 2.4577201844937404, + "grad_norm": 2.235610246658325, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.742754340171814, + "num_tokens": 578970595.0, + "step": 22380 + }, + { + "epoch": 2.457830002196354, + "grad_norm": 2.2497808933258057, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7289062738418579, + "num_tokens": 578999841.0, + "step": 22381 + }, + { + "epoch": 2.457939819898968, + "grad_norm": 2.2778475284576416, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.6968672871589661, + "num_tokens": 579028000.0, + "step": 22382 + }, + { + "epoch": 2.4580496376015812, + "grad_norm": 2.5094666481018066, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7271181344985962, + "num_tokens": 579051268.0, + "step": 22383 + }, + { + "epoch": 2.458159455304195, + "grad_norm": 2.394251585006714, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7092810273170471, + "num_tokens": 579079968.0, + "step": 22384 + }, + { + "epoch": 2.4582692730068088, + "grad_norm": 2.704770803451538, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7309634685516357, + "num_tokens": 579100722.0, + "step": 22385 + }, + { + "epoch": 2.4583790907094225, + "grad_norm": 2.2091147899627686, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7316271066665649, + "num_tokens": 579129843.0, + "step": 22386 + }, + { + "epoch": 2.4584889084120363, + "grad_norm": 2.088757038116455, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7219035625457764, + "num_tokens": 579164875.0, + "step": 22387 + }, + { + "epoch": 2.4585987261146496, + "grad_norm": 2.453622579574585, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7153246402740479, + "num_tokens": 579190306.0, + "step": 22388 + }, + { + "epoch": 2.4587085438172633, + "grad_norm": 2.7390103340148926, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7680220603942871, + "num_tokens": 579209165.0, + "step": 22389 + }, + { + "epoch": 2.458818361519877, + "grad_norm": 2.3688087463378906, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7111448049545288, + "num_tokens": 579236012.0, + "step": 22390 + }, + { + "epoch": 2.458928179222491, + "grad_norm": 2.2023582458496094, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7129511833190918, + "num_tokens": 579266172.0, + "step": 22391 + }, + { + "epoch": 2.459037996925104, + "grad_norm": 2.302347421646118, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7329901456832886, + "num_tokens": 579293066.0, + "step": 22392 + }, + { + "epoch": 2.459147814627718, + "grad_norm": 2.3775742053985596, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7215631008148193, + "num_tokens": 579318838.0, + "step": 22393 + }, + { + "epoch": 2.4592576323303317, + "grad_norm": 2.088822841644287, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7165323495864868, + "num_tokens": 579352649.0, + "step": 22394 + }, + { + "epoch": 2.4593674500329454, + "grad_norm": 2.3365135192871094, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7403522729873657, + "num_tokens": 579376862.0, + "step": 22395 + }, + { + "epoch": 2.4594772677355587, + "grad_norm": 2.243039846420288, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7459806799888611, + "num_tokens": 579404419.0, + "step": 22396 + }, + { + "epoch": 2.4595870854381725, + "grad_norm": 2.3861300945281982, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7296102046966553, + "num_tokens": 579431207.0, + "step": 22397 + }, + { + "epoch": 2.4596969031407863, + "grad_norm": 1.9788782596588135, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7311184406280518, + "num_tokens": 579463351.0, + "step": 22398 + }, + { + "epoch": 2.4598067208434, + "grad_norm": 2.331261157989502, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.743910551071167, + "num_tokens": 579490210.0, + "step": 22399 + }, + { + "epoch": 2.4599165385460138, + "grad_norm": 2.401390790939331, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.7616085410118103, + "num_tokens": 579514549.0, + "step": 22400 + }, + { + "epoch": 2.460026356248627, + "grad_norm": 2.463823080062866, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7360024452209473, + "num_tokens": 579538738.0, + "step": 22401 + }, + { + "epoch": 2.460136173951241, + "grad_norm": 2.172597646713257, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7202911376953125, + "num_tokens": 579568753.0, + "step": 22402 + }, + { + "epoch": 2.4602459916538546, + "grad_norm": 2.4372260570526123, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7472054958343506, + "num_tokens": 579593085.0, + "step": 22403 + }, + { + "epoch": 2.4603558093564684, + "grad_norm": 2.5791983604431152, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7172800302505493, + "num_tokens": 579615851.0, + "step": 22404 + }, + { + "epoch": 2.460465627059082, + "grad_norm": 2.272937059402466, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7506310939788818, + "num_tokens": 579641659.0, + "step": 22405 + }, + { + "epoch": 2.4605754447616954, + "grad_norm": 2.560354709625244, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7390374541282654, + "num_tokens": 579665632.0, + "step": 22406 + }, + { + "epoch": 2.460685262464309, + "grad_norm": 2.377196788787842, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7563541531562805, + "num_tokens": 579689755.0, + "step": 22407 + }, + { + "epoch": 2.460795080166923, + "grad_norm": 2.4296607971191406, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7389179468154907, + "num_tokens": 579713267.0, + "step": 22408 + }, + { + "epoch": 2.4609048978695367, + "grad_norm": 2.3691492080688477, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.710240364074707, + "num_tokens": 579739361.0, + "step": 22409 + }, + { + "epoch": 2.4610147155721505, + "grad_norm": 2.5508577823638916, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7141124606132507, + "num_tokens": 579761589.0, + "step": 22410 + }, + { + "epoch": 2.4611245332747638, + "grad_norm": 2.4832820892333984, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7178430557250977, + "num_tokens": 579788253.0, + "step": 22411 + }, + { + "epoch": 2.4612343509773775, + "grad_norm": 2.2247796058654785, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7055870890617371, + "num_tokens": 579817323.0, + "step": 22412 + }, + { + "epoch": 2.4613441686799913, + "grad_norm": 2.482978582382202, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7313359975814819, + "num_tokens": 579840291.0, + "step": 22413 + }, + { + "epoch": 2.461453986382605, + "grad_norm": 2.2069735527038574, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7040446400642395, + "num_tokens": 579872806.0, + "step": 22414 + }, + { + "epoch": 2.4615638040852184, + "grad_norm": 2.0593817234039307, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7243905067443848, + "num_tokens": 579905609.0, + "step": 22415 + }, + { + "epoch": 2.461673621787832, + "grad_norm": 2.241008996963501, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7275075316429138, + "num_tokens": 579934455.0, + "step": 22416 + }, + { + "epoch": 2.461783439490446, + "grad_norm": 2.6359493732452393, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7541396021842957, + "num_tokens": 579956473.0, + "step": 22417 + }, + { + "epoch": 2.4618932571930596, + "grad_norm": 2.3597922325134277, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7138471603393555, + "num_tokens": 579983569.0, + "step": 22418 + }, + { + "epoch": 2.462003074895673, + "grad_norm": 2.2312352657318115, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7300780415534973, + "num_tokens": 580010210.0, + "step": 22419 + }, + { + "epoch": 2.4621128925982867, + "grad_norm": 2.4652678966522217, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.6996786594390869, + "num_tokens": 580035267.0, + "step": 22420 + }, + { + "epoch": 2.4622227103009005, + "grad_norm": 2.413534641265869, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7395034432411194, + "num_tokens": 580059049.0, + "step": 22421 + }, + { + "epoch": 2.462332528003514, + "grad_norm": 2.351248264312744, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7479876279830933, + "num_tokens": 580083726.0, + "step": 22422 + }, + { + "epoch": 2.462442345706128, + "grad_norm": 2.4680325984954834, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7194654941558838, + "num_tokens": 580109057.0, + "step": 22423 + }, + { + "epoch": 2.4625521634087413, + "grad_norm": 2.5807876586914062, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7313904166221619, + "num_tokens": 580131527.0, + "step": 22424 + }, + { + "epoch": 2.462661981111355, + "grad_norm": 2.409687042236328, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7511876821517944, + "num_tokens": 580158296.0, + "step": 22425 + }, + { + "epoch": 2.462771798813969, + "grad_norm": 2.297475814819336, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7206014394760132, + "num_tokens": 580185587.0, + "step": 22426 + }, + { + "epoch": 2.4628816165165826, + "grad_norm": 2.2997775077819824, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7322259545326233, + "num_tokens": 580211264.0, + "step": 22427 + }, + { + "epoch": 2.4629914342191963, + "grad_norm": 2.2604055404663086, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.744823157787323, + "num_tokens": 580238128.0, + "step": 22428 + }, + { + "epoch": 2.4631012519218096, + "grad_norm": 2.324552536010742, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7321950197219849, + "num_tokens": 580263380.0, + "step": 22429 + }, + { + "epoch": 2.4632110696244234, + "grad_norm": 2.310027837753296, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6976066827774048, + "num_tokens": 580291588.0, + "step": 22430 + }, + { + "epoch": 2.463320887327037, + "grad_norm": 2.310575008392334, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7201483845710754, + "num_tokens": 580321155.0, + "step": 22431 + }, + { + "epoch": 2.463430705029651, + "grad_norm": 2.347829818725586, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7326889038085938, + "num_tokens": 580346786.0, + "step": 22432 + }, + { + "epoch": 2.4635405227322646, + "grad_norm": 3.099519729614258, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7435213327407837, + "num_tokens": 580368009.0, + "step": 22433 + }, + { + "epoch": 2.463650340434878, + "grad_norm": 2.49031400680542, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7463092803955078, + "num_tokens": 580391179.0, + "step": 22434 + }, + { + "epoch": 2.4637601581374917, + "grad_norm": 2.2487988471984863, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.726376473903656, + "num_tokens": 580417400.0, + "step": 22435 + }, + { + "epoch": 2.4638699758401055, + "grad_norm": 2.5399816036224365, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7357387542724609, + "num_tokens": 580440859.0, + "step": 22436 + }, + { + "epoch": 2.4639797935427192, + "grad_norm": 2.362311363220215, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7262479066848755, + "num_tokens": 580465329.0, + "step": 22437 + }, + { + "epoch": 2.464089611245333, + "grad_norm": 2.183685541152954, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7483752965927124, + "num_tokens": 580494802.0, + "step": 22438 + }, + { + "epoch": 2.4641994289479463, + "grad_norm": 2.282682418823242, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7427110075950623, + "num_tokens": 580520357.0, + "step": 22439 + }, + { + "epoch": 2.46430924665056, + "grad_norm": 2.3273415565490723, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7716779708862305, + "num_tokens": 580544172.0, + "step": 22440 + }, + { + "epoch": 2.464419064353174, + "grad_norm": 2.621812343597412, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7478911876678467, + "num_tokens": 580565362.0, + "step": 22441 + }, + { + "epoch": 2.4645288820557876, + "grad_norm": 2.279850721359253, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7344508171081543, + "num_tokens": 580592144.0, + "step": 22442 + }, + { + "epoch": 2.464638699758401, + "grad_norm": 2.322526693344116, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7245016098022461, + "num_tokens": 580619409.0, + "step": 22443 + }, + { + "epoch": 2.4647485174610146, + "grad_norm": 2.214420795440674, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7315628528594971, + "num_tokens": 580647254.0, + "step": 22444 + }, + { + "epoch": 2.4648583351636284, + "grad_norm": 2.524158477783203, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7372862100601196, + "num_tokens": 580669445.0, + "step": 22445 + }, + { + "epoch": 2.464968152866242, + "grad_norm": 2.2246570587158203, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7428425550460815, + "num_tokens": 580698773.0, + "step": 22446 + }, + { + "epoch": 2.4650779705688555, + "grad_norm": 2.1432747840881348, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7090229988098145, + "num_tokens": 580728787.0, + "step": 22447 + }, + { + "epoch": 2.4651877882714692, + "grad_norm": 2.128058671951294, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7287547588348389, + "num_tokens": 580758320.0, + "step": 22448 + }, + { + "epoch": 2.465297605974083, + "grad_norm": 2.1751623153686523, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7291660308837891, + "num_tokens": 580787451.0, + "step": 22449 + }, + { + "epoch": 2.4654074236766967, + "grad_norm": 2.2308664321899414, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7358185052871704, + "num_tokens": 580813681.0, + "step": 22450 + }, + { + "epoch": 2.4655172413793105, + "grad_norm": 2.3546295166015625, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7219313979148865, + "num_tokens": 580838606.0, + "step": 22451 + }, + { + "epoch": 2.465627059081924, + "grad_norm": 2.277508020401001, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7368603944778442, + "num_tokens": 580866549.0, + "step": 22452 + }, + { + "epoch": 2.4657368767845376, + "grad_norm": 2.288404703140259, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7168586254119873, + "num_tokens": 580893306.0, + "step": 22453 + }, + { + "epoch": 2.4658466944871513, + "grad_norm": 2.5510520935058594, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7332473397254944, + "num_tokens": 580916349.0, + "step": 22454 + }, + { + "epoch": 2.465956512189765, + "grad_norm": 2.704026460647583, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7539764642715454, + "num_tokens": 580936238.0, + "step": 22455 + }, + { + "epoch": 2.466066329892379, + "grad_norm": 2.049257755279541, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7056400179862976, + "num_tokens": 580970218.0, + "step": 22456 + }, + { + "epoch": 2.466176147594992, + "grad_norm": 2.599029064178467, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7439636588096619, + "num_tokens": 580993275.0, + "step": 22457 + }, + { + "epoch": 2.466285965297606, + "grad_norm": 1.9940288066864014, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7452747225761414, + "num_tokens": 581023208.0, + "step": 22458 + }, + { + "epoch": 2.4663957830002197, + "grad_norm": 2.3599231243133545, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7612281441688538, + "num_tokens": 581047242.0, + "step": 22459 + }, + { + "epoch": 2.4665056007028334, + "grad_norm": 2.571552276611328, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7493584752082825, + "num_tokens": 581069604.0, + "step": 22460 + }, + { + "epoch": 2.466615418405447, + "grad_norm": 2.5164811611175537, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7402869462966919, + "num_tokens": 581093665.0, + "step": 22461 + }, + { + "epoch": 2.4667252361080605, + "grad_norm": 2.4495675563812256, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7595125436782837, + "num_tokens": 581117503.0, + "step": 22462 + }, + { + "epoch": 2.4668350538106742, + "grad_norm": 2.430213451385498, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7440663576126099, + "num_tokens": 581141075.0, + "step": 22463 + }, + { + "epoch": 2.466944871513288, + "grad_norm": 2.2899107933044434, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7192981243133545, + "num_tokens": 581168850.0, + "step": 22464 + }, + { + "epoch": 2.4670546892159018, + "grad_norm": 2.4017064571380615, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7337974905967712, + "num_tokens": 581194901.0, + "step": 22465 + }, + { + "epoch": 2.467164506918515, + "grad_norm": 2.249525785446167, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7259862422943115, + "num_tokens": 581223253.0, + "step": 22466 + }, + { + "epoch": 2.467274324621129, + "grad_norm": 2.0600955486297607, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7103461623191833, + "num_tokens": 581257657.0, + "step": 22467 + }, + { + "epoch": 2.4673841423237426, + "grad_norm": 3.7379257678985596, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7319800853729248, + "num_tokens": 581280242.0, + "step": 22468 + }, + { + "epoch": 2.4674939600263563, + "grad_norm": 2.265934705734253, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7268122434616089, + "num_tokens": 581309213.0, + "step": 22469 + }, + { + "epoch": 2.4676037777289697, + "grad_norm": 2.8173398971557617, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7353965044021606, + "num_tokens": 581328185.0, + "step": 22470 + }, + { + "epoch": 2.4677135954315834, + "grad_norm": 2.6559510231018066, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7650896310806274, + "num_tokens": 581348910.0, + "step": 22471 + }, + { + "epoch": 2.467823413134197, + "grad_norm": 2.3702147006988525, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7337626218795776, + "num_tokens": 581375189.0, + "step": 22472 + }, + { + "epoch": 2.467933230836811, + "grad_norm": 2.5432794094085693, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7506681680679321, + "num_tokens": 581396012.0, + "step": 22473 + }, + { + "epoch": 2.4680430485394247, + "grad_norm": 2.2641849517822266, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7354613542556763, + "num_tokens": 581423460.0, + "step": 22474 + }, + { + "epoch": 2.468152866242038, + "grad_norm": 2.4378507137298584, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7468557357788086, + "num_tokens": 581449750.0, + "step": 22475 + }, + { + "epoch": 2.4682626839446518, + "grad_norm": 2.52812123298645, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7429935932159424, + "num_tokens": 581470226.0, + "step": 22476 + }, + { + "epoch": 2.4683725016472655, + "grad_norm": 2.361323595046997, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7274469137191772, + "num_tokens": 581496730.0, + "step": 22477 + }, + { + "epoch": 2.4684823193498793, + "grad_norm": 2.341564655303955, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.740007758140564, + "num_tokens": 581520234.0, + "step": 22478 + }, + { + "epoch": 2.468592137052493, + "grad_norm": 2.2668206691741943, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7357673048973083, + "num_tokens": 581548145.0, + "step": 22479 + }, + { + "epoch": 2.4687019547551063, + "grad_norm": 2.640820026397705, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7406662702560425, + "num_tokens": 581570008.0, + "step": 22480 + }, + { + "epoch": 2.46881177245772, + "grad_norm": 2.3177008628845215, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7456022500991821, + "num_tokens": 581594675.0, + "step": 22481 + }, + { + "epoch": 2.468921590160334, + "grad_norm": 2.236070394515991, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.731417715549469, + "num_tokens": 581621903.0, + "step": 22482 + }, + { + "epoch": 2.4690314078629476, + "grad_norm": 2.171468734741211, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7267759442329407, + "num_tokens": 581653424.0, + "step": 22483 + }, + { + "epoch": 2.4691412255655614, + "grad_norm": 2.3888235092163086, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7314993739128113, + "num_tokens": 581680659.0, + "step": 22484 + }, + { + "epoch": 2.4692510432681747, + "grad_norm": 2.307666540145874, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.726984441280365, + "num_tokens": 581706034.0, + "step": 22485 + }, + { + "epoch": 2.4693608609707884, + "grad_norm": 2.087555408477783, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7220892906188965, + "num_tokens": 581736537.0, + "step": 22486 + }, + { + "epoch": 2.469470678673402, + "grad_norm": 2.6836788654327393, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7530958652496338, + "num_tokens": 581758212.0, + "step": 22487 + }, + { + "epoch": 2.469580496376016, + "grad_norm": 2.7630324363708496, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7417370676994324, + "num_tokens": 581777728.0, + "step": 22488 + }, + { + "epoch": 2.4696903140786297, + "grad_norm": 2.4501571655273438, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7596721649169922, + "num_tokens": 581802384.0, + "step": 22489 + }, + { + "epoch": 2.469800131781243, + "grad_norm": 2.2224926948547363, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7190634608268738, + "num_tokens": 581831897.0, + "step": 22490 + }, + { + "epoch": 2.469909949483857, + "grad_norm": 2.7640788555145264, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7497013211250305, + "num_tokens": 581852383.0, + "step": 22491 + }, + { + "epoch": 2.4700197671864705, + "grad_norm": 2.429777145385742, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.731691837310791, + "num_tokens": 581876337.0, + "step": 22492 + }, + { + "epoch": 2.4701295848890843, + "grad_norm": 2.2169911861419678, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7142224311828613, + "num_tokens": 581904782.0, + "step": 22493 + }, + { + "epoch": 2.4702394025916976, + "grad_norm": 2.0643081665039062, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6930094957351685, + "num_tokens": 581937627.0, + "step": 22494 + }, + { + "epoch": 2.4703492202943114, + "grad_norm": 2.4211137294769287, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7296063303947449, + "num_tokens": 581961556.0, + "step": 22495 + }, + { + "epoch": 2.470459037996925, + "grad_norm": 2.5158989429473877, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7554773092269897, + "num_tokens": 581982982.0, + "step": 22496 + }, + { + "epoch": 2.470568855699539, + "grad_norm": 2.5088882446289062, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7472994327545166, + "num_tokens": 582007050.0, + "step": 22497 + }, + { + "epoch": 2.470678673402152, + "grad_norm": 2.1529695987701416, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7389057874679565, + "num_tokens": 582037396.0, + "step": 22498 + }, + { + "epoch": 2.470788491104766, + "grad_norm": 2.5455758571624756, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7108058929443359, + "num_tokens": 582060656.0, + "step": 22499 + }, + { + "epoch": 2.4708983088073797, + "grad_norm": 2.0009782314300537, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7005906105041504, + "num_tokens": 582098147.0, + "step": 22500 + }, + { + "epoch": 2.4710081265099935, + "grad_norm": 2.5461158752441406, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7127797603607178, + "num_tokens": 582120838.0, + "step": 22501 + }, + { + "epoch": 2.471117944212607, + "grad_norm": 2.5886919498443604, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7318710684776306, + "num_tokens": 582142587.0, + "step": 22502 + }, + { + "epoch": 2.4712277619152205, + "grad_norm": 2.675771713256836, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7253824472427368, + "num_tokens": 582164094.0, + "step": 22503 + }, + { + "epoch": 2.4713375796178343, + "grad_norm": 2.1604807376861572, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7232829332351685, + "num_tokens": 582191897.0, + "step": 22504 + }, + { + "epoch": 2.471447397320448, + "grad_norm": 2.468966007232666, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7334860563278198, + "num_tokens": 582215152.0, + "step": 22505 + }, + { + "epoch": 2.471557215023062, + "grad_norm": 2.7711181640625, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7349124550819397, + "num_tokens": 582235363.0, + "step": 22506 + }, + { + "epoch": 2.4716670327256756, + "grad_norm": 2.0493597984313965, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7279955148696899, + "num_tokens": 582268099.0, + "step": 22507 + }, + { + "epoch": 2.471776850428289, + "grad_norm": 2.6871917247772217, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7356330156326294, + "num_tokens": 582287577.0, + "step": 22508 + }, + { + "epoch": 2.4718866681309026, + "grad_norm": 2.2645609378814697, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7305929064750671, + "num_tokens": 582316791.0, + "step": 22509 + }, + { + "epoch": 2.4719964858335164, + "grad_norm": 2.3363137245178223, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.738640546798706, + "num_tokens": 582343571.0, + "step": 22510 + }, + { + "epoch": 2.47210630353613, + "grad_norm": 2.364586114883423, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7284712791442871, + "num_tokens": 582370286.0, + "step": 22511 + }, + { + "epoch": 2.472216121238744, + "grad_norm": 2.2112531661987305, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7502400875091553, + "num_tokens": 582396706.0, + "step": 22512 + }, + { + "epoch": 2.472325938941357, + "grad_norm": 2.6710190773010254, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7241321802139282, + "num_tokens": 582419388.0, + "step": 22513 + }, + { + "epoch": 2.472435756643971, + "grad_norm": 2.166273593902588, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7166692018508911, + "num_tokens": 582449128.0, + "step": 22514 + }, + { + "epoch": 2.4725455743465847, + "grad_norm": 2.108672857284546, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7313143014907837, + "num_tokens": 582481366.0, + "step": 22515 + }, + { + "epoch": 2.4726553920491985, + "grad_norm": 2.2798984050750732, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7168991565704346, + "num_tokens": 582509599.0, + "step": 22516 + }, + { + "epoch": 2.472765209751812, + "grad_norm": 2.347033739089966, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7255547046661377, + "num_tokens": 582537510.0, + "step": 22517 + }, + { + "epoch": 2.4728750274544256, + "grad_norm": 2.1941041946411133, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7305002212524414, + "num_tokens": 582565833.0, + "step": 22518 + }, + { + "epoch": 2.4729848451570393, + "grad_norm": 2.1868269443511963, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7301137447357178, + "num_tokens": 582595069.0, + "step": 22519 + }, + { + "epoch": 2.473094662859653, + "grad_norm": 2.1923630237579346, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7234830856323242, + "num_tokens": 582626140.0, + "step": 22520 + }, + { + "epoch": 2.473204480562267, + "grad_norm": 2.0962610244750977, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7394476532936096, + "num_tokens": 582656988.0, + "step": 22521 + }, + { + "epoch": 2.47331429826488, + "grad_norm": 2.114534378051758, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7026798725128174, + "num_tokens": 582690171.0, + "step": 22522 + }, + { + "epoch": 2.473424115967494, + "grad_norm": 2.340287923812866, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7484670877456665, + "num_tokens": 582714233.0, + "step": 22523 + }, + { + "epoch": 2.4735339336701077, + "grad_norm": 2.3637006282806396, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7512354850769043, + "num_tokens": 582738901.0, + "step": 22524 + }, + { + "epoch": 2.4736437513727214, + "grad_norm": 2.422842264175415, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7522445917129517, + "num_tokens": 582761168.0, + "step": 22525 + }, + { + "epoch": 2.4737535690753347, + "grad_norm": 2.348052978515625, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7292075753211975, + "num_tokens": 582786137.0, + "step": 22526 + }, + { + "epoch": 2.4738633867779485, + "grad_norm": 2.2901885509490967, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.719308614730835, + "num_tokens": 582812937.0, + "step": 22527 + }, + { + "epoch": 2.4739732044805622, + "grad_norm": 2.3831348419189453, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7211395502090454, + "num_tokens": 582840990.0, + "step": 22528 + }, + { + "epoch": 2.474083022183176, + "grad_norm": 2.649683952331543, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7450176477432251, + "num_tokens": 582862034.0, + "step": 22529 + }, + { + "epoch": 2.4741928398857898, + "grad_norm": 2.472759246826172, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7277272939682007, + "num_tokens": 582887359.0, + "step": 22530 + }, + { + "epoch": 2.474302657588403, + "grad_norm": 2.3095386028289795, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7542471289634705, + "num_tokens": 582911724.0, + "step": 22531 + }, + { + "epoch": 2.474412475291017, + "grad_norm": 2.204660415649414, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.733519971370697, + "num_tokens": 582940079.0, + "step": 22532 + }, + { + "epoch": 2.4745222929936306, + "grad_norm": 2.3812286853790283, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7286778688430786, + "num_tokens": 582966645.0, + "step": 22533 + }, + { + "epoch": 2.4746321106962443, + "grad_norm": 2.476490020751953, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7309590578079224, + "num_tokens": 582990337.0, + "step": 22534 + }, + { + "epoch": 2.474741928398858, + "grad_norm": 2.2915263175964355, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7290776371955872, + "num_tokens": 583019007.0, + "step": 22535 + }, + { + "epoch": 2.4748517461014714, + "grad_norm": 2.316249132156372, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7365666627883911, + "num_tokens": 583046525.0, + "step": 22536 + }, + { + "epoch": 2.474961563804085, + "grad_norm": 2.2877109050750732, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7122242450714111, + "num_tokens": 583075457.0, + "step": 22537 + }, + { + "epoch": 2.475071381506699, + "grad_norm": 2.57635760307312, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7336673736572266, + "num_tokens": 583097935.0, + "step": 22538 + }, + { + "epoch": 2.4751811992093127, + "grad_norm": 2.831912040710449, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7470901012420654, + "num_tokens": 583116913.0, + "step": 22539 + }, + { + "epoch": 2.4752910169119264, + "grad_norm": 1.9588658809661865, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7449039220809937, + "num_tokens": 583152270.0, + "step": 22540 + }, + { + "epoch": 2.4754008346145397, + "grad_norm": 2.25651216506958, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7208632826805115, + "num_tokens": 583181986.0, + "step": 22541 + }, + { + "epoch": 2.4755106523171535, + "grad_norm": 2.057076930999756, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7238951921463013, + "num_tokens": 583212767.0, + "step": 22542 + }, + { + "epoch": 2.4756204700197673, + "grad_norm": 2.8108139038085938, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7561394572257996, + "num_tokens": 583230748.0, + "step": 22543 + }, + { + "epoch": 2.475730287722381, + "grad_norm": 2.678231954574585, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7569954991340637, + "num_tokens": 583250752.0, + "step": 22544 + }, + { + "epoch": 2.4758401054249943, + "grad_norm": 2.3798937797546387, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7071807384490967, + "num_tokens": 583278437.0, + "step": 22545 + }, + { + "epoch": 2.475949923127608, + "grad_norm": 2.2207789421081543, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7117563486099243, + "num_tokens": 583305175.0, + "step": 22546 + }, + { + "epoch": 2.476059740830222, + "grad_norm": 2.5521275997161865, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7287745475769043, + "num_tokens": 583327026.0, + "step": 22547 + }, + { + "epoch": 2.4761695585328356, + "grad_norm": 2.219860792160034, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7216029167175293, + "num_tokens": 583356237.0, + "step": 22548 + }, + { + "epoch": 2.476279376235449, + "grad_norm": 2.1698968410491943, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7303189039230347, + "num_tokens": 583384805.0, + "step": 22549 + }, + { + "epoch": 2.4763891939380627, + "grad_norm": 2.37813138961792, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7406977415084839, + "num_tokens": 583409634.0, + "step": 22550 + }, + { + "epoch": 2.4764990116406764, + "grad_norm": 2.137187957763672, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7102832198143005, + "num_tokens": 583439142.0, + "step": 22551 + }, + { + "epoch": 2.47660882934329, + "grad_norm": 2.505526065826416, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7197228670120239, + "num_tokens": 583462208.0, + "step": 22552 + }, + { + "epoch": 2.476718647045904, + "grad_norm": 2.414290428161621, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7384742498397827, + "num_tokens": 583485821.0, + "step": 22553 + }, + { + "epoch": 2.4768284647485173, + "grad_norm": 2.220660924911499, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7087286710739136, + "num_tokens": 583516572.0, + "step": 22554 + }, + { + "epoch": 2.476938282451131, + "grad_norm": 2.290980577468872, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7283996343612671, + "num_tokens": 583542215.0, + "step": 22555 + }, + { + "epoch": 2.4770481001537448, + "grad_norm": 2.4392189979553223, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7142899632453918, + "num_tokens": 583568246.0, + "step": 22556 + }, + { + "epoch": 2.4771579178563585, + "grad_norm": 2.463491678237915, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7416830658912659, + "num_tokens": 583593046.0, + "step": 22557 + }, + { + "epoch": 2.4772677355589723, + "grad_norm": 2.4482407569885254, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7484813332557678, + "num_tokens": 583617182.0, + "step": 22558 + }, + { + "epoch": 2.4773775532615856, + "grad_norm": 2.5055816173553467, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7275873422622681, + "num_tokens": 583640409.0, + "step": 22559 + }, + { + "epoch": 2.4774873709641994, + "grad_norm": 2.343341112136841, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7310361266136169, + "num_tokens": 583667501.0, + "step": 22560 + }, + { + "epoch": 2.477597188666813, + "grad_norm": 2.4434309005737305, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.735051155090332, + "num_tokens": 583691981.0, + "step": 22561 + }, + { + "epoch": 2.477707006369427, + "grad_norm": 2.0730111598968506, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7282239198684692, + "num_tokens": 583723544.0, + "step": 22562 + }, + { + "epoch": 2.4778168240720406, + "grad_norm": 2.6621901988983154, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7364532947540283, + "num_tokens": 583745249.0, + "step": 22563 + }, + { + "epoch": 2.477926641774654, + "grad_norm": 2.4972119331359863, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.746902346611023, + "num_tokens": 583767713.0, + "step": 22564 + }, + { + "epoch": 2.4780364594772677, + "grad_norm": 2.2779958248138428, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7344481945037842, + "num_tokens": 583796351.0, + "step": 22565 + }, + { + "epoch": 2.4781462771798815, + "grad_norm": 2.6568219661712646, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7460266351699829, + "num_tokens": 583818522.0, + "step": 22566 + }, + { + "epoch": 2.478256094882495, + "grad_norm": 2.7524025440216064, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.7487934827804565, + "num_tokens": 583841306.0, + "step": 22567 + }, + { + "epoch": 2.478365912585109, + "grad_norm": 2.6102869510650635, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7587448954582214, + "num_tokens": 583862720.0, + "step": 22568 + }, + { + "epoch": 2.4784757302877223, + "grad_norm": 2.287100315093994, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7009036540985107, + "num_tokens": 583891557.0, + "step": 22569 + }, + { + "epoch": 2.478585547990336, + "grad_norm": 2.4134833812713623, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7325525879859924, + "num_tokens": 583915524.0, + "step": 22570 + }, + { + "epoch": 2.47869536569295, + "grad_norm": 2.060534954071045, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.739140510559082, + "num_tokens": 583947311.0, + "step": 22571 + }, + { + "epoch": 2.4788051833955635, + "grad_norm": 2.3896830081939697, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7255300283432007, + "num_tokens": 583971538.0, + "step": 22572 + }, + { + "epoch": 2.478915001098177, + "grad_norm": 2.8002419471740723, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7489727735519409, + "num_tokens": 583990662.0, + "step": 22573 + }, + { + "epoch": 2.4790248188007906, + "grad_norm": 2.2259018421173096, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7260399460792542, + "num_tokens": 584020069.0, + "step": 22574 + }, + { + "epoch": 2.4791346365034044, + "grad_norm": 2.509218454360962, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7500408887863159, + "num_tokens": 584043942.0, + "step": 22575 + }, + { + "epoch": 2.479244454206018, + "grad_norm": 2.6483588218688965, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7375963926315308, + "num_tokens": 584064359.0, + "step": 22576 + }, + { + "epoch": 2.4793542719086314, + "grad_norm": 2.408015489578247, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7167453765869141, + "num_tokens": 584089446.0, + "step": 22577 + }, + { + "epoch": 2.479464089611245, + "grad_norm": 2.6160995960235596, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7588165402412415, + "num_tokens": 584112276.0, + "step": 22578 + }, + { + "epoch": 2.479573907313859, + "grad_norm": 2.8888168334960938, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7209102511405945, + "num_tokens": 584131282.0, + "step": 22579 + }, + { + "epoch": 2.4796837250164727, + "grad_norm": 2.4776697158813477, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7243318557739258, + "num_tokens": 584157027.0, + "step": 22580 + }, + { + "epoch": 2.4797935427190865, + "grad_norm": 2.210724115371704, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7011681199073792, + "num_tokens": 584187271.0, + "step": 22581 + }, + { + "epoch": 2.4799033604217, + "grad_norm": 2.3501968383789062, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7294520139694214, + "num_tokens": 584214126.0, + "step": 22582 + }, + { + "epoch": 2.4800131781243135, + "grad_norm": 2.4354591369628906, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7459022402763367, + "num_tokens": 584238308.0, + "step": 22583 + }, + { + "epoch": 2.4801229958269273, + "grad_norm": 2.592815637588501, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.708621621131897, + "num_tokens": 584260562.0, + "step": 22584 + }, + { + "epoch": 2.480232813529541, + "grad_norm": 2.5175418853759766, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7259173393249512, + "num_tokens": 584285600.0, + "step": 22585 + }, + { + "epoch": 2.480342631232155, + "grad_norm": 2.4327433109283447, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7371684312820435, + "num_tokens": 584309999.0, + "step": 22586 + }, + { + "epoch": 2.480452448934768, + "grad_norm": 2.7588884830474854, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.743286669254303, + "num_tokens": 584331339.0, + "step": 22587 + }, + { + "epoch": 2.480562266637382, + "grad_norm": 2.6619107723236084, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7697505354881287, + "num_tokens": 584352018.0, + "step": 22588 + }, + { + "epoch": 2.4806720843399956, + "grad_norm": 2.541137218475342, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7670490145683289, + "num_tokens": 584374314.0, + "step": 22589 + }, + { + "epoch": 2.4807819020426094, + "grad_norm": 2.5507864952087402, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7068572044372559, + "num_tokens": 584396902.0, + "step": 22590 + }, + { + "epoch": 2.480891719745223, + "grad_norm": 2.498769760131836, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7455098628997803, + "num_tokens": 584419211.0, + "step": 22591 + }, + { + "epoch": 2.4810015374478365, + "grad_norm": 2.6074249744415283, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7313812971115112, + "num_tokens": 584442109.0, + "step": 22592 + }, + { + "epoch": 2.4811113551504502, + "grad_norm": 2.543194055557251, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7391928434371948, + "num_tokens": 584466087.0, + "step": 22593 + }, + { + "epoch": 2.481221172853064, + "grad_norm": 2.4379913806915283, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7214004993438721, + "num_tokens": 584490736.0, + "step": 22594 + }, + { + "epoch": 2.4813309905556777, + "grad_norm": 2.0060322284698486, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7263358235359192, + "num_tokens": 584525952.0, + "step": 22595 + }, + { + "epoch": 2.481440808258291, + "grad_norm": 2.203937292098999, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7280928492546082, + "num_tokens": 584554175.0, + "step": 22596 + }, + { + "epoch": 2.481550625960905, + "grad_norm": 2.3257224559783936, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7367420196533203, + "num_tokens": 584581818.0, + "step": 22597 + }, + { + "epoch": 2.4816604436635186, + "grad_norm": 2.2915563583374023, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7332074642181396, + "num_tokens": 584608837.0, + "step": 22598 + }, + { + "epoch": 2.4817702613661323, + "grad_norm": 2.273664712905884, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7166936993598938, + "num_tokens": 584637585.0, + "step": 22599 + }, + { + "epoch": 2.4818800790687456, + "grad_norm": 2.4499809741973877, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7335614562034607, + "num_tokens": 584661031.0, + "step": 22600 + }, + { + "epoch": 2.4819898967713594, + "grad_norm": 2.401857852935791, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.744381308555603, + "num_tokens": 584684536.0, + "step": 22601 + }, + { + "epoch": 2.482099714473973, + "grad_norm": 2.447478771209717, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7458754777908325, + "num_tokens": 584708199.0, + "step": 22602 + }, + { + "epoch": 2.482209532176587, + "grad_norm": 2.338324785232544, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7113864421844482, + "num_tokens": 584733161.0, + "step": 22603 + }, + { + "epoch": 2.4823193498792007, + "grad_norm": 2.447561025619507, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7410544753074646, + "num_tokens": 584757816.0, + "step": 22604 + }, + { + "epoch": 2.482429167581814, + "grad_norm": 2.1095387935638428, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7494245767593384, + "num_tokens": 584787031.0, + "step": 22605 + }, + { + "epoch": 2.4825389852844277, + "grad_norm": 2.5717248916625977, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7468701601028442, + "num_tokens": 584808485.0, + "step": 22606 + }, + { + "epoch": 2.4826488029870415, + "grad_norm": 2.7341108322143555, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7351421117782593, + "num_tokens": 584830143.0, + "step": 22607 + }, + { + "epoch": 2.4827586206896552, + "grad_norm": 2.4177823066711426, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7320223450660706, + "num_tokens": 584854495.0, + "step": 22608 + }, + { + "epoch": 2.482868438392269, + "grad_norm": 2.493852376937866, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7319706678390503, + "num_tokens": 584877030.0, + "step": 22609 + }, + { + "epoch": 2.4829782560948823, + "grad_norm": 2.5952954292297363, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7586532235145569, + "num_tokens": 584899034.0, + "step": 22610 + }, + { + "epoch": 2.483088073797496, + "grad_norm": 2.5227954387664795, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.728186845779419, + "num_tokens": 584924205.0, + "step": 22611 + }, + { + "epoch": 2.48319789150011, + "grad_norm": 2.391923427581787, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7117660045623779, + "num_tokens": 584949529.0, + "step": 22612 + }, + { + "epoch": 2.4833077092027236, + "grad_norm": 2.636441230773926, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7373627424240112, + "num_tokens": 584972765.0, + "step": 22613 + }, + { + "epoch": 2.4834175269053373, + "grad_norm": 2.091170072555542, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7104618549346924, + "num_tokens": 585004051.0, + "step": 22614 + }, + { + "epoch": 2.4835273446079507, + "grad_norm": 2.116353988647461, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7005405426025391, + "num_tokens": 585035043.0, + "step": 22615 + }, + { + "epoch": 2.4836371623105644, + "grad_norm": 2.2531917095184326, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7352175116539001, + "num_tokens": 585063833.0, + "step": 22616 + }, + { + "epoch": 2.483746980013178, + "grad_norm": 2.440462589263916, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7174358367919922, + "num_tokens": 585089772.0, + "step": 22617 + }, + { + "epoch": 2.483856797715792, + "grad_norm": 2.382300853729248, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7188695669174194, + "num_tokens": 585113862.0, + "step": 22618 + }, + { + "epoch": 2.4839666154184057, + "grad_norm": 2.249624490737915, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.725153923034668, + "num_tokens": 585141848.0, + "step": 22619 + }, + { + "epoch": 2.484076433121019, + "grad_norm": 2.506884813308716, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.727599024772644, + "num_tokens": 585165372.0, + "step": 22620 + }, + { + "epoch": 2.4841862508236328, + "grad_norm": 2.408216714859009, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7267493009567261, + "num_tokens": 585190186.0, + "step": 22621 + }, + { + "epoch": 2.4842960685262465, + "grad_norm": 2.4604527950286865, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7436929941177368, + "num_tokens": 585215818.0, + "step": 22622 + }, + { + "epoch": 2.4844058862288603, + "grad_norm": 2.3855936527252197, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7532804608345032, + "num_tokens": 585238265.0, + "step": 22623 + }, + { + "epoch": 2.4845157039314736, + "grad_norm": 2.175436496734619, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7435293793678284, + "num_tokens": 585266467.0, + "step": 22624 + }, + { + "epoch": 2.4846255216340873, + "grad_norm": 2.2539055347442627, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7448409199714661, + "num_tokens": 585295652.0, + "step": 22625 + }, + { + "epoch": 2.484735339336701, + "grad_norm": 2.256861686706543, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7418851256370544, + "num_tokens": 585323200.0, + "step": 22626 + }, + { + "epoch": 2.484845157039315, + "grad_norm": 2.4344534873962402, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7352113723754883, + "num_tokens": 585347754.0, + "step": 22627 + }, + { + "epoch": 2.484954974741928, + "grad_norm": 2.3445560932159424, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7340982556343079, + "num_tokens": 585373003.0, + "step": 22628 + }, + { + "epoch": 2.485064792444542, + "grad_norm": 2.2807228565216064, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7243584394454956, + "num_tokens": 585399580.0, + "step": 22629 + }, + { + "epoch": 2.4851746101471557, + "grad_norm": 2.5501468181610107, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7410498857498169, + "num_tokens": 585423217.0, + "step": 22630 + }, + { + "epoch": 2.4852844278497694, + "grad_norm": 2.235623359680176, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7269906997680664, + "num_tokens": 585451117.0, + "step": 22631 + }, + { + "epoch": 2.485394245552383, + "grad_norm": 2.145110845565796, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7288025617599487, + "num_tokens": 585480424.0, + "step": 22632 + }, + { + "epoch": 2.4855040632549965, + "grad_norm": 2.2330033779144287, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7446206212043762, + "num_tokens": 585508401.0, + "step": 22633 + }, + { + "epoch": 2.4856138809576103, + "grad_norm": 2.2141754627227783, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7174874544143677, + "num_tokens": 585537066.0, + "step": 22634 + }, + { + "epoch": 2.485723698660224, + "grad_norm": 2.3213484287261963, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7242031097412109, + "num_tokens": 585564235.0, + "step": 22635 + }, + { + "epoch": 2.485833516362838, + "grad_norm": 2.680401086807251, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7403151988983154, + "num_tokens": 585585897.0, + "step": 22636 + }, + { + "epoch": 2.4859433340654515, + "grad_norm": 2.169041872024536, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7275099754333496, + "num_tokens": 585613902.0, + "step": 22637 + }, + { + "epoch": 2.486053151768065, + "grad_norm": 2.3170969486236572, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7320294380187988, + "num_tokens": 585640191.0, + "step": 22638 + }, + { + "epoch": 2.4861629694706786, + "grad_norm": 2.397526502609253, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7301780581474304, + "num_tokens": 585668203.0, + "step": 22639 + }, + { + "epoch": 2.4862727871732924, + "grad_norm": 2.4946022033691406, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7370526790618896, + "num_tokens": 585692082.0, + "step": 22640 + }, + { + "epoch": 2.486382604875906, + "grad_norm": 2.5295403003692627, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7586684226989746, + "num_tokens": 585713766.0, + "step": 22641 + }, + { + "epoch": 2.48649242257852, + "grad_norm": 2.3330419063568115, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7490679025650024, + "num_tokens": 585739024.0, + "step": 22642 + }, + { + "epoch": 2.486602240281133, + "grad_norm": 2.1903305053710938, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7243999242782593, + "num_tokens": 585768397.0, + "step": 22643 + }, + { + "epoch": 2.486712057983747, + "grad_norm": 2.5777549743652344, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7320719957351685, + "num_tokens": 585790832.0, + "step": 22644 + }, + { + "epoch": 2.4868218756863607, + "grad_norm": 2.3746602535247803, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7207388877868652, + "num_tokens": 585816244.0, + "step": 22645 + }, + { + "epoch": 2.4869316933889745, + "grad_norm": 2.341139316558838, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7454780340194702, + "num_tokens": 585841796.0, + "step": 22646 + }, + { + "epoch": 2.4870415110915878, + "grad_norm": 2.3258862495422363, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7065505981445312, + "num_tokens": 585868691.0, + "step": 22647 + }, + { + "epoch": 2.4871513287942015, + "grad_norm": 2.2559497356414795, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7261062264442444, + "num_tokens": 585896525.0, + "step": 22648 + }, + { + "epoch": 2.4872611464968153, + "grad_norm": 2.386453151702881, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7322793006896973, + "num_tokens": 585922581.0, + "step": 22649 + }, + { + "epoch": 2.487370964199429, + "grad_norm": 2.2238669395446777, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.740459144115448, + "num_tokens": 585952214.0, + "step": 22650 + }, + { + "epoch": 2.487480781902043, + "grad_norm": 2.171778917312622, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7341740727424622, + "num_tokens": 585981483.0, + "step": 22651 + }, + { + "epoch": 2.487590599604656, + "grad_norm": 2.321593999862671, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7363108992576599, + "num_tokens": 586006401.0, + "step": 22652 + }, + { + "epoch": 2.48770041730727, + "grad_norm": 2.1618738174438477, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7182027697563171, + "num_tokens": 586035868.0, + "step": 22653 + }, + { + "epoch": 2.4878102350098836, + "grad_norm": 2.3602781295776367, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7210976481437683, + "num_tokens": 586063210.0, + "step": 22654 + }, + { + "epoch": 2.4879200527124974, + "grad_norm": 2.445444345474243, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7469873428344727, + "num_tokens": 586088749.0, + "step": 22655 + }, + { + "epoch": 2.4880298704151107, + "grad_norm": 2.2900757789611816, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7379637956619263, + "num_tokens": 586114721.0, + "step": 22656 + }, + { + "epoch": 2.4881396881177245, + "grad_norm": 2.385845899581909, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7428253889083862, + "num_tokens": 586138740.0, + "step": 22657 + }, + { + "epoch": 2.488249505820338, + "grad_norm": 2.4195375442504883, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7230603694915771, + "num_tokens": 586163806.0, + "step": 22658 + }, + { + "epoch": 2.488359323522952, + "grad_norm": 2.363718271255493, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7141016721725464, + "num_tokens": 586189894.0, + "step": 22659 + }, + { + "epoch": 2.4884691412255657, + "grad_norm": 2.041778326034546, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7245094776153564, + "num_tokens": 586218294.0, + "step": 22660 + }, + { + "epoch": 2.488578958928179, + "grad_norm": 2.4935872554779053, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7370705604553223, + "num_tokens": 586241428.0, + "step": 22661 + }, + { + "epoch": 2.488688776630793, + "grad_norm": 2.682690143585205, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7478286027908325, + "num_tokens": 586259610.0, + "step": 22662 + }, + { + "epoch": 2.4887985943334066, + "grad_norm": 2.3323745727539062, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7178326845169067, + "num_tokens": 586287264.0, + "step": 22663 + }, + { + "epoch": 2.4889084120360203, + "grad_norm": 2.1619770526885986, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7440364360809326, + "num_tokens": 586315879.0, + "step": 22664 + }, + { + "epoch": 2.489018229738634, + "grad_norm": 2.721174955368042, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7167780995368958, + "num_tokens": 586336976.0, + "step": 22665 + }, + { + "epoch": 2.4891280474412474, + "grad_norm": 2.0940704345703125, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7144270539283752, + "num_tokens": 586369199.0, + "step": 22666 + }, + { + "epoch": 2.489237865143861, + "grad_norm": 2.953273296356201, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7388628721237183, + "num_tokens": 586387568.0, + "step": 22667 + }, + { + "epoch": 2.489347682846475, + "grad_norm": 2.2666234970092773, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7217531800270081, + "num_tokens": 586415242.0, + "step": 22668 + }, + { + "epoch": 2.4894575005490887, + "grad_norm": 2.298816204071045, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7332414388656616, + "num_tokens": 586441756.0, + "step": 22669 + }, + { + "epoch": 2.4895673182517024, + "grad_norm": 2.1870996952056885, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7393878102302551, + "num_tokens": 586468933.0, + "step": 22670 + }, + { + "epoch": 2.4896771359543157, + "grad_norm": 2.429751396179199, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.719034731388092, + "num_tokens": 586495660.0, + "step": 22671 + }, + { + "epoch": 2.4897869536569295, + "grad_norm": 2.2993152141571045, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.721677303314209, + "num_tokens": 586522449.0, + "step": 22672 + }, + { + "epoch": 2.4898967713595432, + "grad_norm": 2.1983559131622314, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7632833123207092, + "num_tokens": 586549179.0, + "step": 22673 + }, + { + "epoch": 2.490006589062157, + "grad_norm": 2.4741227626800537, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7330367565155029, + "num_tokens": 586574623.0, + "step": 22674 + }, + { + "epoch": 2.4901164067647703, + "grad_norm": 2.4796628952026367, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7209882736206055, + "num_tokens": 586600188.0, + "step": 22675 + }, + { + "epoch": 2.490226224467384, + "grad_norm": 2.511744737625122, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7312694787979126, + "num_tokens": 586623978.0, + "step": 22676 + }, + { + "epoch": 2.490336042169998, + "grad_norm": 2.5439226627349854, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7205365300178528, + "num_tokens": 586647401.0, + "step": 22677 + }, + { + "epoch": 2.4904458598726116, + "grad_norm": 2.7141470909118652, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7296456098556519, + "num_tokens": 586668775.0, + "step": 22678 + }, + { + "epoch": 2.490555677575225, + "grad_norm": 2.4117424488067627, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7368336915969849, + "num_tokens": 586694421.0, + "step": 22679 + }, + { + "epoch": 2.4906654952778386, + "grad_norm": 2.2847394943237305, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7249603271484375, + "num_tokens": 586721979.0, + "step": 22680 + }, + { + "epoch": 2.4907753129804524, + "grad_norm": 2.312929153442383, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6912273168563843, + "num_tokens": 586752442.0, + "step": 22681 + }, + { + "epoch": 2.490885130683066, + "grad_norm": 2.318559169769287, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.708316445350647, + "num_tokens": 586780628.0, + "step": 22682 + }, + { + "epoch": 2.49099494838568, + "grad_norm": 2.128319263458252, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7482250332832336, + "num_tokens": 586809136.0, + "step": 22683 + }, + { + "epoch": 2.4911047660882932, + "grad_norm": 2.144821882247925, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7284470796585083, + "num_tokens": 586838550.0, + "step": 22684 + }, + { + "epoch": 2.491214583790907, + "grad_norm": 2.064213991165161, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.6945832967758179, + "num_tokens": 586872032.0, + "step": 22685 + }, + { + "epoch": 2.4913244014935207, + "grad_norm": 2.0710928440093994, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7207304835319519, + "num_tokens": 586906818.0, + "step": 22686 + }, + { + "epoch": 2.4914342191961345, + "grad_norm": 2.7493247985839844, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.740216851234436, + "num_tokens": 586928153.0, + "step": 22687 + }, + { + "epoch": 2.4915440368987483, + "grad_norm": 2.285581588745117, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7219932675361633, + "num_tokens": 586954338.0, + "step": 22688 + }, + { + "epoch": 2.4916538546013616, + "grad_norm": 2.27034854888916, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7251952886581421, + "num_tokens": 586981709.0, + "step": 22689 + }, + { + "epoch": 2.4917636723039753, + "grad_norm": 2.2901418209075928, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7471498250961304, + "num_tokens": 587009605.0, + "step": 22690 + }, + { + "epoch": 2.491873490006589, + "grad_norm": 2.4742043018341064, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.746435284614563, + "num_tokens": 587032841.0, + "step": 22691 + }, + { + "epoch": 2.491983307709203, + "grad_norm": 2.5761325359344482, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7320352792739868, + "num_tokens": 587056958.0, + "step": 22692 + }, + { + "epoch": 2.4920931254118166, + "grad_norm": 2.2983481884002686, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.735045313835144, + "num_tokens": 587083782.0, + "step": 22693 + }, + { + "epoch": 2.49220294311443, + "grad_norm": 2.400930881500244, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7106068134307861, + "num_tokens": 587110823.0, + "step": 22694 + }, + { + "epoch": 2.4923127608170437, + "grad_norm": 2.1283888816833496, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7139238119125366, + "num_tokens": 587140248.0, + "step": 22695 + }, + { + "epoch": 2.4924225785196574, + "grad_norm": 2.315187692642212, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7074314951896667, + "num_tokens": 587169688.0, + "step": 22696 + }, + { + "epoch": 2.492532396222271, + "grad_norm": 2.5943455696105957, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.741716206073761, + "num_tokens": 587192285.0, + "step": 22697 + }, + { + "epoch": 2.4926422139248845, + "grad_norm": 2.5441012382507324, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7434936165809631, + "num_tokens": 587213982.0, + "step": 22698 + }, + { + "epoch": 2.4927520316274983, + "grad_norm": 2.234797477722168, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7341737151145935, + "num_tokens": 587241666.0, + "step": 22699 + }, + { + "epoch": 2.492861849330112, + "grad_norm": 2.3871757984161377, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7396886348724365, + "num_tokens": 587265561.0, + "step": 22700 + }, + { + "epoch": 2.4929716670327258, + "grad_norm": 2.2845773696899414, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7121432423591614, + "num_tokens": 587295396.0, + "step": 22701 + }, + { + "epoch": 2.4930814847353395, + "grad_norm": 2.3494713306427, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7535974979400635, + "num_tokens": 587321031.0, + "step": 22702 + }, + { + "epoch": 2.493191302437953, + "grad_norm": 2.8761115074157715, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7445940971374512, + "num_tokens": 587339850.0, + "step": 22703 + }, + { + "epoch": 2.4933011201405666, + "grad_norm": 2.112579345703125, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.703639566898346, + "num_tokens": 587369105.0, + "step": 22704 + }, + { + "epoch": 2.4934109378431804, + "grad_norm": 2.368530035018921, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7137573957443237, + "num_tokens": 587397579.0, + "step": 22705 + }, + { + "epoch": 2.493520755545794, + "grad_norm": 2.3738837242126465, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7334868907928467, + "num_tokens": 587423378.0, + "step": 22706 + }, + { + "epoch": 2.4936305732484074, + "grad_norm": 2.5178534984588623, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7335538864135742, + "num_tokens": 587446342.0, + "step": 22707 + }, + { + "epoch": 2.493740390951021, + "grad_norm": 2.107104539871216, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7235032320022583, + "num_tokens": 587476983.0, + "step": 22708 + }, + { + "epoch": 2.493850208653635, + "grad_norm": 2.807737112045288, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7335001826286316, + "num_tokens": 587501236.0, + "step": 22709 + }, + { + "epoch": 2.4939600263562487, + "grad_norm": 2.782186269760132, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7358485460281372, + "num_tokens": 587521020.0, + "step": 22710 + }, + { + "epoch": 2.4940698440588625, + "grad_norm": 2.280322313308716, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7301938533782959, + "num_tokens": 587547275.0, + "step": 22711 + }, + { + "epoch": 2.4941796617614758, + "grad_norm": 2.424560785293579, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7416822910308838, + "num_tokens": 587573084.0, + "step": 22712 + }, + { + "epoch": 2.4942894794640895, + "grad_norm": 2.23820161819458, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7215842604637146, + "num_tokens": 587603257.0, + "step": 22713 + }, + { + "epoch": 2.4943992971667033, + "grad_norm": 2.3140957355499268, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7243492603302002, + "num_tokens": 587630187.0, + "step": 22714 + }, + { + "epoch": 2.494509114869317, + "grad_norm": 2.732959747314453, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7355445623397827, + "num_tokens": 587649842.0, + "step": 22715 + }, + { + "epoch": 2.494618932571931, + "grad_norm": 2.3471901416778564, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.725867748260498, + "num_tokens": 587675550.0, + "step": 22716 + }, + { + "epoch": 2.494728750274544, + "grad_norm": 2.5535922050476074, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7360639572143555, + "num_tokens": 587698137.0, + "step": 22717 + }, + { + "epoch": 2.494838567977158, + "grad_norm": 2.286165237426758, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7355431318283081, + "num_tokens": 587726673.0, + "step": 22718 + }, + { + "epoch": 2.4949483856797716, + "grad_norm": 2.126706838607788, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7341119050979614, + "num_tokens": 587756002.0, + "step": 22719 + }, + { + "epoch": 2.4950582033823854, + "grad_norm": 2.3259525299072266, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7181147336959839, + "num_tokens": 587784302.0, + "step": 22720 + }, + { + "epoch": 2.495168021084999, + "grad_norm": 2.210920572280884, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7287590503692627, + "num_tokens": 587813162.0, + "step": 22721 + }, + { + "epoch": 2.4952778387876124, + "grad_norm": 2.0723254680633545, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7426680326461792, + "num_tokens": 587844242.0, + "step": 22722 + }, + { + "epoch": 2.495387656490226, + "grad_norm": 2.249633550643921, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7168842554092407, + "num_tokens": 587872383.0, + "step": 22723 + }, + { + "epoch": 2.49549747419284, + "grad_norm": 2.466437816619873, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7292865514755249, + "num_tokens": 587895424.0, + "step": 22724 + }, + { + "epoch": 2.4956072918954537, + "grad_norm": 2.05110764503479, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7293848395347595, + "num_tokens": 587929377.0, + "step": 22725 + }, + { + "epoch": 2.495717109598067, + "grad_norm": 2.495363235473633, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7391422986984253, + "num_tokens": 587952057.0, + "step": 22726 + }, + { + "epoch": 2.495826927300681, + "grad_norm": 2.508969306945801, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7555199861526489, + "num_tokens": 587973851.0, + "step": 22727 + }, + { + "epoch": 2.4959367450032945, + "grad_norm": 2.4008212089538574, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7342435717582703, + "num_tokens": 588000156.0, + "step": 22728 + }, + { + "epoch": 2.4960465627059083, + "grad_norm": 2.217298984527588, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.732489287853241, + "num_tokens": 588027864.0, + "step": 22729 + }, + { + "epoch": 2.4961563804085216, + "grad_norm": 2.449437379837036, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7068654894828796, + "num_tokens": 588053574.0, + "step": 22730 + }, + { + "epoch": 2.4962661981111354, + "grad_norm": 2.064578056335449, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7424532175064087, + "num_tokens": 588083717.0, + "step": 22731 + }, + { + "epoch": 2.496376015813749, + "grad_norm": 2.683565139770508, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7361480593681335, + "num_tokens": 588106376.0, + "step": 22732 + }, + { + "epoch": 2.496485833516363, + "grad_norm": 2.261526346206665, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7337323427200317, + "num_tokens": 588134677.0, + "step": 22733 + }, + { + "epoch": 2.4965956512189766, + "grad_norm": 2.13271427154541, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.721060037612915, + "num_tokens": 588163054.0, + "step": 22734 + }, + { + "epoch": 2.49670546892159, + "grad_norm": 2.1873860359191895, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7258007526397705, + "num_tokens": 588192624.0, + "step": 22735 + }, + { + "epoch": 2.4968152866242037, + "grad_norm": 2.4009251594543457, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7288724780082703, + "num_tokens": 588217520.0, + "step": 22736 + }, + { + "epoch": 2.4969251043268175, + "grad_norm": 2.284553050994873, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7176783084869385, + "num_tokens": 588246847.0, + "step": 22737 + }, + { + "epoch": 2.4970349220294312, + "grad_norm": 2.516458749771118, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.727394163608551, + "num_tokens": 588271089.0, + "step": 22738 + }, + { + "epoch": 2.497144739732045, + "grad_norm": 2.3493852615356445, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7292091846466064, + "num_tokens": 588296509.0, + "step": 22739 + }, + { + "epoch": 2.4972545574346583, + "grad_norm": 2.546677350997925, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7276994585990906, + "num_tokens": 588320644.0, + "step": 22740 + }, + { + "epoch": 2.497364375137272, + "grad_norm": 2.098156690597534, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7376317977905273, + "num_tokens": 588351365.0, + "step": 22741 + }, + { + "epoch": 2.497474192839886, + "grad_norm": 2.4351682662963867, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.737491250038147, + "num_tokens": 588374301.0, + "step": 22742 + }, + { + "epoch": 2.4975840105424996, + "grad_norm": 2.6347625255584717, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7366870641708374, + "num_tokens": 588394979.0, + "step": 22743 + }, + { + "epoch": 2.4976938282451133, + "grad_norm": 2.5479323863983154, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7283976674079895, + "num_tokens": 588418678.0, + "step": 22744 + }, + { + "epoch": 2.4978036459477266, + "grad_norm": 2.4187679290771484, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7240968942642212, + "num_tokens": 588442045.0, + "step": 22745 + }, + { + "epoch": 2.4979134636503404, + "grad_norm": 2.6205077171325684, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7373347282409668, + "num_tokens": 588464154.0, + "step": 22746 + }, + { + "epoch": 2.498023281352954, + "grad_norm": 2.309765338897705, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.745057225227356, + "num_tokens": 588490177.0, + "step": 22747 + }, + { + "epoch": 2.498133099055568, + "grad_norm": 2.457911729812622, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.722936749458313, + "num_tokens": 588514830.0, + "step": 22748 + }, + { + "epoch": 2.4982429167581817, + "grad_norm": 2.864304542541504, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7457596063613892, + "num_tokens": 588533952.0, + "step": 22749 + }, + { + "epoch": 2.498352734460795, + "grad_norm": 2.79693865776062, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7316954135894775, + "num_tokens": 588553512.0, + "step": 22750 + }, + { + "epoch": 2.4984625521634087, + "grad_norm": 2.8678290843963623, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7644587755203247, + "num_tokens": 588570616.0, + "step": 22751 + }, + { + "epoch": 2.4985723698660225, + "grad_norm": 2.43072247505188, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7215081453323364, + "num_tokens": 588595651.0, + "step": 22752 + }, + { + "epoch": 2.4986821875686362, + "grad_norm": 2.6026697158813477, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7493820786476135, + "num_tokens": 588617392.0, + "step": 22753 + }, + { + "epoch": 2.4987920052712496, + "grad_norm": 2.13017201423645, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7002418041229248, + "num_tokens": 588649690.0, + "step": 22754 + }, + { + "epoch": 2.4989018229738633, + "grad_norm": 2.239483594894409, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7216072082519531, + "num_tokens": 588680395.0, + "step": 22755 + }, + { + "epoch": 2.499011640676477, + "grad_norm": 1.9891928434371948, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7146259546279907, + "num_tokens": 588715486.0, + "step": 22756 + }, + { + "epoch": 2.499121458379091, + "grad_norm": 2.467541217803955, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.723297119140625, + "num_tokens": 588741555.0, + "step": 22757 + }, + { + "epoch": 2.499231276081704, + "grad_norm": 2.1216940879821777, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7333135604858398, + "num_tokens": 588771413.0, + "step": 22758 + }, + { + "epoch": 2.499341093784318, + "grad_norm": 2.6184425354003906, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7554362416267395, + "num_tokens": 588791492.0, + "step": 22759 + }, + { + "epoch": 2.4994509114869317, + "grad_norm": 3.004354953765869, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7394411563873291, + "num_tokens": 588811461.0, + "step": 22760 + }, + { + "epoch": 2.4995607291895454, + "grad_norm": 2.3633270263671875, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7192217707633972, + "num_tokens": 588840279.0, + "step": 22761 + }, + { + "epoch": 2.499670546892159, + "grad_norm": 2.334211826324463, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7343209981918335, + "num_tokens": 588866667.0, + "step": 22762 + }, + { + "epoch": 2.4997803645947725, + "grad_norm": 2.251413106918335, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7247166633605957, + "num_tokens": 588894255.0, + "step": 22763 + }, + { + "epoch": 2.4998901822973862, + "grad_norm": 2.3910534381866455, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7305042147636414, + "num_tokens": 588921334.0, + "step": 22764 + }, + { + "epoch": 2.5, + "grad_norm": 2.6518402099609375, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7258565425872803, + "num_tokens": 588942704.0, + "step": 22765 + }, + { + "epoch": 2.5001098177026138, + "grad_norm": 2.189236879348755, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7198784351348877, + "num_tokens": 588971823.0, + "step": 22766 + }, + { + "epoch": 2.5002196354052275, + "grad_norm": 2.693652868270874, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7464072704315186, + "num_tokens": 588993280.0, + "step": 22767 + }, + { + "epoch": 2.500329453107841, + "grad_norm": 2.5321743488311768, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7168048024177551, + "num_tokens": 589016288.0, + "step": 22768 + }, + { + "epoch": 2.5004392708104546, + "grad_norm": 2.316232919692993, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7352766990661621, + "num_tokens": 589042750.0, + "step": 22769 + }, + { + "epoch": 2.5005490885130683, + "grad_norm": 2.4282004833221436, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.716841995716095, + "num_tokens": 589068047.0, + "step": 22770 + }, + { + "epoch": 2.500658906215682, + "grad_norm": 2.479376792907715, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7405222058296204, + "num_tokens": 589090217.0, + "step": 22771 + }, + { + "epoch": 2.500768723918296, + "grad_norm": 2.2475736141204834, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.72862708568573, + "num_tokens": 589120073.0, + "step": 22772 + }, + { + "epoch": 2.500878541620909, + "grad_norm": 2.3879733085632324, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.754898726940155, + "num_tokens": 589144251.0, + "step": 22773 + }, + { + "epoch": 2.500988359323523, + "grad_norm": 2.449981451034546, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7301722168922424, + "num_tokens": 589168790.0, + "step": 22774 + }, + { + "epoch": 2.5010981770261367, + "grad_norm": 2.3680593967437744, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7417036294937134, + "num_tokens": 589193331.0, + "step": 22775 + }, + { + "epoch": 2.5012079947287504, + "grad_norm": 2.265659809112549, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7284623384475708, + "num_tokens": 589221114.0, + "step": 22776 + }, + { + "epoch": 2.501317812431364, + "grad_norm": 2.292996883392334, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7625063061714172, + "num_tokens": 589246004.0, + "step": 22777 + }, + { + "epoch": 2.5014276301339775, + "grad_norm": 2.9045603275299072, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7355719804763794, + "num_tokens": 589265631.0, + "step": 22778 + }, + { + "epoch": 2.5015374478365913, + "grad_norm": 2.5231680870056152, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7557113170623779, + "num_tokens": 589287728.0, + "step": 22779 + }, + { + "epoch": 2.501647265539205, + "grad_norm": 2.3043789863586426, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7312356233596802, + "num_tokens": 589313655.0, + "step": 22780 + }, + { + "epoch": 2.5017570832418183, + "grad_norm": 2.4681689739227295, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7323100566864014, + "num_tokens": 589337141.0, + "step": 22781 + }, + { + "epoch": 2.501866900944432, + "grad_norm": 2.47417950630188, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7321735620498657, + "num_tokens": 589359967.0, + "step": 22782 + }, + { + "epoch": 2.501976718647046, + "grad_norm": 2.183382749557495, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.756889820098877, + "num_tokens": 589387337.0, + "step": 22783 + }, + { + "epoch": 2.5020865363496596, + "grad_norm": 2.311190366744995, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7287002801895142, + "num_tokens": 589414927.0, + "step": 22784 + }, + { + "epoch": 2.5021963540522734, + "grad_norm": 2.715862274169922, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7415180206298828, + "num_tokens": 589436169.0, + "step": 22785 + }, + { + "epoch": 2.5023061717548867, + "grad_norm": 2.1784677505493164, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7264625430107117, + "num_tokens": 589464166.0, + "step": 22786 + }, + { + "epoch": 2.5024159894575004, + "grad_norm": 2.156071186065674, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7137802839279175, + "num_tokens": 589495499.0, + "step": 22787 + }, + { + "epoch": 2.502525807160114, + "grad_norm": 2.2463951110839844, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7252874374389648, + "num_tokens": 589522298.0, + "step": 22788 + }, + { + "epoch": 2.502635624862728, + "grad_norm": 2.787553310394287, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7421764135360718, + "num_tokens": 589542211.0, + "step": 22789 + }, + { + "epoch": 2.5027454425653417, + "grad_norm": 2.7894632816314697, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7147627472877502, + "num_tokens": 589563880.0, + "step": 22790 + }, + { + "epoch": 2.502855260267955, + "grad_norm": 2.279265880584717, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7726107239723206, + "num_tokens": 589590574.0, + "step": 22791 + }, + { + "epoch": 2.5029650779705688, + "grad_norm": 2.2644710540771484, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7383426427841187, + "num_tokens": 589619617.0, + "step": 22792 + }, + { + "epoch": 2.5030748956731825, + "grad_norm": 2.7187254428863525, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7223761081695557, + "num_tokens": 589640033.0, + "step": 22793 + }, + { + "epoch": 2.5031847133757963, + "grad_norm": 2.288715362548828, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7454016208648682, + "num_tokens": 589665609.0, + "step": 22794 + }, + { + "epoch": 2.50329453107841, + "grad_norm": 2.565211057662964, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7405352592468262, + "num_tokens": 589687734.0, + "step": 22795 + }, + { + "epoch": 2.5034043487810234, + "grad_norm": 2.622519016265869, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.759981095790863, + "num_tokens": 589707580.0, + "step": 22796 + }, + { + "epoch": 2.503514166483637, + "grad_norm": 2.2580442428588867, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7647172212600708, + "num_tokens": 589734307.0, + "step": 22797 + }, + { + "epoch": 2.503623984186251, + "grad_norm": 2.609081745147705, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7383469343185425, + "num_tokens": 589757196.0, + "step": 22798 + }, + { + "epoch": 2.5037338018888646, + "grad_norm": 2.398268938064575, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7434341311454773, + "num_tokens": 589780928.0, + "step": 22799 + }, + { + "epoch": 2.5038436195914784, + "grad_norm": 2.2284457683563232, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7382702231407166, + "num_tokens": 589810612.0, + "step": 22800 + }, + { + "epoch": 2.5039534372940917, + "grad_norm": 2.7025022506713867, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7290290594100952, + "num_tokens": 589831931.0, + "step": 22801 + }, + { + "epoch": 2.5040632549967055, + "grad_norm": 2.1838459968566895, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7247515916824341, + "num_tokens": 589861413.0, + "step": 22802 + }, + { + "epoch": 2.504173072699319, + "grad_norm": 2.1717641353607178, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7255163192749023, + "num_tokens": 589891007.0, + "step": 22803 + }, + { + "epoch": 2.5042828904019325, + "grad_norm": 2.2586307525634766, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7251232266426086, + "num_tokens": 589918620.0, + "step": 22804 + }, + { + "epoch": 2.5043927081045467, + "grad_norm": 2.3151934146881104, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7322015762329102, + "num_tokens": 589946586.0, + "step": 22805 + }, + { + "epoch": 2.50450252580716, + "grad_norm": 2.251756191253662, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7385089993476868, + "num_tokens": 589974246.0, + "step": 22806 + }, + { + "epoch": 2.504612343509774, + "grad_norm": 2.2399375438690186, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7260575294494629, + "num_tokens": 590002770.0, + "step": 22807 + }, + { + "epoch": 2.5047221612123876, + "grad_norm": 2.4850921630859375, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7286349534988403, + "num_tokens": 590027131.0, + "step": 22808 + }, + { + "epoch": 2.504831978915001, + "grad_norm": 2.8427398204803467, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7361860275268555, + "num_tokens": 590046283.0, + "step": 22809 + }, + { + "epoch": 2.5049417966176146, + "grad_norm": 2.0637550354003906, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.70688396692276, + "num_tokens": 590076302.0, + "step": 22810 + }, + { + "epoch": 2.5050516143202284, + "grad_norm": 2.3388607501983643, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7160520553588867, + "num_tokens": 590102808.0, + "step": 22811 + }, + { + "epoch": 2.505161432022842, + "grad_norm": 2.3389205932617188, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7272512912750244, + "num_tokens": 590128825.0, + "step": 22812 + }, + { + "epoch": 2.505271249725456, + "grad_norm": 2.3053102493286133, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.745807409286499, + "num_tokens": 590153176.0, + "step": 22813 + }, + { + "epoch": 2.505381067428069, + "grad_norm": 2.3467702865600586, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7317690849304199, + "num_tokens": 590179251.0, + "step": 22814 + }, + { + "epoch": 2.505490885130683, + "grad_norm": 2.561307430267334, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7393193244934082, + "num_tokens": 590199334.0, + "step": 22815 + }, + { + "epoch": 2.5056007028332967, + "grad_norm": 2.082557201385498, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7302142381668091, + "num_tokens": 590230521.0, + "step": 22816 + }, + { + "epoch": 2.5057105205359105, + "grad_norm": 2.394287586212158, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7189371585845947, + "num_tokens": 590256396.0, + "step": 22817 + }, + { + "epoch": 2.5058203382385242, + "grad_norm": 2.4677484035491943, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7323488593101501, + "num_tokens": 590280641.0, + "step": 22818 + }, + { + "epoch": 2.5059301559411375, + "grad_norm": 2.35744047164917, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7183114290237427, + "num_tokens": 590307933.0, + "step": 22819 + }, + { + "epoch": 2.5060399736437513, + "grad_norm": 2.586982250213623, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7214808464050293, + "num_tokens": 590328960.0, + "step": 22820 + }, + { + "epoch": 2.506149791346365, + "grad_norm": 2.407447099685669, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7489727139472961, + "num_tokens": 590350891.0, + "step": 22821 + }, + { + "epoch": 2.506259609048979, + "grad_norm": 2.6588821411132812, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7468863725662231, + "num_tokens": 590371449.0, + "step": 22822 + }, + { + "epoch": 2.5063694267515926, + "grad_norm": 2.5827112197875977, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7659528851509094, + "num_tokens": 590393247.0, + "step": 22823 + }, + { + "epoch": 2.506479244454206, + "grad_norm": 2.54398775100708, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7334439754486084, + "num_tokens": 590416572.0, + "step": 22824 + }, + { + "epoch": 2.5065890621568196, + "grad_norm": 2.4977054595947266, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7432762384414673, + "num_tokens": 590442330.0, + "step": 22825 + }, + { + "epoch": 2.5066988798594334, + "grad_norm": 2.197138786315918, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7107751965522766, + "num_tokens": 590474719.0, + "step": 22826 + }, + { + "epoch": 2.506808697562047, + "grad_norm": 2.092860698699951, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7697679996490479, + "num_tokens": 590502613.0, + "step": 22827 + }, + { + "epoch": 2.506918515264661, + "grad_norm": 2.3179523944854736, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7348640561103821, + "num_tokens": 590528490.0, + "step": 22828 + }, + { + "epoch": 2.5070283329672742, + "grad_norm": 2.3026256561279297, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7094634771347046, + "num_tokens": 590556084.0, + "step": 22829 + }, + { + "epoch": 2.507138150669888, + "grad_norm": 2.6132090091705322, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.729530930519104, + "num_tokens": 590579798.0, + "step": 22830 + }, + { + "epoch": 2.5072479683725017, + "grad_norm": 2.5898654460906982, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7356526851654053, + "num_tokens": 590601789.0, + "step": 22831 + }, + { + "epoch": 2.507357786075115, + "grad_norm": 2.6066107749938965, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7323281764984131, + "num_tokens": 590626173.0, + "step": 22832 + }, + { + "epoch": 2.507467603777729, + "grad_norm": 2.3868093490600586, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7243784070014954, + "num_tokens": 590653382.0, + "step": 22833 + }, + { + "epoch": 2.5075774214803426, + "grad_norm": 2.2888875007629395, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.752881646156311, + "num_tokens": 590681835.0, + "step": 22834 + }, + { + "epoch": 2.5076872391829563, + "grad_norm": 2.5798287391662598, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7420034408569336, + "num_tokens": 590703115.0, + "step": 22835 + }, + { + "epoch": 2.50779705688557, + "grad_norm": 2.285501003265381, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.742605447769165, + "num_tokens": 590729143.0, + "step": 22836 + }, + { + "epoch": 2.5079068745881834, + "grad_norm": 2.279994249343872, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7156850099563599, + "num_tokens": 590757784.0, + "step": 22837 + }, + { + "epoch": 2.508016692290797, + "grad_norm": 2.9253385066986084, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7455891966819763, + "num_tokens": 590777166.0, + "step": 22838 + }, + { + "epoch": 2.508126509993411, + "grad_norm": 2.5157554149627686, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7615111470222473, + "num_tokens": 590801084.0, + "step": 22839 + }, + { + "epoch": 2.5082363276960247, + "grad_norm": 2.600219488143921, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7364028096199036, + "num_tokens": 590823667.0, + "step": 22840 + }, + { + "epoch": 2.5083461453986384, + "grad_norm": 2.2220845222473145, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7081206440925598, + "num_tokens": 590853423.0, + "step": 22841 + }, + { + "epoch": 2.5084559631012517, + "grad_norm": 2.6224617958068848, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7357633113861084, + "num_tokens": 590876373.0, + "step": 22842 + }, + { + "epoch": 2.5085657808038655, + "grad_norm": 2.4204788208007812, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7318068742752075, + "num_tokens": 590900751.0, + "step": 22843 + }, + { + "epoch": 2.5086755985064793, + "grad_norm": 2.2321057319641113, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7232156991958618, + "num_tokens": 590929624.0, + "step": 22844 + }, + { + "epoch": 2.508785416209093, + "grad_norm": 2.236504554748535, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.727843701839447, + "num_tokens": 590957559.0, + "step": 22845 + }, + { + "epoch": 2.5088952339117068, + "grad_norm": 2.7812447547912598, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7386645078659058, + "num_tokens": 590977189.0, + "step": 22846 + }, + { + "epoch": 2.50900505161432, + "grad_norm": 2.1760170459747314, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7379603981971741, + "num_tokens": 591006782.0, + "step": 22847 + }, + { + "epoch": 2.509114869316934, + "grad_norm": 2.398707389831543, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7428022623062134, + "num_tokens": 591032821.0, + "step": 22848 + }, + { + "epoch": 2.5092246870195476, + "grad_norm": 2.3039774894714355, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7407253980636597, + "num_tokens": 591063078.0, + "step": 22849 + }, + { + "epoch": 2.5093345047221614, + "grad_norm": 2.2350845336914062, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7272542119026184, + "num_tokens": 591093057.0, + "step": 22850 + }, + { + "epoch": 2.509444322424775, + "grad_norm": 2.353633403778076, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7493629455566406, + "num_tokens": 591117707.0, + "step": 22851 + }, + { + "epoch": 2.5095541401273884, + "grad_norm": 2.7324531078338623, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7398762106895447, + "num_tokens": 591139261.0, + "step": 22852 + }, + { + "epoch": 2.509663957830002, + "grad_norm": 2.5301222801208496, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7359347343444824, + "num_tokens": 591163001.0, + "step": 22853 + }, + { + "epoch": 2.509773775532616, + "grad_norm": 2.198051691055298, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7385135889053345, + "num_tokens": 591191088.0, + "step": 22854 + }, + { + "epoch": 2.5098835932352292, + "grad_norm": 2.370445728302002, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7079801559448242, + "num_tokens": 591217394.0, + "step": 22855 + }, + { + "epoch": 2.5099934109378434, + "grad_norm": 2.6310324668884277, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7454224824905396, + "num_tokens": 591238614.0, + "step": 22856 + }, + { + "epoch": 2.5101032286404568, + "grad_norm": 2.5172417163848877, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7268502712249756, + "num_tokens": 591264604.0, + "step": 22857 + }, + { + "epoch": 2.5102130463430705, + "grad_norm": 2.5625360012054443, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7528684735298157, + "num_tokens": 591287228.0, + "step": 22858 + }, + { + "epoch": 2.5103228640456843, + "grad_norm": 2.328669548034668, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7241753339767456, + "num_tokens": 591312753.0, + "step": 22859 + }, + { + "epoch": 2.5104326817482976, + "grad_norm": 2.3937370777130127, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7300925254821777, + "num_tokens": 591338562.0, + "step": 22860 + }, + { + "epoch": 2.5105424994509113, + "grad_norm": 2.2322096824645996, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7045320272445679, + "num_tokens": 591366615.0, + "step": 22861 + }, + { + "epoch": 2.510652317153525, + "grad_norm": 2.496316432952881, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7249647378921509, + "num_tokens": 591390433.0, + "step": 22862 + }, + { + "epoch": 2.510762134856139, + "grad_norm": 2.4761881828308105, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.720512866973877, + "num_tokens": 591412696.0, + "step": 22863 + }, + { + "epoch": 2.5108719525587526, + "grad_norm": 2.2631077766418457, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7364703416824341, + "num_tokens": 591439559.0, + "step": 22864 + }, + { + "epoch": 2.510981770261366, + "grad_norm": 2.6022608280181885, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7657725811004639, + "num_tokens": 591459784.0, + "step": 22865 + }, + { + "epoch": 2.5110915879639797, + "grad_norm": 2.4366464614868164, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7363260984420776, + "num_tokens": 591483730.0, + "step": 22866 + }, + { + "epoch": 2.5112014056665934, + "grad_norm": 2.450672149658203, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7353149652481079, + "num_tokens": 591509318.0, + "step": 22867 + }, + { + "epoch": 2.511311223369207, + "grad_norm": 2.5330471992492676, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7353580594062805, + "num_tokens": 591532774.0, + "step": 22868 + }, + { + "epoch": 2.511421041071821, + "grad_norm": 2.6591134071350098, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7260856628417969, + "num_tokens": 591553811.0, + "step": 22869 + }, + { + "epoch": 2.5115308587744343, + "grad_norm": 2.0186610221862793, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7153512835502625, + "num_tokens": 591587246.0, + "step": 22870 + }, + { + "epoch": 2.511640676477048, + "grad_norm": 2.2875924110412598, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7060835361480713, + "num_tokens": 591616083.0, + "step": 22871 + }, + { + "epoch": 2.511750494179662, + "grad_norm": 2.222935438156128, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7573314309120178, + "num_tokens": 591642943.0, + "step": 22872 + }, + { + "epoch": 2.5118603118822755, + "grad_norm": 2.531107187271118, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7513257265090942, + "num_tokens": 591668235.0, + "step": 22873 + }, + { + "epoch": 2.5119701295848893, + "grad_norm": 2.2095608711242676, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7273876667022705, + "num_tokens": 591696854.0, + "step": 22874 + }, + { + "epoch": 2.5120799472875026, + "grad_norm": 2.3880748748779297, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7228585481643677, + "num_tokens": 591722220.0, + "step": 22875 + }, + { + "epoch": 2.5121897649901164, + "grad_norm": 2.335186719894409, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7169884443283081, + "num_tokens": 591749674.0, + "step": 22876 + }, + { + "epoch": 2.51229958269273, + "grad_norm": 2.340747117996216, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7556362152099609, + "num_tokens": 591777655.0, + "step": 22877 + }, + { + "epoch": 2.512409400395344, + "grad_norm": 2.0870895385742188, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7278212308883667, + "num_tokens": 591810654.0, + "step": 22878 + }, + { + "epoch": 2.5125192180979576, + "grad_norm": 2.261709451675415, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7323259115219116, + "num_tokens": 591838576.0, + "step": 22879 + }, + { + "epoch": 2.512629035800571, + "grad_norm": 2.393803834915161, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7592202425003052, + "num_tokens": 591860586.0, + "step": 22880 + }, + { + "epoch": 2.5127388535031847, + "grad_norm": 2.476506233215332, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7393980622291565, + "num_tokens": 591885342.0, + "step": 22881 + }, + { + "epoch": 2.5128486712057985, + "grad_norm": 2.6037442684173584, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7440115213394165, + "num_tokens": 591907658.0, + "step": 22882 + }, + { + "epoch": 2.512958488908412, + "grad_norm": 2.386810541152954, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.730902910232544, + "num_tokens": 591933944.0, + "step": 22883 + }, + { + "epoch": 2.5130683066110255, + "grad_norm": 2.1465394496917725, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7223718166351318, + "num_tokens": 591965979.0, + "step": 22884 + }, + { + "epoch": 2.5131781243136393, + "grad_norm": 2.799635648727417, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7519416809082031, + "num_tokens": 591985122.0, + "step": 22885 + }, + { + "epoch": 2.513287942016253, + "grad_norm": 2.551769971847534, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7311867475509644, + "num_tokens": 592009719.0, + "step": 22886 + }, + { + "epoch": 2.513397759718867, + "grad_norm": 2.330583333969116, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7322665452957153, + "num_tokens": 592036278.0, + "step": 22887 + }, + { + "epoch": 2.51350757742148, + "grad_norm": 2.2563772201538086, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7236161828041077, + "num_tokens": 592062284.0, + "step": 22888 + }, + { + "epoch": 2.513617395124094, + "grad_norm": 2.01202654838562, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7175930738449097, + "num_tokens": 592098411.0, + "step": 22889 + }, + { + "epoch": 2.5137272128267076, + "grad_norm": 2.7162435054779053, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7427056431770325, + "num_tokens": 592124820.0, + "step": 22890 + }, + { + "epoch": 2.5138370305293214, + "grad_norm": 2.7088358402252197, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7441244125366211, + "num_tokens": 592143372.0, + "step": 22891 + }, + { + "epoch": 2.513946848231935, + "grad_norm": 2.266115427017212, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7264931201934814, + "num_tokens": 592171269.0, + "step": 22892 + }, + { + "epoch": 2.5140566659345485, + "grad_norm": 2.4151229858398438, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7266362905502319, + "num_tokens": 592196793.0, + "step": 22893 + }, + { + "epoch": 2.514166483637162, + "grad_norm": 2.343463659286499, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.740253746509552, + "num_tokens": 592221733.0, + "step": 22894 + }, + { + "epoch": 2.514276301339776, + "grad_norm": 2.374783992767334, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7207426428794861, + "num_tokens": 592247370.0, + "step": 22895 + }, + { + "epoch": 2.5143861190423897, + "grad_norm": 2.188772201538086, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7052438855171204, + "num_tokens": 592276498.0, + "step": 22896 + }, + { + "epoch": 2.5144959367450035, + "grad_norm": 2.368128538131714, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7339943647384644, + "num_tokens": 592300544.0, + "step": 22897 + }, + { + "epoch": 2.514605754447617, + "grad_norm": 2.2036349773406982, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7121027708053589, + "num_tokens": 592330526.0, + "step": 22898 + }, + { + "epoch": 2.5147155721502306, + "grad_norm": 2.264200210571289, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7401242256164551, + "num_tokens": 592358300.0, + "step": 22899 + }, + { + "epoch": 2.5148253898528443, + "grad_norm": 2.1992316246032715, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7357556819915771, + "num_tokens": 592385102.0, + "step": 22900 + }, + { + "epoch": 2.514935207555458, + "grad_norm": 2.2103753089904785, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7403695583343506, + "num_tokens": 592415240.0, + "step": 22901 + }, + { + "epoch": 2.515045025258072, + "grad_norm": 2.8955154418945312, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7517046928405762, + "num_tokens": 592432834.0, + "step": 22902 + }, + { + "epoch": 2.515154842960685, + "grad_norm": 2.405959367752075, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7494565844535828, + "num_tokens": 592457682.0, + "step": 22903 + }, + { + "epoch": 2.515264660663299, + "grad_norm": 2.385563611984253, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7592959403991699, + "num_tokens": 592481379.0, + "step": 22904 + }, + { + "epoch": 2.5153744783659127, + "grad_norm": 2.146595001220703, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7534657120704651, + "num_tokens": 592511498.0, + "step": 22905 + }, + { + "epoch": 2.515484296068526, + "grad_norm": 2.522819995880127, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7300220727920532, + "num_tokens": 592535201.0, + "step": 22906 + }, + { + "epoch": 2.51559411377114, + "grad_norm": 2.0008161067962646, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7266594171524048, + "num_tokens": 592569210.0, + "step": 22907 + }, + { + "epoch": 2.5157039314737535, + "grad_norm": 1.8834487199783325, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7145596742630005, + "num_tokens": 592606097.0, + "step": 22908 + }, + { + "epoch": 2.5158137491763672, + "grad_norm": 2.8935277462005615, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7341779470443726, + "num_tokens": 592625761.0, + "step": 22909 + }, + { + "epoch": 2.515923566878981, + "grad_norm": 2.3981878757476807, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7521783113479614, + "num_tokens": 592649864.0, + "step": 22910 + }, + { + "epoch": 2.5160333845815943, + "grad_norm": 2.42346453666687, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7274971604347229, + "num_tokens": 592674280.0, + "step": 22911 + }, + { + "epoch": 2.516143202284208, + "grad_norm": 2.5785059928894043, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7155919671058655, + "num_tokens": 592696943.0, + "step": 22912 + }, + { + "epoch": 2.516253019986822, + "grad_norm": 2.384305953979492, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7363276481628418, + "num_tokens": 592722978.0, + "step": 22913 + }, + { + "epoch": 2.5163628376894356, + "grad_norm": 2.4898369312286377, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7505142688751221, + "num_tokens": 592746688.0, + "step": 22914 + }, + { + "epoch": 2.5164726553920493, + "grad_norm": 2.207624912261963, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7313206195831299, + "num_tokens": 592776387.0, + "step": 22915 + }, + { + "epoch": 2.5165824730946627, + "grad_norm": 2.436429500579834, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7210925817489624, + "num_tokens": 592800075.0, + "step": 22916 + }, + { + "epoch": 2.5166922907972764, + "grad_norm": 2.1883718967437744, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7504381537437439, + "num_tokens": 592826613.0, + "step": 22917 + }, + { + "epoch": 2.51680210849989, + "grad_norm": 2.3433473110198975, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7381608486175537, + "num_tokens": 592851609.0, + "step": 22918 + }, + { + "epoch": 2.516911926202504, + "grad_norm": 2.467747449874878, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7277911901473999, + "num_tokens": 592879143.0, + "step": 22919 + }, + { + "epoch": 2.5170217439051177, + "grad_norm": 2.303309679031372, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7358793020248413, + "num_tokens": 592905359.0, + "step": 22920 + }, + { + "epoch": 2.517131561607731, + "grad_norm": 2.197345018386841, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7201038599014282, + "num_tokens": 592934634.0, + "step": 22921 + }, + { + "epoch": 2.5172413793103448, + "grad_norm": 2.3631327152252197, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7438431978225708, + "num_tokens": 592961039.0, + "step": 22922 + }, + { + "epoch": 2.5173511970129585, + "grad_norm": 2.345327138900757, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7208153009414673, + "num_tokens": 592988378.0, + "step": 22923 + }, + { + "epoch": 2.5174610147155723, + "grad_norm": 2.2583329677581787, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7201515436172485, + "num_tokens": 593017157.0, + "step": 22924 + }, + { + "epoch": 2.517570832418186, + "grad_norm": 2.1890032291412354, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7053346037864685, + "num_tokens": 593052808.0, + "step": 22925 + }, + { + "epoch": 2.5176806501207993, + "grad_norm": 2.4794692993164062, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7537909746170044, + "num_tokens": 593077480.0, + "step": 22926 + }, + { + "epoch": 2.517790467823413, + "grad_norm": 2.033755302429199, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7370691299438477, + "num_tokens": 593111996.0, + "step": 22927 + }, + { + "epoch": 2.517900285526027, + "grad_norm": 2.2796289920806885, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7085901498794556, + "num_tokens": 593139114.0, + "step": 22928 + }, + { + "epoch": 2.5180101032286406, + "grad_norm": 2.256096601486206, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7299966812133789, + "num_tokens": 593167042.0, + "step": 22929 + }, + { + "epoch": 2.5181199209312544, + "grad_norm": 2.392051935195923, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7236731052398682, + "num_tokens": 593191849.0, + "step": 22930 + }, + { + "epoch": 2.5182297386338677, + "grad_norm": 2.4066107273101807, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7357152700424194, + "num_tokens": 593218714.0, + "step": 22931 + }, + { + "epoch": 2.5183395563364814, + "grad_norm": 2.240915060043335, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7276563048362732, + "num_tokens": 593247208.0, + "step": 22932 + }, + { + "epoch": 2.518449374039095, + "grad_norm": 2.4442217350006104, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7171441316604614, + "num_tokens": 593272635.0, + "step": 22933 + }, + { + "epoch": 2.5185591917417085, + "grad_norm": 2.459956169128418, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7279865741729736, + "num_tokens": 593297594.0, + "step": 22934 + }, + { + "epoch": 2.5186690094443223, + "grad_norm": 2.225431203842163, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7252141833305359, + "num_tokens": 593329123.0, + "step": 22935 + }, + { + "epoch": 2.518778827146936, + "grad_norm": 2.276808738708496, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7286772131919861, + "num_tokens": 593358256.0, + "step": 22936 + }, + { + "epoch": 2.5188886448495498, + "grad_norm": 2.6710898876190186, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7572234869003296, + "num_tokens": 593379323.0, + "step": 22937 + }, + { + "epoch": 2.5189984625521635, + "grad_norm": 2.092099666595459, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7406620979309082, + "num_tokens": 593412384.0, + "step": 22938 + }, + { + "epoch": 2.519108280254777, + "grad_norm": 2.381270408630371, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7244371175765991, + "num_tokens": 593439200.0, + "step": 22939 + }, + { + "epoch": 2.5192180979573906, + "grad_norm": 2.247575521469116, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7117247581481934, + "num_tokens": 593467494.0, + "step": 22940 + }, + { + "epoch": 2.5193279156600044, + "grad_norm": 2.0755491256713867, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7317754030227661, + "num_tokens": 593497676.0, + "step": 22941 + }, + { + "epoch": 2.519437733362618, + "grad_norm": 2.638634443283081, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7527631521224976, + "num_tokens": 593519999.0, + "step": 22942 + }, + { + "epoch": 2.519547551065232, + "grad_norm": 2.338730812072754, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7107868194580078, + "num_tokens": 593546497.0, + "step": 22943 + }, + { + "epoch": 2.519657368767845, + "grad_norm": 2.2617201805114746, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7343836426734924, + "num_tokens": 593571869.0, + "step": 22944 + }, + { + "epoch": 2.519767186470459, + "grad_norm": 2.4584710597991943, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7355631589889526, + "num_tokens": 593595902.0, + "step": 22945 + }, + { + "epoch": 2.5198770041730727, + "grad_norm": 2.819704294204712, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7542840242385864, + "num_tokens": 593616021.0, + "step": 22946 + }, + { + "epoch": 2.5199868218756865, + "grad_norm": 2.373852014541626, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7177021503448486, + "num_tokens": 593640677.0, + "step": 22947 + }, + { + "epoch": 2.5200966395783, + "grad_norm": 1.9938610792160034, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7123115062713623, + "num_tokens": 593675273.0, + "step": 22948 + }, + { + "epoch": 2.5202064572809135, + "grad_norm": 2.6036055088043213, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7249331474304199, + "num_tokens": 593698371.0, + "step": 22949 + }, + { + "epoch": 2.5203162749835273, + "grad_norm": 2.6176722049713135, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7124074697494507, + "num_tokens": 593719487.0, + "step": 22950 + }, + { + "epoch": 2.520426092686141, + "grad_norm": 2.215498685836792, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7327101826667786, + "num_tokens": 593748209.0, + "step": 22951 + }, + { + "epoch": 2.520535910388755, + "grad_norm": 2.093137741088867, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7217488884925842, + "num_tokens": 593779992.0, + "step": 22952 + }, + { + "epoch": 2.5206457280913686, + "grad_norm": 2.3166110515594482, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7383170127868652, + "num_tokens": 593805162.0, + "step": 22953 + }, + { + "epoch": 2.520755545793982, + "grad_norm": 2.5011274814605713, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7510138750076294, + "num_tokens": 593826809.0, + "step": 22954 + }, + { + "epoch": 2.5208653634965956, + "grad_norm": 2.3556249141693115, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.731410801410675, + "num_tokens": 593854502.0, + "step": 22955 + }, + { + "epoch": 2.5209751811992094, + "grad_norm": 2.3293724060058594, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7100058794021606, + "num_tokens": 593882364.0, + "step": 22956 + }, + { + "epoch": 2.521084998901823, + "grad_norm": 2.0805976390838623, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7237555980682373, + "num_tokens": 593914002.0, + "step": 22957 + }, + { + "epoch": 2.521194816604437, + "grad_norm": 2.2372970581054688, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.749798059463501, + "num_tokens": 593942545.0, + "step": 22958 + }, + { + "epoch": 2.52130463430705, + "grad_norm": 2.2758636474609375, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7297290563583374, + "num_tokens": 593969663.0, + "step": 22959 + }, + { + "epoch": 2.521414452009664, + "grad_norm": 2.465031623840332, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.719789445400238, + "num_tokens": 593993735.0, + "step": 22960 + }, + { + "epoch": 2.5215242697122777, + "grad_norm": 2.1783735752105713, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7311906814575195, + "num_tokens": 594022149.0, + "step": 22961 + }, + { + "epoch": 2.521634087414891, + "grad_norm": 2.3316543102264404, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7376508712768555, + "num_tokens": 594048254.0, + "step": 22962 + }, + { + "epoch": 2.521743905117505, + "grad_norm": 2.6544883251190186, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7437810897827148, + "num_tokens": 594069964.0, + "step": 22963 + }, + { + "epoch": 2.5218537228201185, + "grad_norm": 2.342191457748413, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7453261017799377, + "num_tokens": 594094639.0, + "step": 22964 + }, + { + "epoch": 2.5219635405227323, + "grad_norm": 2.695162534713745, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7498717308044434, + "num_tokens": 594115174.0, + "step": 22965 + }, + { + "epoch": 2.522073358225346, + "grad_norm": 2.402040481567383, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7360661029815674, + "num_tokens": 594140095.0, + "step": 22966 + }, + { + "epoch": 2.5221831759279594, + "grad_norm": 2.5958285331726074, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7261761426925659, + "num_tokens": 594162787.0, + "step": 22967 + }, + { + "epoch": 2.522292993630573, + "grad_norm": 2.5408542156219482, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7321075797080994, + "num_tokens": 594188094.0, + "step": 22968 + }, + { + "epoch": 2.522402811333187, + "grad_norm": 2.3628435134887695, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7314009666442871, + "num_tokens": 594213783.0, + "step": 22969 + }, + { + "epoch": 2.5225126290358006, + "grad_norm": 2.254140853881836, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7215890884399414, + "num_tokens": 594240306.0, + "step": 22970 + }, + { + "epoch": 2.5226224467384144, + "grad_norm": 2.1284875869750977, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7305265665054321, + "num_tokens": 594269709.0, + "step": 22971 + }, + { + "epoch": 2.5227322644410277, + "grad_norm": 2.6956801414489746, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7530646324157715, + "num_tokens": 594289564.0, + "step": 22972 + }, + { + "epoch": 2.5228420821436415, + "grad_norm": 2.281747579574585, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7292790412902832, + "num_tokens": 594317962.0, + "step": 22973 + }, + { + "epoch": 2.5229518998462552, + "grad_norm": 2.531548023223877, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.748176634311676, + "num_tokens": 594340334.0, + "step": 22974 + }, + { + "epoch": 2.523061717548869, + "grad_norm": 2.5278100967407227, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.720093846321106, + "num_tokens": 594364305.0, + "step": 22975 + }, + { + "epoch": 2.5231715352514827, + "grad_norm": 2.6780941486358643, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7565100193023682, + "num_tokens": 594383972.0, + "step": 22976 + }, + { + "epoch": 2.523281352954096, + "grad_norm": 2.1155383586883545, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7368888258934021, + "num_tokens": 594413752.0, + "step": 22977 + }, + { + "epoch": 2.52339117065671, + "grad_norm": 2.3048348426818848, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7153519988059998, + "num_tokens": 594442079.0, + "step": 22978 + }, + { + "epoch": 2.5235009883593236, + "grad_norm": 2.58978271484375, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7081004977226257, + "num_tokens": 594465901.0, + "step": 22979 + }, + { + "epoch": 2.5236108060619373, + "grad_norm": 2.201063632965088, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7192155122756958, + "num_tokens": 594496278.0, + "step": 22980 + }, + { + "epoch": 2.523720623764551, + "grad_norm": 2.3818953037261963, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7384827733039856, + "num_tokens": 594520948.0, + "step": 22981 + }, + { + "epoch": 2.5238304414671644, + "grad_norm": 2.241732597351074, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7417459487915039, + "num_tokens": 594548810.0, + "step": 22982 + }, + { + "epoch": 2.523940259169778, + "grad_norm": 2.4572765827178955, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7061662673950195, + "num_tokens": 594574861.0, + "step": 22983 + }, + { + "epoch": 2.524050076872392, + "grad_norm": 2.240396738052368, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7142496109008789, + "num_tokens": 594604307.0, + "step": 22984 + }, + { + "epoch": 2.5241598945750052, + "grad_norm": 2.3737854957580566, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7162634134292603, + "num_tokens": 594629850.0, + "step": 22985 + }, + { + "epoch": 2.5242697122776194, + "grad_norm": 2.9220755100250244, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7398018836975098, + "num_tokens": 594649519.0, + "step": 22986 + }, + { + "epoch": 2.5243795299802327, + "grad_norm": 2.7574806213378906, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7350704669952393, + "num_tokens": 594669664.0, + "step": 22987 + }, + { + "epoch": 2.5244893476828465, + "grad_norm": 2.234323740005493, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.721003532409668, + "num_tokens": 594698318.0, + "step": 22988 + }, + { + "epoch": 2.5245991653854603, + "grad_norm": 2.469538927078247, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7238390445709229, + "num_tokens": 594724229.0, + "step": 22989 + }, + { + "epoch": 2.5247089830880736, + "grad_norm": 2.453263998031616, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7267037630081177, + "num_tokens": 594750082.0, + "step": 22990 + }, + { + "epoch": 2.5248188007906873, + "grad_norm": 2.8471157550811768, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7455602288246155, + "num_tokens": 594769454.0, + "step": 22991 + }, + { + "epoch": 2.524928618493301, + "grad_norm": 2.631028652191162, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7467911243438721, + "num_tokens": 594790967.0, + "step": 22992 + }, + { + "epoch": 2.525038436195915, + "grad_norm": 2.6913769245147705, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7190659046173096, + "num_tokens": 594813218.0, + "step": 22993 + }, + { + "epoch": 2.5251482538985286, + "grad_norm": 2.049806833267212, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7199263572692871, + "num_tokens": 594844603.0, + "step": 22994 + }, + { + "epoch": 2.525258071601142, + "grad_norm": 2.2997982501983643, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7268460392951965, + "num_tokens": 594871719.0, + "step": 22995 + }, + { + "epoch": 2.5253678893037557, + "grad_norm": 2.4725992679595947, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7265197038650513, + "num_tokens": 594895488.0, + "step": 22996 + }, + { + "epoch": 2.5254777070063694, + "grad_norm": 2.5111818313598633, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7392077445983887, + "num_tokens": 594919364.0, + "step": 22997 + }, + { + "epoch": 2.525587524708983, + "grad_norm": 2.4699747562408447, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7315713167190552, + "num_tokens": 594946118.0, + "step": 22998 + }, + { + "epoch": 2.525697342411597, + "grad_norm": 2.451894998550415, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7197632193565369, + "num_tokens": 594969450.0, + "step": 22999 + }, + { + "epoch": 2.5258071601142102, + "grad_norm": 2.8044116497039795, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7381687760353088, + "num_tokens": 594988795.0, + "step": 23000 + }, + { + "epoch": 2.525916977816824, + "grad_norm": 2.3283016681671143, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7470909953117371, + "num_tokens": 595016239.0, + "step": 23001 + }, + { + "epoch": 2.5260267955194378, + "grad_norm": 2.3821327686309814, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7466254830360413, + "num_tokens": 595041564.0, + "step": 23002 + }, + { + "epoch": 2.5261366132220515, + "grad_norm": 2.4609375, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7417696118354797, + "num_tokens": 595064933.0, + "step": 23003 + }, + { + "epoch": 2.5262464309246653, + "grad_norm": 2.237879753112793, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7201840281486511, + "num_tokens": 595094820.0, + "step": 23004 + }, + { + "epoch": 2.5263562486272786, + "grad_norm": 2.453677177429199, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7550086975097656, + "num_tokens": 595120478.0, + "step": 23005 + }, + { + "epoch": 2.5264660663298923, + "grad_norm": 2.375417470932007, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7450094223022461, + "num_tokens": 595144689.0, + "step": 23006 + }, + { + "epoch": 2.526575884032506, + "grad_norm": 2.228684425354004, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7312862277030945, + "num_tokens": 595171859.0, + "step": 23007 + }, + { + "epoch": 2.52668570173512, + "grad_norm": 2.41398024559021, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.738982081413269, + "num_tokens": 595196206.0, + "step": 23008 + }, + { + "epoch": 2.5267955194377336, + "grad_norm": 2.506662130355835, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7158111333847046, + "num_tokens": 595219237.0, + "step": 23009 + }, + { + "epoch": 2.526905337140347, + "grad_norm": 2.2381386756896973, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6968182921409607, + "num_tokens": 595249726.0, + "step": 23010 + }, + { + "epoch": 2.5270151548429607, + "grad_norm": 2.3827664852142334, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7324001789093018, + "num_tokens": 595274027.0, + "step": 23011 + }, + { + "epoch": 2.5271249725455744, + "grad_norm": 2.5481274127960205, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7187230587005615, + "num_tokens": 595296661.0, + "step": 23012 + }, + { + "epoch": 2.5272347902481878, + "grad_norm": 2.5405526161193848, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7416641712188721, + "num_tokens": 595319437.0, + "step": 23013 + }, + { + "epoch": 2.5273446079508015, + "grad_norm": 2.0879220962524414, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7101669311523438, + "num_tokens": 595351096.0, + "step": 23014 + }, + { + "epoch": 2.5274544256534153, + "grad_norm": 2.4665939807891846, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7284851670265198, + "num_tokens": 595376735.0, + "step": 23015 + }, + { + "epoch": 2.527564243356029, + "grad_norm": 2.3085153102874756, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7257360219955444, + "num_tokens": 595404045.0, + "step": 23016 + }, + { + "epoch": 2.527674061058643, + "grad_norm": 2.4278132915496826, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7259268760681152, + "num_tokens": 595429491.0, + "step": 23017 + }, + { + "epoch": 2.527783878761256, + "grad_norm": 2.3278088569641113, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7123552560806274, + "num_tokens": 595457524.0, + "step": 23018 + }, + { + "epoch": 2.52789369646387, + "grad_norm": 2.013213634490967, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.73503178358078, + "num_tokens": 595490527.0, + "step": 23019 + }, + { + "epoch": 2.5280035141664836, + "grad_norm": 2.475292205810547, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7267063856124878, + "num_tokens": 595515326.0, + "step": 23020 + }, + { + "epoch": 2.5281133318690974, + "grad_norm": 2.06343936920166, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7199232578277588, + "num_tokens": 595547889.0, + "step": 23021 + }, + { + "epoch": 2.528223149571711, + "grad_norm": 2.2294492721557617, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7194647192955017, + "num_tokens": 595576016.0, + "step": 23022 + }, + { + "epoch": 2.5283329672743244, + "grad_norm": 2.581294536590576, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.720077395439148, + "num_tokens": 595599972.0, + "step": 23023 + }, + { + "epoch": 2.528442784976938, + "grad_norm": 2.2021312713623047, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.739440381526947, + "num_tokens": 595629268.0, + "step": 23024 + }, + { + "epoch": 2.528552602679552, + "grad_norm": 2.470203399658203, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.733725368976593, + "num_tokens": 595653835.0, + "step": 23025 + }, + { + "epoch": 2.5286624203821657, + "grad_norm": 2.3934166431427, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7262675762176514, + "num_tokens": 595679525.0, + "step": 23026 + }, + { + "epoch": 2.5287722380847795, + "grad_norm": 2.5235414505004883, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7179594039916992, + "num_tokens": 595704461.0, + "step": 23027 + }, + { + "epoch": 2.528882055787393, + "grad_norm": 2.5868818759918213, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7347924709320068, + "num_tokens": 595726467.0, + "step": 23028 + }, + { + "epoch": 2.5289918734900065, + "grad_norm": 2.5231432914733887, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7424083352088928, + "num_tokens": 595749917.0, + "step": 23029 + }, + { + "epoch": 2.5291016911926203, + "grad_norm": 2.1713569164276123, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7160981297492981, + "num_tokens": 595778799.0, + "step": 23030 + }, + { + "epoch": 2.529211508895234, + "grad_norm": 2.646994113922119, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7331579327583313, + "num_tokens": 595798731.0, + "step": 23031 + }, + { + "epoch": 2.529321326597848, + "grad_norm": 2.4916398525238037, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7198140621185303, + "num_tokens": 595824073.0, + "step": 23032 + }, + { + "epoch": 2.529431144300461, + "grad_norm": 2.2995247840881348, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7238636016845703, + "num_tokens": 595849564.0, + "step": 23033 + }, + { + "epoch": 2.529540962003075, + "grad_norm": 2.298727512359619, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7286319732666016, + "num_tokens": 595878108.0, + "step": 23034 + }, + { + "epoch": 2.5296507797056886, + "grad_norm": 2.1667373180389404, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7195471525192261, + "num_tokens": 595906890.0, + "step": 23035 + }, + { + "epoch": 2.529760597408302, + "grad_norm": 2.7872023582458496, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7372645139694214, + "num_tokens": 595926522.0, + "step": 23036 + }, + { + "epoch": 2.529870415110916, + "grad_norm": 2.404287338256836, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7253974676132202, + "num_tokens": 595949645.0, + "step": 23037 + }, + { + "epoch": 2.5299802328135295, + "grad_norm": 2.2314884662628174, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7239968776702881, + "num_tokens": 595977614.0, + "step": 23038 + }, + { + "epoch": 2.530090050516143, + "grad_norm": 2.1156771183013916, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7202705144882202, + "num_tokens": 596007884.0, + "step": 23039 + }, + { + "epoch": 2.530199868218757, + "grad_norm": 2.4218509197235107, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7512595057487488, + "num_tokens": 596031739.0, + "step": 23040 + }, + { + "epoch": 2.5303096859213703, + "grad_norm": 2.825911521911621, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7400267720222473, + "num_tokens": 596050106.0, + "step": 23041 + }, + { + "epoch": 2.530419503623984, + "grad_norm": 2.7034037113189697, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7724695801734924, + "num_tokens": 596069132.0, + "step": 23042 + }, + { + "epoch": 2.530529321326598, + "grad_norm": 2.204251766204834, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7370504140853882, + "num_tokens": 596097303.0, + "step": 23043 + }, + { + "epoch": 2.5306391390292116, + "grad_norm": 2.439319610595703, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7490619421005249, + "num_tokens": 596121083.0, + "step": 23044 + }, + { + "epoch": 2.5307489567318253, + "grad_norm": 2.323408842086792, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7225306034088135, + "num_tokens": 596148564.0, + "step": 23045 + }, + { + "epoch": 2.5308587744344386, + "grad_norm": 2.2803220748901367, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7443059682846069, + "num_tokens": 596174298.0, + "step": 23046 + }, + { + "epoch": 2.5309685921370524, + "grad_norm": 2.3941614627838135, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7093138694763184, + "num_tokens": 596198003.0, + "step": 23047 + }, + { + "epoch": 2.531078409839666, + "grad_norm": 2.7080507278442383, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7571045756340027, + "num_tokens": 596220404.0, + "step": 23048 + }, + { + "epoch": 2.53118822754228, + "grad_norm": 2.1494102478027344, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7135285139083862, + "num_tokens": 596253105.0, + "step": 23049 + }, + { + "epoch": 2.5312980452448937, + "grad_norm": 2.4060628414154053, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7304353713989258, + "num_tokens": 596277084.0, + "step": 23050 + }, + { + "epoch": 2.531407862947507, + "grad_norm": 2.592275619506836, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7146056890487671, + "num_tokens": 596300341.0, + "step": 23051 + }, + { + "epoch": 2.5315176806501207, + "grad_norm": 2.571791887283325, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7482162714004517, + "num_tokens": 596322781.0, + "step": 23052 + }, + { + "epoch": 2.5316274983527345, + "grad_norm": 2.523618221282959, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7335814237594604, + "num_tokens": 596345388.0, + "step": 23053 + }, + { + "epoch": 2.5317373160553482, + "grad_norm": 2.2045962810516357, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7258964776992798, + "num_tokens": 596373827.0, + "step": 23054 + }, + { + "epoch": 2.531847133757962, + "grad_norm": 2.446451425552368, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7484789490699768, + "num_tokens": 596396903.0, + "step": 23055 + }, + { + "epoch": 2.5319569514605753, + "grad_norm": 2.3087081909179688, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.716320276260376, + "num_tokens": 596424384.0, + "step": 23056 + }, + { + "epoch": 2.532066769163189, + "grad_norm": 2.4352588653564453, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7299924492835999, + "num_tokens": 596448517.0, + "step": 23057 + }, + { + "epoch": 2.532176586865803, + "grad_norm": 2.203896999359131, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7089042663574219, + "num_tokens": 596476470.0, + "step": 23058 + }, + { + "epoch": 2.5322864045684166, + "grad_norm": 2.2232370376586914, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7357131242752075, + "num_tokens": 596504634.0, + "step": 23059 + }, + { + "epoch": 2.5323962222710303, + "grad_norm": 2.1303036212921143, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7167092561721802, + "num_tokens": 596533937.0, + "step": 23060 + }, + { + "epoch": 2.5325060399736437, + "grad_norm": 2.1061222553253174, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7031960487365723, + "num_tokens": 596566767.0, + "step": 23061 + }, + { + "epoch": 2.5326158576762574, + "grad_norm": 2.403409481048584, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7215642929077148, + "num_tokens": 596592004.0, + "step": 23062 + }, + { + "epoch": 2.532725675378871, + "grad_norm": 2.3739609718322754, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7367578148841858, + "num_tokens": 596617623.0, + "step": 23063 + }, + { + "epoch": 2.5328354930814845, + "grad_norm": 2.1814162731170654, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7339974641799927, + "num_tokens": 596647337.0, + "step": 23064 + }, + { + "epoch": 2.5329453107840982, + "grad_norm": 2.623624086380005, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7357801795005798, + "num_tokens": 596668598.0, + "step": 23065 + }, + { + "epoch": 2.533055128486712, + "grad_norm": 2.156404495239258, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.726282000541687, + "num_tokens": 596699776.0, + "step": 23066 + }, + { + "epoch": 2.5331649461893258, + "grad_norm": 2.407928228378296, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7181169390678406, + "num_tokens": 596725281.0, + "step": 23067 + }, + { + "epoch": 2.5332747638919395, + "grad_norm": 2.700314521789551, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7475884556770325, + "num_tokens": 596747043.0, + "step": 23068 + }, + { + "epoch": 2.533384581594553, + "grad_norm": 2.642781972885132, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7085715532302856, + "num_tokens": 596771535.0, + "step": 23069 + }, + { + "epoch": 2.5334943992971666, + "grad_norm": 2.3324084281921387, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7219581604003906, + "num_tokens": 596798443.0, + "step": 23070 + }, + { + "epoch": 2.5336042169997803, + "grad_norm": 2.688704490661621, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7469222545623779, + "num_tokens": 596819292.0, + "step": 23071 + }, + { + "epoch": 2.533714034702394, + "grad_norm": 2.2729711532592773, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7422688007354736, + "num_tokens": 596848840.0, + "step": 23072 + }, + { + "epoch": 2.533823852405008, + "grad_norm": 2.2618157863616943, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7340788841247559, + "num_tokens": 596876828.0, + "step": 23073 + }, + { + "epoch": 2.533933670107621, + "grad_norm": 2.306198835372925, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.720072865486145, + "num_tokens": 596903668.0, + "step": 23074 + }, + { + "epoch": 2.534043487810235, + "grad_norm": 2.0281593799591064, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7155386805534363, + "num_tokens": 596939752.0, + "step": 23075 + }, + { + "epoch": 2.5341533055128487, + "grad_norm": 2.386874198913574, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.721836268901825, + "num_tokens": 596965149.0, + "step": 23076 + }, + { + "epoch": 2.5342631232154624, + "grad_norm": 2.3765060901641846, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7374171614646912, + "num_tokens": 596988327.0, + "step": 23077 + }, + { + "epoch": 2.534372940918076, + "grad_norm": 2.5332422256469727, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7293299436569214, + "num_tokens": 597013645.0, + "step": 23078 + }, + { + "epoch": 2.5344827586206895, + "grad_norm": 2.480203151702881, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7365115880966187, + "num_tokens": 597036964.0, + "step": 23079 + }, + { + "epoch": 2.5345925763233033, + "grad_norm": 2.68863582611084, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7507922649383545, + "num_tokens": 597055558.0, + "step": 23080 + }, + { + "epoch": 2.534702394025917, + "grad_norm": 2.3114283084869385, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7178239822387695, + "num_tokens": 597081249.0, + "step": 23081 + }, + { + "epoch": 2.5348122117285308, + "grad_norm": 2.428187370300293, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.74268639087677, + "num_tokens": 597104520.0, + "step": 23082 + }, + { + "epoch": 2.5349220294311445, + "grad_norm": 2.031292200088501, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7328959107398987, + "num_tokens": 597136510.0, + "step": 23083 + }, + { + "epoch": 2.535031847133758, + "grad_norm": 2.6808576583862305, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7261587977409363, + "num_tokens": 597157658.0, + "step": 23084 + }, + { + "epoch": 2.5351416648363716, + "grad_norm": 2.064932107925415, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7284969687461853, + "num_tokens": 597189398.0, + "step": 23085 + }, + { + "epoch": 2.5352514825389854, + "grad_norm": 2.6337480545043945, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7497200965881348, + "num_tokens": 597210921.0, + "step": 23086 + }, + { + "epoch": 2.5353613002415987, + "grad_norm": 2.2727041244506836, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7452017068862915, + "num_tokens": 597237123.0, + "step": 23087 + }, + { + "epoch": 2.535471117944213, + "grad_norm": 2.4796223640441895, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7403444051742554, + "num_tokens": 597260578.0, + "step": 23088 + }, + { + "epoch": 2.535580935646826, + "grad_norm": 2.229449987411499, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7419073581695557, + "num_tokens": 597289215.0, + "step": 23089 + }, + { + "epoch": 2.53569075334944, + "grad_norm": 2.2423524856567383, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7292178869247437, + "num_tokens": 597316786.0, + "step": 23090 + }, + { + "epoch": 2.5358005710520537, + "grad_norm": 2.4007418155670166, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7223578691482544, + "num_tokens": 597340930.0, + "step": 23091 + }, + { + "epoch": 2.535910388754667, + "grad_norm": 2.552468776702881, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7282038927078247, + "num_tokens": 597364706.0, + "step": 23092 + }, + { + "epoch": 2.5360202064572808, + "grad_norm": 2.4774036407470703, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7262154817581177, + "num_tokens": 597386630.0, + "step": 23093 + }, + { + "epoch": 2.5361300241598945, + "grad_norm": 2.4341564178466797, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7360880374908447, + "num_tokens": 597410659.0, + "step": 23094 + }, + { + "epoch": 2.5362398418625083, + "grad_norm": 2.4063282012939453, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7304861545562744, + "num_tokens": 597437572.0, + "step": 23095 + }, + { + "epoch": 2.536349659565122, + "grad_norm": 2.4035489559173584, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7324284315109253, + "num_tokens": 597460736.0, + "step": 23096 + }, + { + "epoch": 2.5364594772677354, + "grad_norm": 2.4308032989501953, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7319471836090088, + "num_tokens": 597485885.0, + "step": 23097 + }, + { + "epoch": 2.536569294970349, + "grad_norm": 2.381147623062134, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7339731454849243, + "num_tokens": 597511911.0, + "step": 23098 + }, + { + "epoch": 2.536679112672963, + "grad_norm": 2.2581288814544678, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7354679107666016, + "num_tokens": 597540549.0, + "step": 23099 + }, + { + "epoch": 2.5367889303755766, + "grad_norm": 2.1936635971069336, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7296456694602966, + "num_tokens": 597568988.0, + "step": 23100 + }, + { + "epoch": 2.5368987480781904, + "grad_norm": 2.128436803817749, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7244607210159302, + "num_tokens": 597599355.0, + "step": 23101 + }, + { + "epoch": 2.5370085657808037, + "grad_norm": 2.214759588241577, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7326337099075317, + "num_tokens": 597626892.0, + "step": 23102 + }, + { + "epoch": 2.5371183834834174, + "grad_norm": 2.067190647125244, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7210357785224915, + "num_tokens": 597656696.0, + "step": 23103 + }, + { + "epoch": 2.537228201186031, + "grad_norm": 2.1431994438171387, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6956443786621094, + "num_tokens": 597686492.0, + "step": 23104 + }, + { + "epoch": 2.537338018888645, + "grad_norm": 2.1523001194000244, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.70732581615448, + "num_tokens": 597720272.0, + "step": 23105 + }, + { + "epoch": 2.5374478365912587, + "grad_norm": 2.510546922683716, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7315057516098022, + "num_tokens": 597745245.0, + "step": 23106 + }, + { + "epoch": 2.537557654293872, + "grad_norm": 2.531420946121216, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7337231636047363, + "num_tokens": 597769483.0, + "step": 23107 + }, + { + "epoch": 2.537667471996486, + "grad_norm": 2.2887685298919678, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7398737072944641, + "num_tokens": 597794789.0, + "step": 23108 + }, + { + "epoch": 2.5377772896990995, + "grad_norm": 2.199693441390991, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7309905886650085, + "num_tokens": 597823774.0, + "step": 23109 + }, + { + "epoch": 2.5378871074017133, + "grad_norm": 2.199012517929077, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.725437581539154, + "num_tokens": 597854226.0, + "step": 23110 + }, + { + "epoch": 2.537996925104327, + "grad_norm": 2.398038864135742, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7449240684509277, + "num_tokens": 597879666.0, + "step": 23111 + }, + { + "epoch": 2.5381067428069404, + "grad_norm": 2.4981894493103027, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7417600154876709, + "num_tokens": 597903409.0, + "step": 23112 + }, + { + "epoch": 2.538216560509554, + "grad_norm": 2.2829225063323975, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7419158816337585, + "num_tokens": 597932638.0, + "step": 23113 + }, + { + "epoch": 2.538326378212168, + "grad_norm": 2.5366485118865967, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7452393770217896, + "num_tokens": 597954380.0, + "step": 23114 + }, + { + "epoch": 2.538436195914781, + "grad_norm": 2.444042205810547, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.725456953048706, + "num_tokens": 597980225.0, + "step": 23115 + }, + { + "epoch": 2.538546013617395, + "grad_norm": 2.5552420616149902, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7380518317222595, + "num_tokens": 598001818.0, + "step": 23116 + }, + { + "epoch": 2.5386558313200087, + "grad_norm": 2.170741319656372, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.740085244178772, + "num_tokens": 598031330.0, + "step": 23117 + }, + { + "epoch": 2.5387656490226225, + "grad_norm": 2.578778028488159, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7427446842193604, + "num_tokens": 598052679.0, + "step": 23118 + }, + { + "epoch": 2.5388754667252362, + "grad_norm": 2.599825859069824, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7413724660873413, + "num_tokens": 598073395.0, + "step": 23119 + }, + { + "epoch": 2.5389852844278495, + "grad_norm": 2.3915798664093018, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7312309145927429, + "num_tokens": 598099590.0, + "step": 23120 + }, + { + "epoch": 2.5390951021304633, + "grad_norm": 2.9603993892669678, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7571800947189331, + "num_tokens": 598117357.0, + "step": 23121 + }, + { + "epoch": 2.539204919833077, + "grad_norm": 2.1231632232666016, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7472096681594849, + "num_tokens": 598146407.0, + "step": 23122 + }, + { + "epoch": 2.539314737535691, + "grad_norm": 2.639209270477295, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7320512533187866, + "num_tokens": 598168807.0, + "step": 23123 + }, + { + "epoch": 2.5394245552383046, + "grad_norm": 2.4792933464050293, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7349725961685181, + "num_tokens": 598193948.0, + "step": 23124 + }, + { + "epoch": 2.539534372940918, + "grad_norm": 2.311357021331787, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.734809160232544, + "num_tokens": 598220707.0, + "step": 23125 + }, + { + "epoch": 2.5396441906435316, + "grad_norm": 2.33642315864563, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.745020866394043, + "num_tokens": 598246641.0, + "step": 23126 + }, + { + "epoch": 2.5397540083461454, + "grad_norm": 2.3266589641571045, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7350387573242188, + "num_tokens": 598275816.0, + "step": 23127 + }, + { + "epoch": 2.539863826048759, + "grad_norm": 2.290614128112793, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7287759780883789, + "num_tokens": 598304609.0, + "step": 23128 + }, + { + "epoch": 2.539973643751373, + "grad_norm": 2.139355182647705, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7166234254837036, + "num_tokens": 598332952.0, + "step": 23129 + }, + { + "epoch": 2.5400834614539862, + "grad_norm": 2.18574595451355, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7140309810638428, + "num_tokens": 598361065.0, + "step": 23130 + }, + { + "epoch": 2.5401932791566, + "grad_norm": 2.166128635406494, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7069807648658752, + "num_tokens": 598394400.0, + "step": 23131 + }, + { + "epoch": 2.5403030968592137, + "grad_norm": 2.3760361671447754, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7366081476211548, + "num_tokens": 598421433.0, + "step": 23132 + }, + { + "epoch": 2.5404129145618275, + "grad_norm": 2.673138380050659, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7302208542823792, + "num_tokens": 598441169.0, + "step": 23133 + }, + { + "epoch": 2.5405227322644413, + "grad_norm": 2.5507237911224365, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7352856397628784, + "num_tokens": 598463886.0, + "step": 23134 + }, + { + "epoch": 2.5406325499670546, + "grad_norm": 2.7285499572753906, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7253250479698181, + "num_tokens": 598485379.0, + "step": 23135 + }, + { + "epoch": 2.5407423676696683, + "grad_norm": 2.4770052433013916, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7343671917915344, + "num_tokens": 598511676.0, + "step": 23136 + }, + { + "epoch": 2.540852185372282, + "grad_norm": 2.631906747817993, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7177385687828064, + "num_tokens": 598533470.0, + "step": 23137 + }, + { + "epoch": 2.540962003074896, + "grad_norm": 2.3144960403442383, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7201260328292847, + "num_tokens": 598560612.0, + "step": 23138 + }, + { + "epoch": 2.5410718207775096, + "grad_norm": 2.5326473712921143, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.7465924024581909, + "num_tokens": 598583473.0, + "step": 23139 + }, + { + "epoch": 2.541181638480123, + "grad_norm": 2.211101770401001, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7300999760627747, + "num_tokens": 598611489.0, + "step": 23140 + }, + { + "epoch": 2.5412914561827367, + "grad_norm": 2.4014902114868164, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7329803109169006, + "num_tokens": 598633506.0, + "step": 23141 + }, + { + "epoch": 2.5414012738853504, + "grad_norm": 2.549689769744873, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7461395263671875, + "num_tokens": 598655695.0, + "step": 23142 + }, + { + "epoch": 2.5415110915879637, + "grad_norm": 2.4199752807617188, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.7484186887741089, + "num_tokens": 598678323.0, + "step": 23143 + }, + { + "epoch": 2.5416209092905775, + "grad_norm": 2.3219759464263916, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7413002252578735, + "num_tokens": 598702544.0, + "step": 23144 + }, + { + "epoch": 2.5417307269931912, + "grad_norm": 2.363813638687134, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7130728960037231, + "num_tokens": 598728721.0, + "step": 23145 + }, + { + "epoch": 2.541840544695805, + "grad_norm": 2.7322864532470703, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7531664967536926, + "num_tokens": 598748373.0, + "step": 23146 + }, + { + "epoch": 2.5419503623984188, + "grad_norm": 2.4279768466949463, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7318630218505859, + "num_tokens": 598772238.0, + "step": 23147 + }, + { + "epoch": 2.542060180101032, + "grad_norm": 2.3751273155212402, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7323124408721924, + "num_tokens": 598798213.0, + "step": 23148 + }, + { + "epoch": 2.542169997803646, + "grad_norm": 2.3097293376922607, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7113303542137146, + "num_tokens": 598825979.0, + "step": 23149 + }, + { + "epoch": 2.5422798155062596, + "grad_norm": 2.6098222732543945, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7439365386962891, + "num_tokens": 598848689.0, + "step": 23150 + }, + { + "epoch": 2.5423896332088733, + "grad_norm": 2.145663022994995, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7314082980155945, + "num_tokens": 598879821.0, + "step": 23151 + }, + { + "epoch": 2.542499450911487, + "grad_norm": 2.05058217048645, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7141019701957703, + "num_tokens": 598912920.0, + "step": 23152 + }, + { + "epoch": 2.5426092686141004, + "grad_norm": 2.1766209602355957, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7330754995346069, + "num_tokens": 598941236.0, + "step": 23153 + }, + { + "epoch": 2.542719086316714, + "grad_norm": 2.0556840896606445, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7382579445838928, + "num_tokens": 598974363.0, + "step": 23154 + }, + { + "epoch": 2.542828904019328, + "grad_norm": 2.446322441101074, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7226517796516418, + "num_tokens": 598998809.0, + "step": 23155 + }, + { + "epoch": 2.5429387217219417, + "grad_norm": 2.078423500061035, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7257304191589355, + "num_tokens": 599030383.0, + "step": 23156 + }, + { + "epoch": 2.5430485394245554, + "grad_norm": 2.163116216659546, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7261961698532104, + "num_tokens": 599061889.0, + "step": 23157 + }, + { + "epoch": 2.5431583571271688, + "grad_norm": 2.378842353820801, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7166321277618408, + "num_tokens": 599087343.0, + "step": 23158 + }, + { + "epoch": 2.5432681748297825, + "grad_norm": 2.2283453941345215, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7207790613174438, + "num_tokens": 599116440.0, + "step": 23159 + }, + { + "epoch": 2.5433779925323963, + "grad_norm": 2.746983766555786, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7516758441925049, + "num_tokens": 599137004.0, + "step": 23160 + }, + { + "epoch": 2.54348781023501, + "grad_norm": 2.3218019008636475, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7235878705978394, + "num_tokens": 599164010.0, + "step": 23161 + }, + { + "epoch": 2.543597627937624, + "grad_norm": 2.4885852336883545, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7490262985229492, + "num_tokens": 599186751.0, + "step": 23162 + }, + { + "epoch": 2.543707445640237, + "grad_norm": 2.2917239665985107, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7145904302597046, + "num_tokens": 599216747.0, + "step": 23163 + }, + { + "epoch": 2.543817263342851, + "grad_norm": 2.3451342582702637, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7338773012161255, + "num_tokens": 599242956.0, + "step": 23164 + }, + { + "epoch": 2.5439270810454646, + "grad_norm": 2.4196982383728027, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7324428558349609, + "num_tokens": 599268162.0, + "step": 23165 + }, + { + "epoch": 2.544036898748078, + "grad_norm": 2.4822864532470703, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7328019142150879, + "num_tokens": 599292277.0, + "step": 23166 + }, + { + "epoch": 2.544146716450692, + "grad_norm": 2.7133524417877197, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7448095679283142, + "num_tokens": 599314496.0, + "step": 23167 + }, + { + "epoch": 2.5442565341533054, + "grad_norm": 2.432025194168091, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7107548117637634, + "num_tokens": 599341428.0, + "step": 23168 + }, + { + "epoch": 2.544366351855919, + "grad_norm": 2.6959824562072754, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.76121985912323, + "num_tokens": 599362376.0, + "step": 23169 + }, + { + "epoch": 2.544476169558533, + "grad_norm": 2.621516704559326, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7543423771858215, + "num_tokens": 599382189.0, + "step": 23170 + }, + { + "epoch": 2.5445859872611463, + "grad_norm": 2.6422336101531982, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7436627149581909, + "num_tokens": 599404613.0, + "step": 23171 + }, + { + "epoch": 2.54469580496376, + "grad_norm": 2.91743540763855, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7335026264190674, + "num_tokens": 599422510.0, + "step": 23172 + }, + { + "epoch": 2.544805622666374, + "grad_norm": 2.4775123596191406, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7259712219238281, + "num_tokens": 599448293.0, + "step": 23173 + }, + { + "epoch": 2.5449154403689875, + "grad_norm": 2.230700731277466, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7141905426979065, + "num_tokens": 599478848.0, + "step": 23174 + }, + { + "epoch": 2.5450252580716013, + "grad_norm": 2.4387309551239014, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7528059482574463, + "num_tokens": 599502001.0, + "step": 23175 + }, + { + "epoch": 2.5451350757742146, + "grad_norm": 2.2908639907836914, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7292051315307617, + "num_tokens": 599528639.0, + "step": 23176 + }, + { + "epoch": 2.5452448934768284, + "grad_norm": 2.345895767211914, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7017635703086853, + "num_tokens": 599556815.0, + "step": 23177 + }, + { + "epoch": 2.545354711179442, + "grad_norm": 2.4123973846435547, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7502701282501221, + "num_tokens": 599580546.0, + "step": 23178 + }, + { + "epoch": 2.545464528882056, + "grad_norm": 2.1669747829437256, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7236806154251099, + "num_tokens": 599610573.0, + "step": 23179 + }, + { + "epoch": 2.5455743465846696, + "grad_norm": 2.5829551219940186, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7208049297332764, + "num_tokens": 599632809.0, + "step": 23180 + }, + { + "epoch": 2.545684164287283, + "grad_norm": 2.381861686706543, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7148512601852417, + "num_tokens": 599660754.0, + "step": 23181 + }, + { + "epoch": 2.5457939819898967, + "grad_norm": 2.123310089111328, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7237800359725952, + "num_tokens": 599692272.0, + "step": 23182 + }, + { + "epoch": 2.5459037996925105, + "grad_norm": 2.2679731845855713, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.717867374420166, + "num_tokens": 599722418.0, + "step": 23183 + }, + { + "epoch": 2.546013617395124, + "grad_norm": 2.2557716369628906, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.73582923412323, + "num_tokens": 599750615.0, + "step": 23184 + }, + { + "epoch": 2.546123435097738, + "grad_norm": 2.33278226852417, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7430019378662109, + "num_tokens": 599779424.0, + "step": 23185 + }, + { + "epoch": 2.5462332528003513, + "grad_norm": 2.385831117630005, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7380340099334717, + "num_tokens": 599804042.0, + "step": 23186 + }, + { + "epoch": 2.546343070502965, + "grad_norm": 2.4149012565612793, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6934716701507568, + "num_tokens": 599831383.0, + "step": 23187 + }, + { + "epoch": 2.546452888205579, + "grad_norm": 2.4758002758026123, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7441414594650269, + "num_tokens": 599855286.0, + "step": 23188 + }, + { + "epoch": 2.5465627059081926, + "grad_norm": 2.579106569290161, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7420748472213745, + "num_tokens": 599878271.0, + "step": 23189 + }, + { + "epoch": 2.5466725236108063, + "grad_norm": 2.316866874694824, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7329137325286865, + "num_tokens": 599905729.0, + "step": 23190 + }, + { + "epoch": 2.5467823413134196, + "grad_norm": 2.6158885955810547, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.75989830493927, + "num_tokens": 599928077.0, + "step": 23191 + }, + { + "epoch": 2.5468921590160334, + "grad_norm": 2.5049872398376465, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7463371753692627, + "num_tokens": 599951375.0, + "step": 23192 + }, + { + "epoch": 2.547001976718647, + "grad_norm": 2.2754478454589844, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7287871241569519, + "num_tokens": 599976493.0, + "step": 23193 + }, + { + "epoch": 2.5471117944212605, + "grad_norm": 2.5299007892608643, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7406058311462402, + "num_tokens": 600000689.0, + "step": 23194 + }, + { + "epoch": 2.547221612123874, + "grad_norm": 2.162698984146118, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.696222186088562, + "num_tokens": 600031697.0, + "step": 23195 + }, + { + "epoch": 2.547331429826488, + "grad_norm": 2.2788431644439697, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7472248673439026, + "num_tokens": 600056997.0, + "step": 23196 + }, + { + "epoch": 2.5474412475291017, + "grad_norm": 2.4674832820892334, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.7443180084228516, + "num_tokens": 600081186.0, + "step": 23197 + }, + { + "epoch": 2.5475510652317155, + "grad_norm": 2.3602328300476074, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7196480631828308, + "num_tokens": 600106506.0, + "step": 23198 + }, + { + "epoch": 2.547660882934329, + "grad_norm": 2.1782360076904297, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.714011549949646, + "num_tokens": 600137370.0, + "step": 23199 + }, + { + "epoch": 2.5477707006369426, + "grad_norm": 2.296478748321533, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7385268211364746, + "num_tokens": 600164594.0, + "step": 23200 + }, + { + "epoch": 2.5478805183395563, + "grad_norm": 2.3556079864501953, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7294330596923828, + "num_tokens": 600189231.0, + "step": 23201 + }, + { + "epoch": 2.54799033604217, + "grad_norm": 2.3823659420013428, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7320082187652588, + "num_tokens": 600214379.0, + "step": 23202 + }, + { + "epoch": 2.548100153744784, + "grad_norm": 2.209304094314575, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.747821569442749, + "num_tokens": 600241873.0, + "step": 23203 + }, + { + "epoch": 2.548209971447397, + "grad_norm": 2.38783597946167, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7425755262374878, + "num_tokens": 600265780.0, + "step": 23204 + }, + { + "epoch": 2.548319789150011, + "grad_norm": 2.2726755142211914, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7373305559158325, + "num_tokens": 600294184.0, + "step": 23205 + }, + { + "epoch": 2.5484296068526247, + "grad_norm": 2.2462265491485596, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7447404861450195, + "num_tokens": 600322499.0, + "step": 23206 + }, + { + "epoch": 2.5485394245552384, + "grad_norm": 2.2717301845550537, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7254111766815186, + "num_tokens": 600351212.0, + "step": 23207 + }, + { + "epoch": 2.548649242257852, + "grad_norm": 2.3482449054718018, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7214524149894714, + "num_tokens": 600376114.0, + "step": 23208 + }, + { + "epoch": 2.5487590599604655, + "grad_norm": 2.6788501739501953, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7296099662780762, + "num_tokens": 600397781.0, + "step": 23209 + }, + { + "epoch": 2.5488688776630792, + "grad_norm": 2.2230241298675537, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7493300437927246, + "num_tokens": 600425370.0, + "step": 23210 + }, + { + "epoch": 2.548978695365693, + "grad_norm": 2.390646457672119, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7147795557975769, + "num_tokens": 600451785.0, + "step": 23211 + }, + { + "epoch": 2.5490885130683067, + "grad_norm": 2.0655903816223145, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7350099086761475, + "num_tokens": 600485089.0, + "step": 23212 + }, + { + "epoch": 2.5491983307709205, + "grad_norm": 2.741558790206909, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7470405101776123, + "num_tokens": 600505764.0, + "step": 23213 + }, + { + "epoch": 2.549308148473534, + "grad_norm": 2.571650981903076, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.738836944103241, + "num_tokens": 600529607.0, + "step": 23214 + }, + { + "epoch": 2.5494179661761476, + "grad_norm": 2.736985683441162, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7618764638900757, + "num_tokens": 600549122.0, + "step": 23215 + }, + { + "epoch": 2.5495277838787613, + "grad_norm": 2.135410785675049, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7339293956756592, + "num_tokens": 600578980.0, + "step": 23216 + }, + { + "epoch": 2.5496376015813746, + "grad_norm": 2.4814164638519287, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7633960843086243, + "num_tokens": 600602436.0, + "step": 23217 + }, + { + "epoch": 2.549747419283989, + "grad_norm": 2.0912439823150635, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7453324794769287, + "num_tokens": 600631297.0, + "step": 23218 + }, + { + "epoch": 2.549857236986602, + "grad_norm": 2.3605434894561768, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7297141551971436, + "num_tokens": 600656923.0, + "step": 23219 + }, + { + "epoch": 2.549967054689216, + "grad_norm": 2.246372938156128, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7283779382705688, + "num_tokens": 600683939.0, + "step": 23220 + }, + { + "epoch": 2.5500768723918297, + "grad_norm": 2.182582378387451, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.724220871925354, + "num_tokens": 600712750.0, + "step": 23221 + }, + { + "epoch": 2.550186690094443, + "grad_norm": 2.451582670211792, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7228612899780273, + "num_tokens": 600736385.0, + "step": 23222 + }, + { + "epoch": 2.5502965077970567, + "grad_norm": 2.682431936264038, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7459713220596313, + "num_tokens": 600756943.0, + "step": 23223 + }, + { + "epoch": 2.5504063254996705, + "grad_norm": 2.300762176513672, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7265071868896484, + "num_tokens": 600784891.0, + "step": 23224 + }, + { + "epoch": 2.5505161432022843, + "grad_norm": 2.108334541320801, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7248901128768921, + "num_tokens": 600815435.0, + "step": 23225 + }, + { + "epoch": 2.550625960904898, + "grad_norm": 2.3414371013641357, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7105488777160645, + "num_tokens": 600842585.0, + "step": 23226 + }, + { + "epoch": 2.5507357786075113, + "grad_norm": 2.2016308307647705, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7532979249954224, + "num_tokens": 600870637.0, + "step": 23227 + }, + { + "epoch": 2.550845596310125, + "grad_norm": 2.4856116771698, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.737372875213623, + "num_tokens": 600894183.0, + "step": 23228 + }, + { + "epoch": 2.550955414012739, + "grad_norm": 2.4819912910461426, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7344662547111511, + "num_tokens": 600917597.0, + "step": 23229 + }, + { + "epoch": 2.5510652317153526, + "grad_norm": 2.0308547019958496, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7240958213806152, + "num_tokens": 600950127.0, + "step": 23230 + }, + { + "epoch": 2.5511750494179664, + "grad_norm": 2.297086238861084, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7313705086708069, + "num_tokens": 600976942.0, + "step": 23231 + }, + { + "epoch": 2.5512848671205797, + "grad_norm": 2.483888864517212, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7349166870117188, + "num_tokens": 601000981.0, + "step": 23232 + }, + { + "epoch": 2.5513946848231934, + "grad_norm": 2.3848540782928467, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7486975193023682, + "num_tokens": 601027115.0, + "step": 23233 + }, + { + "epoch": 2.551504502525807, + "grad_norm": 2.3806705474853516, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7609421014785767, + "num_tokens": 601050975.0, + "step": 23234 + }, + { + "epoch": 2.551614320228421, + "grad_norm": 2.2018885612487793, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7405399084091187, + "num_tokens": 601079541.0, + "step": 23235 + }, + { + "epoch": 2.5517241379310347, + "grad_norm": 2.272559881210327, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7209963202476501, + "num_tokens": 601109464.0, + "step": 23236 + }, + { + "epoch": 2.551833955633648, + "grad_norm": 2.2244937419891357, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7472953796386719, + "num_tokens": 601137656.0, + "step": 23237 + }, + { + "epoch": 2.5519437733362618, + "grad_norm": 2.5043106079101562, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7376962900161743, + "num_tokens": 601160857.0, + "step": 23238 + }, + { + "epoch": 2.5520535910388755, + "grad_norm": 2.2731056213378906, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7459099888801575, + "num_tokens": 601187777.0, + "step": 23239 + }, + { + "epoch": 2.5521634087414893, + "grad_norm": 2.3873202800750732, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7480167150497437, + "num_tokens": 601211846.0, + "step": 23240 + }, + { + "epoch": 2.552273226444103, + "grad_norm": 2.4466350078582764, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7266456484794617, + "num_tokens": 601236545.0, + "step": 23241 + }, + { + "epoch": 2.5523830441467164, + "grad_norm": 2.2065019607543945, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7315753698348999, + "num_tokens": 601264423.0, + "step": 23242 + }, + { + "epoch": 2.55249286184933, + "grad_norm": 2.294544219970703, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7228386402130127, + "num_tokens": 601290790.0, + "step": 23243 + }, + { + "epoch": 2.552602679551944, + "grad_norm": 2.5416195392608643, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7416630983352661, + "num_tokens": 601311762.0, + "step": 23244 + }, + { + "epoch": 2.552712497254557, + "grad_norm": 2.268951654434204, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7360395193099976, + "num_tokens": 601338352.0, + "step": 23245 + }, + { + "epoch": 2.552822314957171, + "grad_norm": 2.570301055908203, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7532292604446411, + "num_tokens": 601360971.0, + "step": 23246 + }, + { + "epoch": 2.5529321326597847, + "grad_norm": 2.5659759044647217, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7418155670166016, + "num_tokens": 601382171.0, + "step": 23247 + }, + { + "epoch": 2.5530419503623984, + "grad_norm": 2.458371162414551, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7384903430938721, + "num_tokens": 601405621.0, + "step": 23248 + }, + { + "epoch": 2.553151768065012, + "grad_norm": 2.2795560359954834, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7255496978759766, + "num_tokens": 601433299.0, + "step": 23249 + }, + { + "epoch": 2.5532615857676255, + "grad_norm": 2.1597471237182617, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7369967103004456, + "num_tokens": 601461285.0, + "step": 23250 + }, + { + "epoch": 2.5533714034702393, + "grad_norm": 2.2054827213287354, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7122113704681396, + "num_tokens": 601491882.0, + "step": 23251 + }, + { + "epoch": 2.553481221172853, + "grad_norm": 2.7048568725585938, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7333704233169556, + "num_tokens": 601513290.0, + "step": 23252 + }, + { + "epoch": 2.553591038875467, + "grad_norm": 2.400165319442749, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7082394361495972, + "num_tokens": 601538138.0, + "step": 23253 + }, + { + "epoch": 2.5537008565780805, + "grad_norm": 2.2453370094299316, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7099776268005371, + "num_tokens": 601568430.0, + "step": 23254 + }, + { + "epoch": 2.553810674280694, + "grad_norm": 2.3911962509155273, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7341634631156921, + "num_tokens": 601594573.0, + "step": 23255 + }, + { + "epoch": 2.5539204919833076, + "grad_norm": 2.192267417907715, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7434680461883545, + "num_tokens": 601622798.0, + "step": 23256 + }, + { + "epoch": 2.5540303096859214, + "grad_norm": 2.5339841842651367, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7287089824676514, + "num_tokens": 601645628.0, + "step": 23257 + }, + { + "epoch": 2.554140127388535, + "grad_norm": 2.766934871673584, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7233418226242065, + "num_tokens": 601666359.0, + "step": 23258 + }, + { + "epoch": 2.554249945091149, + "grad_norm": 2.5772323608398438, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7262616157531738, + "num_tokens": 601687415.0, + "step": 23259 + }, + { + "epoch": 2.554359762793762, + "grad_norm": 2.542696475982666, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7077860236167908, + "num_tokens": 601712944.0, + "step": 23260 + }, + { + "epoch": 2.554469580496376, + "grad_norm": 2.258394241333008, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7284928560256958, + "num_tokens": 601740554.0, + "step": 23261 + }, + { + "epoch": 2.5545793981989897, + "grad_norm": 2.7271881103515625, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7295597791671753, + "num_tokens": 601762863.0, + "step": 23262 + }, + { + "epoch": 2.5546892159016035, + "grad_norm": 2.373304605484009, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7247951030731201, + "num_tokens": 601790171.0, + "step": 23263 + }, + { + "epoch": 2.5547990336042172, + "grad_norm": 2.209390640258789, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.752052903175354, + "num_tokens": 601817728.0, + "step": 23264 + }, + { + "epoch": 2.5549088513068305, + "grad_norm": 2.5495686531066895, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7362139225006104, + "num_tokens": 601840399.0, + "step": 23265 + }, + { + "epoch": 2.5550186690094443, + "grad_norm": 2.456294298171997, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7482406497001648, + "num_tokens": 601864120.0, + "step": 23266 + }, + { + "epoch": 2.555128486712058, + "grad_norm": 2.5590038299560547, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7285435199737549, + "num_tokens": 601887840.0, + "step": 23267 + }, + { + "epoch": 2.5552383044146714, + "grad_norm": 2.5665464401245117, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7180827260017395, + "num_tokens": 601911940.0, + "step": 23268 + }, + { + "epoch": 2.5553481221172856, + "grad_norm": 2.755620241165161, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7287113666534424, + "num_tokens": 601932390.0, + "step": 23269 + }, + { + "epoch": 2.555457939819899, + "grad_norm": 2.348182439804077, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7311251163482666, + "num_tokens": 601961872.0, + "step": 23270 + }, + { + "epoch": 2.5555677575225126, + "grad_norm": 2.245542526245117, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7216407656669617, + "num_tokens": 601990854.0, + "step": 23271 + }, + { + "epoch": 2.5556775752251264, + "grad_norm": 2.6905627250671387, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7391760349273682, + "num_tokens": 602010988.0, + "step": 23272 + }, + { + "epoch": 2.5557873929277397, + "grad_norm": 2.3939414024353027, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7418464422225952, + "num_tokens": 602035175.0, + "step": 23273 + }, + { + "epoch": 2.5558972106303535, + "grad_norm": 2.4158854484558105, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7216880917549133, + "num_tokens": 602057633.0, + "step": 23274 + }, + { + "epoch": 2.5560070283329672, + "grad_norm": 2.390493392944336, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7473032474517822, + "num_tokens": 602082040.0, + "step": 23275 + }, + { + "epoch": 2.556116846035581, + "grad_norm": 2.5733723640441895, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7327628135681152, + "num_tokens": 602104266.0, + "step": 23276 + }, + { + "epoch": 2.5562266637381947, + "grad_norm": 2.225830316543579, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7148780822753906, + "num_tokens": 602131165.0, + "step": 23277 + }, + { + "epoch": 2.556336481440808, + "grad_norm": 2.410907745361328, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7591687440872192, + "num_tokens": 602155657.0, + "step": 23278 + }, + { + "epoch": 2.556446299143422, + "grad_norm": 2.352001428604126, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7330119609832764, + "num_tokens": 602181965.0, + "step": 23279 + }, + { + "epoch": 2.5565561168460356, + "grad_norm": 2.216761589050293, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.721001923084259, + "num_tokens": 602211976.0, + "step": 23280 + }, + { + "epoch": 2.5566659345486493, + "grad_norm": 2.122840166091919, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7548513412475586, + "num_tokens": 602239196.0, + "step": 23281 + }, + { + "epoch": 2.556775752251263, + "grad_norm": 2.2351067066192627, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.734959602355957, + "num_tokens": 602267444.0, + "step": 23282 + }, + { + "epoch": 2.5568855699538764, + "grad_norm": 2.0545125007629395, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7295072078704834, + "num_tokens": 602298803.0, + "step": 23283 + }, + { + "epoch": 2.55699538765649, + "grad_norm": 2.209087371826172, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7340223789215088, + "num_tokens": 602328423.0, + "step": 23284 + }, + { + "epoch": 2.557105205359104, + "grad_norm": 2.5781779289245605, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7235602736473083, + "num_tokens": 602350943.0, + "step": 23285 + }, + { + "epoch": 2.5572150230617177, + "grad_norm": 2.239530563354492, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7367990612983704, + "num_tokens": 602378886.0, + "step": 23286 + }, + { + "epoch": 2.5573248407643314, + "grad_norm": 2.522714376449585, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7560316324234009, + "num_tokens": 602400804.0, + "step": 23287 + }, + { + "epoch": 2.5574346584669447, + "grad_norm": 2.3921542167663574, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7223206758499146, + "num_tokens": 602426240.0, + "step": 23288 + }, + { + "epoch": 2.5575444761695585, + "grad_norm": 2.7258951663970947, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7329562306404114, + "num_tokens": 602447651.0, + "step": 23289 + }, + { + "epoch": 2.5576542938721722, + "grad_norm": 2.8174667358398438, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7282455563545227, + "num_tokens": 602469324.0, + "step": 23290 + }, + { + "epoch": 2.557764111574786, + "grad_norm": 2.366279125213623, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7557344436645508, + "num_tokens": 602493436.0, + "step": 23291 + }, + { + "epoch": 2.5578739292773998, + "grad_norm": 2.214771032333374, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.720741331577301, + "num_tokens": 602524451.0, + "step": 23292 + }, + { + "epoch": 2.557983746980013, + "grad_norm": 2.266408681869507, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7374701499938965, + "num_tokens": 602552139.0, + "step": 23293 + }, + { + "epoch": 2.558093564682627, + "grad_norm": 2.115241527557373, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7341664433479309, + "num_tokens": 602584609.0, + "step": 23294 + }, + { + "epoch": 2.5582033823852406, + "grad_norm": 2.2534658908843994, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7090142369270325, + "num_tokens": 602613567.0, + "step": 23295 + }, + { + "epoch": 2.558313200087854, + "grad_norm": 3.033681631088257, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.749657154083252, + "num_tokens": 602630565.0, + "step": 23296 + }, + { + "epoch": 2.5584230177904677, + "grad_norm": 2.4566779136657715, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7193324565887451, + "num_tokens": 602655516.0, + "step": 23297 + }, + { + "epoch": 2.5585328354930814, + "grad_norm": 2.2499117851257324, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7357758283615112, + "num_tokens": 602681620.0, + "step": 23298 + }, + { + "epoch": 2.558642653195695, + "grad_norm": 2.239220142364502, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.747786283493042, + "num_tokens": 602707308.0, + "step": 23299 + }, + { + "epoch": 2.558752470898309, + "grad_norm": 2.3186986446380615, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.735019862651825, + "num_tokens": 602731336.0, + "step": 23300 + }, + { + "epoch": 2.5588622886009222, + "grad_norm": 2.877264976501465, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7540589570999146, + "num_tokens": 602751478.0, + "step": 23301 + }, + { + "epoch": 2.558972106303536, + "grad_norm": 2.4348347187042236, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7133762836456299, + "num_tokens": 602776797.0, + "step": 23302 + }, + { + "epoch": 2.5590819240061498, + "grad_norm": 2.4028518199920654, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7453517913818359, + "num_tokens": 602799985.0, + "step": 23303 + }, + { + "epoch": 2.5591917417087635, + "grad_norm": 2.4352526664733887, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7305601835250854, + "num_tokens": 602825496.0, + "step": 23304 + }, + { + "epoch": 2.5593015594113773, + "grad_norm": 2.4558300971984863, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7212837934494019, + "num_tokens": 602849205.0, + "step": 23305 + }, + { + "epoch": 2.5594113771139906, + "grad_norm": 2.1806130409240723, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7141199707984924, + "num_tokens": 602879621.0, + "step": 23306 + }, + { + "epoch": 2.5595211948166043, + "grad_norm": 2.199761390686035, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7406421899795532, + "num_tokens": 602908426.0, + "step": 23307 + }, + { + "epoch": 2.559631012519218, + "grad_norm": 2.3007850646972656, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.740128219127655, + "num_tokens": 602934702.0, + "step": 23308 + }, + { + "epoch": 2.559740830221832, + "grad_norm": 2.585602045059204, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.73563152551651, + "num_tokens": 602956640.0, + "step": 23309 + }, + { + "epoch": 2.5598506479244456, + "grad_norm": 2.362687587738037, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7482787370681763, + "num_tokens": 602980767.0, + "step": 23310 + }, + { + "epoch": 2.559960465627059, + "grad_norm": 2.4376819133758545, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7372673749923706, + "num_tokens": 603004907.0, + "step": 23311 + }, + { + "epoch": 2.5600702833296727, + "grad_norm": 2.260251045227051, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7389081716537476, + "num_tokens": 603031709.0, + "step": 23312 + }, + { + "epoch": 2.5601801010322864, + "grad_norm": 2.225454092025757, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7163098454475403, + "num_tokens": 603061858.0, + "step": 23313 + }, + { + "epoch": 2.5602899187349, + "grad_norm": 2.4501097202301025, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7387638688087463, + "num_tokens": 603085881.0, + "step": 23314 + }, + { + "epoch": 2.560399736437514, + "grad_norm": 2.9848170280456543, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7392393946647644, + "num_tokens": 603103768.0, + "step": 23315 + }, + { + "epoch": 2.5605095541401273, + "grad_norm": 2.228734016418457, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7332315444946289, + "num_tokens": 603132650.0, + "step": 23316 + }, + { + "epoch": 2.560619371842741, + "grad_norm": 2.4797251224517822, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7395332455635071, + "num_tokens": 603157782.0, + "step": 23317 + }, + { + "epoch": 2.560729189545355, + "grad_norm": 2.877896308898926, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7296756505966187, + "num_tokens": 603177854.0, + "step": 23318 + }, + { + "epoch": 2.5608390072479685, + "grad_norm": 2.717656373977661, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7487999200820923, + "num_tokens": 603199065.0, + "step": 23319 + }, + { + "epoch": 2.5609488249505823, + "grad_norm": 2.3581197261810303, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7338772416114807, + "num_tokens": 603225984.0, + "step": 23320 + }, + { + "epoch": 2.5610586426531956, + "grad_norm": 2.327427864074707, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7351837158203125, + "num_tokens": 603253866.0, + "step": 23321 + }, + { + "epoch": 2.5611684603558094, + "grad_norm": 2.5080208778381348, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7465066909790039, + "num_tokens": 603276823.0, + "step": 23322 + }, + { + "epoch": 2.561278278058423, + "grad_norm": 2.499319076538086, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7541946172714233, + "num_tokens": 603298508.0, + "step": 23323 + }, + { + "epoch": 2.5613880957610364, + "grad_norm": 2.2576777935028076, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7199195623397827, + "num_tokens": 603329742.0, + "step": 23324 + }, + { + "epoch": 2.56149791346365, + "grad_norm": 2.2725870609283447, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7093280553817749, + "num_tokens": 603358198.0, + "step": 23325 + }, + { + "epoch": 2.561607731166264, + "grad_norm": 2.4354465007781982, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7407124042510986, + "num_tokens": 603383642.0, + "step": 23326 + }, + { + "epoch": 2.5617175488688777, + "grad_norm": 2.3386647701263428, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7464587688446045, + "num_tokens": 603409275.0, + "step": 23327 + }, + { + "epoch": 2.5618273665714915, + "grad_norm": 2.105177164077759, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.714388370513916, + "num_tokens": 603440126.0, + "step": 23328 + }, + { + "epoch": 2.5619371842741048, + "grad_norm": 2.768824577331543, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7401891350746155, + "num_tokens": 603459754.0, + "step": 23329 + }, + { + "epoch": 2.5620470019767185, + "grad_norm": 2.0359833240509033, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.713571310043335, + "num_tokens": 603491305.0, + "step": 23330 + }, + { + "epoch": 2.5621568196793323, + "grad_norm": 2.335048198699951, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7407106757164001, + "num_tokens": 603516967.0, + "step": 23331 + }, + { + "epoch": 2.562266637381946, + "grad_norm": 2.2489969730377197, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7377426028251648, + "num_tokens": 603543598.0, + "step": 23332 + }, + { + "epoch": 2.56237645508456, + "grad_norm": 2.1557416915893555, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7301192283630371, + "num_tokens": 603572869.0, + "step": 23333 + }, + { + "epoch": 2.562486272787173, + "grad_norm": 2.6002774238586426, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7216033339500427, + "num_tokens": 603595639.0, + "step": 23334 + }, + { + "epoch": 2.562596090489787, + "grad_norm": 2.6504721641540527, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7152932286262512, + "num_tokens": 603619042.0, + "step": 23335 + }, + { + "epoch": 2.5627059081924006, + "grad_norm": 2.043032169342041, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7125177383422852, + "num_tokens": 603654904.0, + "step": 23336 + }, + { + "epoch": 2.5628157258950144, + "grad_norm": 2.4029407501220703, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7292509078979492, + "num_tokens": 603679865.0, + "step": 23337 + }, + { + "epoch": 2.562925543597628, + "grad_norm": 2.501946449279785, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7341914176940918, + "num_tokens": 603705678.0, + "step": 23338 + }, + { + "epoch": 2.5630353613002415, + "grad_norm": 2.541590929031372, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.711194634437561, + "num_tokens": 603729771.0, + "step": 23339 + }, + { + "epoch": 2.563145179002855, + "grad_norm": 2.7665553092956543, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7246943712234497, + "num_tokens": 603755805.0, + "step": 23340 + }, + { + "epoch": 2.563254996705469, + "grad_norm": 2.3678250312805176, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7194637060165405, + "num_tokens": 603784069.0, + "step": 23341 + }, + { + "epoch": 2.5633648144080827, + "grad_norm": 2.3939385414123535, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7422571182250977, + "num_tokens": 603809966.0, + "step": 23342 + }, + { + "epoch": 2.5634746321106965, + "grad_norm": 2.532959222793579, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7162237167358398, + "num_tokens": 603833997.0, + "step": 23343 + }, + { + "epoch": 2.56358444981331, + "grad_norm": 2.2319281101226807, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7138457298278809, + "num_tokens": 603862599.0, + "step": 23344 + }, + { + "epoch": 2.5636942675159236, + "grad_norm": 2.390636920928955, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7126643657684326, + "num_tokens": 603889002.0, + "step": 23345 + }, + { + "epoch": 2.5638040852185373, + "grad_norm": 2.3618388175964355, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7188177704811096, + "num_tokens": 603915766.0, + "step": 23346 + }, + { + "epoch": 2.5639139029211506, + "grad_norm": 2.099848508834839, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7081564664840698, + "num_tokens": 603949935.0, + "step": 23347 + }, + { + "epoch": 2.564023720623765, + "grad_norm": 2.417402982711792, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.751045823097229, + "num_tokens": 603976908.0, + "step": 23348 + }, + { + "epoch": 2.564133538326378, + "grad_norm": 2.5121545791625977, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7244054079055786, + "num_tokens": 604002638.0, + "step": 23349 + }, + { + "epoch": 2.564243356028992, + "grad_norm": 2.2326879501342773, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7153714895248413, + "num_tokens": 604032425.0, + "step": 23350 + }, + { + "epoch": 2.5643531737316057, + "grad_norm": 2.2362253665924072, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7290303111076355, + "num_tokens": 604062291.0, + "step": 23351 + }, + { + "epoch": 2.564462991434219, + "grad_norm": 2.493511915206909, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7304596900939941, + "num_tokens": 604085436.0, + "step": 23352 + }, + { + "epoch": 2.5645728091368327, + "grad_norm": 2.5613529682159424, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7203394174575806, + "num_tokens": 604108637.0, + "step": 23353 + }, + { + "epoch": 2.5646826268394465, + "grad_norm": 2.42793607711792, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.721900224685669, + "num_tokens": 604133226.0, + "step": 23354 + }, + { + "epoch": 2.5647924445420602, + "grad_norm": 2.3872623443603516, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7299154996871948, + "num_tokens": 604159178.0, + "step": 23355 + }, + { + "epoch": 2.564902262244674, + "grad_norm": 2.622765064239502, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7423169612884521, + "num_tokens": 604181713.0, + "step": 23356 + }, + { + "epoch": 2.5650120799472873, + "grad_norm": 2.1361968517303467, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7247312068939209, + "num_tokens": 604211696.0, + "step": 23357 + }, + { + "epoch": 2.565121897649901, + "grad_norm": 2.299476385116577, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7386846542358398, + "num_tokens": 604238132.0, + "step": 23358 + }, + { + "epoch": 2.565231715352515, + "grad_norm": 2.218686819076538, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7316220998764038, + "num_tokens": 604267759.0, + "step": 23359 + }, + { + "epoch": 2.5653415330551286, + "grad_norm": 2.355705499649048, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7293404936790466, + "num_tokens": 604295286.0, + "step": 23360 + }, + { + "epoch": 2.5654513507577423, + "grad_norm": 2.5154945850372314, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7389026880264282, + "num_tokens": 604318751.0, + "step": 23361 + }, + { + "epoch": 2.5655611684603556, + "grad_norm": 2.2692415714263916, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.756492555141449, + "num_tokens": 604343713.0, + "step": 23362 + }, + { + "epoch": 2.5656709861629694, + "grad_norm": 2.0795998573303223, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7159842252731323, + "num_tokens": 604376454.0, + "step": 23363 + }, + { + "epoch": 2.565780803865583, + "grad_norm": 2.6048636436462402, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7444114089012146, + "num_tokens": 604397121.0, + "step": 23364 + }, + { + "epoch": 2.565890621568197, + "grad_norm": 2.5597944259643555, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7651079893112183, + "num_tokens": 604417671.0, + "step": 23365 + }, + { + "epoch": 2.5660004392708107, + "grad_norm": 2.500239849090576, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7258691191673279, + "num_tokens": 604441561.0, + "step": 23366 + }, + { + "epoch": 2.566110256973424, + "grad_norm": 2.410438060760498, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7248785495758057, + "num_tokens": 604467657.0, + "step": 23367 + }, + { + "epoch": 2.5662200746760377, + "grad_norm": 2.3709118366241455, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7166669368743896, + "num_tokens": 604497964.0, + "step": 23368 + }, + { + "epoch": 2.5663298923786515, + "grad_norm": 2.4362895488739014, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7309610843658447, + "num_tokens": 604523342.0, + "step": 23369 + }, + { + "epoch": 2.5664397100812653, + "grad_norm": 2.4447364807128906, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7224345207214355, + "num_tokens": 604548761.0, + "step": 23370 + }, + { + "epoch": 2.566549527783879, + "grad_norm": 2.4496569633483887, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7161442041397095, + "num_tokens": 604574531.0, + "step": 23371 + }, + { + "epoch": 2.5666593454864923, + "grad_norm": 2.0643763542175293, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7134462594985962, + "num_tokens": 604605288.0, + "step": 23372 + }, + { + "epoch": 2.566769163189106, + "grad_norm": 2.890338897705078, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7570385932922363, + "num_tokens": 604624112.0, + "step": 23373 + }, + { + "epoch": 2.56687898089172, + "grad_norm": 2.4892051219940186, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7454359531402588, + "num_tokens": 604647904.0, + "step": 23374 + }, + { + "epoch": 2.566988798594333, + "grad_norm": 2.5141470432281494, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7304196953773499, + "num_tokens": 604674123.0, + "step": 23375 + }, + { + "epoch": 2.567098616296947, + "grad_norm": 2.5198686122894287, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.744797945022583, + "num_tokens": 604698545.0, + "step": 23376 + }, + { + "epoch": 2.5672084339995607, + "grad_norm": 2.2405567169189453, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7268190383911133, + "num_tokens": 604728830.0, + "step": 23377 + }, + { + "epoch": 2.5673182517021744, + "grad_norm": 2.439744472503662, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7530533671379089, + "num_tokens": 604753387.0, + "step": 23378 + }, + { + "epoch": 2.567428069404788, + "grad_norm": 2.316580295562744, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7476364374160767, + "num_tokens": 604778137.0, + "step": 23379 + }, + { + "epoch": 2.5675378871074015, + "grad_norm": 1.8731441497802734, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7331615686416626, + "num_tokens": 604814087.0, + "step": 23380 + }, + { + "epoch": 2.5676477048100153, + "grad_norm": 2.5023458003997803, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.759157657623291, + "num_tokens": 604836600.0, + "step": 23381 + }, + { + "epoch": 2.567757522512629, + "grad_norm": 2.704119920730591, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7552964687347412, + "num_tokens": 604855732.0, + "step": 23382 + }, + { + "epoch": 2.5678673402152428, + "grad_norm": 2.3255255222320557, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7512562274932861, + "num_tokens": 604883780.0, + "step": 23383 + }, + { + "epoch": 2.5679771579178565, + "grad_norm": 2.4767038822174072, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7508276700973511, + "num_tokens": 604906772.0, + "step": 23384 + }, + { + "epoch": 2.56808697562047, + "grad_norm": 2.1841163635253906, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7013347744941711, + "num_tokens": 604937195.0, + "step": 23385 + }, + { + "epoch": 2.5681967933230836, + "grad_norm": 2.384355306625366, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7276334762573242, + "num_tokens": 604964163.0, + "step": 23386 + }, + { + "epoch": 2.5683066110256974, + "grad_norm": 2.2301523685455322, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7179998159408569, + "num_tokens": 604992272.0, + "step": 23387 + }, + { + "epoch": 2.568416428728311, + "grad_norm": 2.4087460041046143, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7306015491485596, + "num_tokens": 605016750.0, + "step": 23388 + }, + { + "epoch": 2.568526246430925, + "grad_norm": 2.290923833847046, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.754574716091156, + "num_tokens": 605041357.0, + "step": 23389 + }, + { + "epoch": 2.568636064133538, + "grad_norm": 2.2871108055114746, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7453306317329407, + "num_tokens": 605067758.0, + "step": 23390 + }, + { + "epoch": 2.568745881836152, + "grad_norm": 2.38362193107605, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7391936182975769, + "num_tokens": 605092976.0, + "step": 23391 + }, + { + "epoch": 2.5688556995387657, + "grad_norm": 2.214221477508545, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7260038256645203, + "num_tokens": 605121236.0, + "step": 23392 + }, + { + "epoch": 2.5689655172413794, + "grad_norm": 2.1022560596466064, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7277207374572754, + "num_tokens": 605152048.0, + "step": 23393 + }, + { + "epoch": 2.569075334943993, + "grad_norm": 2.6547296047210693, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7480838298797607, + "num_tokens": 605172717.0, + "step": 23394 + }, + { + "epoch": 2.5691851526466065, + "grad_norm": 2.473137378692627, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7532255053520203, + "num_tokens": 605193782.0, + "step": 23395 + }, + { + "epoch": 2.5692949703492203, + "grad_norm": 2.4288320541381836, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7300760746002197, + "num_tokens": 605218931.0, + "step": 23396 + }, + { + "epoch": 2.569404788051834, + "grad_norm": 2.3682844638824463, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.724429726600647, + "num_tokens": 605246891.0, + "step": 23397 + }, + { + "epoch": 2.5695146057544473, + "grad_norm": 2.294705867767334, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7408761978149414, + "num_tokens": 605272368.0, + "step": 23398 + }, + { + "epoch": 2.5696244234570615, + "grad_norm": 2.6902315616607666, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7479200959205627, + "num_tokens": 605292360.0, + "step": 23399 + }, + { + "epoch": 2.569734241159675, + "grad_norm": 2.303054094314575, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7155570387840271, + "num_tokens": 605322031.0, + "step": 23400 + }, + { + "epoch": 2.5698440588622886, + "grad_norm": 2.3280911445617676, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7402230501174927, + "num_tokens": 605347994.0, + "step": 23401 + }, + { + "epoch": 2.5699538765649024, + "grad_norm": 2.5143327713012695, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7266678810119629, + "num_tokens": 605372584.0, + "step": 23402 + }, + { + "epoch": 2.5700636942675157, + "grad_norm": 2.35567307472229, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7366300225257874, + "num_tokens": 605398031.0, + "step": 23403 + }, + { + "epoch": 2.5701735119701294, + "grad_norm": 2.719353675842285, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7290852069854736, + "num_tokens": 605420220.0, + "step": 23404 + }, + { + "epoch": 2.570283329672743, + "grad_norm": 2.2850635051727295, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7141849994659424, + "num_tokens": 605447632.0, + "step": 23405 + }, + { + "epoch": 2.570393147375357, + "grad_norm": 2.564967393875122, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7356358766555786, + "num_tokens": 605469430.0, + "step": 23406 + }, + { + "epoch": 2.5705029650779707, + "grad_norm": 2.1790881156921387, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7081869840621948, + "num_tokens": 605499226.0, + "step": 23407 + }, + { + "epoch": 2.570612782780584, + "grad_norm": 2.2500557899475098, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7214987874031067, + "num_tokens": 605527397.0, + "step": 23408 + }, + { + "epoch": 2.570722600483198, + "grad_norm": 2.2127532958984375, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7210294604301453, + "num_tokens": 605553009.0, + "step": 23409 + }, + { + "epoch": 2.5708324181858115, + "grad_norm": 2.7994585037231445, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7545932531356812, + "num_tokens": 605572554.0, + "step": 23410 + }, + { + "epoch": 2.5709422358884253, + "grad_norm": 2.3153231143951416, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.725648045539856, + "num_tokens": 605600291.0, + "step": 23411 + }, + { + "epoch": 2.571052053591039, + "grad_norm": 2.4615275859832764, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7723863124847412, + "num_tokens": 605624462.0, + "step": 23412 + }, + { + "epoch": 2.5711618712936524, + "grad_norm": 2.1606507301330566, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7386207580566406, + "num_tokens": 605654859.0, + "step": 23413 + }, + { + "epoch": 2.571271688996266, + "grad_norm": 2.5150060653686523, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7519105672836304, + "num_tokens": 605677494.0, + "step": 23414 + }, + { + "epoch": 2.57138150669888, + "grad_norm": 2.271066904067993, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7243945598602295, + "num_tokens": 605703942.0, + "step": 23415 + }, + { + "epoch": 2.5714913244014936, + "grad_norm": 2.487555503845215, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7084456086158752, + "num_tokens": 605728956.0, + "step": 23416 + }, + { + "epoch": 2.5716011421041074, + "grad_norm": 2.4634649753570557, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.7511407136917114, + "num_tokens": 605752376.0, + "step": 23417 + }, + { + "epoch": 2.5717109598067207, + "grad_norm": 2.4658374786376953, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7261699438095093, + "num_tokens": 605776295.0, + "step": 23418 + }, + { + "epoch": 2.5718207775093345, + "grad_norm": 2.5527377128601074, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7554842233657837, + "num_tokens": 605798727.0, + "step": 23419 + }, + { + "epoch": 2.5719305952119482, + "grad_norm": 2.1873905658721924, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7063392400741577, + "num_tokens": 605830369.0, + "step": 23420 + }, + { + "epoch": 2.572040412914562, + "grad_norm": 2.5000205039978027, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7317337989807129, + "num_tokens": 605853381.0, + "step": 23421 + }, + { + "epoch": 2.5721502306171757, + "grad_norm": 2.2740116119384766, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7226696014404297, + "num_tokens": 605881397.0, + "step": 23422 + }, + { + "epoch": 2.572260048319789, + "grad_norm": 2.4986062049865723, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7287808656692505, + "num_tokens": 605905674.0, + "step": 23423 + }, + { + "epoch": 2.572369866022403, + "grad_norm": 2.400178909301758, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7280653715133667, + "num_tokens": 605931293.0, + "step": 23424 + }, + { + "epoch": 2.5724796837250166, + "grad_norm": 2.868149995803833, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7314729690551758, + "num_tokens": 605949840.0, + "step": 23425 + }, + { + "epoch": 2.57258950142763, + "grad_norm": 2.7541797161102295, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7544077634811401, + "num_tokens": 605969918.0, + "step": 23426 + }, + { + "epoch": 2.5726993191302436, + "grad_norm": 2.3876986503601074, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7055704593658447, + "num_tokens": 605997400.0, + "step": 23427 + }, + { + "epoch": 2.5728091368328574, + "grad_norm": 2.4304959774017334, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7139888405799866, + "num_tokens": 606022978.0, + "step": 23428 + }, + { + "epoch": 2.572918954535471, + "grad_norm": 2.6490728855133057, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7655321359634399, + "num_tokens": 606043132.0, + "step": 23429 + }, + { + "epoch": 2.573028772238085, + "grad_norm": 2.473531723022461, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7127981185913086, + "num_tokens": 606070287.0, + "step": 23430 + }, + { + "epoch": 2.573138589940698, + "grad_norm": 2.1639533042907715, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7220233678817749, + "num_tokens": 606100701.0, + "step": 23431 + }, + { + "epoch": 2.573248407643312, + "grad_norm": 2.2922396659851074, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7342106103897095, + "num_tokens": 606128149.0, + "step": 23432 + }, + { + "epoch": 2.5733582253459257, + "grad_norm": 2.6084933280944824, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7215508818626404, + "num_tokens": 606150533.0, + "step": 23433 + }, + { + "epoch": 2.5734680430485395, + "grad_norm": 2.749826192855835, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7694406509399414, + "num_tokens": 606169482.0, + "step": 23434 + }, + { + "epoch": 2.5735778607511532, + "grad_norm": 2.309999942779541, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7387281060218811, + "num_tokens": 606195832.0, + "step": 23435 + }, + { + "epoch": 2.5736876784537666, + "grad_norm": 2.183976650238037, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7317382097244263, + "num_tokens": 606225471.0, + "step": 23436 + }, + { + "epoch": 2.5737974961563803, + "grad_norm": 2.3420300483703613, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7156988382339478, + "num_tokens": 606251218.0, + "step": 23437 + }, + { + "epoch": 2.573907313858994, + "grad_norm": 1.9920849800109863, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7114952802658081, + "num_tokens": 606287133.0, + "step": 23438 + }, + { + "epoch": 2.574017131561608, + "grad_norm": 2.039600133895874, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7154936790466309, + "num_tokens": 606319512.0, + "step": 23439 + }, + { + "epoch": 2.5741269492642216, + "grad_norm": 2.0786163806915283, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.731922447681427, + "num_tokens": 606351132.0, + "step": 23440 + }, + { + "epoch": 2.574236766966835, + "grad_norm": 2.2371394634246826, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7674118280410767, + "num_tokens": 606377231.0, + "step": 23441 + }, + { + "epoch": 2.5743465846694487, + "grad_norm": 2.434932231903076, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7254589796066284, + "num_tokens": 606401545.0, + "step": 23442 + }, + { + "epoch": 2.5744564023720624, + "grad_norm": 2.302384614944458, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7447391152381897, + "num_tokens": 606426396.0, + "step": 23443 + }, + { + "epoch": 2.574566220074676, + "grad_norm": 2.2923424243927, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7391670942306519, + "num_tokens": 606455352.0, + "step": 23444 + }, + { + "epoch": 2.57467603777729, + "grad_norm": 2.539475917816162, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7380527853965759, + "num_tokens": 606478255.0, + "step": 23445 + }, + { + "epoch": 2.5747858554799032, + "grad_norm": 2.5723307132720947, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7191563844680786, + "num_tokens": 606502966.0, + "step": 23446 + }, + { + "epoch": 2.574895673182517, + "grad_norm": 2.8624277114868164, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7403851747512817, + "num_tokens": 606520752.0, + "step": 23447 + }, + { + "epoch": 2.5750054908851308, + "grad_norm": 2.154744863510132, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7390236258506775, + "num_tokens": 606549079.0, + "step": 23448 + }, + { + "epoch": 2.575115308587744, + "grad_norm": 2.502673864364624, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7362093925476074, + "num_tokens": 606571157.0, + "step": 23449 + }, + { + "epoch": 2.5752251262903583, + "grad_norm": 2.3577945232391357, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7229543924331665, + "num_tokens": 606596777.0, + "step": 23450 + }, + { + "epoch": 2.5753349439929716, + "grad_norm": 2.5206706523895264, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7346914410591125, + "num_tokens": 606619862.0, + "step": 23451 + }, + { + "epoch": 2.5754447616955853, + "grad_norm": 2.5586001873016357, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7338534593582153, + "num_tokens": 606643526.0, + "step": 23452 + }, + { + "epoch": 2.575554579398199, + "grad_norm": 2.6556954383850098, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7440984845161438, + "num_tokens": 606665192.0, + "step": 23453 + }, + { + "epoch": 2.5756643971008124, + "grad_norm": 2.269289255142212, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7099822759628296, + "num_tokens": 606694666.0, + "step": 23454 + }, + { + "epoch": 2.575774214803426, + "grad_norm": 2.3015034198760986, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7166067361831665, + "num_tokens": 606721178.0, + "step": 23455 + }, + { + "epoch": 2.57588403250604, + "grad_norm": 2.167043924331665, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7319923639297485, + "num_tokens": 606751290.0, + "step": 23456 + }, + { + "epoch": 2.5759938502086537, + "grad_norm": 2.373964309692383, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7220457196235657, + "num_tokens": 606777418.0, + "step": 23457 + }, + { + "epoch": 2.5761036679112674, + "grad_norm": 2.4046943187713623, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7353618144989014, + "num_tokens": 606802705.0, + "step": 23458 + }, + { + "epoch": 2.5762134856138807, + "grad_norm": 2.3952229022979736, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7409024238586426, + "num_tokens": 606826176.0, + "step": 23459 + }, + { + "epoch": 2.5763233033164945, + "grad_norm": 2.435288429260254, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7562966346740723, + "num_tokens": 606851494.0, + "step": 23460 + }, + { + "epoch": 2.5764331210191083, + "grad_norm": 2.18057918548584, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7198702692985535, + "num_tokens": 606880711.0, + "step": 23461 + }, + { + "epoch": 2.576542938721722, + "grad_norm": 2.251547336578369, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7129346132278442, + "num_tokens": 606909157.0, + "step": 23462 + }, + { + "epoch": 2.5766527564243358, + "grad_norm": 2.4957077503204346, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7236969470977783, + "num_tokens": 606934181.0, + "step": 23463 + }, + { + "epoch": 2.576762574126949, + "grad_norm": 2.1970417499542236, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7011430263519287, + "num_tokens": 606967102.0, + "step": 23464 + }, + { + "epoch": 2.576872391829563, + "grad_norm": 2.143878221511841, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7439175844192505, + "num_tokens": 606994772.0, + "step": 23465 + }, + { + "epoch": 2.5769822095321766, + "grad_norm": 2.2526755332946777, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7081731557846069, + "num_tokens": 607024344.0, + "step": 23466 + }, + { + "epoch": 2.5770920272347904, + "grad_norm": 2.605189800262451, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7336525917053223, + "num_tokens": 607045444.0, + "step": 23467 + }, + { + "epoch": 2.577201844937404, + "grad_norm": 2.2469863891601562, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7454859018325806, + "num_tokens": 607072488.0, + "step": 23468 + }, + { + "epoch": 2.5773116626400174, + "grad_norm": 2.5356101989746094, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7189357280731201, + "num_tokens": 607095180.0, + "step": 23469 + }, + { + "epoch": 2.577421480342631, + "grad_norm": 2.4939072132110596, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7473840713500977, + "num_tokens": 607118068.0, + "step": 23470 + }, + { + "epoch": 2.577531298045245, + "grad_norm": 2.1626992225646973, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.725665271282196, + "num_tokens": 607151100.0, + "step": 23471 + }, + { + "epoch": 2.5776411157478587, + "grad_norm": 2.4417898654937744, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7264708876609802, + "num_tokens": 607176109.0, + "step": 23472 + }, + { + "epoch": 2.5777509334504725, + "grad_norm": 2.525604724884033, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7448117136955261, + "num_tokens": 607200271.0, + "step": 23473 + }, + { + "epoch": 2.5778607511530858, + "grad_norm": 2.1366629600524902, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.714330792427063, + "num_tokens": 607231627.0, + "step": 23474 + }, + { + "epoch": 2.5779705688556995, + "grad_norm": 2.194112539291382, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7332879304885864, + "num_tokens": 607261567.0, + "step": 23475 + }, + { + "epoch": 2.5780803865583133, + "grad_norm": 2.5210416316986084, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7408028841018677, + "num_tokens": 607283953.0, + "step": 23476 + }, + { + "epoch": 2.5781902042609266, + "grad_norm": 2.1522576808929443, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7009099125862122, + "num_tokens": 607314055.0, + "step": 23477 + }, + { + "epoch": 2.578300021963541, + "grad_norm": 2.245992422103882, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7389176487922668, + "num_tokens": 607341826.0, + "step": 23478 + }, + { + "epoch": 2.578409839666154, + "grad_norm": 2.4995532035827637, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7464181184768677, + "num_tokens": 607362328.0, + "step": 23479 + }, + { + "epoch": 2.578519657368768, + "grad_norm": 2.6517765522003174, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7348995208740234, + "num_tokens": 607385401.0, + "step": 23480 + }, + { + "epoch": 2.5786294750713816, + "grad_norm": 2.256969690322876, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.733292818069458, + "num_tokens": 607411585.0, + "step": 23481 + }, + { + "epoch": 2.578739292773995, + "grad_norm": 2.4081056118011475, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7227206230163574, + "num_tokens": 607438405.0, + "step": 23482 + }, + { + "epoch": 2.5788491104766087, + "grad_norm": 2.3809332847595215, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7429772019386292, + "num_tokens": 607464513.0, + "step": 23483 + }, + { + "epoch": 2.5789589281792225, + "grad_norm": 2.394753932952881, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7417674660682678, + "num_tokens": 607489036.0, + "step": 23484 + }, + { + "epoch": 2.579068745881836, + "grad_norm": 2.132204294204712, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7236971855163574, + "num_tokens": 607518788.0, + "step": 23485 + }, + { + "epoch": 2.57917856358445, + "grad_norm": 2.248162269592285, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7415459156036377, + "num_tokens": 607545199.0, + "step": 23486 + }, + { + "epoch": 2.5792883812870633, + "grad_norm": 2.319397449493408, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7143658399581909, + "num_tokens": 607571705.0, + "step": 23487 + }, + { + "epoch": 2.579398198989677, + "grad_norm": 2.200742483139038, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7453525066375732, + "num_tokens": 607600658.0, + "step": 23488 + }, + { + "epoch": 2.579508016692291, + "grad_norm": 2.5344295501708984, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.748160719871521, + "num_tokens": 607622523.0, + "step": 23489 + }, + { + "epoch": 2.5796178343949046, + "grad_norm": 2.2475762367248535, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7511028051376343, + "num_tokens": 607649235.0, + "step": 23490 + }, + { + "epoch": 2.5797276520975183, + "grad_norm": 2.3504374027252197, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7383350133895874, + "num_tokens": 607674183.0, + "step": 23491 + }, + { + "epoch": 2.5798374698001316, + "grad_norm": 2.5513429641723633, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7197402715682983, + "num_tokens": 607697653.0, + "step": 23492 + }, + { + "epoch": 2.5799472875027454, + "grad_norm": 2.4310078620910645, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7341137528419495, + "num_tokens": 607720883.0, + "step": 23493 + }, + { + "epoch": 2.580057105205359, + "grad_norm": 2.623318910598755, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.7434128522872925, + "num_tokens": 607742152.0, + "step": 23494 + }, + { + "epoch": 2.580166922907973, + "grad_norm": 2.3301024436950684, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7326436042785645, + "num_tokens": 607768335.0, + "step": 23495 + }, + { + "epoch": 2.5802767406105866, + "grad_norm": 2.2691047191619873, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.725467324256897, + "num_tokens": 607795566.0, + "step": 23496 + }, + { + "epoch": 2.5803865583132, + "grad_norm": 2.242567539215088, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7140666246414185, + "num_tokens": 607825518.0, + "step": 23497 + }, + { + "epoch": 2.5804963760158137, + "grad_norm": 2.363104820251465, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7408764958381653, + "num_tokens": 607850631.0, + "step": 23498 + }, + { + "epoch": 2.5806061937184275, + "grad_norm": 2.1637895107269287, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.702042818069458, + "num_tokens": 607880161.0, + "step": 23499 + }, + { + "epoch": 2.5807160114210412, + "grad_norm": 1.9886986017227173, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7092989683151245, + "num_tokens": 607912535.0, + "step": 23500 + }, + { + "epoch": 2.580825829123655, + "grad_norm": 1.9570674896240234, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7283079624176025, + "num_tokens": 607949154.0, + "step": 23501 + }, + { + "epoch": 2.5809356468262683, + "grad_norm": 2.278536558151245, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7156999111175537, + "num_tokens": 607978259.0, + "step": 23502 + }, + { + "epoch": 2.581045464528882, + "grad_norm": 2.5859246253967285, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.727032482624054, + "num_tokens": 607999828.0, + "step": 23503 + }, + { + "epoch": 2.581155282231496, + "grad_norm": 2.4483683109283447, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7243267297744751, + "num_tokens": 608026370.0, + "step": 23504 + }, + { + "epoch": 2.581265099934109, + "grad_norm": 2.327820301055908, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7227351069450378, + "num_tokens": 608052186.0, + "step": 23505 + }, + { + "epoch": 2.581374917636723, + "grad_norm": 2.7020161151885986, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7523810863494873, + "num_tokens": 608071921.0, + "step": 23506 + }, + { + "epoch": 2.5814847353393366, + "grad_norm": 2.694742441177368, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7360852956771851, + "num_tokens": 608094605.0, + "step": 23507 + }, + { + "epoch": 2.5815945530419504, + "grad_norm": 2.2686517238616943, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.703857421875, + "num_tokens": 608121527.0, + "step": 23508 + }, + { + "epoch": 2.581704370744564, + "grad_norm": 2.3976831436157227, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7247676849365234, + "num_tokens": 608148738.0, + "step": 23509 + }, + { + "epoch": 2.5818141884471775, + "grad_norm": 2.2704317569732666, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7172384262084961, + "num_tokens": 608178076.0, + "step": 23510 + }, + { + "epoch": 2.5819240061497912, + "grad_norm": 2.557687520980835, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7521148920059204, + "num_tokens": 608200723.0, + "step": 23511 + }, + { + "epoch": 2.582033823852405, + "grad_norm": 2.5578317642211914, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7388861179351807, + "num_tokens": 608223552.0, + "step": 23512 + }, + { + "epoch": 2.5821436415550187, + "grad_norm": 2.350865364074707, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7199157476425171, + "num_tokens": 608248956.0, + "step": 23513 + }, + { + "epoch": 2.5822534592576325, + "grad_norm": 2.0929861068725586, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7290878295898438, + "num_tokens": 608281144.0, + "step": 23514 + }, + { + "epoch": 2.582363276960246, + "grad_norm": 2.917410373687744, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7419096827507019, + "num_tokens": 608298919.0, + "step": 23515 + }, + { + "epoch": 2.5824730946628596, + "grad_norm": 2.784074068069458, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7624821066856384, + "num_tokens": 608317932.0, + "step": 23516 + }, + { + "epoch": 2.5825829123654733, + "grad_norm": 2.4138567447662354, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7329748868942261, + "num_tokens": 608343042.0, + "step": 23517 + }, + { + "epoch": 2.582692730068087, + "grad_norm": 2.344555139541626, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7600256204605103, + "num_tokens": 608366394.0, + "step": 23518 + }, + { + "epoch": 2.582802547770701, + "grad_norm": 2.5451860427856445, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7199902534484863, + "num_tokens": 608391877.0, + "step": 23519 + }, + { + "epoch": 2.582912365473314, + "grad_norm": 2.1940720081329346, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7137162685394287, + "num_tokens": 608422930.0, + "step": 23520 + }, + { + "epoch": 2.583022183175928, + "grad_norm": 2.2137346267700195, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7003998160362244, + "num_tokens": 608452170.0, + "step": 23521 + }, + { + "epoch": 2.5831320008785417, + "grad_norm": 2.5632989406585693, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7489303350448608, + "num_tokens": 608474138.0, + "step": 23522 + }, + { + "epoch": 2.5832418185811554, + "grad_norm": 2.311185836791992, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7473410964012146, + "num_tokens": 608500710.0, + "step": 23523 + }, + { + "epoch": 2.583351636283769, + "grad_norm": 2.350862503051758, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.745947003364563, + "num_tokens": 608526318.0, + "step": 23524 + }, + { + "epoch": 2.5834614539863825, + "grad_norm": 2.515124797821045, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.736653208732605, + "num_tokens": 608549646.0, + "step": 23525 + }, + { + "epoch": 2.5835712716889963, + "grad_norm": 2.37129807472229, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7129449844360352, + "num_tokens": 608576690.0, + "step": 23526 + }, + { + "epoch": 2.58368108939161, + "grad_norm": 2.2307610511779785, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7213915586471558, + "num_tokens": 608605824.0, + "step": 23527 + }, + { + "epoch": 2.5837909070942233, + "grad_norm": 2.3644661903381348, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.731138288974762, + "num_tokens": 608631410.0, + "step": 23528 + }, + { + "epoch": 2.5839007247968375, + "grad_norm": 2.3305795192718506, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7279091477394104, + "num_tokens": 608659695.0, + "step": 23529 + }, + { + "epoch": 2.584010542499451, + "grad_norm": 2.3931593894958496, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.722373366355896, + "num_tokens": 608684659.0, + "step": 23530 + }, + { + "epoch": 2.5841203602020646, + "grad_norm": 2.329551935195923, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7319443821907043, + "num_tokens": 608712998.0, + "step": 23531 + }, + { + "epoch": 2.5842301779046783, + "grad_norm": 2.0999491214752197, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7226358652114868, + "num_tokens": 608742565.0, + "step": 23532 + }, + { + "epoch": 2.5843399956072917, + "grad_norm": 2.163764238357544, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7349025011062622, + "num_tokens": 608772595.0, + "step": 23533 + }, + { + "epoch": 2.5844498133099054, + "grad_norm": 2.3210291862487793, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.716000497341156, + "num_tokens": 608800380.0, + "step": 23534 + }, + { + "epoch": 2.584559631012519, + "grad_norm": 2.330068588256836, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7110918760299683, + "num_tokens": 608826483.0, + "step": 23535 + }, + { + "epoch": 2.584669448715133, + "grad_norm": 2.4434022903442383, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.72315514087677, + "num_tokens": 608851677.0, + "step": 23536 + }, + { + "epoch": 2.5847792664177467, + "grad_norm": 2.385683059692383, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7252880334854126, + "num_tokens": 608877458.0, + "step": 23537 + }, + { + "epoch": 2.58488908412036, + "grad_norm": 2.4079508781433105, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7476669549942017, + "num_tokens": 608903584.0, + "step": 23538 + }, + { + "epoch": 2.5849989018229738, + "grad_norm": 2.2632768154144287, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7283092141151428, + "num_tokens": 608931746.0, + "step": 23539 + }, + { + "epoch": 2.5851087195255875, + "grad_norm": 2.2194042205810547, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7026450037956238, + "num_tokens": 608960365.0, + "step": 23540 + }, + { + "epoch": 2.5852185372282013, + "grad_norm": 2.1899304389953613, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7248079776763916, + "num_tokens": 608990885.0, + "step": 23541 + }, + { + "epoch": 2.585328354930815, + "grad_norm": 2.1758010387420654, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7280128598213196, + "num_tokens": 609018899.0, + "step": 23542 + }, + { + "epoch": 2.5854381726334283, + "grad_norm": 2.4596786499023438, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7237297296524048, + "num_tokens": 609042820.0, + "step": 23543 + }, + { + "epoch": 2.585547990336042, + "grad_norm": 2.1671345233917236, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7304027080535889, + "num_tokens": 609071626.0, + "step": 23544 + }, + { + "epoch": 2.585657808038656, + "grad_norm": 2.6049482822418213, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7258359789848328, + "num_tokens": 609094227.0, + "step": 23545 + }, + { + "epoch": 2.5857676257412696, + "grad_norm": 2.1749117374420166, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7274298667907715, + "num_tokens": 609125138.0, + "step": 23546 + }, + { + "epoch": 2.5858774434438834, + "grad_norm": 2.295983076095581, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7507643103599548, + "num_tokens": 609153763.0, + "step": 23547 + }, + { + "epoch": 2.5859872611464967, + "grad_norm": 2.328948497772217, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7213037014007568, + "num_tokens": 609181468.0, + "step": 23548 + }, + { + "epoch": 2.5860970788491104, + "grad_norm": 2.4944417476654053, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7508666515350342, + "num_tokens": 609203204.0, + "step": 23549 + }, + { + "epoch": 2.586206896551724, + "grad_norm": 2.501060724258423, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7442250847816467, + "num_tokens": 609226339.0, + "step": 23550 + }, + { + "epoch": 2.586316714254338, + "grad_norm": 2.463578701019287, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7249821424484253, + "num_tokens": 609251273.0, + "step": 23551 + }, + { + "epoch": 2.5864265319569517, + "grad_norm": 2.5317800045013428, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7263115644454956, + "num_tokens": 609275377.0, + "step": 23552 + }, + { + "epoch": 2.586536349659565, + "grad_norm": 2.4093759059906006, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7294808626174927, + "num_tokens": 609302709.0, + "step": 23553 + }, + { + "epoch": 2.586646167362179, + "grad_norm": 2.476386785507202, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7142741084098816, + "num_tokens": 609328684.0, + "step": 23554 + }, + { + "epoch": 2.5867559850647925, + "grad_norm": 2.2215354442596436, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.6962825655937195, + "num_tokens": 609358995.0, + "step": 23555 + }, + { + "epoch": 2.586865802767406, + "grad_norm": 2.400538444519043, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7171125411987305, + "num_tokens": 609383638.0, + "step": 23556 + }, + { + "epoch": 2.5869756204700196, + "grad_norm": 2.5183773040771484, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.718449592590332, + "num_tokens": 609406471.0, + "step": 23557 + }, + { + "epoch": 2.5870854381726334, + "grad_norm": 2.3553028106689453, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7141379117965698, + "num_tokens": 609432105.0, + "step": 23558 + }, + { + "epoch": 2.587195255875247, + "grad_norm": 2.2980964183807373, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7141722440719604, + "num_tokens": 609460303.0, + "step": 23559 + }, + { + "epoch": 2.587305073577861, + "grad_norm": 2.5199007987976074, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7116038799285889, + "num_tokens": 609482805.0, + "step": 23560 + }, + { + "epoch": 2.587414891280474, + "grad_norm": 2.518428325653076, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7402850389480591, + "num_tokens": 609508378.0, + "step": 23561 + }, + { + "epoch": 2.587524708983088, + "grad_norm": 2.423482656478882, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7301079034805298, + "num_tokens": 609532991.0, + "step": 23562 + }, + { + "epoch": 2.5876345266857017, + "grad_norm": 2.151562213897705, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7081655263900757, + "num_tokens": 609562786.0, + "step": 23563 + }, + { + "epoch": 2.5877443443883155, + "grad_norm": 2.511667490005493, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7594447135925293, + "num_tokens": 609584798.0, + "step": 23564 + }, + { + "epoch": 2.587854162090929, + "grad_norm": 2.1477785110473633, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7325323820114136, + "num_tokens": 609613866.0, + "step": 23565 + }, + { + "epoch": 2.5879639797935425, + "grad_norm": 2.364434003829956, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7654881477355957, + "num_tokens": 609637725.0, + "step": 23566 + }, + { + "epoch": 2.5880737974961563, + "grad_norm": 2.1850404739379883, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7349411845207214, + "num_tokens": 609666176.0, + "step": 23567 + }, + { + "epoch": 2.58818361519877, + "grad_norm": 2.5570664405822754, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7435169219970703, + "num_tokens": 609688773.0, + "step": 23568 + }, + { + "epoch": 2.588293432901384, + "grad_norm": 2.5267841815948486, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7362816333770752, + "num_tokens": 609713472.0, + "step": 23569 + }, + { + "epoch": 2.5884032506039976, + "grad_norm": 2.300523042678833, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7472938299179077, + "num_tokens": 609740360.0, + "step": 23570 + }, + { + "epoch": 2.588513068306611, + "grad_norm": 2.477492570877075, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7433916330337524, + "num_tokens": 609767501.0, + "step": 23571 + }, + { + "epoch": 2.5886228860092246, + "grad_norm": 2.494410753250122, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.738018274307251, + "num_tokens": 609791624.0, + "step": 23572 + }, + { + "epoch": 2.5887327037118384, + "grad_norm": 2.731900453567505, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7553115487098694, + "num_tokens": 609811024.0, + "step": 23573 + }, + { + "epoch": 2.588842521414452, + "grad_norm": 2.3188517093658447, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7412662506103516, + "num_tokens": 609835938.0, + "step": 23574 + }, + { + "epoch": 2.588952339117066, + "grad_norm": 2.1724255084991455, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7320874929428101, + "num_tokens": 609865852.0, + "step": 23575 + }, + { + "epoch": 2.589062156819679, + "grad_norm": 2.2334983348846436, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7255873680114746, + "num_tokens": 609892294.0, + "step": 23576 + }, + { + "epoch": 2.589171974522293, + "grad_norm": 2.252920389175415, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7321345806121826, + "num_tokens": 609918345.0, + "step": 23577 + }, + { + "epoch": 2.5892817922249067, + "grad_norm": 2.5111937522888184, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.726065993309021, + "num_tokens": 609943155.0, + "step": 23578 + }, + { + "epoch": 2.58939160992752, + "grad_norm": 2.319415330886841, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7586601972579956, + "num_tokens": 609967669.0, + "step": 23579 + }, + { + "epoch": 2.5895014276301342, + "grad_norm": 2.056408643722534, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7231308221817017, + "num_tokens": 609999713.0, + "step": 23580 + }, + { + "epoch": 2.5896112453327476, + "grad_norm": 2.128804922103882, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7153691053390503, + "num_tokens": 610029806.0, + "step": 23581 + }, + { + "epoch": 2.5897210630353613, + "grad_norm": 2.129762649536133, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7082636952400208, + "num_tokens": 610060118.0, + "step": 23582 + }, + { + "epoch": 2.589830880737975, + "grad_norm": 2.37772536277771, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7208738923072815, + "num_tokens": 610086735.0, + "step": 23583 + }, + { + "epoch": 2.5899406984405884, + "grad_norm": 2.1925504207611084, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7094115614891052, + "num_tokens": 610115401.0, + "step": 23584 + }, + { + "epoch": 2.590050516143202, + "grad_norm": 2.511270761489868, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7430406808853149, + "num_tokens": 610138102.0, + "step": 23585 + }, + { + "epoch": 2.590160333845816, + "grad_norm": 2.139910936355591, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7216694355010986, + "num_tokens": 610169734.0, + "step": 23586 + }, + { + "epoch": 2.5902701515484297, + "grad_norm": 2.3818726539611816, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7280990481376648, + "num_tokens": 610194219.0, + "step": 23587 + }, + { + "epoch": 2.5903799692510434, + "grad_norm": 2.431025266647339, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7084956765174866, + "num_tokens": 610218464.0, + "step": 23588 + }, + { + "epoch": 2.5904897869536567, + "grad_norm": 2.424520969390869, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.73684161901474, + "num_tokens": 610244318.0, + "step": 23589 + }, + { + "epoch": 2.5905996046562705, + "grad_norm": 2.6468095779418945, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7310683727264404, + "num_tokens": 610265950.0, + "step": 23590 + }, + { + "epoch": 2.5907094223588842, + "grad_norm": 2.543064832687378, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7316036224365234, + "num_tokens": 610289450.0, + "step": 23591 + }, + { + "epoch": 2.590819240061498, + "grad_norm": 2.191483974456787, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7282275557518005, + "num_tokens": 610318546.0, + "step": 23592 + }, + { + "epoch": 2.5909290577641118, + "grad_norm": 2.384526252746582, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7304826378822327, + "num_tokens": 610344578.0, + "step": 23593 + }, + { + "epoch": 2.591038875466725, + "grad_norm": 2.5574471950531006, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7467020750045776, + "num_tokens": 610368562.0, + "step": 23594 + }, + { + "epoch": 2.591148693169339, + "grad_norm": 2.426358461380005, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7319743633270264, + "num_tokens": 610393880.0, + "step": 23595 + }, + { + "epoch": 2.5912585108719526, + "grad_norm": 2.1011464595794678, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.6978499889373779, + "num_tokens": 610427393.0, + "step": 23596 + }, + { + "epoch": 2.5913683285745663, + "grad_norm": 2.684678077697754, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7474298477172852, + "num_tokens": 610449326.0, + "step": 23597 + }, + { + "epoch": 2.59147814627718, + "grad_norm": 2.367697238922119, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7172740697860718, + "num_tokens": 610477099.0, + "step": 23598 + }, + { + "epoch": 2.5915879639797934, + "grad_norm": 2.5038106441497803, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.718132495880127, + "num_tokens": 610502104.0, + "step": 23599 + }, + { + "epoch": 2.591697781682407, + "grad_norm": 2.483670711517334, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7248870134353638, + "num_tokens": 610526704.0, + "step": 23600 + }, + { + "epoch": 2.591807599385021, + "grad_norm": 2.3817763328552246, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7348837852478027, + "num_tokens": 610552280.0, + "step": 23601 + }, + { + "epoch": 2.5919174170876347, + "grad_norm": 2.941516399383545, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7519638538360596, + "num_tokens": 610570278.0, + "step": 23602 + }, + { + "epoch": 2.5920272347902484, + "grad_norm": 2.380643606185913, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7211400270462036, + "num_tokens": 610594914.0, + "step": 23603 + }, + { + "epoch": 2.5921370524928617, + "grad_norm": 2.4679248332977295, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7468061447143555, + "num_tokens": 610618667.0, + "step": 23604 + }, + { + "epoch": 2.5922468701954755, + "grad_norm": 2.4468283653259277, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.747797429561615, + "num_tokens": 610641748.0, + "step": 23605 + }, + { + "epoch": 2.5923566878980893, + "grad_norm": 1.9252386093139648, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6921469569206238, + "num_tokens": 610678830.0, + "step": 23606 + }, + { + "epoch": 2.5924665056007026, + "grad_norm": 2.174920082092285, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7529523372650146, + "num_tokens": 610706433.0, + "step": 23607 + }, + { + "epoch": 2.5925763233033163, + "grad_norm": 2.3558762073516846, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7295687794685364, + "num_tokens": 610732306.0, + "step": 23608 + }, + { + "epoch": 2.59268614100593, + "grad_norm": 2.5174593925476074, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7404276132583618, + "num_tokens": 610755381.0, + "step": 23609 + }, + { + "epoch": 2.592795958708544, + "grad_norm": 2.5349621772766113, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7113332748413086, + "num_tokens": 610779631.0, + "step": 23610 + }, + { + "epoch": 2.5929057764111576, + "grad_norm": 2.428105354309082, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7020881175994873, + "num_tokens": 610807265.0, + "step": 23611 + }, + { + "epoch": 2.593015594113771, + "grad_norm": 2.0959935188293457, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7358012199401855, + "num_tokens": 610837654.0, + "step": 23612 + }, + { + "epoch": 2.5931254118163847, + "grad_norm": 2.2355377674102783, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7436889410018921, + "num_tokens": 610863735.0, + "step": 23613 + }, + { + "epoch": 2.5932352295189984, + "grad_norm": 2.1333889961242676, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7306509613990784, + "num_tokens": 610894086.0, + "step": 23614 + }, + { + "epoch": 2.593345047221612, + "grad_norm": 2.5769078731536865, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7360092401504517, + "num_tokens": 610916455.0, + "step": 23615 + }, + { + "epoch": 2.593454864924226, + "grad_norm": 2.605879306793213, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7252488136291504, + "num_tokens": 610939189.0, + "step": 23616 + }, + { + "epoch": 2.5935646826268393, + "grad_norm": 2.490220785140991, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7446357011795044, + "num_tokens": 610962545.0, + "step": 23617 + }, + { + "epoch": 2.593674500329453, + "grad_norm": 2.97660756111145, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7405940294265747, + "num_tokens": 610979802.0, + "step": 23618 + }, + { + "epoch": 2.5937843180320668, + "grad_norm": 2.2616047859191895, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7287768125534058, + "num_tokens": 611007790.0, + "step": 23619 + }, + { + "epoch": 2.5938941357346805, + "grad_norm": 2.680246591567993, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7269294261932373, + "num_tokens": 611028766.0, + "step": 23620 + }, + { + "epoch": 2.5940039534372943, + "grad_norm": 2.229090690612793, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7149468660354614, + "num_tokens": 611060140.0, + "step": 23621 + }, + { + "epoch": 2.5941137711399076, + "grad_norm": 2.400885581970215, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7373092174530029, + "num_tokens": 611084267.0, + "step": 23622 + }, + { + "epoch": 2.5942235888425214, + "grad_norm": 2.398836612701416, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7171213030815125, + "num_tokens": 611110761.0, + "step": 23623 + }, + { + "epoch": 2.594333406545135, + "grad_norm": 2.4714457988739014, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7042112350463867, + "num_tokens": 611136743.0, + "step": 23624 + }, + { + "epoch": 2.594443224247749, + "grad_norm": 2.392617702484131, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7312638759613037, + "num_tokens": 611162697.0, + "step": 23625 + }, + { + "epoch": 2.5945530419503626, + "grad_norm": 2.286343812942505, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7235156297683716, + "num_tokens": 611190809.0, + "step": 23626 + }, + { + "epoch": 2.594662859652976, + "grad_norm": 2.337510108947754, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.700355052947998, + "num_tokens": 611219321.0, + "step": 23627 + }, + { + "epoch": 2.5947726773555897, + "grad_norm": 2.6993398666381836, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7371792197227478, + "num_tokens": 611240407.0, + "step": 23628 + }, + { + "epoch": 2.5948824950582035, + "grad_norm": 2.2115955352783203, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7234631180763245, + "num_tokens": 611269693.0, + "step": 23629 + }, + { + "epoch": 2.594992312760817, + "grad_norm": 2.3923068046569824, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7073253393173218, + "num_tokens": 611294073.0, + "step": 23630 + }, + { + "epoch": 2.595102130463431, + "grad_norm": 2.252114772796631, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7207430601119995, + "num_tokens": 611322370.0, + "step": 23631 + }, + { + "epoch": 2.5952119481660443, + "grad_norm": 2.413742780685425, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7222172021865845, + "num_tokens": 611346265.0, + "step": 23632 + }, + { + "epoch": 2.595321765868658, + "grad_norm": 2.42364764213562, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7354271411895752, + "num_tokens": 611371607.0, + "step": 23633 + }, + { + "epoch": 2.595431583571272, + "grad_norm": 2.3158059120178223, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7474010586738586, + "num_tokens": 611399678.0, + "step": 23634 + }, + { + "epoch": 2.595541401273885, + "grad_norm": 2.389378547668457, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7266247868537903, + "num_tokens": 611428886.0, + "step": 23635 + }, + { + "epoch": 2.595651218976499, + "grad_norm": 2.6066596508026123, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7267129421234131, + "num_tokens": 611451301.0, + "step": 23636 + }, + { + "epoch": 2.5957610366791126, + "grad_norm": 2.2102653980255127, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7303991317749023, + "num_tokens": 611478880.0, + "step": 23637 + }, + { + "epoch": 2.5958708543817264, + "grad_norm": 2.2400460243225098, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7271445989608765, + "num_tokens": 611506458.0, + "step": 23638 + }, + { + "epoch": 2.59598067208434, + "grad_norm": 2.325615406036377, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7470362782478333, + "num_tokens": 611532584.0, + "step": 23639 + }, + { + "epoch": 2.5960904897869534, + "grad_norm": 1.9377195835113525, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7335251569747925, + "num_tokens": 611568513.0, + "step": 23640 + }, + { + "epoch": 2.596200307489567, + "grad_norm": 2.593402147293091, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7121884822845459, + "num_tokens": 611595791.0, + "step": 23641 + }, + { + "epoch": 2.596310125192181, + "grad_norm": 2.3438212871551514, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7161870002746582, + "num_tokens": 611623092.0, + "step": 23642 + }, + { + "epoch": 2.5964199428947947, + "grad_norm": 2.469857692718506, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.738578736782074, + "num_tokens": 611646677.0, + "step": 23643 + }, + { + "epoch": 2.5965297605974085, + "grad_norm": 2.2992427349090576, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7394310235977173, + "num_tokens": 611672403.0, + "step": 23644 + }, + { + "epoch": 2.596639578300022, + "grad_norm": 2.225119113922119, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7326594591140747, + "num_tokens": 611702719.0, + "step": 23645 + }, + { + "epoch": 2.5967493960026355, + "grad_norm": 1.9284361600875854, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7079497575759888, + "num_tokens": 611740960.0, + "step": 23646 + }, + { + "epoch": 2.5968592137052493, + "grad_norm": 2.253089427947998, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7312275171279907, + "num_tokens": 611770401.0, + "step": 23647 + }, + { + "epoch": 2.596969031407863, + "grad_norm": 2.5451877117156982, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7496635913848877, + "num_tokens": 611791834.0, + "step": 23648 + }, + { + "epoch": 2.597078849110477, + "grad_norm": 2.460122585296631, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7289930582046509, + "num_tokens": 611818006.0, + "step": 23649 + }, + { + "epoch": 2.59718866681309, + "grad_norm": 2.7937941551208496, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.735797643661499, + "num_tokens": 611836610.0, + "step": 23650 + }, + { + "epoch": 2.597298484515704, + "grad_norm": 2.13916015625, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7173289656639099, + "num_tokens": 611868349.0, + "step": 23651 + }, + { + "epoch": 2.5974083022183176, + "grad_norm": 2.6486144065856934, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7542370557785034, + "num_tokens": 611888622.0, + "step": 23652 + }, + { + "epoch": 2.5975181199209314, + "grad_norm": 2.0374791622161865, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7467654943466187, + "num_tokens": 611919472.0, + "step": 23653 + }, + { + "epoch": 2.597627937623545, + "grad_norm": 2.104623317718506, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7286903262138367, + "num_tokens": 611949302.0, + "step": 23654 + }, + { + "epoch": 2.5977377553261585, + "grad_norm": 2.2498960494995117, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7356635928153992, + "num_tokens": 611977106.0, + "step": 23655 + }, + { + "epoch": 2.5978475730287722, + "grad_norm": 2.3898632526397705, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.732626736164093, + "num_tokens": 612002776.0, + "step": 23656 + }, + { + "epoch": 2.597957390731386, + "grad_norm": 2.3758809566497803, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7155642509460449, + "num_tokens": 612028747.0, + "step": 23657 + }, + { + "epoch": 2.5980672084339993, + "grad_norm": 2.7110328674316406, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7466724514961243, + "num_tokens": 612051398.0, + "step": 23658 + }, + { + "epoch": 2.5981770261366135, + "grad_norm": 2.5126543045043945, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7380549311637878, + "num_tokens": 612076770.0, + "step": 23659 + }, + { + "epoch": 2.598286843839227, + "grad_norm": 2.420576810836792, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7414687871932983, + "num_tokens": 612100720.0, + "step": 23660 + }, + { + "epoch": 2.5983966615418406, + "grad_norm": 2.5221898555755615, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7520153522491455, + "num_tokens": 612124654.0, + "step": 23661 + }, + { + "epoch": 2.5985064792444543, + "grad_norm": 2.614255428314209, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7592201232910156, + "num_tokens": 612144913.0, + "step": 23662 + }, + { + "epoch": 2.5986162969470676, + "grad_norm": 2.28751277923584, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7537384033203125, + "num_tokens": 612171849.0, + "step": 23663 + }, + { + "epoch": 2.5987261146496814, + "grad_norm": 2.1224594116210938, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7096855640411377, + "num_tokens": 612203713.0, + "step": 23664 + }, + { + "epoch": 2.598835932352295, + "grad_norm": 2.1761014461517334, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7288535237312317, + "num_tokens": 612232019.0, + "step": 23665 + }, + { + "epoch": 2.598945750054909, + "grad_norm": 2.671069622039795, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.730202853679657, + "num_tokens": 612254444.0, + "step": 23666 + }, + { + "epoch": 2.5990555677575227, + "grad_norm": 2.322571039199829, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7515032291412354, + "num_tokens": 612280856.0, + "step": 23667 + }, + { + "epoch": 2.599165385460136, + "grad_norm": 2.4723026752471924, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7253066301345825, + "num_tokens": 612304793.0, + "step": 23668 + }, + { + "epoch": 2.5992752031627497, + "grad_norm": 2.3767786026000977, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7152503728866577, + "num_tokens": 612330615.0, + "step": 23669 + }, + { + "epoch": 2.5993850208653635, + "grad_norm": 2.350958824157715, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7450372576713562, + "num_tokens": 612356540.0, + "step": 23670 + }, + { + "epoch": 2.5994948385679773, + "grad_norm": 2.6033148765563965, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7397027611732483, + "num_tokens": 612378341.0, + "step": 23671 + }, + { + "epoch": 2.599604656270591, + "grad_norm": 2.255418300628662, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7222639322280884, + "num_tokens": 612409321.0, + "step": 23672 + }, + { + "epoch": 2.5997144739732043, + "grad_norm": 2.3389716148376465, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7090542912483215, + "num_tokens": 612436828.0, + "step": 23673 + }, + { + "epoch": 2.599824291675818, + "grad_norm": 2.4339680671691895, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7267113327980042, + "num_tokens": 612461194.0, + "step": 23674 + }, + { + "epoch": 2.599934109378432, + "grad_norm": 2.587087869644165, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7293964624404907, + "num_tokens": 612484439.0, + "step": 23675 + }, + { + "epoch": 2.6000439270810456, + "grad_norm": 2.30023455619812, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7357414364814758, + "num_tokens": 612510719.0, + "step": 23676 + }, + { + "epoch": 2.6001537447836593, + "grad_norm": 2.5444390773773193, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7420994639396667, + "num_tokens": 612531870.0, + "step": 23677 + }, + { + "epoch": 2.6002635624862727, + "grad_norm": 2.337183952331543, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7435377836227417, + "num_tokens": 612559693.0, + "step": 23678 + }, + { + "epoch": 2.6003733801888864, + "grad_norm": 2.4999983310699463, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7356365919113159, + "num_tokens": 612582916.0, + "step": 23679 + }, + { + "epoch": 2.6004831978915, + "grad_norm": 2.5950961112976074, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7349115014076233, + "num_tokens": 612607430.0, + "step": 23680 + }, + { + "epoch": 2.600593015594114, + "grad_norm": 2.4193053245544434, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7319836020469666, + "num_tokens": 612632203.0, + "step": 23681 + }, + { + "epoch": 2.6007028332967277, + "grad_norm": 2.2249326705932617, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7047649025917053, + "num_tokens": 612663563.0, + "step": 23682 + }, + { + "epoch": 2.600812650999341, + "grad_norm": 2.5686609745025635, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7354782819747925, + "num_tokens": 612687162.0, + "step": 23683 + }, + { + "epoch": 2.6009224687019548, + "grad_norm": 2.740882635116577, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7293449640274048, + "num_tokens": 612708150.0, + "step": 23684 + }, + { + "epoch": 2.6010322864045685, + "grad_norm": 2.4883627891540527, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7153831720352173, + "num_tokens": 612732273.0, + "step": 23685 + }, + { + "epoch": 2.601142104107182, + "grad_norm": 2.1210577487945557, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7229134440422058, + "num_tokens": 612762367.0, + "step": 23686 + }, + { + "epoch": 2.6012519218097956, + "grad_norm": 2.775346279144287, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7421809434890747, + "num_tokens": 612784479.0, + "step": 23687 + }, + { + "epoch": 2.6013617395124093, + "grad_norm": 2.485236406326294, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7554491758346558, + "num_tokens": 612807784.0, + "step": 23688 + }, + { + "epoch": 2.601471557215023, + "grad_norm": 2.6107640266418457, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7388569712638855, + "num_tokens": 612830441.0, + "step": 23689 + }, + { + "epoch": 2.601581374917637, + "grad_norm": 2.044684410095215, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7528884410858154, + "num_tokens": 612860287.0, + "step": 23690 + }, + { + "epoch": 2.60169119262025, + "grad_norm": 2.4993066787719727, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7238821983337402, + "num_tokens": 612885853.0, + "step": 23691 + }, + { + "epoch": 2.601801010322864, + "grad_norm": 2.3351101875305176, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7238543033599854, + "num_tokens": 612913970.0, + "step": 23692 + }, + { + "epoch": 2.6019108280254777, + "grad_norm": 2.555049180984497, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.7453837394714355, + "num_tokens": 612936848.0, + "step": 23693 + }, + { + "epoch": 2.6020206457280914, + "grad_norm": 2.301434278488159, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7271007299423218, + "num_tokens": 612962683.0, + "step": 23694 + }, + { + "epoch": 2.602130463430705, + "grad_norm": 2.2114408016204834, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7552578449249268, + "num_tokens": 612990744.0, + "step": 23695 + }, + { + "epoch": 2.6022402811333185, + "grad_norm": 2.292710304260254, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7414333820343018, + "num_tokens": 613017058.0, + "step": 23696 + }, + { + "epoch": 2.6023500988359323, + "grad_norm": 2.609825849533081, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7510181665420532, + "num_tokens": 613038577.0, + "step": 23697 + }, + { + "epoch": 2.602459916538546, + "grad_norm": 2.255967378616333, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7186660170555115, + "num_tokens": 613067846.0, + "step": 23698 + }, + { + "epoch": 2.60256973424116, + "grad_norm": 2.700075387954712, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7252085208892822, + "num_tokens": 613089819.0, + "step": 23699 + }, + { + "epoch": 2.6026795519437735, + "grad_norm": 2.6310670375823975, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7312750816345215, + "num_tokens": 613111268.0, + "step": 23700 + }, + { + "epoch": 2.602789369646387, + "grad_norm": 2.428842544555664, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.737028956413269, + "num_tokens": 613138421.0, + "step": 23701 + }, + { + "epoch": 2.6028991873490006, + "grad_norm": 2.4959607124328613, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7097258567810059, + "num_tokens": 613163254.0, + "step": 23702 + }, + { + "epoch": 2.6030090050516144, + "grad_norm": 2.1035854816436768, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7239636182785034, + "num_tokens": 613196287.0, + "step": 23703 + }, + { + "epoch": 2.603118822754228, + "grad_norm": 2.5712273120880127, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7212649583816528, + "num_tokens": 613220044.0, + "step": 23704 + }, + { + "epoch": 2.603228640456842, + "grad_norm": 2.613799810409546, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7190473079681396, + "num_tokens": 613241739.0, + "step": 23705 + }, + { + "epoch": 2.603338458159455, + "grad_norm": 2.3647449016571045, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7061099410057068, + "num_tokens": 613269674.0, + "step": 23706 + }, + { + "epoch": 2.603448275862069, + "grad_norm": 2.2729923725128174, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7277594804763794, + "num_tokens": 613299200.0, + "step": 23707 + }, + { + "epoch": 2.6035580935646827, + "grad_norm": 2.214449405670166, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.74924236536026, + "num_tokens": 613326806.0, + "step": 23708 + }, + { + "epoch": 2.603667911267296, + "grad_norm": 2.037729501724243, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7039721012115479, + "num_tokens": 613363218.0, + "step": 23709 + }, + { + "epoch": 2.60377772896991, + "grad_norm": 2.1535377502441406, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7538643479347229, + "num_tokens": 613392091.0, + "step": 23710 + }, + { + "epoch": 2.6038875466725235, + "grad_norm": 2.41902756690979, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7512618899345398, + "num_tokens": 613415949.0, + "step": 23711 + }, + { + "epoch": 2.6039973643751373, + "grad_norm": 2.274078845977783, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7216285467147827, + "num_tokens": 613442712.0, + "step": 23712 + }, + { + "epoch": 2.604107182077751, + "grad_norm": 2.4275879859924316, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.722437858581543, + "num_tokens": 613468117.0, + "step": 23713 + }, + { + "epoch": 2.6042169997803644, + "grad_norm": 2.1571390628814697, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7077335715293884, + "num_tokens": 613497539.0, + "step": 23714 + }, + { + "epoch": 2.604326817482978, + "grad_norm": 2.3336520195007324, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7391937971115112, + "num_tokens": 613522873.0, + "step": 23715 + }, + { + "epoch": 2.604436635185592, + "grad_norm": 2.361717462539673, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7036449909210205, + "num_tokens": 613549805.0, + "step": 23716 + }, + { + "epoch": 2.6045464528882056, + "grad_norm": 2.313234329223633, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7173371315002441, + "num_tokens": 613576634.0, + "step": 23717 + }, + { + "epoch": 2.6046562705908194, + "grad_norm": 2.534386396408081, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7262272834777832, + "num_tokens": 613600814.0, + "step": 23718 + }, + { + "epoch": 2.6047660882934327, + "grad_norm": 2.3467471599578857, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7281057238578796, + "num_tokens": 613626305.0, + "step": 23719 + }, + { + "epoch": 2.6048759059960465, + "grad_norm": 2.2634668350219727, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7391259670257568, + "num_tokens": 613654566.0, + "step": 23720 + }, + { + "epoch": 2.60498572369866, + "grad_norm": 2.3256328105926514, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7427452802658081, + "num_tokens": 613678702.0, + "step": 23721 + }, + { + "epoch": 2.605095541401274, + "grad_norm": 2.5529093742370605, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7028267979621887, + "num_tokens": 613705791.0, + "step": 23722 + }, + { + "epoch": 2.6052053591038877, + "grad_norm": 2.258758544921875, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7209295630455017, + "num_tokens": 613734905.0, + "step": 23723 + }, + { + "epoch": 2.605315176806501, + "grad_norm": 2.482839584350586, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7501544952392578, + "num_tokens": 613758211.0, + "step": 23724 + }, + { + "epoch": 2.605424994509115, + "grad_norm": 2.713148355484009, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7236262559890747, + "num_tokens": 613778836.0, + "step": 23725 + }, + { + "epoch": 2.6055348122117286, + "grad_norm": 2.650649309158325, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7265734076499939, + "num_tokens": 613801624.0, + "step": 23726 + }, + { + "epoch": 2.6056446299143423, + "grad_norm": 2.3451321125030518, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7287479043006897, + "num_tokens": 613829644.0, + "step": 23727 + }, + { + "epoch": 2.605754447616956, + "grad_norm": 2.44547438621521, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.738397479057312, + "num_tokens": 613853936.0, + "step": 23728 + }, + { + "epoch": 2.6058642653195694, + "grad_norm": 2.684673309326172, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7512087821960449, + "num_tokens": 613873765.0, + "step": 23729 + }, + { + "epoch": 2.605974083022183, + "grad_norm": 2.1298739910125732, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7264037132263184, + "num_tokens": 613901446.0, + "step": 23730 + }, + { + "epoch": 2.606083900724797, + "grad_norm": 2.415341377258301, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.740067720413208, + "num_tokens": 613926484.0, + "step": 23731 + }, + { + "epoch": 2.6061937184274107, + "grad_norm": 2.335108757019043, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7281440496444702, + "num_tokens": 613955178.0, + "step": 23732 + }, + { + "epoch": 2.6063035361300244, + "grad_norm": 2.1337876319885254, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7183074355125427, + "num_tokens": 613986262.0, + "step": 23733 + }, + { + "epoch": 2.6064133538326377, + "grad_norm": 2.8097615242004395, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7566079497337341, + "num_tokens": 614004905.0, + "step": 23734 + }, + { + "epoch": 2.6065231715352515, + "grad_norm": 2.5277936458587646, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7389405965805054, + "num_tokens": 614028283.0, + "step": 23735 + }, + { + "epoch": 2.6066329892378652, + "grad_norm": 2.918213367462158, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7439508438110352, + "num_tokens": 614046571.0, + "step": 23736 + }, + { + "epoch": 2.6067428069404786, + "grad_norm": 2.7595272064208984, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7455881834030151, + "num_tokens": 614067079.0, + "step": 23737 + }, + { + "epoch": 2.6068526246430923, + "grad_norm": 2.457218885421753, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7238421440124512, + "num_tokens": 614092766.0, + "step": 23738 + }, + { + "epoch": 2.606962442345706, + "grad_norm": 2.573578119277954, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7375403642654419, + "num_tokens": 614115213.0, + "step": 23739 + }, + { + "epoch": 2.60707226004832, + "grad_norm": 2.686723470687866, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7399694919586182, + "num_tokens": 614137069.0, + "step": 23740 + }, + { + "epoch": 2.6071820777509336, + "grad_norm": 2.751235246658325, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7563711404800415, + "num_tokens": 614160859.0, + "step": 23741 + }, + { + "epoch": 2.607291895453547, + "grad_norm": 2.2567496299743652, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7243194580078125, + "num_tokens": 614188900.0, + "step": 23742 + }, + { + "epoch": 2.6074017131561606, + "grad_norm": 2.3699724674224854, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7303500175476074, + "num_tokens": 614217035.0, + "step": 23743 + }, + { + "epoch": 2.6075115308587744, + "grad_norm": 2.5514614582061768, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7288005352020264, + "num_tokens": 614239744.0, + "step": 23744 + }, + { + "epoch": 2.607621348561388, + "grad_norm": 2.291482448577881, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7356427907943726, + "num_tokens": 614266888.0, + "step": 23745 + }, + { + "epoch": 2.607731166264002, + "grad_norm": 2.2077383995056152, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7125585079193115, + "num_tokens": 614295459.0, + "step": 23746 + }, + { + "epoch": 2.6078409839666152, + "grad_norm": 2.393075466156006, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7422699332237244, + "num_tokens": 614319657.0, + "step": 23747 + }, + { + "epoch": 2.607950801669229, + "grad_norm": 2.4983956813812256, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7327433824539185, + "num_tokens": 614343494.0, + "step": 23748 + }, + { + "epoch": 2.6080606193718427, + "grad_norm": 2.267301559448242, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7315291166305542, + "num_tokens": 614371395.0, + "step": 23749 + }, + { + "epoch": 2.6081704370744565, + "grad_norm": 2.4002301692962646, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7350703477859497, + "num_tokens": 614398280.0, + "step": 23750 + }, + { + "epoch": 2.6082802547770703, + "grad_norm": 2.434586524963379, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7378377914428711, + "num_tokens": 614423501.0, + "step": 23751 + }, + { + "epoch": 2.6083900724796836, + "grad_norm": 2.793454170227051, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7514376640319824, + "num_tokens": 614443447.0, + "step": 23752 + }, + { + "epoch": 2.6084998901822973, + "grad_norm": 2.391063928604126, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7124929428100586, + "num_tokens": 614470168.0, + "step": 23753 + }, + { + "epoch": 2.608609707884911, + "grad_norm": 2.2203352451324463, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.733128547668457, + "num_tokens": 614496569.0, + "step": 23754 + }, + { + "epoch": 2.608719525587525, + "grad_norm": 2.4123713970184326, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7208264470100403, + "num_tokens": 614522567.0, + "step": 23755 + }, + { + "epoch": 2.6088293432901386, + "grad_norm": 2.404839277267456, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7241365909576416, + "num_tokens": 614547755.0, + "step": 23756 + }, + { + "epoch": 2.608939160992752, + "grad_norm": 2.563854932785034, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7200130224227905, + "num_tokens": 614574353.0, + "step": 23757 + }, + { + "epoch": 2.6090489786953657, + "grad_norm": 2.397348165512085, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7362238168716431, + "num_tokens": 614600494.0, + "step": 23758 + }, + { + "epoch": 2.6091587963979794, + "grad_norm": 2.593559980392456, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7282471060752869, + "num_tokens": 614624797.0, + "step": 23759 + }, + { + "epoch": 2.6092686141005927, + "grad_norm": 2.2616026401519775, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.6993581056594849, + "num_tokens": 614655913.0, + "step": 23760 + }, + { + "epoch": 2.609378431803207, + "grad_norm": 2.5207204818725586, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7509957551956177, + "num_tokens": 614679388.0, + "step": 23761 + }, + { + "epoch": 2.6094882495058203, + "grad_norm": 2.1467056274414062, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7412113547325134, + "num_tokens": 614709117.0, + "step": 23762 + }, + { + "epoch": 2.609598067208434, + "grad_norm": 2.919896125793457, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.751343846321106, + "num_tokens": 614727989.0, + "step": 23763 + }, + { + "epoch": 2.6097078849110478, + "grad_norm": 2.5529847145080566, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7434523105621338, + "num_tokens": 614750503.0, + "step": 23764 + }, + { + "epoch": 2.609817702613661, + "grad_norm": 2.5619513988494873, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7297253012657166, + "num_tokens": 614772555.0, + "step": 23765 + }, + { + "epoch": 2.609927520316275, + "grad_norm": 2.403310775756836, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7168251276016235, + "num_tokens": 614798477.0, + "step": 23766 + }, + { + "epoch": 2.6100373380188886, + "grad_norm": 2.3723433017730713, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7435659766197205, + "num_tokens": 614821899.0, + "step": 23767 + }, + { + "epoch": 2.6101471557215024, + "grad_norm": 2.2288408279418945, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7388201951980591, + "num_tokens": 614853929.0, + "step": 23768 + }, + { + "epoch": 2.610256973424116, + "grad_norm": 2.2291479110717773, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7511821985244751, + "num_tokens": 614880210.0, + "step": 23769 + }, + { + "epoch": 2.6103667911267294, + "grad_norm": 2.4708640575408936, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7522679567337036, + "num_tokens": 614903738.0, + "step": 23770 + }, + { + "epoch": 2.610476608829343, + "grad_norm": 2.2101988792419434, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7557065486907959, + "num_tokens": 614929478.0, + "step": 23771 + }, + { + "epoch": 2.610586426531957, + "grad_norm": 2.271923542022705, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7251431941986084, + "num_tokens": 614955346.0, + "step": 23772 + }, + { + "epoch": 2.6106962442345707, + "grad_norm": 2.2166526317596436, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7384928464889526, + "num_tokens": 614982587.0, + "step": 23773 + }, + { + "epoch": 2.6108060619371845, + "grad_norm": 2.4562957286834717, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.71946120262146, + "num_tokens": 615006632.0, + "step": 23774 + }, + { + "epoch": 2.6109158796397978, + "grad_norm": 2.3139126300811768, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.733216404914856, + "num_tokens": 615035959.0, + "step": 23775 + }, + { + "epoch": 2.6110256973424115, + "grad_norm": 2.2288565635681152, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7136551737785339, + "num_tokens": 615064681.0, + "step": 23776 + }, + { + "epoch": 2.6111355150450253, + "grad_norm": 2.3772168159484863, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7313770651817322, + "num_tokens": 615090058.0, + "step": 23777 + }, + { + "epoch": 2.611245332747639, + "grad_norm": 2.612884998321533, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7437991499900818, + "num_tokens": 615115131.0, + "step": 23778 + }, + { + "epoch": 2.611355150450253, + "grad_norm": 2.4382598400115967, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7545180320739746, + "num_tokens": 615137660.0, + "step": 23779 + }, + { + "epoch": 2.611464968152866, + "grad_norm": 2.2785117626190186, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7401463985443115, + "num_tokens": 615164764.0, + "step": 23780 + }, + { + "epoch": 2.61157478585548, + "grad_norm": 2.2567970752716064, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7352365851402283, + "num_tokens": 615193518.0, + "step": 23781 + }, + { + "epoch": 2.6116846035580936, + "grad_norm": 2.5756311416625977, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.749311089515686, + "num_tokens": 615214833.0, + "step": 23782 + }, + { + "epoch": 2.6117944212607074, + "grad_norm": 2.2821462154388428, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7364895343780518, + "num_tokens": 615240146.0, + "step": 23783 + }, + { + "epoch": 2.611904238963321, + "grad_norm": 2.2949585914611816, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7363276481628418, + "num_tokens": 615266486.0, + "step": 23784 + }, + { + "epoch": 2.6120140566659344, + "grad_norm": 2.2355401515960693, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7316620349884033, + "num_tokens": 615295792.0, + "step": 23785 + }, + { + "epoch": 2.612123874368548, + "grad_norm": 2.232064962387085, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7645040154457092, + "num_tokens": 615322940.0, + "step": 23786 + }, + { + "epoch": 2.612233692071162, + "grad_norm": 2.6221864223480225, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7375447750091553, + "num_tokens": 615344774.0, + "step": 23787 + }, + { + "epoch": 2.6123435097737753, + "grad_norm": 2.6658287048339844, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7441774606704712, + "num_tokens": 615365924.0, + "step": 23788 + }, + { + "epoch": 2.612453327476389, + "grad_norm": 2.4660308361053467, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7367197871208191, + "num_tokens": 615390288.0, + "step": 23789 + }, + { + "epoch": 2.612563145179003, + "grad_norm": 2.4720373153686523, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7542476058006287, + "num_tokens": 615413048.0, + "step": 23790 + }, + { + "epoch": 2.6126729628816165, + "grad_norm": 2.3425066471099854, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.7445321679115295, + "num_tokens": 615436740.0, + "step": 23791 + }, + { + "epoch": 2.6127827805842303, + "grad_norm": 2.237555742263794, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7358640432357788, + "num_tokens": 615462034.0, + "step": 23792 + }, + { + "epoch": 2.6128925982868436, + "grad_norm": 2.3657565116882324, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7260501384735107, + "num_tokens": 615488700.0, + "step": 23793 + }, + { + "epoch": 2.6130024159894574, + "grad_norm": 2.4867615699768066, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7335036993026733, + "num_tokens": 615513606.0, + "step": 23794 + }, + { + "epoch": 2.613112233692071, + "grad_norm": 2.4858436584472656, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7294608950614929, + "num_tokens": 615535227.0, + "step": 23795 + }, + { + "epoch": 2.613222051394685, + "grad_norm": 2.2362236976623535, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7172703742980957, + "num_tokens": 615562831.0, + "step": 23796 + }, + { + "epoch": 2.6133318690972986, + "grad_norm": 2.2357771396636963, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7060552835464478, + "num_tokens": 615592648.0, + "step": 23797 + }, + { + "epoch": 2.613441686799912, + "grad_norm": 2.590766191482544, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7334891557693481, + "num_tokens": 615614363.0, + "step": 23798 + }, + { + "epoch": 2.6135515045025257, + "grad_norm": 2.108882188796997, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.702727198600769, + "num_tokens": 615647265.0, + "step": 23799 + }, + { + "epoch": 2.6136613222051395, + "grad_norm": 2.44459867477417, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7505444288253784, + "num_tokens": 615671285.0, + "step": 23800 + }, + { + "epoch": 2.6137711399077532, + "grad_norm": 2.2551429271698, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7150266170501709, + "num_tokens": 615704263.0, + "step": 23801 + }, + { + "epoch": 2.613880957610367, + "grad_norm": 2.4342947006225586, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7307707667350769, + "num_tokens": 615728820.0, + "step": 23802 + }, + { + "epoch": 2.6139907753129803, + "grad_norm": 2.1377222537994385, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7173868417739868, + "num_tokens": 615758594.0, + "step": 23803 + }, + { + "epoch": 2.614100593015594, + "grad_norm": 2.228257417678833, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7071460485458374, + "num_tokens": 615789841.0, + "step": 23804 + }, + { + "epoch": 2.614210410718208, + "grad_norm": 2.568082809448242, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7290851473808289, + "num_tokens": 615811739.0, + "step": 23805 + }, + { + "epoch": 2.6143202284208216, + "grad_norm": 2.1304099559783936, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7207183241844177, + "num_tokens": 615843106.0, + "step": 23806 + }, + { + "epoch": 2.6144300461234353, + "grad_norm": 2.242734909057617, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7407463192939758, + "num_tokens": 615870852.0, + "step": 23807 + }, + { + "epoch": 2.6145398638260486, + "grad_norm": 2.471334457397461, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7592068314552307, + "num_tokens": 615893427.0, + "step": 23808 + }, + { + "epoch": 2.6146496815286624, + "grad_norm": 2.2201828956604004, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7081088423728943, + "num_tokens": 615922853.0, + "step": 23809 + }, + { + "epoch": 2.614759499231276, + "grad_norm": 2.6205387115478516, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7296757102012634, + "num_tokens": 615944594.0, + "step": 23810 + }, + { + "epoch": 2.61486931693389, + "grad_norm": 2.5326197147369385, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7372047305107117, + "num_tokens": 615968025.0, + "step": 23811 + }, + { + "epoch": 2.6149791346365037, + "grad_norm": 2.4501137733459473, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7123140096664429, + "num_tokens": 615996177.0, + "step": 23812 + }, + { + "epoch": 2.615088952339117, + "grad_norm": 2.273391008377075, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7298015356063843, + "num_tokens": 616023342.0, + "step": 23813 + }, + { + "epoch": 2.6151987700417307, + "grad_norm": 2.0666160583496094, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7293255925178528, + "num_tokens": 616052964.0, + "step": 23814 + }, + { + "epoch": 2.6153085877443445, + "grad_norm": 2.5076351165771484, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7235620021820068, + "num_tokens": 616078639.0, + "step": 23815 + }, + { + "epoch": 2.615418405446958, + "grad_norm": 2.097200393676758, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.737921953201294, + "num_tokens": 616108430.0, + "step": 23816 + }, + { + "epoch": 2.6155282231495716, + "grad_norm": 2.263540029525757, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7450723648071289, + "num_tokens": 616132651.0, + "step": 23817 + }, + { + "epoch": 2.6156380408521853, + "grad_norm": 2.288764476776123, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.713271975517273, + "num_tokens": 616160477.0, + "step": 23818 + }, + { + "epoch": 2.615747858554799, + "grad_norm": 2.751751661300659, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7401891946792603, + "num_tokens": 616181366.0, + "step": 23819 + }, + { + "epoch": 2.615857676257413, + "grad_norm": 2.701422691345215, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7254019975662231, + "num_tokens": 616203339.0, + "step": 23820 + }, + { + "epoch": 2.615967493960026, + "grad_norm": 2.4268743991851807, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7456023097038269, + "num_tokens": 616226918.0, + "step": 23821 + }, + { + "epoch": 2.61607731166264, + "grad_norm": 2.3736093044281006, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7120292782783508, + "num_tokens": 616253500.0, + "step": 23822 + }, + { + "epoch": 2.6161871293652537, + "grad_norm": 2.5476346015930176, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7520493268966675, + "num_tokens": 616275443.0, + "step": 23823 + }, + { + "epoch": 2.6162969470678674, + "grad_norm": 2.4760661125183105, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7315866351127625, + "num_tokens": 616298483.0, + "step": 23824 + }, + { + "epoch": 2.616406764770481, + "grad_norm": 2.554271697998047, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.730345606803894, + "num_tokens": 616321924.0, + "step": 23825 + }, + { + "epoch": 2.6165165824730945, + "grad_norm": 2.3389036655426025, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7294163703918457, + "num_tokens": 616349063.0, + "step": 23826 + }, + { + "epoch": 2.6166264001757082, + "grad_norm": 2.2484629154205322, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7259498834609985, + "num_tokens": 616375379.0, + "step": 23827 + }, + { + "epoch": 2.616736217878322, + "grad_norm": 2.3550100326538086, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7227858901023865, + "num_tokens": 616401438.0, + "step": 23828 + }, + { + "epoch": 2.6168460355809358, + "grad_norm": 2.2053072452545166, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7318214178085327, + "num_tokens": 616430323.0, + "step": 23829 + }, + { + "epoch": 2.6169558532835495, + "grad_norm": 2.2742884159088135, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.744032621383667, + "num_tokens": 616457365.0, + "step": 23830 + }, + { + "epoch": 2.617065670986163, + "grad_norm": 2.382157325744629, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7380025386810303, + "num_tokens": 616484086.0, + "step": 23831 + }, + { + "epoch": 2.6171754886887766, + "grad_norm": 2.1243226528167725, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7157031893730164, + "num_tokens": 616515260.0, + "step": 23832 + }, + { + "epoch": 2.6172853063913903, + "grad_norm": 2.069654703140259, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7229750752449036, + "num_tokens": 616545524.0, + "step": 23833 + }, + { + "epoch": 2.617395124094004, + "grad_norm": 2.438201904296875, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7346392869949341, + "num_tokens": 616570629.0, + "step": 23834 + }, + { + "epoch": 2.617504941796618, + "grad_norm": 2.1696248054504395, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7290963530540466, + "num_tokens": 616600138.0, + "step": 23835 + }, + { + "epoch": 2.617614759499231, + "grad_norm": 2.752819061279297, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.717336893081665, + "num_tokens": 616621489.0, + "step": 23836 + }, + { + "epoch": 2.617724577201845, + "grad_norm": 2.7589664459228516, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7403124570846558, + "num_tokens": 616642458.0, + "step": 23837 + }, + { + "epoch": 2.6178343949044587, + "grad_norm": 2.393925666809082, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.739807665348053, + "num_tokens": 616668748.0, + "step": 23838 + }, + { + "epoch": 2.617944212607072, + "grad_norm": 2.136284112930298, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7310240864753723, + "num_tokens": 616700767.0, + "step": 23839 + }, + { + "epoch": 2.618054030309686, + "grad_norm": 2.4181203842163086, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7494255304336548, + "num_tokens": 616725784.0, + "step": 23840 + }, + { + "epoch": 2.6181638480122995, + "grad_norm": 2.505202293395996, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7520109415054321, + "num_tokens": 616749963.0, + "step": 23841 + }, + { + "epoch": 2.6182736657149133, + "grad_norm": 2.4451215267181396, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7324600219726562, + "num_tokens": 616773116.0, + "step": 23842 + }, + { + "epoch": 2.618383483417527, + "grad_norm": 2.4314513206481934, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7015032768249512, + "num_tokens": 616799662.0, + "step": 23843 + }, + { + "epoch": 2.6184933011201403, + "grad_norm": 2.094961166381836, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7261996269226074, + "num_tokens": 616832181.0, + "step": 23844 + }, + { + "epoch": 2.618603118822754, + "grad_norm": 2.480334997177124, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.768231987953186, + "num_tokens": 616856032.0, + "step": 23845 + }, + { + "epoch": 2.618712936525368, + "grad_norm": 2.3739213943481445, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7241036891937256, + "num_tokens": 616882922.0, + "step": 23846 + }, + { + "epoch": 2.6188227542279816, + "grad_norm": 2.5898821353912354, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7362067103385925, + "num_tokens": 616905003.0, + "step": 23847 + }, + { + "epoch": 2.6189325719305954, + "grad_norm": 2.2631189823150635, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7272414565086365, + "num_tokens": 616934182.0, + "step": 23848 + }, + { + "epoch": 2.6190423896332087, + "grad_norm": 2.35422682762146, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7139959335327148, + "num_tokens": 616962792.0, + "step": 23849 + }, + { + "epoch": 2.6191522073358224, + "grad_norm": 2.205636978149414, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7226969599723816, + "num_tokens": 616993715.0, + "step": 23850 + }, + { + "epoch": 2.619262025038436, + "grad_norm": 2.5130574703216553, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7400046586990356, + "num_tokens": 617016131.0, + "step": 23851 + }, + { + "epoch": 2.61937184274105, + "grad_norm": 2.034121513366699, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7186679840087891, + "num_tokens": 617047681.0, + "step": 23852 + }, + { + "epoch": 2.6194816604436637, + "grad_norm": 2.248135566711426, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7490108013153076, + "num_tokens": 617074011.0, + "step": 23853 + }, + { + "epoch": 2.619591478146277, + "grad_norm": 2.372626304626465, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7425589561462402, + "num_tokens": 617098524.0, + "step": 23854 + }, + { + "epoch": 2.6197012958488908, + "grad_norm": 2.207304000854492, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7251280546188354, + "num_tokens": 617125190.0, + "step": 23855 + }, + { + "epoch": 2.6198111135515045, + "grad_norm": 2.4808473587036133, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7455953359603882, + "num_tokens": 617149480.0, + "step": 23856 + }, + { + "epoch": 2.6199209312541183, + "grad_norm": 2.351436138153076, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7298771739006042, + "num_tokens": 617175751.0, + "step": 23857 + }, + { + "epoch": 2.620030748956732, + "grad_norm": 2.1230502128601074, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7286111116409302, + "num_tokens": 617207187.0, + "step": 23858 + }, + { + "epoch": 2.6201405666593454, + "grad_norm": 2.4988269805908203, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7376021146774292, + "num_tokens": 617230627.0, + "step": 23859 + }, + { + "epoch": 2.620250384361959, + "grad_norm": 2.466731309890747, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.6997274160385132, + "num_tokens": 617255474.0, + "step": 23860 + }, + { + "epoch": 2.620360202064573, + "grad_norm": 2.2574691772460938, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7353107929229736, + "num_tokens": 617283950.0, + "step": 23861 + }, + { + "epoch": 2.6204700197671866, + "grad_norm": 2.234795331954956, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.726394772529602, + "num_tokens": 617311662.0, + "step": 23862 + }, + { + "epoch": 2.6205798374698004, + "grad_norm": 2.2269928455352783, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7476384043693542, + "num_tokens": 617337524.0, + "step": 23863 + }, + { + "epoch": 2.6206896551724137, + "grad_norm": 2.478945732116699, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7310081720352173, + "num_tokens": 617362717.0, + "step": 23864 + }, + { + "epoch": 2.6207994728750275, + "grad_norm": 2.3138575553894043, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7245371341705322, + "num_tokens": 617390882.0, + "step": 23865 + }, + { + "epoch": 2.620909290577641, + "grad_norm": 2.294214963912964, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7137569189071655, + "num_tokens": 617419244.0, + "step": 23866 + }, + { + "epoch": 2.6210191082802545, + "grad_norm": 2.6922223567962646, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7428094148635864, + "num_tokens": 617440822.0, + "step": 23867 + }, + { + "epoch": 2.6211289259828683, + "grad_norm": 2.198740005493164, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7240662574768066, + "num_tokens": 617472444.0, + "step": 23868 + }, + { + "epoch": 2.621238743685482, + "grad_norm": 2.2768023014068604, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.742790937423706, + "num_tokens": 617498360.0, + "step": 23869 + }, + { + "epoch": 2.621348561388096, + "grad_norm": 2.353597402572632, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.733238935470581, + "num_tokens": 617524776.0, + "step": 23870 + }, + { + "epoch": 2.6214583790907096, + "grad_norm": 2.0642685890197754, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7139235734939575, + "num_tokens": 617559133.0, + "step": 23871 + }, + { + "epoch": 2.621568196793323, + "grad_norm": 2.011258840560913, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7205908894538879, + "num_tokens": 617591498.0, + "step": 23872 + }, + { + "epoch": 2.6216780144959366, + "grad_norm": 2.194756507873535, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7300083637237549, + "num_tokens": 617622204.0, + "step": 23873 + }, + { + "epoch": 2.6217878321985504, + "grad_norm": 2.7665646076202393, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7231893539428711, + "num_tokens": 617645255.0, + "step": 23874 + }, + { + "epoch": 2.621897649901164, + "grad_norm": 2.334440231323242, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7434371113777161, + "num_tokens": 617668651.0, + "step": 23875 + }, + { + "epoch": 2.622007467603778, + "grad_norm": 2.395627737045288, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7052925229072571, + "num_tokens": 617693808.0, + "step": 23876 + }, + { + "epoch": 2.622117285306391, + "grad_norm": 2.6258983612060547, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7538761496543884, + "num_tokens": 617714789.0, + "step": 23877 + }, + { + "epoch": 2.622227103009005, + "grad_norm": 2.4490864276885986, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7318998575210571, + "num_tokens": 617740029.0, + "step": 23878 + }, + { + "epoch": 2.6223369207116187, + "grad_norm": 2.5740842819213867, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7462376952171326, + "num_tokens": 617763305.0, + "step": 23879 + }, + { + "epoch": 2.6224467384142325, + "grad_norm": 2.3512728214263916, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7541844844818115, + "num_tokens": 617788975.0, + "step": 23880 + }, + { + "epoch": 2.6225565561168462, + "grad_norm": 2.09997296333313, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7270853519439697, + "num_tokens": 617821155.0, + "step": 23881 + }, + { + "epoch": 2.6226663738194596, + "grad_norm": 2.58264422416687, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7543754577636719, + "num_tokens": 617843279.0, + "step": 23882 + }, + { + "epoch": 2.6227761915220733, + "grad_norm": 2.255176067352295, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7203915119171143, + "num_tokens": 617871053.0, + "step": 23883 + }, + { + "epoch": 2.622886009224687, + "grad_norm": 2.0572662353515625, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7397556900978088, + "num_tokens": 617901241.0, + "step": 23884 + }, + { + "epoch": 2.622995826927301, + "grad_norm": 2.4306490421295166, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7139898538589478, + "num_tokens": 617925776.0, + "step": 23885 + }, + { + "epoch": 2.6231056446299146, + "grad_norm": 2.520014762878418, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7668148279190063, + "num_tokens": 617947516.0, + "step": 23886 + }, + { + "epoch": 2.623215462332528, + "grad_norm": 2.881895065307617, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7440512180328369, + "num_tokens": 617966419.0, + "step": 23887 + }, + { + "epoch": 2.6233252800351416, + "grad_norm": 2.3835980892181396, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7352832555770874, + "num_tokens": 617991109.0, + "step": 23888 + }, + { + "epoch": 2.6234350977377554, + "grad_norm": 2.2498106956481934, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7231435775756836, + "num_tokens": 618020680.0, + "step": 23889 + }, + { + "epoch": 2.6235449154403687, + "grad_norm": 2.4164843559265137, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7489103674888611, + "num_tokens": 618043607.0, + "step": 23890 + }, + { + "epoch": 2.623654733142983, + "grad_norm": 2.1909878253936768, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7293024063110352, + "num_tokens": 618072111.0, + "step": 23891 + }, + { + "epoch": 2.6237645508455962, + "grad_norm": 2.521047830581665, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.727968692779541, + "num_tokens": 618095676.0, + "step": 23892 + }, + { + "epoch": 2.62387436854821, + "grad_norm": 2.231423854827881, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7239959239959717, + "num_tokens": 618122311.0, + "step": 23893 + }, + { + "epoch": 2.6239841862508237, + "grad_norm": 2.4462950229644775, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7383538484573364, + "num_tokens": 618146648.0, + "step": 23894 + }, + { + "epoch": 2.624094003953437, + "grad_norm": 2.5650107860565186, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7501268982887268, + "num_tokens": 618167716.0, + "step": 23895 + }, + { + "epoch": 2.624203821656051, + "grad_norm": 2.2707314491271973, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7349485158920288, + "num_tokens": 618194484.0, + "step": 23896 + }, + { + "epoch": 2.6243136393586646, + "grad_norm": 2.5738685131073, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7402349710464478, + "num_tokens": 618215614.0, + "step": 23897 + }, + { + "epoch": 2.6244234570612783, + "grad_norm": 2.467620372772217, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7221128940582275, + "num_tokens": 618241722.0, + "step": 23898 + }, + { + "epoch": 2.624533274763892, + "grad_norm": 2.324474334716797, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7260250449180603, + "num_tokens": 618270986.0, + "step": 23899 + }, + { + "epoch": 2.6246430924665054, + "grad_norm": 2.3948287963867188, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7368157505989075, + "num_tokens": 618298025.0, + "step": 23900 + }, + { + "epoch": 2.624752910169119, + "grad_norm": 2.6604199409484863, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7294617295265198, + "num_tokens": 618320589.0, + "step": 23901 + }, + { + "epoch": 2.624862727871733, + "grad_norm": 2.4174106121063232, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7579452991485596, + "num_tokens": 618343548.0, + "step": 23902 + }, + { + "epoch": 2.6249725455743467, + "grad_norm": 2.521615505218506, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7552492022514343, + "num_tokens": 618366871.0, + "step": 23903 + }, + { + "epoch": 2.6250823632769604, + "grad_norm": 2.2043755054473877, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7298069596290588, + "num_tokens": 618397442.0, + "step": 23904 + }, + { + "epoch": 2.6251921809795737, + "grad_norm": 2.14105224609375, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7251303791999817, + "num_tokens": 618425474.0, + "step": 23905 + }, + { + "epoch": 2.6253019986821875, + "grad_norm": 2.24349308013916, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7581666111946106, + "num_tokens": 618451059.0, + "step": 23906 + }, + { + "epoch": 2.6254118163848013, + "grad_norm": 2.1708407402038574, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7188496589660645, + "num_tokens": 618481342.0, + "step": 23907 + }, + { + "epoch": 2.625521634087415, + "grad_norm": 2.107703447341919, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7361451983451843, + "num_tokens": 618512365.0, + "step": 23908 + }, + { + "epoch": 2.6256314517900288, + "grad_norm": 2.2225348949432373, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.735531747341156, + "num_tokens": 618540595.0, + "step": 23909 + }, + { + "epoch": 2.625741269492642, + "grad_norm": 2.6020023822784424, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7309671640396118, + "num_tokens": 618562278.0, + "step": 23910 + }, + { + "epoch": 2.625851087195256, + "grad_norm": 2.3868675231933594, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.740120530128479, + "num_tokens": 618587828.0, + "step": 23911 + }, + { + "epoch": 2.6259609048978696, + "grad_norm": 2.2895009517669678, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7166976928710938, + "num_tokens": 618615196.0, + "step": 23912 + }, + { + "epoch": 2.6260707226004834, + "grad_norm": 2.1387734413146973, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7534517645835876, + "num_tokens": 618643927.0, + "step": 23913 + }, + { + "epoch": 2.626180540303097, + "grad_norm": 2.089378595352173, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7124111652374268, + "num_tokens": 618676520.0, + "step": 23914 + }, + { + "epoch": 2.6262903580057104, + "grad_norm": 2.4859726428985596, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7257854342460632, + "num_tokens": 618701042.0, + "step": 23915 + }, + { + "epoch": 2.626400175708324, + "grad_norm": 2.3086581230163574, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.742440938949585, + "num_tokens": 618727378.0, + "step": 23916 + }, + { + "epoch": 2.626509993410938, + "grad_norm": 2.2261202335357666, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7276608347892761, + "num_tokens": 618754896.0, + "step": 23917 + }, + { + "epoch": 2.6266198111135513, + "grad_norm": 2.3890933990478516, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.738713800907135, + "num_tokens": 618778671.0, + "step": 23918 + }, + { + "epoch": 2.626729628816165, + "grad_norm": 2.3082170486450195, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7226024866104126, + "num_tokens": 618804593.0, + "step": 23919 + }, + { + "epoch": 2.6268394465187788, + "grad_norm": 2.4206581115722656, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7321311235427856, + "num_tokens": 618829558.0, + "step": 23920 + }, + { + "epoch": 2.6269492642213925, + "grad_norm": 2.384127616882324, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7312203645706177, + "num_tokens": 618856096.0, + "step": 23921 + }, + { + "epoch": 2.6270590819240063, + "grad_norm": 2.164097547531128, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7243353128433228, + "num_tokens": 618887366.0, + "step": 23922 + }, + { + "epoch": 2.6271688996266196, + "grad_norm": 2.1265130043029785, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7219598293304443, + "num_tokens": 618918351.0, + "step": 23923 + }, + { + "epoch": 2.6272787173292333, + "grad_norm": 2.4737536907196045, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7256293892860413, + "num_tokens": 618944233.0, + "step": 23924 + }, + { + "epoch": 2.627388535031847, + "grad_norm": 2.4375572204589844, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7369886636734009, + "num_tokens": 618968428.0, + "step": 23925 + }, + { + "epoch": 2.627498352734461, + "grad_norm": 2.787196397781372, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7465710639953613, + "num_tokens": 618989435.0, + "step": 23926 + }, + { + "epoch": 2.6276081704370746, + "grad_norm": 2.55196213722229, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7265437841415405, + "num_tokens": 619011294.0, + "step": 23927 + }, + { + "epoch": 2.627717988139688, + "grad_norm": 2.681359052658081, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7407671809196472, + "num_tokens": 619031707.0, + "step": 23928 + }, + { + "epoch": 2.6278278058423017, + "grad_norm": 2.3458211421966553, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7474515438079834, + "num_tokens": 619056051.0, + "step": 23929 + }, + { + "epoch": 2.6279376235449154, + "grad_norm": 2.453075885772705, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7288100123405457, + "num_tokens": 619078179.0, + "step": 23930 + }, + { + "epoch": 2.628047441247529, + "grad_norm": 2.241109848022461, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7299808263778687, + "num_tokens": 619107033.0, + "step": 23931 + }, + { + "epoch": 2.628157258950143, + "grad_norm": 2.261023998260498, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.722132682800293, + "num_tokens": 619134208.0, + "step": 23932 + }, + { + "epoch": 2.6282670766527563, + "grad_norm": 2.400275230407715, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7279144525527954, + "num_tokens": 619160739.0, + "step": 23933 + }, + { + "epoch": 2.62837689435537, + "grad_norm": 2.4859960079193115, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7287358045578003, + "num_tokens": 619182600.0, + "step": 23934 + }, + { + "epoch": 2.628486712057984, + "grad_norm": 2.456897020339966, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7224355340003967, + "num_tokens": 619207172.0, + "step": 23935 + }, + { + "epoch": 2.6285965297605975, + "grad_norm": 2.2995476722717285, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7395056486129761, + "num_tokens": 619234152.0, + "step": 23936 + }, + { + "epoch": 2.6287063474632113, + "grad_norm": 2.100193977355957, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.726494312286377, + "num_tokens": 619264240.0, + "step": 23937 + }, + { + "epoch": 2.6288161651658246, + "grad_norm": 2.2232041358947754, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7272739410400391, + "num_tokens": 619292047.0, + "step": 23938 + }, + { + "epoch": 2.6289259828684384, + "grad_norm": 2.5756242275238037, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7476940751075745, + "num_tokens": 619315084.0, + "step": 23939 + }, + { + "epoch": 2.629035800571052, + "grad_norm": 2.9184250831604004, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7279558777809143, + "num_tokens": 619335129.0, + "step": 23940 + }, + { + "epoch": 2.6291456182736654, + "grad_norm": 2.2714858055114746, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7268987894058228, + "num_tokens": 619362365.0, + "step": 23941 + }, + { + "epoch": 2.6292554359762796, + "grad_norm": 2.4463448524475098, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7246870994567871, + "num_tokens": 619385774.0, + "step": 23942 + }, + { + "epoch": 2.629365253678893, + "grad_norm": 2.3284027576446533, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7640371322631836, + "num_tokens": 619411958.0, + "step": 23943 + }, + { + "epoch": 2.6294750713815067, + "grad_norm": 2.1187939643859863, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7138765454292297, + "num_tokens": 619446957.0, + "step": 23944 + }, + { + "epoch": 2.6295848890841205, + "grad_norm": 2.223126173019409, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7325500249862671, + "num_tokens": 619477566.0, + "step": 23945 + }, + { + "epoch": 2.629694706786734, + "grad_norm": 2.496042251586914, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7415955066680908, + "num_tokens": 619501131.0, + "step": 23946 + }, + { + "epoch": 2.6298045244893475, + "grad_norm": 2.0005459785461426, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7286651134490967, + "num_tokens": 619535315.0, + "step": 23947 + }, + { + "epoch": 2.6299143421919613, + "grad_norm": 2.6188504695892334, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7373087406158447, + "num_tokens": 619556423.0, + "step": 23948 + }, + { + "epoch": 2.630024159894575, + "grad_norm": 2.0854508876800537, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7118074893951416, + "num_tokens": 619587941.0, + "step": 23949 + }, + { + "epoch": 2.630133977597189, + "grad_norm": 2.41377329826355, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7371882200241089, + "num_tokens": 619612379.0, + "step": 23950 + }, + { + "epoch": 2.630243795299802, + "grad_norm": 2.524954080581665, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.745735764503479, + "num_tokens": 619634800.0, + "step": 23951 + }, + { + "epoch": 2.630353613002416, + "grad_norm": 2.2746853828430176, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7337499856948853, + "num_tokens": 619661653.0, + "step": 23952 + }, + { + "epoch": 2.6304634307050296, + "grad_norm": 2.239856243133545, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7337165474891663, + "num_tokens": 619688426.0, + "step": 23953 + }, + { + "epoch": 2.6305732484076434, + "grad_norm": 2.5381741523742676, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.729894757270813, + "num_tokens": 619711954.0, + "step": 23954 + }, + { + "epoch": 2.630683066110257, + "grad_norm": 2.2146685123443604, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.719793975353241, + "num_tokens": 619740104.0, + "step": 23955 + }, + { + "epoch": 2.6307928838128705, + "grad_norm": 2.4063074588775635, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7294877767562866, + "num_tokens": 619763446.0, + "step": 23956 + }, + { + "epoch": 2.630902701515484, + "grad_norm": 2.3756003379821777, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7133581638336182, + "num_tokens": 619789381.0, + "step": 23957 + }, + { + "epoch": 2.631012519218098, + "grad_norm": 2.482861042022705, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7353352904319763, + "num_tokens": 619815071.0, + "step": 23958 + }, + { + "epoch": 2.6311223369207117, + "grad_norm": 2.575167179107666, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7443702220916748, + "num_tokens": 619836300.0, + "step": 23959 + }, + { + "epoch": 2.6312321546233255, + "grad_norm": 2.290377616882324, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7176617383956909, + "num_tokens": 619864827.0, + "step": 23960 + }, + { + "epoch": 2.631341972325939, + "grad_norm": 2.6466526985168457, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7462900280952454, + "num_tokens": 619886514.0, + "step": 23961 + }, + { + "epoch": 2.6314517900285526, + "grad_norm": 2.352266311645508, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7325992584228516, + "num_tokens": 619911601.0, + "step": 23962 + }, + { + "epoch": 2.6315616077311663, + "grad_norm": 2.680161476135254, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.737566351890564, + "num_tokens": 619933542.0, + "step": 23963 + }, + { + "epoch": 2.63167142543378, + "grad_norm": 2.222734212875366, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7337586879730225, + "num_tokens": 619962953.0, + "step": 23964 + }, + { + "epoch": 2.631781243136394, + "grad_norm": 2.324456214904785, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7309965491294861, + "num_tokens": 619990685.0, + "step": 23965 + }, + { + "epoch": 2.631891060839007, + "grad_norm": 2.4088542461395264, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7301771640777588, + "num_tokens": 620016539.0, + "step": 23966 + }, + { + "epoch": 2.632000878541621, + "grad_norm": 2.0369834899902344, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.717974066734314, + "num_tokens": 620048256.0, + "step": 23967 + }, + { + "epoch": 2.6321106962442347, + "grad_norm": 2.626305103302002, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7355144023895264, + "num_tokens": 620071741.0, + "step": 23968 + }, + { + "epoch": 2.632220513946848, + "grad_norm": 2.4247095584869385, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7277060747146606, + "num_tokens": 620096596.0, + "step": 23969 + }, + { + "epoch": 2.6323303316494617, + "grad_norm": 2.6325676441192627, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.73407381772995, + "num_tokens": 620118757.0, + "step": 23970 + }, + { + "epoch": 2.6324401493520755, + "grad_norm": 2.189753532409668, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.6990712881088257, + "num_tokens": 620152284.0, + "step": 23971 + }, + { + "epoch": 2.6325499670546892, + "grad_norm": 2.627332925796509, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7257065176963806, + "num_tokens": 620174262.0, + "step": 23972 + }, + { + "epoch": 2.632659784757303, + "grad_norm": 2.124208927154541, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6888203024864197, + "num_tokens": 620205473.0, + "step": 23973 + }, + { + "epoch": 2.6327696024599163, + "grad_norm": 2.302823543548584, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7164837121963501, + "num_tokens": 620231736.0, + "step": 23974 + }, + { + "epoch": 2.63287942016253, + "grad_norm": 2.192307472229004, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7525413632392883, + "num_tokens": 620260991.0, + "step": 23975 + }, + { + "epoch": 2.632989237865144, + "grad_norm": 2.167422294616699, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7321364879608154, + "num_tokens": 620288854.0, + "step": 23976 + }, + { + "epoch": 2.6330990555677576, + "grad_norm": 2.289052963256836, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.70949387550354, + "num_tokens": 620317596.0, + "step": 23977 + }, + { + "epoch": 2.6332088732703713, + "grad_norm": 2.3455886840820312, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7554504871368408, + "num_tokens": 620341641.0, + "step": 23978 + }, + { + "epoch": 2.6333186909729847, + "grad_norm": 2.2148427963256836, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7317972183227539, + "num_tokens": 620371222.0, + "step": 23979 + }, + { + "epoch": 2.6334285086755984, + "grad_norm": 2.5312998294830322, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7437710165977478, + "num_tokens": 620395234.0, + "step": 23980 + }, + { + "epoch": 2.633538326378212, + "grad_norm": 2.227025032043457, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7344355583190918, + "num_tokens": 620423716.0, + "step": 23981 + }, + { + "epoch": 2.633648144080826, + "grad_norm": 2.6648480892181396, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7302168607711792, + "num_tokens": 620445412.0, + "step": 23982 + }, + { + "epoch": 2.6337579617834397, + "grad_norm": 2.5131478309631348, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7504587173461914, + "num_tokens": 620468398.0, + "step": 23983 + }, + { + "epoch": 2.633867779486053, + "grad_norm": 2.409438371658325, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7318929433822632, + "num_tokens": 620492287.0, + "step": 23984 + }, + { + "epoch": 2.6339775971886668, + "grad_norm": 2.504819631576538, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7402552366256714, + "num_tokens": 620514595.0, + "step": 23985 + }, + { + "epoch": 2.6340874148912805, + "grad_norm": 2.216040849685669, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.73357754945755, + "num_tokens": 620543280.0, + "step": 23986 + }, + { + "epoch": 2.6341972325938943, + "grad_norm": 2.5775680541992188, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.733066201210022, + "num_tokens": 620564893.0, + "step": 23987 + }, + { + "epoch": 2.634307050296508, + "grad_norm": 2.3293867111206055, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.719199001789093, + "num_tokens": 620592556.0, + "step": 23988 + }, + { + "epoch": 2.6344168679991213, + "grad_norm": 2.7853763103485107, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7514883279800415, + "num_tokens": 620612564.0, + "step": 23989 + }, + { + "epoch": 2.634526685701735, + "grad_norm": 2.365042209625244, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7464669942855835, + "num_tokens": 620637624.0, + "step": 23990 + }, + { + "epoch": 2.634636503404349, + "grad_norm": 2.4105706214904785, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7279589772224426, + "num_tokens": 620662855.0, + "step": 23991 + }, + { + "epoch": 2.6347463211069626, + "grad_norm": 2.57267689704895, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7523995637893677, + "num_tokens": 620684595.0, + "step": 23992 + }, + { + "epoch": 2.6348561388095764, + "grad_norm": 2.392310857772827, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7325199842453003, + "num_tokens": 620710732.0, + "step": 23993 + }, + { + "epoch": 2.6349659565121897, + "grad_norm": 2.1977317333221436, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7123361229896545, + "num_tokens": 620740712.0, + "step": 23994 + }, + { + "epoch": 2.6350757742148034, + "grad_norm": 2.449533462524414, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7381833791732788, + "num_tokens": 620764298.0, + "step": 23995 + }, + { + "epoch": 2.635185591917417, + "grad_norm": 2.2875936031341553, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7225899696350098, + "num_tokens": 620790394.0, + "step": 23996 + }, + { + "epoch": 2.6352954096200305, + "grad_norm": 2.2790091037750244, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7128711938858032, + "num_tokens": 620817893.0, + "step": 23997 + }, + { + "epoch": 2.6354052273226443, + "grad_norm": 2.136460065841675, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7468008995056152, + "num_tokens": 620844533.0, + "step": 23998 + }, + { + "epoch": 2.635515045025258, + "grad_norm": 2.7443225383758545, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7262486219406128, + "num_tokens": 620867060.0, + "step": 23999 + }, + { + "epoch": 2.6356248627278718, + "grad_norm": 2.3316493034362793, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7310175895690918, + "num_tokens": 620893365.0, + "step": 24000 + }, + { + "epoch": 2.6357346804304855, + "grad_norm": 2.238260269165039, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7442382574081421, + "num_tokens": 620920208.0, + "step": 24001 + }, + { + "epoch": 2.635844498133099, + "grad_norm": 2.6482317447662354, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7236698269844055, + "num_tokens": 620942974.0, + "step": 24002 + }, + { + "epoch": 2.6359543158357126, + "grad_norm": 2.260403633117676, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7166119813919067, + "num_tokens": 620970769.0, + "step": 24003 + }, + { + "epoch": 2.6360641335383264, + "grad_norm": 2.158034563064575, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7068229913711548, + "num_tokens": 621003008.0, + "step": 24004 + }, + { + "epoch": 2.63617395124094, + "grad_norm": 2.2080628871917725, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7266756296157837, + "num_tokens": 621033307.0, + "step": 24005 + }, + { + "epoch": 2.636283768943554, + "grad_norm": 2.171011209487915, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7300893068313599, + "num_tokens": 621063311.0, + "step": 24006 + }, + { + "epoch": 2.636393586646167, + "grad_norm": 2.2838821411132812, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7262144088745117, + "num_tokens": 621091081.0, + "step": 24007 + }, + { + "epoch": 2.636503404348781, + "grad_norm": 2.865191698074341, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7178658246994019, + "num_tokens": 621111113.0, + "step": 24008 + }, + { + "epoch": 2.6366132220513947, + "grad_norm": 2.338144063949585, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7283042669296265, + "num_tokens": 621137657.0, + "step": 24009 + }, + { + "epoch": 2.6367230397540085, + "grad_norm": 2.6308512687683105, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7477328777313232, + "num_tokens": 621157945.0, + "step": 24010 + }, + { + "epoch": 2.636832857456622, + "grad_norm": 2.7007758617401123, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7163044810295105, + "num_tokens": 621178900.0, + "step": 24011 + }, + { + "epoch": 2.6369426751592355, + "grad_norm": 2.390505075454712, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7331551909446716, + "num_tokens": 621204561.0, + "step": 24012 + }, + { + "epoch": 2.6370524928618493, + "grad_norm": 2.5733466148376465, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7349245548248291, + "num_tokens": 621226032.0, + "step": 24013 + }, + { + "epoch": 2.637162310564463, + "grad_norm": 2.767073631286621, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7253534197807312, + "num_tokens": 621245472.0, + "step": 24014 + }, + { + "epoch": 2.637272128267077, + "grad_norm": 2.5919203758239746, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7254535555839539, + "num_tokens": 621266922.0, + "step": 24015 + }, + { + "epoch": 2.6373819459696906, + "grad_norm": 2.240405797958374, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7202622890472412, + "num_tokens": 621296209.0, + "step": 24016 + }, + { + "epoch": 2.637491763672304, + "grad_norm": 2.6537704467773438, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7208502292633057, + "num_tokens": 621318693.0, + "step": 24017 + }, + { + "epoch": 2.6376015813749176, + "grad_norm": 2.2622413635253906, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.6998282670974731, + "num_tokens": 621347752.0, + "step": 24018 + }, + { + "epoch": 2.6377113990775314, + "grad_norm": 2.4087142944335938, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.758268415927887, + "num_tokens": 621373631.0, + "step": 24019 + }, + { + "epoch": 2.6378212167801447, + "grad_norm": 2.525679111480713, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.741564154624939, + "num_tokens": 621398262.0, + "step": 24020 + }, + { + "epoch": 2.637931034482759, + "grad_norm": 2.242696523666382, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7570334076881409, + "num_tokens": 621423360.0, + "step": 24021 + }, + { + "epoch": 2.638040852185372, + "grad_norm": 2.224766731262207, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7335089445114136, + "num_tokens": 621451716.0, + "step": 24022 + }, + { + "epoch": 2.638150669887986, + "grad_norm": 2.5429298877716064, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7207236886024475, + "num_tokens": 621475198.0, + "step": 24023 + }, + { + "epoch": 2.6382604875905997, + "grad_norm": 2.424600839614868, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.727859616279602, + "num_tokens": 621499649.0, + "step": 24024 + }, + { + "epoch": 2.638370305293213, + "grad_norm": 2.3369317054748535, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7392818331718445, + "num_tokens": 621524844.0, + "step": 24025 + }, + { + "epoch": 2.638480122995827, + "grad_norm": 2.1656076908111572, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.705271303653717, + "num_tokens": 621556794.0, + "step": 24026 + }, + { + "epoch": 2.6385899406984406, + "grad_norm": 2.7453832626342773, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7307495474815369, + "num_tokens": 621576931.0, + "step": 24027 + }, + { + "epoch": 2.6386997584010543, + "grad_norm": 2.2371630668640137, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7261339426040649, + "num_tokens": 621603944.0, + "step": 24028 + }, + { + "epoch": 2.638809576103668, + "grad_norm": 2.178807497024536, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7142622470855713, + "num_tokens": 621635189.0, + "step": 24029 + }, + { + "epoch": 2.6389193938062814, + "grad_norm": 2.642199993133545, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.751825213432312, + "num_tokens": 621656516.0, + "step": 24030 + }, + { + "epoch": 2.639029211508895, + "grad_norm": 2.405573606491089, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7290704250335693, + "num_tokens": 621681421.0, + "step": 24031 + }, + { + "epoch": 2.639139029211509, + "grad_norm": 2.4285144805908203, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7028815746307373, + "num_tokens": 621705867.0, + "step": 24032 + }, + { + "epoch": 2.6392488469141226, + "grad_norm": 2.222935914993286, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7364974021911621, + "num_tokens": 621731823.0, + "step": 24033 + }, + { + "epoch": 2.6393586646167364, + "grad_norm": 2.7674038410186768, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7494604587554932, + "num_tokens": 621751871.0, + "step": 24034 + }, + { + "epoch": 2.6394684823193497, + "grad_norm": 2.24051833152771, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7351105809211731, + "num_tokens": 621778959.0, + "step": 24035 + }, + { + "epoch": 2.6395783000219635, + "grad_norm": 2.3206536769866943, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7200788259506226, + "num_tokens": 621806512.0, + "step": 24036 + }, + { + "epoch": 2.6396881177245772, + "grad_norm": 4.472209930419922, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7425976991653442, + "num_tokens": 621823949.0, + "step": 24037 + }, + { + "epoch": 2.639797935427191, + "grad_norm": 2.1963815689086914, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7429497241973877, + "num_tokens": 621851867.0, + "step": 24038 + }, + { + "epoch": 2.6399077531298047, + "grad_norm": 2.3951101303100586, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7368673086166382, + "num_tokens": 621877830.0, + "step": 24039 + }, + { + "epoch": 2.640017570832418, + "grad_norm": 2.3616857528686523, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7132965326309204, + "num_tokens": 621904075.0, + "step": 24040 + }, + { + "epoch": 2.640127388535032, + "grad_norm": 2.7419819831848145, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7469707727432251, + "num_tokens": 621923382.0, + "step": 24041 + }, + { + "epoch": 2.6402372062376456, + "grad_norm": 2.2400386333465576, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7298848628997803, + "num_tokens": 621953238.0, + "step": 24042 + }, + { + "epoch": 2.6403470239402593, + "grad_norm": 2.77469539642334, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7453349828720093, + "num_tokens": 621972582.0, + "step": 24043 + }, + { + "epoch": 2.640456841642873, + "grad_norm": 2.450939178466797, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.73062664270401, + "num_tokens": 621997752.0, + "step": 24044 + }, + { + "epoch": 2.6405666593454864, + "grad_norm": 2.614567518234253, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7428122758865356, + "num_tokens": 622020119.0, + "step": 24045 + }, + { + "epoch": 2.6406764770481, + "grad_norm": 2.551260232925415, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7351539134979248, + "num_tokens": 622043496.0, + "step": 24046 + }, + { + "epoch": 2.640786294750714, + "grad_norm": 2.3002402782440186, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7193336486816406, + "num_tokens": 622070300.0, + "step": 24047 + }, + { + "epoch": 2.6408961124533272, + "grad_norm": 2.5282068252563477, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7303273677825928, + "num_tokens": 622094411.0, + "step": 24048 + }, + { + "epoch": 2.641005930155941, + "grad_norm": 2.0549397468566895, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7093212604522705, + "num_tokens": 622128698.0, + "step": 24049 + }, + { + "epoch": 2.6411157478585547, + "grad_norm": 2.604673147201538, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7330440282821655, + "num_tokens": 622148936.0, + "step": 24050 + }, + { + "epoch": 2.6412255655611685, + "grad_norm": 2.190582036972046, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.736018180847168, + "num_tokens": 622176301.0, + "step": 24051 + }, + { + "epoch": 2.6413353832637823, + "grad_norm": 2.082564115524292, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7132647037506104, + "num_tokens": 622208911.0, + "step": 24052 + }, + { + "epoch": 2.6414452009663956, + "grad_norm": 2.358396530151367, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7144085168838501, + "num_tokens": 622236356.0, + "step": 24053 + }, + { + "epoch": 2.6415550186690093, + "grad_norm": 2.2784178256988525, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7292042970657349, + "num_tokens": 622262816.0, + "step": 24054 + }, + { + "epoch": 2.641664836371623, + "grad_norm": 2.5898444652557373, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7158750295639038, + "num_tokens": 622284963.0, + "step": 24055 + }, + { + "epoch": 2.641774654074237, + "grad_norm": 2.7099807262420654, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7570797204971313, + "num_tokens": 622305749.0, + "step": 24056 + }, + { + "epoch": 2.6418844717768506, + "grad_norm": 2.316009759902954, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7495308518409729, + "num_tokens": 622331647.0, + "step": 24057 + }, + { + "epoch": 2.641994289479464, + "grad_norm": 2.491542339324951, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7243041396141052, + "num_tokens": 622354461.0, + "step": 24058 + }, + { + "epoch": 2.6421041071820777, + "grad_norm": 2.4485182762145996, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7247529625892639, + "num_tokens": 622378606.0, + "step": 24059 + }, + { + "epoch": 2.6422139248846914, + "grad_norm": 2.3092868328094482, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.732205867767334, + "num_tokens": 622404972.0, + "step": 24060 + }, + { + "epoch": 2.642323742587305, + "grad_norm": 2.215000629425049, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7149645686149597, + "num_tokens": 622437224.0, + "step": 24061 + }, + { + "epoch": 2.642433560289919, + "grad_norm": 2.090324878692627, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6912523508071899, + "num_tokens": 622470711.0, + "step": 24062 + }, + { + "epoch": 2.6425433779925322, + "grad_norm": 2.4697139263153076, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7445492744445801, + "num_tokens": 622495174.0, + "step": 24063 + }, + { + "epoch": 2.642653195695146, + "grad_norm": 2.5344295501708984, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7509943246841431, + "num_tokens": 622518425.0, + "step": 24064 + }, + { + "epoch": 2.6427630133977598, + "grad_norm": 2.1049585342407227, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7216336727142334, + "num_tokens": 622549915.0, + "step": 24065 + }, + { + "epoch": 2.6428728311003735, + "grad_norm": 2.6004486083984375, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7057117819786072, + "num_tokens": 622572444.0, + "step": 24066 + }, + { + "epoch": 2.6429826488029873, + "grad_norm": 2.6742358207702637, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7504808902740479, + "num_tokens": 622593190.0, + "step": 24067 + }, + { + "epoch": 2.6430924665056006, + "grad_norm": 2.466249704360962, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7296360731124878, + "num_tokens": 622620392.0, + "step": 24068 + }, + { + "epoch": 2.6432022842082143, + "grad_norm": 2.349209785461426, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7261993288993835, + "num_tokens": 622649585.0, + "step": 24069 + }, + { + "epoch": 2.643312101910828, + "grad_norm": 2.233999729156494, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7388502955436707, + "num_tokens": 622676984.0, + "step": 24070 + }, + { + "epoch": 2.6434219196134414, + "grad_norm": 2.2366089820861816, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7218170166015625, + "num_tokens": 622705084.0, + "step": 24071 + }, + { + "epoch": 2.6435317373160556, + "grad_norm": 2.4056060314178467, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.724818229675293, + "num_tokens": 622732563.0, + "step": 24072 + }, + { + "epoch": 2.643641555018669, + "grad_norm": 2.5472939014434814, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7458606958389282, + "num_tokens": 622757536.0, + "step": 24073 + }, + { + "epoch": 2.6437513727212827, + "grad_norm": 2.2099833488464355, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7273802757263184, + "num_tokens": 622787075.0, + "step": 24074 + }, + { + "epoch": 2.6438611904238964, + "grad_norm": 2.4893853664398193, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7246541380882263, + "num_tokens": 622811946.0, + "step": 24075 + }, + { + "epoch": 2.6439710081265098, + "grad_norm": 2.2850375175476074, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7349951267242432, + "num_tokens": 622839809.0, + "step": 24076 + }, + { + "epoch": 2.6440808258291235, + "grad_norm": 2.7639150619506836, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7430742979049683, + "num_tokens": 622858369.0, + "step": 24077 + }, + { + "epoch": 2.6441906435317373, + "grad_norm": 2.607510566711426, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7434120774269104, + "num_tokens": 622879353.0, + "step": 24078 + }, + { + "epoch": 2.644300461234351, + "grad_norm": 2.5405819416046143, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7150978446006775, + "num_tokens": 622906053.0, + "step": 24079 + }, + { + "epoch": 2.644410278936965, + "grad_norm": 2.214710235595703, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7198252081871033, + "num_tokens": 622935393.0, + "step": 24080 + }, + { + "epoch": 2.644520096639578, + "grad_norm": 2.3433845043182373, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7222938537597656, + "num_tokens": 622961271.0, + "step": 24081 + }, + { + "epoch": 2.644629914342192, + "grad_norm": 2.269015073776245, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7455114126205444, + "num_tokens": 622988400.0, + "step": 24082 + }, + { + "epoch": 2.6447397320448056, + "grad_norm": 2.5802805423736572, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.729731559753418, + "num_tokens": 623011022.0, + "step": 24083 + }, + { + "epoch": 2.6448495497474194, + "grad_norm": 2.725715160369873, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7694786787033081, + "num_tokens": 623030796.0, + "step": 24084 + }, + { + "epoch": 2.644959367450033, + "grad_norm": 2.1609103679656982, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.715329647064209, + "num_tokens": 623060043.0, + "step": 24085 + }, + { + "epoch": 2.6450691851526464, + "grad_norm": 2.38763427734375, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7485384941101074, + "num_tokens": 623082033.0, + "step": 24086 + }, + { + "epoch": 2.64517900285526, + "grad_norm": 2.381230354309082, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7193930149078369, + "num_tokens": 623109445.0, + "step": 24087 + }, + { + "epoch": 2.645288820557874, + "grad_norm": 2.254926919937134, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7169394493103027, + "num_tokens": 623138438.0, + "step": 24088 + }, + { + "epoch": 2.6453986382604877, + "grad_norm": 2.4099345207214355, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.73939049243927, + "num_tokens": 623162892.0, + "step": 24089 + }, + { + "epoch": 2.6455084559631015, + "grad_norm": 3.023082733154297, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7437724471092224, + "num_tokens": 623181200.0, + "step": 24090 + }, + { + "epoch": 2.645618273665715, + "grad_norm": 2.2869324684143066, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7299448251724243, + "num_tokens": 623209572.0, + "step": 24091 + }, + { + "epoch": 2.6457280913683285, + "grad_norm": 2.5410633087158203, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7191694974899292, + "num_tokens": 623233462.0, + "step": 24092 + }, + { + "epoch": 2.6458379090709423, + "grad_norm": 2.2118804454803467, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7290099859237671, + "num_tokens": 623260152.0, + "step": 24093 + }, + { + "epoch": 2.645947726773556, + "grad_norm": 2.153764247894287, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7215484380722046, + "num_tokens": 623290976.0, + "step": 24094 + }, + { + "epoch": 2.64605754447617, + "grad_norm": 2.566629409790039, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7151666283607483, + "num_tokens": 623314911.0, + "step": 24095 + }, + { + "epoch": 2.646167362178783, + "grad_norm": 2.2952895164489746, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7363585233688354, + "num_tokens": 623342413.0, + "step": 24096 + }, + { + "epoch": 2.646277179881397, + "grad_norm": 2.1780498027801514, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7161306142807007, + "num_tokens": 623371308.0, + "step": 24097 + }, + { + "epoch": 2.6463869975840106, + "grad_norm": 2.1473965644836426, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7614191770553589, + "num_tokens": 623398913.0, + "step": 24098 + }, + { + "epoch": 2.646496815286624, + "grad_norm": 2.137570858001709, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7136961221694946, + "num_tokens": 623430023.0, + "step": 24099 + }, + { + "epoch": 2.6466066329892377, + "grad_norm": 2.1035916805267334, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7168519496917725, + "num_tokens": 623461192.0, + "step": 24100 + }, + { + "epoch": 2.6467164506918515, + "grad_norm": 2.416546106338501, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7427185773849487, + "num_tokens": 623487347.0, + "step": 24101 + }, + { + "epoch": 2.646826268394465, + "grad_norm": 2.4235360622406006, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7211316227912903, + "num_tokens": 623511378.0, + "step": 24102 + }, + { + "epoch": 2.646936086097079, + "grad_norm": 2.400723934173584, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7224644422531128, + "num_tokens": 623535263.0, + "step": 24103 + }, + { + "epoch": 2.6470459037996923, + "grad_norm": 2.3769032955169678, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7190119028091431, + "num_tokens": 623562398.0, + "step": 24104 + }, + { + "epoch": 2.647155721502306, + "grad_norm": 2.5497024059295654, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7341296672821045, + "num_tokens": 623586222.0, + "step": 24105 + }, + { + "epoch": 2.64726553920492, + "grad_norm": 2.6689035892486572, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7268270254135132, + "num_tokens": 623608154.0, + "step": 24106 + }, + { + "epoch": 2.6473753569075336, + "grad_norm": 2.4617350101470947, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.705429196357727, + "num_tokens": 623634219.0, + "step": 24107 + }, + { + "epoch": 2.6474851746101473, + "grad_norm": 2.4236631393432617, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7253965139389038, + "num_tokens": 623661239.0, + "step": 24108 + }, + { + "epoch": 2.6475949923127606, + "grad_norm": 2.2239673137664795, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7522094249725342, + "num_tokens": 623687698.0, + "step": 24109 + }, + { + "epoch": 2.6477048100153744, + "grad_norm": 2.401759624481201, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7344287037849426, + "num_tokens": 623712239.0, + "step": 24110 + }, + { + "epoch": 2.647814627717988, + "grad_norm": 2.582245111465454, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7466317415237427, + "num_tokens": 623734006.0, + "step": 24111 + }, + { + "epoch": 2.647924445420602, + "grad_norm": 2.3012173175811768, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7269577383995056, + "num_tokens": 623761280.0, + "step": 24112 + }, + { + "epoch": 2.6480342631232157, + "grad_norm": 2.2761125564575195, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7195342779159546, + "num_tokens": 623788872.0, + "step": 24113 + }, + { + "epoch": 2.648144080825829, + "grad_norm": 2.3675436973571777, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7096438407897949, + "num_tokens": 623815823.0, + "step": 24114 + }, + { + "epoch": 2.6482538985284427, + "grad_norm": 2.437673807144165, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7258898019790649, + "num_tokens": 623841142.0, + "step": 24115 + }, + { + "epoch": 2.6483637162310565, + "grad_norm": 2.287564277648926, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.6947998404502869, + "num_tokens": 623868966.0, + "step": 24116 + }, + { + "epoch": 2.6484735339336702, + "grad_norm": 2.5634498596191406, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7361586689949036, + "num_tokens": 623889833.0, + "step": 24117 + }, + { + "epoch": 2.648583351636284, + "grad_norm": 2.3700497150421143, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7171332836151123, + "num_tokens": 623916392.0, + "step": 24118 + }, + { + "epoch": 2.6486931693388973, + "grad_norm": 2.392493963241577, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7296129465103149, + "num_tokens": 623941110.0, + "step": 24119 + }, + { + "epoch": 2.648802987041511, + "grad_norm": 2.340097427368164, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7404664754867554, + "num_tokens": 623965608.0, + "step": 24120 + }, + { + "epoch": 2.648912804744125, + "grad_norm": 2.409349203109741, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7324326038360596, + "num_tokens": 623992636.0, + "step": 24121 + }, + { + "epoch": 2.649022622446738, + "grad_norm": 2.3804562091827393, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7445415258407593, + "num_tokens": 624016363.0, + "step": 24122 + }, + { + "epoch": 2.6491324401493523, + "grad_norm": 2.2421629428863525, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7459496259689331, + "num_tokens": 624043771.0, + "step": 24123 + }, + { + "epoch": 2.6492422578519657, + "grad_norm": 2.1875851154327393, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7228422164916992, + "num_tokens": 624072386.0, + "step": 24124 + }, + { + "epoch": 2.6493520755545794, + "grad_norm": 2.389347553253174, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7310940623283386, + "num_tokens": 624097076.0, + "step": 24125 + }, + { + "epoch": 2.649461893257193, + "grad_norm": 2.391857862472534, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7498956322669983, + "num_tokens": 624120131.0, + "step": 24126 + }, + { + "epoch": 2.6495717109598065, + "grad_norm": 2.2931697368621826, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7416646480560303, + "num_tokens": 624148254.0, + "step": 24127 + }, + { + "epoch": 2.6496815286624202, + "grad_norm": 2.218780755996704, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7138774394989014, + "num_tokens": 624176985.0, + "step": 24128 + }, + { + "epoch": 2.649791346365034, + "grad_norm": 2.428518533706665, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7349075078964233, + "num_tokens": 624202640.0, + "step": 24129 + }, + { + "epoch": 2.6499011640676478, + "grad_norm": 2.4995272159576416, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7481141090393066, + "num_tokens": 624225318.0, + "step": 24130 + }, + { + "epoch": 2.6500109817702615, + "grad_norm": 2.4853155612945557, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.734930157661438, + "num_tokens": 624249215.0, + "step": 24131 + }, + { + "epoch": 2.650120799472875, + "grad_norm": 2.5551400184631348, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7344425916671753, + "num_tokens": 624271947.0, + "step": 24132 + }, + { + "epoch": 2.6502306171754886, + "grad_norm": 2.4075334072113037, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7387906312942505, + "num_tokens": 624294819.0, + "step": 24133 + }, + { + "epoch": 2.6503404348781023, + "grad_norm": 2.470518112182617, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7324173450469971, + "num_tokens": 624319693.0, + "step": 24134 + }, + { + "epoch": 2.650450252580716, + "grad_norm": 2.093724250793457, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.730027437210083, + "num_tokens": 624350989.0, + "step": 24135 + }, + { + "epoch": 2.65056007028333, + "grad_norm": 2.149045705795288, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7571351528167725, + "num_tokens": 624380414.0, + "step": 24136 + }, + { + "epoch": 2.650669887985943, + "grad_norm": 2.3643815517425537, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7210080027580261, + "num_tokens": 624406399.0, + "step": 24137 + }, + { + "epoch": 2.650779705688557, + "grad_norm": 2.285712480545044, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.710174560546875, + "num_tokens": 624433901.0, + "step": 24138 + }, + { + "epoch": 2.6508895233911707, + "grad_norm": 2.357503652572632, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7408220767974854, + "num_tokens": 624458660.0, + "step": 24139 + }, + { + "epoch": 2.6509993410937844, + "grad_norm": 2.563943862915039, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7364087700843811, + "num_tokens": 624481456.0, + "step": 24140 + }, + { + "epoch": 2.651109158796398, + "grad_norm": 2.5684218406677246, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7332345247268677, + "num_tokens": 624504676.0, + "step": 24141 + }, + { + "epoch": 2.6512189764990115, + "grad_norm": 2.1993517875671387, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6969292759895325, + "num_tokens": 624535033.0, + "step": 24142 + }, + { + "epoch": 2.6513287942016253, + "grad_norm": 2.5810811519622803, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7422411441802979, + "num_tokens": 624557535.0, + "step": 24143 + }, + { + "epoch": 2.651438611904239, + "grad_norm": 2.358426332473755, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.731065571308136, + "num_tokens": 624583930.0, + "step": 24144 + }, + { + "epoch": 2.6515484296068528, + "grad_norm": 2.6132986545562744, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7316502332687378, + "num_tokens": 624607216.0, + "step": 24145 + }, + { + "epoch": 2.6516582473094665, + "grad_norm": 2.3900225162506104, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7447320222854614, + "num_tokens": 624629725.0, + "step": 24146 + }, + { + "epoch": 2.65176806501208, + "grad_norm": 2.363708019256592, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7272974252700806, + "num_tokens": 624654995.0, + "step": 24147 + }, + { + "epoch": 2.6518778827146936, + "grad_norm": 2.536069869995117, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7515422105789185, + "num_tokens": 624677595.0, + "step": 24148 + }, + { + "epoch": 2.6519877004173074, + "grad_norm": 2.3974428176879883, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7185845971107483, + "num_tokens": 624705954.0, + "step": 24149 + }, + { + "epoch": 2.6520975181199207, + "grad_norm": 2.232900857925415, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.71675705909729, + "num_tokens": 624735000.0, + "step": 24150 + }, + { + "epoch": 2.6522073358225344, + "grad_norm": 2.748892068862915, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7219945192337036, + "num_tokens": 624756511.0, + "step": 24151 + }, + { + "epoch": 2.652317153525148, + "grad_norm": 2.2127764225006104, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7130684852600098, + "num_tokens": 624784308.0, + "step": 24152 + }, + { + "epoch": 2.652426971227762, + "grad_norm": 2.4770119190216064, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7298866510391235, + "num_tokens": 624809534.0, + "step": 24153 + }, + { + "epoch": 2.6525367889303757, + "grad_norm": 2.3320040702819824, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7440085411071777, + "num_tokens": 624834936.0, + "step": 24154 + }, + { + "epoch": 2.652646606632989, + "grad_norm": 2.175233840942383, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7154617309570312, + "num_tokens": 624864487.0, + "step": 24155 + }, + { + "epoch": 2.6527564243356028, + "grad_norm": 2.464542865753174, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7120923399925232, + "num_tokens": 624889573.0, + "step": 24156 + }, + { + "epoch": 2.6528662420382165, + "grad_norm": 2.2205772399902344, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7088878154754639, + "num_tokens": 624918369.0, + "step": 24157 + }, + { + "epoch": 2.6529760597408303, + "grad_norm": 2.3550901412963867, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7388707995414734, + "num_tokens": 624944122.0, + "step": 24158 + }, + { + "epoch": 2.653085877443444, + "grad_norm": 2.7613186836242676, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7605615258216858, + "num_tokens": 624964762.0, + "step": 24159 + }, + { + "epoch": 2.6531956951460574, + "grad_norm": 2.4280738830566406, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7179081439971924, + "num_tokens": 624989315.0, + "step": 24160 + }, + { + "epoch": 2.653305512848671, + "grad_norm": 2.427111864089966, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.711227536201477, + "num_tokens": 625016020.0, + "step": 24161 + }, + { + "epoch": 2.653415330551285, + "grad_norm": 2.0422608852386475, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7325313091278076, + "num_tokens": 625046600.0, + "step": 24162 + }, + { + "epoch": 2.6535251482538986, + "grad_norm": 2.3413054943084717, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7276138663291931, + "num_tokens": 625073179.0, + "step": 24163 + }, + { + "epoch": 2.6536349659565124, + "grad_norm": 2.18733549118042, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7453120946884155, + "num_tokens": 625100817.0, + "step": 24164 + }, + { + "epoch": 2.6537447836591257, + "grad_norm": 2.0929627418518066, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7068056464195251, + "num_tokens": 625132091.0, + "step": 24165 + }, + { + "epoch": 2.6538546013617395, + "grad_norm": 2.5254876613616943, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7317115068435669, + "num_tokens": 625156293.0, + "step": 24166 + }, + { + "epoch": 2.653964419064353, + "grad_norm": 2.273225784301758, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7096298933029175, + "num_tokens": 625185119.0, + "step": 24167 + }, + { + "epoch": 2.654074236766967, + "grad_norm": 2.257263660430908, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.721294105052948, + "num_tokens": 625216115.0, + "step": 24168 + }, + { + "epoch": 2.6541840544695807, + "grad_norm": 2.3016867637634277, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7289886474609375, + "num_tokens": 625243558.0, + "step": 24169 + }, + { + "epoch": 2.654293872172194, + "grad_norm": 2.5018441677093506, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7381280660629272, + "num_tokens": 625266309.0, + "step": 24170 + }, + { + "epoch": 2.654403689874808, + "grad_norm": 2.1769018173217773, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7121202945709229, + "num_tokens": 625295082.0, + "step": 24171 + }, + { + "epoch": 2.6545135075774215, + "grad_norm": 2.484398365020752, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7512869238853455, + "num_tokens": 625319200.0, + "step": 24172 + }, + { + "epoch": 2.6546233252800353, + "grad_norm": 2.356677532196045, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7562648057937622, + "num_tokens": 625344812.0, + "step": 24173 + }, + { + "epoch": 2.654733142982649, + "grad_norm": 2.2516093254089355, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7324950695037842, + "num_tokens": 625371736.0, + "step": 24174 + }, + { + "epoch": 2.6548429606852624, + "grad_norm": 2.2459585666656494, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7377819418907166, + "num_tokens": 625400843.0, + "step": 24175 + }, + { + "epoch": 2.654952778387876, + "grad_norm": 2.760993242263794, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.724808931350708, + "num_tokens": 625421941.0, + "step": 24176 + }, + { + "epoch": 2.65506259609049, + "grad_norm": 2.721097707748413, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7362891435623169, + "num_tokens": 625443964.0, + "step": 24177 + }, + { + "epoch": 2.655172413793103, + "grad_norm": 2.7310409545898438, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7457760572433472, + "num_tokens": 625463927.0, + "step": 24178 + }, + { + "epoch": 2.655282231495717, + "grad_norm": 2.1725568771362305, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7389179468154907, + "num_tokens": 625492348.0, + "step": 24179 + }, + { + "epoch": 2.6553920491983307, + "grad_norm": 2.695734977722168, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7390829920768738, + "num_tokens": 625513604.0, + "step": 24180 + }, + { + "epoch": 2.6555018669009445, + "grad_norm": 2.635040283203125, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7328145503997803, + "num_tokens": 625534927.0, + "step": 24181 + }, + { + "epoch": 2.6556116846035582, + "grad_norm": 2.5339531898498535, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7476778030395508, + "num_tokens": 625558477.0, + "step": 24182 + }, + { + "epoch": 2.6557215023061715, + "grad_norm": 2.2573015689849854, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7371354103088379, + "num_tokens": 625585865.0, + "step": 24183 + }, + { + "epoch": 2.6558313200087853, + "grad_norm": 2.3293542861938477, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7261123061180115, + "num_tokens": 625610298.0, + "step": 24184 + }, + { + "epoch": 2.655941137711399, + "grad_norm": 2.3775250911712646, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.735844612121582, + "num_tokens": 625635271.0, + "step": 24185 + }, + { + "epoch": 2.656050955414013, + "grad_norm": 2.2168521881103516, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7027081847190857, + "num_tokens": 625667778.0, + "step": 24186 + }, + { + "epoch": 2.6561607731166266, + "grad_norm": 2.270425796508789, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7242426872253418, + "num_tokens": 625693015.0, + "step": 24187 + }, + { + "epoch": 2.65627059081924, + "grad_norm": 2.2388720512390137, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7177121639251709, + "num_tokens": 625726287.0, + "step": 24188 + }, + { + "epoch": 2.6563804085218536, + "grad_norm": 2.7154765129089355, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7317243218421936, + "num_tokens": 625747179.0, + "step": 24189 + }, + { + "epoch": 2.6564902262244674, + "grad_norm": 2.3922674655914307, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7247198820114136, + "num_tokens": 625772030.0, + "step": 24190 + }, + { + "epoch": 2.656600043927081, + "grad_norm": 2.370195150375366, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7246928811073303, + "num_tokens": 625800036.0, + "step": 24191 + }, + { + "epoch": 2.656709861629695, + "grad_norm": 2.2452540397644043, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7229030132293701, + "num_tokens": 625828947.0, + "step": 24192 + }, + { + "epoch": 2.6568196793323082, + "grad_norm": 2.3507590293884277, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7202180624008179, + "num_tokens": 625855482.0, + "step": 24193 + }, + { + "epoch": 2.656929497034922, + "grad_norm": 2.1962497234344482, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7470824718475342, + "num_tokens": 625881288.0, + "step": 24194 + }, + { + "epoch": 2.6570393147375357, + "grad_norm": 2.4539408683776855, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7176254987716675, + "num_tokens": 625905574.0, + "step": 24195 + }, + { + "epoch": 2.6571491324401495, + "grad_norm": 2.125555992126465, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7306544780731201, + "num_tokens": 625934805.0, + "step": 24196 + }, + { + "epoch": 2.6572589501427633, + "grad_norm": 2.6040799617767334, + "learning_rate": 1e-06, + "loss": 0.7954, + "mean_token_accuracy": 0.7410429120063782, + "num_tokens": 625957195.0, + "step": 24197 + }, + { + "epoch": 2.6573687678453766, + "grad_norm": 2.636267900466919, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7730352878570557, + "num_tokens": 625976035.0, + "step": 24198 + }, + { + "epoch": 2.6574785855479903, + "grad_norm": 2.6199135780334473, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7226625680923462, + "num_tokens": 625999073.0, + "step": 24199 + }, + { + "epoch": 2.657588403250604, + "grad_norm": 2.2875819206237793, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7291481494903564, + "num_tokens": 626026380.0, + "step": 24200 + }, + { + "epoch": 2.6576982209532174, + "grad_norm": 2.3980538845062256, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7216085195541382, + "num_tokens": 626052800.0, + "step": 24201 + }, + { + "epoch": 2.6578080386558316, + "grad_norm": 2.4763598442077637, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7286306619644165, + "num_tokens": 626076458.0, + "step": 24202 + }, + { + "epoch": 2.657917856358445, + "grad_norm": 2.6250522136688232, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7481381893157959, + "num_tokens": 626097548.0, + "step": 24203 + }, + { + "epoch": 2.6580276740610587, + "grad_norm": 2.440953016281128, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7429962754249573, + "num_tokens": 626121697.0, + "step": 24204 + }, + { + "epoch": 2.6581374917636724, + "grad_norm": 2.543031692504883, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7370932102203369, + "num_tokens": 626143533.0, + "step": 24205 + }, + { + "epoch": 2.6582473094662857, + "grad_norm": 2.296290874481201, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7185794115066528, + "num_tokens": 626169981.0, + "step": 24206 + }, + { + "epoch": 2.6583571271688995, + "grad_norm": 2.302525281906128, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7241275310516357, + "num_tokens": 626197217.0, + "step": 24207 + }, + { + "epoch": 2.6584669448715132, + "grad_norm": 1.995101809501648, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7335914969444275, + "num_tokens": 626229495.0, + "step": 24208 + }, + { + "epoch": 2.658576762574127, + "grad_norm": 2.1470837593078613, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7250500917434692, + "num_tokens": 626257022.0, + "step": 24209 + }, + { + "epoch": 2.6586865802767408, + "grad_norm": 2.4773688316345215, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7214866280555725, + "num_tokens": 626280771.0, + "step": 24210 + }, + { + "epoch": 2.658796397979354, + "grad_norm": 2.485699415206909, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7215012311935425, + "num_tokens": 626305088.0, + "step": 24211 + }, + { + "epoch": 2.658906215681968, + "grad_norm": 2.7370762825012207, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7194423079490662, + "num_tokens": 626327391.0, + "step": 24212 + }, + { + "epoch": 2.6590160333845816, + "grad_norm": 2.3299014568328857, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7398040294647217, + "num_tokens": 626354456.0, + "step": 24213 + }, + { + "epoch": 2.6591258510871953, + "grad_norm": 2.2246949672698975, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7413173913955688, + "num_tokens": 626380975.0, + "step": 24214 + }, + { + "epoch": 2.659235668789809, + "grad_norm": 2.200968027114868, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.748511016368866, + "num_tokens": 626408294.0, + "step": 24215 + }, + { + "epoch": 2.6593454864924224, + "grad_norm": 2.2571513652801514, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7361751198768616, + "num_tokens": 626436301.0, + "step": 24216 + }, + { + "epoch": 2.659455304195036, + "grad_norm": 2.29327654838562, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7227632403373718, + "num_tokens": 626461838.0, + "step": 24217 + }, + { + "epoch": 2.65956512189765, + "grad_norm": 2.1190640926361084, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7171489596366882, + "num_tokens": 626491759.0, + "step": 24218 + }, + { + "epoch": 2.6596749396002637, + "grad_norm": 2.4193060398101807, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7374588251113892, + "num_tokens": 626516598.0, + "step": 24219 + }, + { + "epoch": 2.6597847573028774, + "grad_norm": 2.499394416809082, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7132982015609741, + "num_tokens": 626542760.0, + "step": 24220 + }, + { + "epoch": 2.6598945750054908, + "grad_norm": 2.3291378021240234, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7366833090782166, + "num_tokens": 626568063.0, + "step": 24221 + }, + { + "epoch": 2.6600043927081045, + "grad_norm": 1.9173787832260132, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7353025674819946, + "num_tokens": 626602420.0, + "step": 24222 + }, + { + "epoch": 2.6601142104107183, + "grad_norm": 2.6261606216430664, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.754464864730835, + "num_tokens": 626622888.0, + "step": 24223 + }, + { + "epoch": 2.660224028113332, + "grad_norm": 1.9774532318115234, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7426818609237671, + "num_tokens": 626655392.0, + "step": 24224 + }, + { + "epoch": 2.660333845815946, + "grad_norm": 2.0614829063415527, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7182374000549316, + "num_tokens": 626687978.0, + "step": 24225 + }, + { + "epoch": 2.660443663518559, + "grad_norm": 2.4239397048950195, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7329877614974976, + "num_tokens": 626714453.0, + "step": 24226 + }, + { + "epoch": 2.660553481221173, + "grad_norm": 2.584296703338623, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7350175976753235, + "num_tokens": 626738129.0, + "step": 24227 + }, + { + "epoch": 2.6606632989237866, + "grad_norm": 2.5234920978546143, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7340001463890076, + "num_tokens": 626762084.0, + "step": 24228 + }, + { + "epoch": 2.6607731166264, + "grad_norm": 2.473843812942505, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7294232845306396, + "num_tokens": 626786533.0, + "step": 24229 + }, + { + "epoch": 2.6608829343290137, + "grad_norm": 2.621784210205078, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7363379597663879, + "num_tokens": 626811012.0, + "step": 24230 + }, + { + "epoch": 2.6609927520316274, + "grad_norm": 2.418818473815918, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7547305822372437, + "num_tokens": 626835039.0, + "step": 24231 + }, + { + "epoch": 2.661102569734241, + "grad_norm": 2.3649675846099854, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7305687069892883, + "num_tokens": 626859162.0, + "step": 24232 + }, + { + "epoch": 2.661212387436855, + "grad_norm": 2.2280330657958984, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7095687985420227, + "num_tokens": 626890271.0, + "step": 24233 + }, + { + "epoch": 2.6613222051394683, + "grad_norm": 2.499850273132324, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7245966196060181, + "num_tokens": 626916816.0, + "step": 24234 + }, + { + "epoch": 2.661432022842082, + "grad_norm": 2.0843353271484375, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7128233909606934, + "num_tokens": 626948544.0, + "step": 24235 + }, + { + "epoch": 2.661541840544696, + "grad_norm": 2.663663625717163, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7184351086616516, + "num_tokens": 626969038.0, + "step": 24236 + }, + { + "epoch": 2.6616516582473095, + "grad_norm": 2.090623617172241, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7226171493530273, + "num_tokens": 627001751.0, + "step": 24237 + }, + { + "epoch": 2.6617614759499233, + "grad_norm": 2.442504405975342, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.729393720626831, + "num_tokens": 627025947.0, + "step": 24238 + }, + { + "epoch": 2.6618712936525366, + "grad_norm": 2.7160937786102295, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7375967502593994, + "num_tokens": 627045642.0, + "step": 24239 + }, + { + "epoch": 2.6619811113551504, + "grad_norm": 2.1953165531158447, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7262901067733765, + "num_tokens": 627072345.0, + "step": 24240 + }, + { + "epoch": 2.662090929057764, + "grad_norm": 2.283376932144165, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7377579808235168, + "num_tokens": 627100585.0, + "step": 24241 + }, + { + "epoch": 2.662200746760378, + "grad_norm": 2.325423240661621, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7154539823532104, + "num_tokens": 627126318.0, + "step": 24242 + }, + { + "epoch": 2.6623105644629916, + "grad_norm": 2.2872314453125, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7150456309318542, + "num_tokens": 627153684.0, + "step": 24243 + }, + { + "epoch": 2.662420382165605, + "grad_norm": 2.2915689945220947, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7291291356086731, + "num_tokens": 627182507.0, + "step": 24244 + }, + { + "epoch": 2.6625301998682187, + "grad_norm": 2.213280200958252, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7706303596496582, + "num_tokens": 627210042.0, + "step": 24245 + }, + { + "epoch": 2.6626400175708325, + "grad_norm": 2.245211124420166, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7620216608047485, + "num_tokens": 627235725.0, + "step": 24246 + }, + { + "epoch": 2.662749835273446, + "grad_norm": 2.2116081714630127, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7193753123283386, + "num_tokens": 627264374.0, + "step": 24247 + }, + { + "epoch": 2.66285965297606, + "grad_norm": 2.163159132003784, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7152736186981201, + "num_tokens": 627292718.0, + "step": 24248 + }, + { + "epoch": 2.6629694706786733, + "grad_norm": 2.1791722774505615, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7039729952812195, + "num_tokens": 627323275.0, + "step": 24249 + }, + { + "epoch": 2.663079288381287, + "grad_norm": 2.5596115589141846, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7463515400886536, + "num_tokens": 627347055.0, + "step": 24250 + }, + { + "epoch": 2.663189106083901, + "grad_norm": 2.2595577239990234, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7172542810440063, + "num_tokens": 627375292.0, + "step": 24251 + }, + { + "epoch": 2.663298923786514, + "grad_norm": 2.2860541343688965, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7109143733978271, + "num_tokens": 627404543.0, + "step": 24252 + }, + { + "epoch": 2.6634087414891283, + "grad_norm": 2.1784181594848633, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7213928699493408, + "num_tokens": 627435946.0, + "step": 24253 + }, + { + "epoch": 2.6635185591917416, + "grad_norm": 2.330601930618286, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7210631370544434, + "num_tokens": 627463374.0, + "step": 24254 + }, + { + "epoch": 2.6636283768943554, + "grad_norm": 2.6130995750427246, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7344733476638794, + "num_tokens": 627486138.0, + "step": 24255 + }, + { + "epoch": 2.663738194596969, + "grad_norm": 2.3130998611450195, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7313007116317749, + "num_tokens": 627512833.0, + "step": 24256 + }, + { + "epoch": 2.6638480122995825, + "grad_norm": 2.084550142288208, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7161298394203186, + "num_tokens": 627545399.0, + "step": 24257 + }, + { + "epoch": 2.663957830002196, + "grad_norm": 2.273467779159546, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.720799446105957, + "num_tokens": 627571910.0, + "step": 24258 + }, + { + "epoch": 2.66406764770481, + "grad_norm": 2.3615241050720215, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7212982177734375, + "num_tokens": 627600017.0, + "step": 24259 + }, + { + "epoch": 2.6641774654074237, + "grad_norm": 2.1923716068267822, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7269721031188965, + "num_tokens": 627630488.0, + "step": 24260 + }, + { + "epoch": 2.6642872831100375, + "grad_norm": 2.526616334915161, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7366938591003418, + "num_tokens": 627654978.0, + "step": 24261 + }, + { + "epoch": 2.664397100812651, + "grad_norm": 2.661112070083618, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.723275363445282, + "num_tokens": 627676512.0, + "step": 24262 + }, + { + "epoch": 2.6645069185152646, + "grad_norm": 2.4495561122894287, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7091176509857178, + "num_tokens": 627702865.0, + "step": 24263 + }, + { + "epoch": 2.6646167362178783, + "grad_norm": 2.3697099685668945, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7380086183547974, + "num_tokens": 627729759.0, + "step": 24264 + }, + { + "epoch": 2.664726553920492, + "grad_norm": 2.0862386226654053, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7150430083274841, + "num_tokens": 627761297.0, + "step": 24265 + }, + { + "epoch": 2.664836371623106, + "grad_norm": 2.5660858154296875, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7398314476013184, + "num_tokens": 627783809.0, + "step": 24266 + }, + { + "epoch": 2.664946189325719, + "grad_norm": 2.460167169570923, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7400470972061157, + "num_tokens": 627808469.0, + "step": 24267 + }, + { + "epoch": 2.665056007028333, + "grad_norm": 2.328792095184326, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7451848387718201, + "num_tokens": 627832696.0, + "step": 24268 + }, + { + "epoch": 2.6651658247309467, + "grad_norm": 2.4490222930908203, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.74817955493927, + "num_tokens": 627857509.0, + "step": 24269 + }, + { + "epoch": 2.6652756424335604, + "grad_norm": 2.281172037124634, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7532768845558167, + "num_tokens": 627883020.0, + "step": 24270 + }, + { + "epoch": 2.665385460136174, + "grad_norm": 2.3980085849761963, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.729417085647583, + "num_tokens": 627907355.0, + "step": 24271 + }, + { + "epoch": 2.6654952778387875, + "grad_norm": 2.507136344909668, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7325868606567383, + "num_tokens": 627934318.0, + "step": 24272 + }, + { + "epoch": 2.6656050955414012, + "grad_norm": 2.365157127380371, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7409723997116089, + "num_tokens": 627962342.0, + "step": 24273 + }, + { + "epoch": 2.665714913244015, + "grad_norm": 2.425943613052368, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7083925604820251, + "num_tokens": 627989770.0, + "step": 24274 + }, + { + "epoch": 2.6658247309466288, + "grad_norm": 2.630035877227783, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7625414729118347, + "num_tokens": 628009019.0, + "step": 24275 + }, + { + "epoch": 2.6659345486492425, + "grad_norm": 2.0293660163879395, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7020502090454102, + "num_tokens": 628044962.0, + "step": 24276 + }, + { + "epoch": 2.666044366351856, + "grad_norm": 2.4418444633483887, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7440319061279297, + "num_tokens": 628069155.0, + "step": 24277 + }, + { + "epoch": 2.6661541840544696, + "grad_norm": 2.471522569656372, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7248228192329407, + "num_tokens": 628094640.0, + "step": 24278 + }, + { + "epoch": 2.6662640017570833, + "grad_norm": 2.422710657119751, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7220767140388489, + "num_tokens": 628118746.0, + "step": 24279 + }, + { + "epoch": 2.6663738194596966, + "grad_norm": 2.2396163940429688, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7605595588684082, + "num_tokens": 628143374.0, + "step": 24280 + }, + { + "epoch": 2.6664836371623104, + "grad_norm": 2.0705652236938477, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7438319325447083, + "num_tokens": 628175194.0, + "step": 24281 + }, + { + "epoch": 2.666593454864924, + "grad_norm": 2.4299328327178955, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7401199340820312, + "num_tokens": 628199375.0, + "step": 24282 + }, + { + "epoch": 2.666703272567538, + "grad_norm": 2.361401319503784, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7386242747306824, + "num_tokens": 628224584.0, + "step": 24283 + }, + { + "epoch": 2.6668130902701517, + "grad_norm": 2.5474588871002197, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7235234975814819, + "num_tokens": 628246848.0, + "step": 24284 + }, + { + "epoch": 2.666922907972765, + "grad_norm": 2.3617734909057617, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7032948732376099, + "num_tokens": 628274470.0, + "step": 24285 + }, + { + "epoch": 2.6670327256753787, + "grad_norm": 2.3712573051452637, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7496470808982849, + "num_tokens": 628299899.0, + "step": 24286 + }, + { + "epoch": 2.6671425433779925, + "grad_norm": 2.4455935955047607, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7144503593444824, + "num_tokens": 628323819.0, + "step": 24287 + }, + { + "epoch": 2.6672523610806063, + "grad_norm": 2.5049781799316406, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7360204458236694, + "num_tokens": 628346219.0, + "step": 24288 + }, + { + "epoch": 2.66736217878322, + "grad_norm": 2.3498003482818604, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7325524091720581, + "num_tokens": 628370534.0, + "step": 24289 + }, + { + "epoch": 2.6674719964858333, + "grad_norm": 2.3159499168395996, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7212097644805908, + "num_tokens": 628397247.0, + "step": 24290 + }, + { + "epoch": 2.667581814188447, + "grad_norm": 2.206643581390381, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7346034646034241, + "num_tokens": 628424134.0, + "step": 24291 + }, + { + "epoch": 2.667691631891061, + "grad_norm": 2.2731471061706543, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7530056834220886, + "num_tokens": 628451886.0, + "step": 24292 + }, + { + "epoch": 2.6678014495936746, + "grad_norm": 2.3881964683532715, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7359430193901062, + "num_tokens": 628475401.0, + "step": 24293 + }, + { + "epoch": 2.6679112672962884, + "grad_norm": 2.6239848136901855, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7298946380615234, + "num_tokens": 628496463.0, + "step": 24294 + }, + { + "epoch": 2.6680210849989017, + "grad_norm": 2.66385817527771, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7439869046211243, + "num_tokens": 628518543.0, + "step": 24295 + }, + { + "epoch": 2.6681309027015154, + "grad_norm": 2.2907040119171143, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7259087562561035, + "num_tokens": 628546183.0, + "step": 24296 + }, + { + "epoch": 2.668240720404129, + "grad_norm": 2.4687137603759766, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7340710163116455, + "num_tokens": 628569383.0, + "step": 24297 + }, + { + "epoch": 2.668350538106743, + "grad_norm": 2.446838140487671, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7426806688308716, + "num_tokens": 628592940.0, + "step": 24298 + }, + { + "epoch": 2.6684603558093567, + "grad_norm": 2.3374528884887695, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7252713441848755, + "num_tokens": 628620055.0, + "step": 24299 + }, + { + "epoch": 2.66857017351197, + "grad_norm": 2.3342175483703613, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7356507182121277, + "num_tokens": 628645968.0, + "step": 24300 + }, + { + "epoch": 2.6686799912145838, + "grad_norm": 2.487316131591797, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7275037169456482, + "num_tokens": 628667921.0, + "step": 24301 + }, + { + "epoch": 2.6687898089171975, + "grad_norm": 2.4392013549804688, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7414347529411316, + "num_tokens": 628693110.0, + "step": 24302 + }, + { + "epoch": 2.668899626619811, + "grad_norm": 2.4822983741760254, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7210061550140381, + "num_tokens": 628716271.0, + "step": 24303 + }, + { + "epoch": 2.669009444322425, + "grad_norm": 2.3405985832214355, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7263604998588562, + "num_tokens": 628741671.0, + "step": 24304 + }, + { + "epoch": 2.6691192620250384, + "grad_norm": 2.3373184204101562, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7137514352798462, + "num_tokens": 628768746.0, + "step": 24305 + }, + { + "epoch": 2.669229079727652, + "grad_norm": 2.5028178691864014, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7527247667312622, + "num_tokens": 628792673.0, + "step": 24306 + }, + { + "epoch": 2.669338897430266, + "grad_norm": 2.3149983882904053, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7201178073883057, + "num_tokens": 628817965.0, + "step": 24307 + }, + { + "epoch": 2.669448715132879, + "grad_norm": 2.017322540283203, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.705310583114624, + "num_tokens": 628852368.0, + "step": 24308 + }, + { + "epoch": 2.669558532835493, + "grad_norm": 2.6386799812316895, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7345975637435913, + "num_tokens": 628876487.0, + "step": 24309 + }, + { + "epoch": 2.6696683505381067, + "grad_norm": 2.1945202350616455, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7294754981994629, + "num_tokens": 628904968.0, + "step": 24310 + }, + { + "epoch": 2.6697781682407205, + "grad_norm": 2.0801124572753906, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7055667638778687, + "num_tokens": 628936609.0, + "step": 24311 + }, + { + "epoch": 2.669887985943334, + "grad_norm": 2.6384799480438232, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7613301277160645, + "num_tokens": 628956272.0, + "step": 24312 + }, + { + "epoch": 2.6699978036459475, + "grad_norm": 2.5400848388671875, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7361712455749512, + "num_tokens": 628978794.0, + "step": 24313 + }, + { + "epoch": 2.6701076213485613, + "grad_norm": 2.5119259357452393, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7220677733421326, + "num_tokens": 629004138.0, + "step": 24314 + }, + { + "epoch": 2.670217439051175, + "grad_norm": 2.3890507221221924, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7276901602745056, + "num_tokens": 629030168.0, + "step": 24315 + }, + { + "epoch": 2.670327256753789, + "grad_norm": 2.3136842250823975, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.728168249130249, + "num_tokens": 629057981.0, + "step": 24316 + }, + { + "epoch": 2.6704370744564025, + "grad_norm": 2.197174549102783, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7370433807373047, + "num_tokens": 629086282.0, + "step": 24317 + }, + { + "epoch": 2.670546892159016, + "grad_norm": 2.4548563957214355, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.739546000957489, + "num_tokens": 629111573.0, + "step": 24318 + }, + { + "epoch": 2.6706567098616296, + "grad_norm": 2.5624678134918213, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7332624197006226, + "num_tokens": 629134667.0, + "step": 24319 + }, + { + "epoch": 2.6707665275642434, + "grad_norm": 2.0818257331848145, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7183021306991577, + "num_tokens": 629169421.0, + "step": 24320 + }, + { + "epoch": 2.670876345266857, + "grad_norm": 2.69203519821167, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7444597482681274, + "num_tokens": 629190802.0, + "step": 24321 + }, + { + "epoch": 2.670986162969471, + "grad_norm": 2.5389928817749023, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7264224886894226, + "num_tokens": 629214455.0, + "step": 24322 + }, + { + "epoch": 2.671095980672084, + "grad_norm": 2.3477213382720947, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7412249445915222, + "num_tokens": 629239574.0, + "step": 24323 + }, + { + "epoch": 2.671205798374698, + "grad_norm": 2.2853031158447266, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7122129797935486, + "num_tokens": 629266362.0, + "step": 24324 + }, + { + "epoch": 2.6713156160773117, + "grad_norm": 2.413407802581787, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7347569465637207, + "num_tokens": 629290907.0, + "step": 24325 + }, + { + "epoch": 2.6714254337799255, + "grad_norm": 2.344303846359253, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7357621788978577, + "num_tokens": 629317945.0, + "step": 24326 + }, + { + "epoch": 2.6715352514825392, + "grad_norm": 2.4851090908050537, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7361838817596436, + "num_tokens": 629340176.0, + "step": 24327 + }, + { + "epoch": 2.6716450691851525, + "grad_norm": 2.359565019607544, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.6988798379898071, + "num_tokens": 629367833.0, + "step": 24328 + }, + { + "epoch": 2.6717548868877663, + "grad_norm": 2.352123498916626, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7222066521644592, + "num_tokens": 629395065.0, + "step": 24329 + }, + { + "epoch": 2.67186470459038, + "grad_norm": 2.444397449493408, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7351095676422119, + "num_tokens": 629420042.0, + "step": 24330 + }, + { + "epoch": 2.6719745222929934, + "grad_norm": 2.8765645027160645, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7607091665267944, + "num_tokens": 629438858.0, + "step": 24331 + }, + { + "epoch": 2.672084339995607, + "grad_norm": 2.360746383666992, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7367958426475525, + "num_tokens": 629463482.0, + "step": 24332 + }, + { + "epoch": 2.672194157698221, + "grad_norm": 2.4026808738708496, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7240777015686035, + "num_tokens": 629489477.0, + "step": 24333 + }, + { + "epoch": 2.6723039754008346, + "grad_norm": 2.1217312812805176, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7478492856025696, + "num_tokens": 629516528.0, + "step": 24334 + }, + { + "epoch": 2.6724137931034484, + "grad_norm": 2.6010143756866455, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7512787580490112, + "num_tokens": 629538327.0, + "step": 24335 + }, + { + "epoch": 2.6725236108060617, + "grad_norm": 2.804600477218628, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7409049868583679, + "num_tokens": 629558109.0, + "step": 24336 + }, + { + "epoch": 2.6726334285086755, + "grad_norm": 2.48018741607666, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7520538568496704, + "num_tokens": 629581221.0, + "step": 24337 + }, + { + "epoch": 2.6727432462112892, + "grad_norm": 2.261629104614258, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7208980321884155, + "num_tokens": 629610989.0, + "step": 24338 + }, + { + "epoch": 2.672853063913903, + "grad_norm": 2.2980453968048096, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.6982632875442505, + "num_tokens": 629639140.0, + "step": 24339 + }, + { + "epoch": 2.6729628816165167, + "grad_norm": 2.772563934326172, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7457113265991211, + "num_tokens": 629658917.0, + "step": 24340 + }, + { + "epoch": 2.67307269931913, + "grad_norm": 2.5291924476623535, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7407041788101196, + "num_tokens": 629683062.0, + "step": 24341 + }, + { + "epoch": 2.673182517021744, + "grad_norm": 2.2586774826049805, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7070058584213257, + "num_tokens": 629717786.0, + "step": 24342 + }, + { + "epoch": 2.6732923347243576, + "grad_norm": 2.658555746078491, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7436336278915405, + "num_tokens": 629738750.0, + "step": 24343 + }, + { + "epoch": 2.6734021524269713, + "grad_norm": 2.7056946754455566, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7208335399627686, + "num_tokens": 629759552.0, + "step": 24344 + }, + { + "epoch": 2.673511970129585, + "grad_norm": 2.327646017074585, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7442066073417664, + "num_tokens": 629787628.0, + "step": 24345 + }, + { + "epoch": 2.6736217878321984, + "grad_norm": 2.360393762588501, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7375705242156982, + "num_tokens": 629812867.0, + "step": 24346 + }, + { + "epoch": 2.673731605534812, + "grad_norm": 2.3610928058624268, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7251377105712891, + "num_tokens": 629839150.0, + "step": 24347 + }, + { + "epoch": 2.673841423237426, + "grad_norm": 2.388791799545288, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.727060079574585, + "num_tokens": 629863973.0, + "step": 24348 + }, + { + "epoch": 2.6739512409400397, + "grad_norm": 2.49249267578125, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7307918071746826, + "num_tokens": 629888021.0, + "step": 24349 + }, + { + "epoch": 2.6740610586426534, + "grad_norm": 2.097663640975952, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.71135413646698, + "num_tokens": 629920231.0, + "step": 24350 + }, + { + "epoch": 2.6741708763452667, + "grad_norm": 2.237122058868408, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7270380854606628, + "num_tokens": 629947745.0, + "step": 24351 + }, + { + "epoch": 2.6742806940478805, + "grad_norm": 2.6187381744384766, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7316824793815613, + "num_tokens": 629969679.0, + "step": 24352 + }, + { + "epoch": 2.6743905117504942, + "grad_norm": 2.269643783569336, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7095316648483276, + "num_tokens": 630001784.0, + "step": 24353 + }, + { + "epoch": 2.674500329453108, + "grad_norm": 2.2329654693603516, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7148100137710571, + "num_tokens": 630033023.0, + "step": 24354 + }, + { + "epoch": 2.6746101471557218, + "grad_norm": 2.6339924335479736, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7245025634765625, + "num_tokens": 630059274.0, + "step": 24355 + }, + { + "epoch": 2.674719964858335, + "grad_norm": 2.451167345046997, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.720587432384491, + "num_tokens": 630085439.0, + "step": 24356 + }, + { + "epoch": 2.674829782560949, + "grad_norm": 2.348877191543579, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7166191339492798, + "num_tokens": 630111706.0, + "step": 24357 + }, + { + "epoch": 2.6749396002635626, + "grad_norm": 2.208890914916992, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7275984287261963, + "num_tokens": 630140800.0, + "step": 24358 + }, + { + "epoch": 2.675049417966176, + "grad_norm": 2.150038242340088, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7211451530456543, + "num_tokens": 630170430.0, + "step": 24359 + }, + { + "epoch": 2.6751592356687897, + "grad_norm": 2.493671178817749, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7427067756652832, + "num_tokens": 630192434.0, + "step": 24360 + }, + { + "epoch": 2.6752690533714034, + "grad_norm": 2.4034154415130615, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7374546527862549, + "num_tokens": 630216439.0, + "step": 24361 + }, + { + "epoch": 2.675378871074017, + "grad_norm": 2.2500269412994385, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7145241498947144, + "num_tokens": 630244907.0, + "step": 24362 + }, + { + "epoch": 2.675488688776631, + "grad_norm": 2.2099807262420654, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.731658935546875, + "num_tokens": 630273273.0, + "step": 24363 + }, + { + "epoch": 2.6755985064792442, + "grad_norm": 2.3118178844451904, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7279006242752075, + "num_tokens": 630299721.0, + "step": 24364 + }, + { + "epoch": 2.675708324181858, + "grad_norm": 2.422238349914551, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7230173349380493, + "num_tokens": 630325033.0, + "step": 24365 + }, + { + "epoch": 2.6758181418844718, + "grad_norm": 2.376343250274658, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7218323945999146, + "num_tokens": 630351318.0, + "step": 24366 + }, + { + "epoch": 2.6759279595870855, + "grad_norm": 2.282835006713867, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7242778539657593, + "num_tokens": 630378922.0, + "step": 24367 + }, + { + "epoch": 2.6760377772896993, + "grad_norm": 2.1345295906066895, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.71683669090271, + "num_tokens": 630408757.0, + "step": 24368 + }, + { + "epoch": 2.6761475949923126, + "grad_norm": 2.3945343494415283, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7433604598045349, + "num_tokens": 630435441.0, + "step": 24369 + }, + { + "epoch": 2.6762574126949263, + "grad_norm": 2.31335711479187, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.714819610118866, + "num_tokens": 630462981.0, + "step": 24370 + }, + { + "epoch": 2.67636723039754, + "grad_norm": 2.612236976623535, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7497649192810059, + "num_tokens": 630484003.0, + "step": 24371 + }, + { + "epoch": 2.676477048100154, + "grad_norm": 2.4285964965820312, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7282058000564575, + "num_tokens": 630510191.0, + "step": 24372 + }, + { + "epoch": 2.6765868658027676, + "grad_norm": 2.590867042541504, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7368427515029907, + "num_tokens": 630532773.0, + "step": 24373 + }, + { + "epoch": 2.676696683505381, + "grad_norm": 2.4112088680267334, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7434043884277344, + "num_tokens": 630557174.0, + "step": 24374 + }, + { + "epoch": 2.6768065012079947, + "grad_norm": 2.4459228515625, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7431560754776001, + "num_tokens": 630580489.0, + "step": 24375 + }, + { + "epoch": 2.6769163189106084, + "grad_norm": 2.276557207107544, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.72881019115448, + "num_tokens": 630609870.0, + "step": 24376 + }, + { + "epoch": 2.677026136613222, + "grad_norm": 2.615751028060913, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7303003072738647, + "num_tokens": 630630976.0, + "step": 24377 + }, + { + "epoch": 2.677135954315836, + "grad_norm": 2.723731279373169, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7465263605117798, + "num_tokens": 630651152.0, + "step": 24378 + }, + { + "epoch": 2.6772457720184493, + "grad_norm": 2.2798688411712646, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7100954055786133, + "num_tokens": 630679285.0, + "step": 24379 + }, + { + "epoch": 2.677355589721063, + "grad_norm": 2.2819910049438477, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7159159779548645, + "num_tokens": 630706399.0, + "step": 24380 + }, + { + "epoch": 2.677465407423677, + "grad_norm": 2.2618513107299805, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.765912652015686, + "num_tokens": 630732261.0, + "step": 24381 + }, + { + "epoch": 2.67757522512629, + "grad_norm": 2.4353716373443604, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7436292767524719, + "num_tokens": 630755814.0, + "step": 24382 + }, + { + "epoch": 2.6776850428289043, + "grad_norm": 2.2996785640716553, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7209792137145996, + "num_tokens": 630782783.0, + "step": 24383 + }, + { + "epoch": 2.6777948605315176, + "grad_norm": 2.0508620738983154, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.709606945514679, + "num_tokens": 630815627.0, + "step": 24384 + }, + { + "epoch": 2.6779046782341314, + "grad_norm": 3.0311105251312256, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7521543502807617, + "num_tokens": 630832948.0, + "step": 24385 + }, + { + "epoch": 2.678014495936745, + "grad_norm": 2.308889865875244, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7384080290794373, + "num_tokens": 630860164.0, + "step": 24386 + }, + { + "epoch": 2.6781243136393584, + "grad_norm": 2.287034749984741, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7258426547050476, + "num_tokens": 630889955.0, + "step": 24387 + }, + { + "epoch": 2.678234131341972, + "grad_norm": 2.2894127368927, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7165408134460449, + "num_tokens": 630917747.0, + "step": 24388 + }, + { + "epoch": 2.678343949044586, + "grad_norm": 2.230069637298584, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7198067903518677, + "num_tokens": 630945446.0, + "step": 24389 + }, + { + "epoch": 2.6784537667471997, + "grad_norm": 2.313717842102051, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7064205408096313, + "num_tokens": 630973759.0, + "step": 24390 + }, + { + "epoch": 2.6785635844498135, + "grad_norm": 2.3362433910369873, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7198780179023743, + "num_tokens": 631001578.0, + "step": 24391 + }, + { + "epoch": 2.6786734021524268, + "grad_norm": 2.050122022628784, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7298116683959961, + "num_tokens": 631032517.0, + "step": 24392 + }, + { + "epoch": 2.6787832198550405, + "grad_norm": 2.643529176712036, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7294812798500061, + "num_tokens": 631054209.0, + "step": 24393 + }, + { + "epoch": 2.6788930375576543, + "grad_norm": 2.9180822372436523, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7530050277709961, + "num_tokens": 631073210.0, + "step": 24394 + }, + { + "epoch": 2.679002855260268, + "grad_norm": 2.2728757858276367, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7349177598953247, + "num_tokens": 631099917.0, + "step": 24395 + }, + { + "epoch": 2.679112672962882, + "grad_norm": 2.3203654289245605, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7018128633499146, + "num_tokens": 631128902.0, + "step": 24396 + }, + { + "epoch": 2.679222490665495, + "grad_norm": 2.2290945053100586, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7088608145713806, + "num_tokens": 631157803.0, + "step": 24397 + }, + { + "epoch": 2.679332308368109, + "grad_norm": 2.6999130249023438, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7389156818389893, + "num_tokens": 631177743.0, + "step": 24398 + }, + { + "epoch": 2.6794421260707226, + "grad_norm": 2.396819829940796, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7456106543540955, + "num_tokens": 631201585.0, + "step": 24399 + }, + { + "epoch": 2.6795519437733364, + "grad_norm": 1.9636374711990356, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7404806613922119, + "num_tokens": 631235262.0, + "step": 24400 + }, + { + "epoch": 2.67966176147595, + "grad_norm": 2.5982367992401123, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7226009368896484, + "num_tokens": 631257057.0, + "step": 24401 + }, + { + "epoch": 2.6797715791785635, + "grad_norm": 2.042616367340088, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.7149199843406677, + "num_tokens": 631287863.0, + "step": 24402 + }, + { + "epoch": 2.679881396881177, + "grad_norm": 2.393380880355835, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.724504828453064, + "num_tokens": 631312985.0, + "step": 24403 + }, + { + "epoch": 2.679991214583791, + "grad_norm": 2.538365364074707, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7308337092399597, + "num_tokens": 631337562.0, + "step": 24404 + }, + { + "epoch": 2.6801010322864047, + "grad_norm": 2.7407050132751465, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7453356385231018, + "num_tokens": 631357719.0, + "step": 24405 + }, + { + "epoch": 2.6802108499890185, + "grad_norm": 2.080960750579834, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7284880876541138, + "num_tokens": 631389130.0, + "step": 24406 + }, + { + "epoch": 2.680320667691632, + "grad_norm": 2.645416259765625, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.738267719745636, + "num_tokens": 631410423.0, + "step": 24407 + }, + { + "epoch": 2.6804304853942456, + "grad_norm": 2.1958212852478027, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7498217225074768, + "num_tokens": 631437968.0, + "step": 24408 + }, + { + "epoch": 2.6805403030968593, + "grad_norm": 2.2237565517425537, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7612849473953247, + "num_tokens": 631463678.0, + "step": 24409 + }, + { + "epoch": 2.6806501207994726, + "grad_norm": 2.1697824001312256, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7368679046630859, + "num_tokens": 631494062.0, + "step": 24410 + }, + { + "epoch": 2.6807599385020864, + "grad_norm": 2.4025843143463135, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7348347902297974, + "num_tokens": 631519307.0, + "step": 24411 + }, + { + "epoch": 2.6808697562047, + "grad_norm": 2.5693295001983643, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7423909306526184, + "num_tokens": 631542890.0, + "step": 24412 + }, + { + "epoch": 2.680979573907314, + "grad_norm": 2.0820975303649902, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.702796459197998, + "num_tokens": 631577584.0, + "step": 24413 + }, + { + "epoch": 2.6810893916099277, + "grad_norm": 2.5598738193511963, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7199819684028625, + "num_tokens": 631602628.0, + "step": 24414 + }, + { + "epoch": 2.681199209312541, + "grad_norm": 2.471848249435425, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7422370314598083, + "num_tokens": 631625565.0, + "step": 24415 + }, + { + "epoch": 2.6813090270151547, + "grad_norm": 2.5644850730895996, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7393261194229126, + "num_tokens": 631648466.0, + "step": 24416 + }, + { + "epoch": 2.6814188447177685, + "grad_norm": 2.381169557571411, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7228856086730957, + "num_tokens": 631672970.0, + "step": 24417 + }, + { + "epoch": 2.6815286624203822, + "grad_norm": 2.2931160926818848, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7186452746391296, + "num_tokens": 631700133.0, + "step": 24418 + }, + { + "epoch": 2.681638480122996, + "grad_norm": 2.353255271911621, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7054467797279358, + "num_tokens": 631726860.0, + "step": 24419 + }, + { + "epoch": 2.6817482978256093, + "grad_norm": 2.158374309539795, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7248570919036865, + "num_tokens": 631754956.0, + "step": 24420 + }, + { + "epoch": 2.681858115528223, + "grad_norm": 2.5399811267852783, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7588231563568115, + "num_tokens": 631775773.0, + "step": 24421 + }, + { + "epoch": 2.681967933230837, + "grad_norm": 2.5173521041870117, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7379229664802551, + "num_tokens": 631798551.0, + "step": 24422 + }, + { + "epoch": 2.6820777509334506, + "grad_norm": 2.247870683670044, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7295215725898743, + "num_tokens": 631826646.0, + "step": 24423 + }, + { + "epoch": 2.6821875686360643, + "grad_norm": 1.9529894590377808, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.7596926689147949, + "num_tokens": 631858987.0, + "step": 24424 + }, + { + "epoch": 2.6822973863386776, + "grad_norm": 2.499474048614502, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7269031405448914, + "num_tokens": 631882069.0, + "step": 24425 + }, + { + "epoch": 2.6824072040412914, + "grad_norm": 2.17453932762146, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7283890247344971, + "num_tokens": 631911029.0, + "step": 24426 + }, + { + "epoch": 2.682517021743905, + "grad_norm": 2.4166409969329834, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7019869685173035, + "num_tokens": 631937513.0, + "step": 24427 + }, + { + "epoch": 2.682626839446519, + "grad_norm": 2.2002079486846924, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7459132075309753, + "num_tokens": 631964628.0, + "step": 24428 + }, + { + "epoch": 2.6827366571491327, + "grad_norm": 2.6301231384277344, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7275643944740295, + "num_tokens": 631987694.0, + "step": 24429 + }, + { + "epoch": 2.682846474851746, + "grad_norm": 2.3160459995269775, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7149043083190918, + "num_tokens": 632018211.0, + "step": 24430 + }, + { + "epoch": 2.6829562925543597, + "grad_norm": 2.2162227630615234, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7206428647041321, + "num_tokens": 632047901.0, + "step": 24431 + }, + { + "epoch": 2.6830661102569735, + "grad_norm": 2.654025077819824, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7574372291564941, + "num_tokens": 632068329.0, + "step": 24432 + }, + { + "epoch": 2.683175927959587, + "grad_norm": 2.36944317817688, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7241060137748718, + "num_tokens": 632094069.0, + "step": 24433 + }, + { + "epoch": 2.683285745662201, + "grad_norm": 2.1668596267700195, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7284035086631775, + "num_tokens": 632123887.0, + "step": 24434 + }, + { + "epoch": 2.6833955633648143, + "grad_norm": 2.29901385307312, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6979891061782837, + "num_tokens": 632151927.0, + "step": 24435 + }, + { + "epoch": 2.683505381067428, + "grad_norm": 2.732978582382202, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7434983253479004, + "num_tokens": 632172273.0, + "step": 24436 + }, + { + "epoch": 2.683615198770042, + "grad_norm": 2.376049041748047, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7235405445098877, + "num_tokens": 632200268.0, + "step": 24437 + }, + { + "epoch": 2.683725016472655, + "grad_norm": 2.13976788520813, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7124067544937134, + "num_tokens": 632232815.0, + "step": 24438 + }, + { + "epoch": 2.683834834175269, + "grad_norm": 2.412034273147583, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7308077216148376, + "num_tokens": 632256901.0, + "step": 24439 + }, + { + "epoch": 2.6839446518778827, + "grad_norm": 2.3197712898254395, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7286255955696106, + "num_tokens": 632282897.0, + "step": 24440 + }, + { + "epoch": 2.6840544695804964, + "grad_norm": 2.5171456336975098, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7326685786247253, + "num_tokens": 632307063.0, + "step": 24441 + }, + { + "epoch": 2.68416428728311, + "grad_norm": 2.21055269241333, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7335078120231628, + "num_tokens": 632336231.0, + "step": 24442 + }, + { + "epoch": 2.6842741049857235, + "grad_norm": 2.183445692062378, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7364264130592346, + "num_tokens": 632365393.0, + "step": 24443 + }, + { + "epoch": 2.6843839226883373, + "grad_norm": 2.3595523834228516, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7039327621459961, + "num_tokens": 632392173.0, + "step": 24444 + }, + { + "epoch": 2.684493740390951, + "grad_norm": 2.3550195693969727, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.715282142162323, + "num_tokens": 632419194.0, + "step": 24445 + }, + { + "epoch": 2.6846035580935648, + "grad_norm": 2.010507345199585, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7285967469215393, + "num_tokens": 632452581.0, + "step": 24446 + }, + { + "epoch": 2.6847133757961785, + "grad_norm": 2.0418708324432373, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7070183753967285, + "num_tokens": 632485790.0, + "step": 24447 + }, + { + "epoch": 2.684823193498792, + "grad_norm": 2.7769358158111572, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7528796195983887, + "num_tokens": 632504942.0, + "step": 24448 + }, + { + "epoch": 2.6849330112014056, + "grad_norm": 2.336913824081421, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7481794953346252, + "num_tokens": 632528163.0, + "step": 24449 + }, + { + "epoch": 2.6850428289040194, + "grad_norm": 2.2159245014190674, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.720561146736145, + "num_tokens": 632556386.0, + "step": 24450 + }, + { + "epoch": 2.685152646606633, + "grad_norm": 2.664400339126587, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7368719577789307, + "num_tokens": 632578355.0, + "step": 24451 + }, + { + "epoch": 2.685262464309247, + "grad_norm": 2.33451509475708, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7573267221450806, + "num_tokens": 632603086.0, + "step": 24452 + }, + { + "epoch": 2.68537228201186, + "grad_norm": 2.311314105987549, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7327386140823364, + "num_tokens": 632631197.0, + "step": 24453 + }, + { + "epoch": 2.685482099714474, + "grad_norm": 2.23108172416687, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7211527824401855, + "num_tokens": 632659157.0, + "step": 24454 + }, + { + "epoch": 2.6855919174170877, + "grad_norm": 2.415642499923706, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.709539532661438, + "num_tokens": 632684876.0, + "step": 24455 + }, + { + "epoch": 2.6857017351197014, + "grad_norm": 2.8083832263946533, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7493166327476501, + "num_tokens": 632704324.0, + "step": 24456 + }, + { + "epoch": 2.685811552822315, + "grad_norm": 2.186631202697754, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.736156702041626, + "num_tokens": 632732603.0, + "step": 24457 + }, + { + "epoch": 2.6859213705249285, + "grad_norm": 2.0329582691192627, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7237897515296936, + "num_tokens": 632766272.0, + "step": 24458 + }, + { + "epoch": 2.6860311882275423, + "grad_norm": 2.28467059135437, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7284048795700073, + "num_tokens": 632793284.0, + "step": 24459 + }, + { + "epoch": 2.686141005930156, + "grad_norm": 2.2235665321350098, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6995095014572144, + "num_tokens": 632822643.0, + "step": 24460 + }, + { + "epoch": 2.6862508236327693, + "grad_norm": 2.358295202255249, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7226605415344238, + "num_tokens": 632848903.0, + "step": 24461 + }, + { + "epoch": 2.686360641335383, + "grad_norm": 2.4847981929779053, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.752900242805481, + "num_tokens": 632871012.0, + "step": 24462 + }, + { + "epoch": 2.686470459037997, + "grad_norm": 2.5716500282287598, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7435865998268127, + "num_tokens": 632893130.0, + "step": 24463 + }, + { + "epoch": 2.6865802767406106, + "grad_norm": 2.2173914909362793, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7130461931228638, + "num_tokens": 632921317.0, + "step": 24464 + }, + { + "epoch": 2.6866900944432244, + "grad_norm": 2.2461373805999756, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7373749613761902, + "num_tokens": 632949586.0, + "step": 24465 + }, + { + "epoch": 2.6867999121458377, + "grad_norm": 2.490288019180298, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7103938460350037, + "num_tokens": 632975352.0, + "step": 24466 + }, + { + "epoch": 2.6869097298484514, + "grad_norm": 2.1063194274902344, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7489408850669861, + "num_tokens": 633004096.0, + "step": 24467 + }, + { + "epoch": 2.687019547551065, + "grad_norm": 2.0515289306640625, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.709155261516571, + "num_tokens": 633035934.0, + "step": 24468 + }, + { + "epoch": 2.687129365253679, + "grad_norm": 2.1188220977783203, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7390161156654358, + "num_tokens": 633066967.0, + "step": 24469 + }, + { + "epoch": 2.6872391829562927, + "grad_norm": 2.4827933311462402, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7276893854141235, + "num_tokens": 633091102.0, + "step": 24470 + }, + { + "epoch": 2.687349000658906, + "grad_norm": 2.3302109241485596, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7501434087753296, + "num_tokens": 633115293.0, + "step": 24471 + }, + { + "epoch": 2.68745881836152, + "grad_norm": 2.632891893386841, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7395064830780029, + "num_tokens": 633138373.0, + "step": 24472 + }, + { + "epoch": 2.6875686360641335, + "grad_norm": 2.398918628692627, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7388426065444946, + "num_tokens": 633162004.0, + "step": 24473 + }, + { + "epoch": 2.6876784537667473, + "grad_norm": 2.4340248107910156, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7376761436462402, + "num_tokens": 633185919.0, + "step": 24474 + }, + { + "epoch": 2.687788271469361, + "grad_norm": 2.2510573863983154, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7297375202178955, + "num_tokens": 633214269.0, + "step": 24475 + }, + { + "epoch": 2.6878980891719744, + "grad_norm": 2.271406650543213, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7235056161880493, + "num_tokens": 633242678.0, + "step": 24476 + }, + { + "epoch": 2.688007906874588, + "grad_norm": 2.584843397140503, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7188980579376221, + "num_tokens": 633264820.0, + "step": 24477 + }, + { + "epoch": 2.688117724577202, + "grad_norm": 1.9418383836746216, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.714093804359436, + "num_tokens": 633298596.0, + "step": 24478 + }, + { + "epoch": 2.6882275422798156, + "grad_norm": 2.229778528213501, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7428571581840515, + "num_tokens": 633325792.0, + "step": 24479 + }, + { + "epoch": 2.6883373599824294, + "grad_norm": 2.55059814453125, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7392112016677856, + "num_tokens": 633348353.0, + "step": 24480 + }, + { + "epoch": 2.6884471776850427, + "grad_norm": 2.483381748199463, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7411429286003113, + "num_tokens": 633371318.0, + "step": 24481 + }, + { + "epoch": 2.6885569953876565, + "grad_norm": 2.0974011421203613, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7290425300598145, + "num_tokens": 633400630.0, + "step": 24482 + }, + { + "epoch": 2.6886668130902702, + "grad_norm": 2.356541872024536, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7552466988563538, + "num_tokens": 633425392.0, + "step": 24483 + }, + { + "epoch": 2.6887766307928835, + "grad_norm": 2.565781354904175, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7306145429611206, + "num_tokens": 633450601.0, + "step": 24484 + }, + { + "epoch": 2.6888864484954977, + "grad_norm": 2.393791913986206, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7367370128631592, + "num_tokens": 633475397.0, + "step": 24485 + }, + { + "epoch": 2.688996266198111, + "grad_norm": 2.4198009967803955, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7300441265106201, + "num_tokens": 633500874.0, + "step": 24486 + }, + { + "epoch": 2.689106083900725, + "grad_norm": 2.1971006393432617, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7128199338912964, + "num_tokens": 633530432.0, + "step": 24487 + }, + { + "epoch": 2.6892159016033386, + "grad_norm": 2.3391647338867188, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7389387488365173, + "num_tokens": 633557693.0, + "step": 24488 + }, + { + "epoch": 2.689325719305952, + "grad_norm": 2.275338649749756, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7395968437194824, + "num_tokens": 633584872.0, + "step": 24489 + }, + { + "epoch": 2.6894355370085656, + "grad_norm": 2.752659320831299, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7533156871795654, + "num_tokens": 633605096.0, + "step": 24490 + }, + { + "epoch": 2.6895453547111794, + "grad_norm": 2.528980016708374, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7396952509880066, + "num_tokens": 633628223.0, + "step": 24491 + }, + { + "epoch": 2.689655172413793, + "grad_norm": 2.3805503845214844, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7356665134429932, + "num_tokens": 633654837.0, + "step": 24492 + }, + { + "epoch": 2.689764990116407, + "grad_norm": 2.500758171081543, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7263945937156677, + "num_tokens": 633678396.0, + "step": 24493 + }, + { + "epoch": 2.68987480781902, + "grad_norm": 2.304060459136963, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7194397449493408, + "num_tokens": 633709747.0, + "step": 24494 + }, + { + "epoch": 2.689984625521634, + "grad_norm": 2.3273322582244873, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7028575539588928, + "num_tokens": 633738224.0, + "step": 24495 + }, + { + "epoch": 2.6900944432242477, + "grad_norm": 2.2320878505706787, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7042508125305176, + "num_tokens": 633765948.0, + "step": 24496 + }, + { + "epoch": 2.6902042609268615, + "grad_norm": 2.553819417953491, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7416048049926758, + "num_tokens": 633789191.0, + "step": 24497 + }, + { + "epoch": 2.6903140786294752, + "grad_norm": 2.871878147125244, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7648224234580994, + "num_tokens": 633806814.0, + "step": 24498 + }, + { + "epoch": 2.6904238963320886, + "grad_norm": 2.290764570236206, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.735386312007904, + "num_tokens": 633836908.0, + "step": 24499 + }, + { + "epoch": 2.6905337140347023, + "grad_norm": 2.5388343334198, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7557413578033447, + "num_tokens": 633859780.0, + "step": 24500 + }, + { + "epoch": 2.690643531737316, + "grad_norm": 2.3688995838165283, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7058708667755127, + "num_tokens": 633885859.0, + "step": 24501 + }, + { + "epoch": 2.69075334943993, + "grad_norm": 2.527974843978882, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7323799133300781, + "num_tokens": 633909121.0, + "step": 24502 + }, + { + "epoch": 2.6908631671425436, + "grad_norm": 2.2971835136413574, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7000510096549988, + "num_tokens": 633938167.0, + "step": 24503 + }, + { + "epoch": 2.690972984845157, + "grad_norm": 2.2091517448425293, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7453250288963318, + "num_tokens": 633965109.0, + "step": 24504 + }, + { + "epoch": 2.6910828025477707, + "grad_norm": 2.384436845779419, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7330725193023682, + "num_tokens": 633988185.0, + "step": 24505 + }, + { + "epoch": 2.6911926202503844, + "grad_norm": 2.5190038681030273, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7159774899482727, + "num_tokens": 634012442.0, + "step": 24506 + }, + { + "epoch": 2.691302437952998, + "grad_norm": 2.4417996406555176, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7395837306976318, + "num_tokens": 634036849.0, + "step": 24507 + }, + { + "epoch": 2.691412255655612, + "grad_norm": 2.481762647628784, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7196241617202759, + "num_tokens": 634061913.0, + "step": 24508 + }, + { + "epoch": 2.6915220733582252, + "grad_norm": 2.349479913711548, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7045525312423706, + "num_tokens": 634088505.0, + "step": 24509 + }, + { + "epoch": 2.691631891060839, + "grad_norm": 2.2049472332000732, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7247294187545776, + "num_tokens": 634119812.0, + "step": 24510 + }, + { + "epoch": 2.6917417087634528, + "grad_norm": 2.3512442111968994, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7355176210403442, + "num_tokens": 634144470.0, + "step": 24511 + }, + { + "epoch": 2.691851526466066, + "grad_norm": 2.009803533554077, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7204680442810059, + "num_tokens": 634178419.0, + "step": 24512 + }, + { + "epoch": 2.69196134416868, + "grad_norm": 2.394340991973877, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7220208644866943, + "num_tokens": 634204048.0, + "step": 24513 + }, + { + "epoch": 2.6920711618712936, + "grad_norm": 2.6459712982177734, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7551265954971313, + "num_tokens": 634224837.0, + "step": 24514 + }, + { + "epoch": 2.6921809795739073, + "grad_norm": 2.2414510250091553, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7251904010772705, + "num_tokens": 634251640.0, + "step": 24515 + }, + { + "epoch": 2.692290797276521, + "grad_norm": 2.1846489906311035, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6854625940322876, + "num_tokens": 634285192.0, + "step": 24516 + }, + { + "epoch": 2.6924006149791344, + "grad_norm": 2.222621440887451, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7305631637573242, + "num_tokens": 634313160.0, + "step": 24517 + }, + { + "epoch": 2.692510432681748, + "grad_norm": 2.746037721633911, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7591493129730225, + "num_tokens": 634334500.0, + "step": 24518 + }, + { + "epoch": 2.692620250384362, + "grad_norm": 2.27237868309021, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.72517991065979, + "num_tokens": 634361753.0, + "step": 24519 + }, + { + "epoch": 2.6927300680869757, + "grad_norm": 2.330508232116699, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7424304485321045, + "num_tokens": 634387044.0, + "step": 24520 + }, + { + "epoch": 2.6928398857895894, + "grad_norm": 2.197922468185425, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.714888334274292, + "num_tokens": 634416653.0, + "step": 24521 + }, + { + "epoch": 2.6929497034922028, + "grad_norm": 2.539034843444824, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7409475445747375, + "num_tokens": 634440434.0, + "step": 24522 + }, + { + "epoch": 2.6930595211948165, + "grad_norm": 2.2470405101776123, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7248848676681519, + "num_tokens": 634466287.0, + "step": 24523 + }, + { + "epoch": 2.6931693388974303, + "grad_norm": 2.357959032058716, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7437992095947266, + "num_tokens": 634491980.0, + "step": 24524 + }, + { + "epoch": 2.693279156600044, + "grad_norm": 2.4226160049438477, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.739609956741333, + "num_tokens": 634515161.0, + "step": 24525 + }, + { + "epoch": 2.693388974302658, + "grad_norm": 2.6626830101013184, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7317805290222168, + "num_tokens": 634536852.0, + "step": 24526 + }, + { + "epoch": 2.693498792005271, + "grad_norm": 2.415294647216797, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7411860823631287, + "num_tokens": 634561649.0, + "step": 24527 + }, + { + "epoch": 2.693608609707885, + "grad_norm": 2.2028636932373047, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7470651865005493, + "num_tokens": 634586384.0, + "step": 24528 + }, + { + "epoch": 2.6937184274104986, + "grad_norm": 2.1108500957489014, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7285798788070679, + "num_tokens": 634617927.0, + "step": 24529 + }, + { + "epoch": 2.6938282451131124, + "grad_norm": 2.26371693611145, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7428504228591919, + "num_tokens": 634644921.0, + "step": 24530 + }, + { + "epoch": 2.693938062815726, + "grad_norm": 2.704528331756592, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.752971887588501, + "num_tokens": 634665892.0, + "step": 24531 + }, + { + "epoch": 2.6940478805183394, + "grad_norm": 2.194734573364258, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7368290424346924, + "num_tokens": 634696236.0, + "step": 24532 + }, + { + "epoch": 2.694157698220953, + "grad_norm": 2.207240104675293, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7406295537948608, + "num_tokens": 634725664.0, + "step": 24533 + }, + { + "epoch": 2.694267515923567, + "grad_norm": 2.4902539253234863, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7358402013778687, + "num_tokens": 634749180.0, + "step": 24534 + }, + { + "epoch": 2.6943773336261807, + "grad_norm": 2.665916919708252, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.737185537815094, + "num_tokens": 634769158.0, + "step": 24535 + }, + { + "epoch": 2.6944871513287945, + "grad_norm": 2.4812684059143066, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7311371564865112, + "num_tokens": 634791663.0, + "step": 24536 + }, + { + "epoch": 2.6945969690314078, + "grad_norm": 2.4116506576538086, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7348846197128296, + "num_tokens": 634817045.0, + "step": 24537 + }, + { + "epoch": 2.6947067867340215, + "grad_norm": 2.750777006149292, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.735327422618866, + "num_tokens": 634838364.0, + "step": 24538 + }, + { + "epoch": 2.6948166044366353, + "grad_norm": 2.262047052383423, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7120659351348877, + "num_tokens": 634867696.0, + "step": 24539 + }, + { + "epoch": 2.6949264221392486, + "grad_norm": 2.7227935791015625, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7272341251373291, + "num_tokens": 634887761.0, + "step": 24540 + }, + { + "epoch": 2.6950362398418624, + "grad_norm": 2.586549758911133, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7367772459983826, + "num_tokens": 634910623.0, + "step": 24541 + }, + { + "epoch": 2.695146057544476, + "grad_norm": 2.502406120300293, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7306697368621826, + "num_tokens": 634933794.0, + "step": 24542 + }, + { + "epoch": 2.69525587524709, + "grad_norm": 2.243311882019043, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7033864855766296, + "num_tokens": 634963445.0, + "step": 24543 + }, + { + "epoch": 2.6953656929497036, + "grad_norm": 2.532484531402588, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7424410581588745, + "num_tokens": 634984866.0, + "step": 24544 + }, + { + "epoch": 2.695475510652317, + "grad_norm": 2.3374712467193604, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7389603853225708, + "num_tokens": 635012109.0, + "step": 24545 + }, + { + "epoch": 2.6955853283549307, + "grad_norm": 2.260054588317871, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7201368808746338, + "num_tokens": 635040466.0, + "step": 24546 + }, + { + "epoch": 2.6956951460575445, + "grad_norm": 2.500040054321289, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7379188537597656, + "num_tokens": 635064335.0, + "step": 24547 + }, + { + "epoch": 2.695804963760158, + "grad_norm": 2.3791699409484863, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7207780480384827, + "num_tokens": 635091295.0, + "step": 24548 + }, + { + "epoch": 2.695914781462772, + "grad_norm": 2.2058069705963135, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7168041467666626, + "num_tokens": 635122405.0, + "step": 24549 + }, + { + "epoch": 2.6960245991653853, + "grad_norm": 2.9730172157287598, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7554515600204468, + "num_tokens": 635143309.0, + "step": 24550 + }, + { + "epoch": 2.696134416867999, + "grad_norm": 2.2539327144622803, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7359544634819031, + "num_tokens": 635172218.0, + "step": 24551 + }, + { + "epoch": 2.696244234570613, + "grad_norm": 2.031059980392456, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7068760395050049, + "num_tokens": 635206805.0, + "step": 24552 + }, + { + "epoch": 2.6963540522732266, + "grad_norm": 2.320159673690796, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7277244329452515, + "num_tokens": 635233987.0, + "step": 24553 + }, + { + "epoch": 2.6964638699758403, + "grad_norm": 2.5507652759552, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7105635404586792, + "num_tokens": 635257790.0, + "step": 24554 + }, + { + "epoch": 2.6965736876784536, + "grad_norm": 2.573148727416992, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7377321720123291, + "num_tokens": 635280021.0, + "step": 24555 + }, + { + "epoch": 2.6966835053810674, + "grad_norm": 2.219031572341919, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7132729291915894, + "num_tokens": 635309456.0, + "step": 24556 + }, + { + "epoch": 2.696793323083681, + "grad_norm": 2.477954626083374, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7358444929122925, + "num_tokens": 635331597.0, + "step": 24557 + }, + { + "epoch": 2.696903140786295, + "grad_norm": 2.8724536895751953, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7351967096328735, + "num_tokens": 635352324.0, + "step": 24558 + }, + { + "epoch": 2.6970129584889087, + "grad_norm": 2.562141180038452, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7429459095001221, + "num_tokens": 635373940.0, + "step": 24559 + }, + { + "epoch": 2.697122776191522, + "grad_norm": 2.0238218307495117, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7226373553276062, + "num_tokens": 635407931.0, + "step": 24560 + }, + { + "epoch": 2.6972325938941357, + "grad_norm": 2.1243224143981934, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7254804968833923, + "num_tokens": 635437344.0, + "step": 24561 + }, + { + "epoch": 2.6973424115967495, + "grad_norm": 2.4520585536956787, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7187075614929199, + "num_tokens": 635462421.0, + "step": 24562 + }, + { + "epoch": 2.697452229299363, + "grad_norm": 2.799884557723999, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7463960647583008, + "num_tokens": 635481273.0, + "step": 24563 + }, + { + "epoch": 2.697562047001977, + "grad_norm": 2.4518253803253174, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7263321876525879, + "num_tokens": 635505387.0, + "step": 24564 + }, + { + "epoch": 2.6976718647045903, + "grad_norm": 2.3000009059906006, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7108807563781738, + "num_tokens": 635533811.0, + "step": 24565 + }, + { + "epoch": 2.697781682407204, + "grad_norm": 2.2689523696899414, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7311354875564575, + "num_tokens": 635561271.0, + "step": 24566 + }, + { + "epoch": 2.697891500109818, + "grad_norm": 2.4338674545288086, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.720655083656311, + "num_tokens": 635584208.0, + "step": 24567 + }, + { + "epoch": 2.698001317812431, + "grad_norm": 2.234409809112549, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7546982765197754, + "num_tokens": 635611111.0, + "step": 24568 + }, + { + "epoch": 2.698111135515045, + "grad_norm": 2.248493194580078, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7379605770111084, + "num_tokens": 635638897.0, + "step": 24569 + }, + { + "epoch": 2.6982209532176586, + "grad_norm": 2.104668617248535, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7314172983169556, + "num_tokens": 635670072.0, + "step": 24570 + }, + { + "epoch": 2.6983307709202724, + "grad_norm": 2.816201686859131, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7489050030708313, + "num_tokens": 635688613.0, + "step": 24571 + }, + { + "epoch": 2.698440588622886, + "grad_norm": 2.5337319374084473, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7432497143745422, + "num_tokens": 635711296.0, + "step": 24572 + }, + { + "epoch": 2.6985504063254995, + "grad_norm": 2.185202121734619, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7321959137916565, + "num_tokens": 635741297.0, + "step": 24573 + }, + { + "epoch": 2.6986602240281132, + "grad_norm": 2.081609010696411, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7182954549789429, + "num_tokens": 635776310.0, + "step": 24574 + }, + { + "epoch": 2.698770041730727, + "grad_norm": 2.3855409622192383, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7276002168655396, + "num_tokens": 635799526.0, + "step": 24575 + }, + { + "epoch": 2.6988798594333407, + "grad_norm": 2.3616414070129395, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7196709513664246, + "num_tokens": 635827342.0, + "step": 24576 + }, + { + "epoch": 2.6989896771359545, + "grad_norm": 2.308743953704834, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7303959131240845, + "num_tokens": 635853904.0, + "step": 24577 + }, + { + "epoch": 2.699099494838568, + "grad_norm": 2.3054916858673096, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7284409999847412, + "num_tokens": 635882300.0, + "step": 24578 + }, + { + "epoch": 2.6992093125411816, + "grad_norm": 2.205148220062256, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7439392805099487, + "num_tokens": 635911264.0, + "step": 24579 + }, + { + "epoch": 2.6993191302437953, + "grad_norm": 2.2167811393737793, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7350192666053772, + "num_tokens": 635937409.0, + "step": 24580 + }, + { + "epoch": 2.699428947946409, + "grad_norm": 2.3412222862243652, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7253934144973755, + "num_tokens": 635962176.0, + "step": 24581 + }, + { + "epoch": 2.699538765649023, + "grad_norm": 2.494832992553711, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7395650744438171, + "num_tokens": 635985408.0, + "step": 24582 + }, + { + "epoch": 2.699648583351636, + "grad_norm": 2.5222580432891846, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7483577728271484, + "num_tokens": 636008460.0, + "step": 24583 + }, + { + "epoch": 2.69975840105425, + "grad_norm": 2.448131799697876, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7170195579528809, + "num_tokens": 636033600.0, + "step": 24584 + }, + { + "epoch": 2.6998682187568637, + "grad_norm": 2.092194080352783, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7039486169815063, + "num_tokens": 636064531.0, + "step": 24585 + }, + { + "epoch": 2.6999780364594774, + "grad_norm": 2.4636290073394775, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7369530200958252, + "num_tokens": 636088199.0, + "step": 24586 + }, + { + "epoch": 2.700087854162091, + "grad_norm": 2.3724818229675293, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7358804941177368, + "num_tokens": 636114960.0, + "step": 24587 + }, + { + "epoch": 2.7001976718647045, + "grad_norm": 2.3649685382843018, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7103878259658813, + "num_tokens": 636142518.0, + "step": 24588 + }, + { + "epoch": 2.7003074895673183, + "grad_norm": 3.225450277328491, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7290979027748108, + "num_tokens": 636170558.0, + "step": 24589 + }, + { + "epoch": 2.700417307269932, + "grad_norm": 2.4353115558624268, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7380422353744507, + "num_tokens": 636195178.0, + "step": 24590 + }, + { + "epoch": 2.7005271249725453, + "grad_norm": 2.553879976272583, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7021388411521912, + "num_tokens": 636219507.0, + "step": 24591 + }, + { + "epoch": 2.700636942675159, + "grad_norm": 2.225318431854248, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7143334150314331, + "num_tokens": 636249262.0, + "step": 24592 + }, + { + "epoch": 2.700746760377773, + "grad_norm": 2.1615006923675537, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6959110498428345, + "num_tokens": 636282270.0, + "step": 24593 + }, + { + "epoch": 2.7008565780803866, + "grad_norm": 2.339974880218506, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7091538310050964, + "num_tokens": 636309437.0, + "step": 24594 + }, + { + "epoch": 2.7009663957830004, + "grad_norm": 2.4797496795654297, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7479910850524902, + "num_tokens": 636334045.0, + "step": 24595 + }, + { + "epoch": 2.7010762134856137, + "grad_norm": 2.508150815963745, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7282836437225342, + "num_tokens": 636356655.0, + "step": 24596 + }, + { + "epoch": 2.7011860311882274, + "grad_norm": 2.4081637859344482, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7538866996765137, + "num_tokens": 636380594.0, + "step": 24597 + }, + { + "epoch": 2.701295848890841, + "grad_norm": 2.2428324222564697, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7272779941558838, + "num_tokens": 636409274.0, + "step": 24598 + }, + { + "epoch": 2.701405666593455, + "grad_norm": 2.512690305709839, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7423954010009766, + "num_tokens": 636431881.0, + "step": 24599 + }, + { + "epoch": 2.7015154842960687, + "grad_norm": 2.5583298206329346, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7200427055358887, + "num_tokens": 636456702.0, + "step": 24600 + }, + { + "epoch": 2.701625301998682, + "grad_norm": 2.4912657737731934, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7644692659378052, + "num_tokens": 636477902.0, + "step": 24601 + }, + { + "epoch": 2.7017351197012958, + "grad_norm": 2.327150344848633, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7274916768074036, + "num_tokens": 636506103.0, + "step": 24602 + }, + { + "epoch": 2.7018449374039095, + "grad_norm": 2.1930925846099854, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7343525886535645, + "num_tokens": 636534483.0, + "step": 24603 + }, + { + "epoch": 2.7019547551065233, + "grad_norm": 2.336961507797241, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7336913347244263, + "num_tokens": 636560499.0, + "step": 24604 + }, + { + "epoch": 2.702064572809137, + "grad_norm": 2.4023804664611816, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7299349904060364, + "num_tokens": 636587021.0, + "step": 24605 + }, + { + "epoch": 2.7021743905117503, + "grad_norm": 2.361673593521118, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7346352934837341, + "num_tokens": 636611808.0, + "step": 24606 + }, + { + "epoch": 2.702284208214364, + "grad_norm": 2.1203789710998535, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.726048469543457, + "num_tokens": 636642235.0, + "step": 24607 + }, + { + "epoch": 2.702394025916978, + "grad_norm": 2.358119487762451, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.723578691482544, + "num_tokens": 636668339.0, + "step": 24608 + }, + { + "epoch": 2.7025038436195916, + "grad_norm": 2.4047956466674805, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7239589095115662, + "num_tokens": 636694031.0, + "step": 24609 + }, + { + "epoch": 2.7026136613222054, + "grad_norm": 2.314770460128784, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7360751032829285, + "num_tokens": 636721196.0, + "step": 24610 + }, + { + "epoch": 2.7027234790248187, + "grad_norm": 2.4682788848876953, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7220442295074463, + "num_tokens": 636744077.0, + "step": 24611 + }, + { + "epoch": 2.7028332967274324, + "grad_norm": 2.3459885120391846, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7266964912414551, + "num_tokens": 636771047.0, + "step": 24612 + }, + { + "epoch": 2.702943114430046, + "grad_norm": 2.2536306381225586, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.744472324848175, + "num_tokens": 636797984.0, + "step": 24613 + }, + { + "epoch": 2.7030529321326595, + "grad_norm": 2.5816924571990967, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7419967651367188, + "num_tokens": 636820040.0, + "step": 24614 + }, + { + "epoch": 2.7031627498352737, + "grad_norm": 2.2734360694885254, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7515252828598022, + "num_tokens": 636847213.0, + "step": 24615 + }, + { + "epoch": 2.703272567537887, + "grad_norm": 2.412151575088501, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7350359559059143, + "num_tokens": 636873425.0, + "step": 24616 + }, + { + "epoch": 2.703382385240501, + "grad_norm": 2.509390354156494, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.736822247505188, + "num_tokens": 636895224.0, + "step": 24617 + }, + { + "epoch": 2.7034922029431145, + "grad_norm": 2.1446869373321533, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7088724970817566, + "num_tokens": 636925505.0, + "step": 24618 + }, + { + "epoch": 2.703602020645728, + "grad_norm": 2.2906558513641357, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7321798801422119, + "num_tokens": 636950985.0, + "step": 24619 + }, + { + "epoch": 2.7037118383483416, + "grad_norm": 2.6025784015655518, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7367574572563171, + "num_tokens": 636972902.0, + "step": 24620 + }, + { + "epoch": 2.7038216560509554, + "grad_norm": 2.522617816925049, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7295069694519043, + "num_tokens": 636995027.0, + "step": 24621 + }, + { + "epoch": 2.703931473753569, + "grad_norm": 2.215909719467163, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7120441794395447, + "num_tokens": 637025443.0, + "step": 24622 + }, + { + "epoch": 2.704041291456183, + "grad_norm": 2.3637888431549072, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7347705364227295, + "num_tokens": 637053295.0, + "step": 24623 + }, + { + "epoch": 2.704151109158796, + "grad_norm": 2.3909263610839844, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7340174317359924, + "num_tokens": 637078599.0, + "step": 24624 + }, + { + "epoch": 2.70426092686141, + "grad_norm": 2.067889451980591, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7227078676223755, + "num_tokens": 637109623.0, + "step": 24625 + }, + { + "epoch": 2.7043707445640237, + "grad_norm": 2.421370267868042, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7282168865203857, + "num_tokens": 637132960.0, + "step": 24626 + }, + { + "epoch": 2.7044805622666375, + "grad_norm": 2.473609447479248, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7407248020172119, + "num_tokens": 637159563.0, + "step": 24627 + }, + { + "epoch": 2.7045903799692512, + "grad_norm": 2.108290433883667, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7453557252883911, + "num_tokens": 637189795.0, + "step": 24628 + }, + { + "epoch": 2.7047001976718645, + "grad_norm": 2.2086143493652344, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7163174748420715, + "num_tokens": 637217854.0, + "step": 24629 + }, + { + "epoch": 2.7048100153744783, + "grad_norm": 2.239427328109741, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7610798478126526, + "num_tokens": 637246068.0, + "step": 24630 + }, + { + "epoch": 2.704919833077092, + "grad_norm": 2.2348477840423584, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7187283039093018, + "num_tokens": 637275907.0, + "step": 24631 + }, + { + "epoch": 2.705029650779706, + "grad_norm": 2.5665383338928223, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7289291620254517, + "num_tokens": 637298231.0, + "step": 24632 + }, + { + "epoch": 2.7051394684823196, + "grad_norm": 2.335313081741333, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7419861555099487, + "num_tokens": 637324818.0, + "step": 24633 + }, + { + "epoch": 2.705249286184933, + "grad_norm": 2.2049005031585693, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7411260008811951, + "num_tokens": 637351693.0, + "step": 24634 + }, + { + "epoch": 2.7053591038875466, + "grad_norm": 2.3901748657226562, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7508264780044556, + "num_tokens": 637376108.0, + "step": 24635 + }, + { + "epoch": 2.7054689215901604, + "grad_norm": 2.5310373306274414, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7269003391265869, + "num_tokens": 637402645.0, + "step": 24636 + }, + { + "epoch": 2.705578739292774, + "grad_norm": 2.391261577606201, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7158179879188538, + "num_tokens": 637427056.0, + "step": 24637 + }, + { + "epoch": 2.705688556995388, + "grad_norm": 2.1645803451538086, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7215621471405029, + "num_tokens": 637457793.0, + "step": 24638 + }, + { + "epoch": 2.705798374698001, + "grad_norm": 2.3436508178710938, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7459314465522766, + "num_tokens": 637481962.0, + "step": 24639 + }, + { + "epoch": 2.705908192400615, + "grad_norm": 2.744135618209839, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7489309906959534, + "num_tokens": 637502153.0, + "step": 24640 + }, + { + "epoch": 2.7060180101032287, + "grad_norm": 2.294893741607666, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7168245315551758, + "num_tokens": 637529072.0, + "step": 24641 + }, + { + "epoch": 2.706127827805842, + "grad_norm": 2.7840793132781982, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7520385980606079, + "num_tokens": 637549332.0, + "step": 24642 + }, + { + "epoch": 2.706237645508456, + "grad_norm": 2.695629835128784, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7580829858779907, + "num_tokens": 637571205.0, + "step": 24643 + }, + { + "epoch": 2.7063474632110696, + "grad_norm": 2.373969793319702, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7204980850219727, + "num_tokens": 637599018.0, + "step": 24644 + }, + { + "epoch": 2.7064572809136833, + "grad_norm": 2.42749285697937, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7144190073013306, + "num_tokens": 637626618.0, + "step": 24645 + }, + { + "epoch": 2.706567098616297, + "grad_norm": 2.2050397396087646, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7247739434242249, + "num_tokens": 637655494.0, + "step": 24646 + }, + { + "epoch": 2.7066769163189104, + "grad_norm": 2.3944551944732666, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7300162315368652, + "num_tokens": 637680457.0, + "step": 24647 + }, + { + "epoch": 2.706786734021524, + "grad_norm": 2.2641706466674805, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7316810488700867, + "num_tokens": 637708205.0, + "step": 24648 + }, + { + "epoch": 2.706896551724138, + "grad_norm": 3.1283771991729736, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7508431077003479, + "num_tokens": 637725835.0, + "step": 24649 + }, + { + "epoch": 2.7070063694267517, + "grad_norm": 2.1063201427459717, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7281409502029419, + "num_tokens": 637757318.0, + "step": 24650 + }, + { + "epoch": 2.7071161871293654, + "grad_norm": 2.1980509757995605, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7242465019226074, + "num_tokens": 637786045.0, + "step": 24651 + }, + { + "epoch": 2.7072260048319787, + "grad_norm": 2.3106648921966553, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7334151268005371, + "num_tokens": 637814344.0, + "step": 24652 + }, + { + "epoch": 2.7073358225345925, + "grad_norm": 1.9546695947647095, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7170385718345642, + "num_tokens": 637851542.0, + "step": 24653 + }, + { + "epoch": 2.7074456402372062, + "grad_norm": 2.2709996700286865, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.6974064111709595, + "num_tokens": 637880353.0, + "step": 24654 + }, + { + "epoch": 2.70755545793982, + "grad_norm": 2.3696980476379395, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7266137599945068, + "num_tokens": 637905596.0, + "step": 24655 + }, + { + "epoch": 2.7076652756424338, + "grad_norm": 2.433497190475464, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7271677255630493, + "num_tokens": 637930958.0, + "step": 24656 + }, + { + "epoch": 2.707775093345047, + "grad_norm": 2.5742437839508057, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7425451874732971, + "num_tokens": 637952486.0, + "step": 24657 + }, + { + "epoch": 2.707884911047661, + "grad_norm": 2.3889389038085938, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7097396850585938, + "num_tokens": 637980785.0, + "step": 24658 + }, + { + "epoch": 2.7079947287502746, + "grad_norm": 2.1588306427001953, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7196515798568726, + "num_tokens": 638012640.0, + "step": 24659 + }, + { + "epoch": 2.7081045464528883, + "grad_norm": 2.620638847351074, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7483614683151245, + "num_tokens": 638035306.0, + "step": 24660 + }, + { + "epoch": 2.708214364155502, + "grad_norm": 2.27058744430542, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7281017899513245, + "num_tokens": 638064193.0, + "step": 24661 + }, + { + "epoch": 2.7083241818581154, + "grad_norm": 2.4064178466796875, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7464281320571899, + "num_tokens": 638088972.0, + "step": 24662 + }, + { + "epoch": 2.708433999560729, + "grad_norm": 2.6750173568725586, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7398015260696411, + "num_tokens": 638110260.0, + "step": 24663 + }, + { + "epoch": 2.708543817263343, + "grad_norm": 2.619292736053467, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7366628646850586, + "num_tokens": 638132932.0, + "step": 24664 + }, + { + "epoch": 2.7086536349659562, + "grad_norm": 2.4081273078918457, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7060656547546387, + "num_tokens": 638159454.0, + "step": 24665 + }, + { + "epoch": 2.7087634526685704, + "grad_norm": 2.4144372940063477, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7385436296463013, + "num_tokens": 638183545.0, + "step": 24666 + }, + { + "epoch": 2.7088732703711838, + "grad_norm": 2.0352516174316406, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7308638095855713, + "num_tokens": 638215187.0, + "step": 24667 + }, + { + "epoch": 2.7089830880737975, + "grad_norm": 2.419668436050415, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7291593551635742, + "num_tokens": 638240218.0, + "step": 24668 + }, + { + "epoch": 2.7090929057764113, + "grad_norm": 2.101595401763916, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7308727502822876, + "num_tokens": 638271344.0, + "step": 24669 + }, + { + "epoch": 2.7092027234790246, + "grad_norm": 2.002380847930908, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7267442941665649, + "num_tokens": 638303577.0, + "step": 24670 + }, + { + "epoch": 2.7093125411816383, + "grad_norm": 2.2379980087280273, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7144151329994202, + "num_tokens": 638330973.0, + "step": 24671 + }, + { + "epoch": 2.709422358884252, + "grad_norm": 2.2331838607788086, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7326836585998535, + "num_tokens": 638357139.0, + "step": 24672 + }, + { + "epoch": 2.709532176586866, + "grad_norm": 3.1653499603271484, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7365742921829224, + "num_tokens": 638373815.0, + "step": 24673 + }, + { + "epoch": 2.7096419942894796, + "grad_norm": 2.493011713027954, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7053765654563904, + "num_tokens": 638399938.0, + "step": 24674 + }, + { + "epoch": 2.709751811992093, + "grad_norm": 2.482483386993408, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7441142797470093, + "num_tokens": 638422808.0, + "step": 24675 + }, + { + "epoch": 2.7098616296947067, + "grad_norm": 2.120898723602295, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7573176026344299, + "num_tokens": 638450788.0, + "step": 24676 + }, + { + "epoch": 2.7099714473973204, + "grad_norm": 2.357860565185547, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.730180025100708, + "num_tokens": 638478890.0, + "step": 24677 + }, + { + "epoch": 2.710081265099934, + "grad_norm": 2.1049835681915283, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7369844913482666, + "num_tokens": 638510335.0, + "step": 24678 + }, + { + "epoch": 2.710191082802548, + "grad_norm": 2.4706292152404785, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.73267662525177, + "num_tokens": 638533854.0, + "step": 24679 + }, + { + "epoch": 2.7103009005051613, + "grad_norm": 2.308154582977295, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7119749784469604, + "num_tokens": 638561507.0, + "step": 24680 + }, + { + "epoch": 2.710410718207775, + "grad_norm": 2.4116604328155518, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7365165948867798, + "num_tokens": 638586013.0, + "step": 24681 + }, + { + "epoch": 2.7105205359103888, + "grad_norm": 2.3590025901794434, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7533125281333923, + "num_tokens": 638609740.0, + "step": 24682 + }, + { + "epoch": 2.7106303536130025, + "grad_norm": 2.4351651668548584, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7221271991729736, + "num_tokens": 638636714.0, + "step": 24683 + }, + { + "epoch": 2.7107401713156163, + "grad_norm": 2.41441011428833, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7433487772941589, + "num_tokens": 638661166.0, + "step": 24684 + }, + { + "epoch": 2.7108499890182296, + "grad_norm": 2.140636682510376, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7196862697601318, + "num_tokens": 638689317.0, + "step": 24685 + }, + { + "epoch": 2.7109598067208434, + "grad_norm": 2.4414222240448, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.710938572883606, + "num_tokens": 638714501.0, + "step": 24686 + }, + { + "epoch": 2.711069624423457, + "grad_norm": 2.4783341884613037, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7167258262634277, + "num_tokens": 638739101.0, + "step": 24687 + }, + { + "epoch": 2.711179442126071, + "grad_norm": 2.3689568042755127, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7161220908164978, + "num_tokens": 638766889.0, + "step": 24688 + }, + { + "epoch": 2.7112892598286846, + "grad_norm": 2.555893898010254, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7347395420074463, + "num_tokens": 638789471.0, + "step": 24689 + }, + { + "epoch": 2.711399077531298, + "grad_norm": 2.6261537075042725, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7358058094978333, + "num_tokens": 638810942.0, + "step": 24690 + }, + { + "epoch": 2.7115088952339117, + "grad_norm": 2.100543260574341, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7320941090583801, + "num_tokens": 638843824.0, + "step": 24691 + }, + { + "epoch": 2.7116187129365255, + "grad_norm": 1.9986674785614014, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7307491898536682, + "num_tokens": 638874574.0, + "step": 24692 + }, + { + "epoch": 2.7117285306391388, + "grad_norm": 2.2136642932891846, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7281836271286011, + "num_tokens": 638903284.0, + "step": 24693 + }, + { + "epoch": 2.7118383483417525, + "grad_norm": 2.502631187438965, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.717677652835846, + "num_tokens": 638928789.0, + "step": 24694 + }, + { + "epoch": 2.7119481660443663, + "grad_norm": 2.096400022506714, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7078859210014343, + "num_tokens": 638960128.0, + "step": 24695 + }, + { + "epoch": 2.71205798374698, + "grad_norm": 2.2310521602630615, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.733211874961853, + "num_tokens": 638986557.0, + "step": 24696 + }, + { + "epoch": 2.712167801449594, + "grad_norm": 2.0300872325897217, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7135992050170898, + "num_tokens": 639019260.0, + "step": 24697 + }, + { + "epoch": 2.712277619152207, + "grad_norm": 2.026632785797119, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7252443432807922, + "num_tokens": 639051229.0, + "step": 24698 + }, + { + "epoch": 2.712387436854821, + "grad_norm": 2.231165885925293, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7258642911911011, + "num_tokens": 639080680.0, + "step": 24699 + }, + { + "epoch": 2.7124972545574346, + "grad_norm": 2.0691773891448975, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7399517297744751, + "num_tokens": 639112770.0, + "step": 24700 + }, + { + "epoch": 2.7126070722600484, + "grad_norm": 2.7950611114501953, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.734085738658905, + "num_tokens": 639132757.0, + "step": 24701 + }, + { + "epoch": 2.712716889962662, + "grad_norm": 2.4692578315734863, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7460888028144836, + "num_tokens": 639155224.0, + "step": 24702 + }, + { + "epoch": 2.7128267076652754, + "grad_norm": 2.3508505821228027, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7290953397750854, + "num_tokens": 639180807.0, + "step": 24703 + }, + { + "epoch": 2.712936525367889, + "grad_norm": 2.8213753700256348, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7242105603218079, + "num_tokens": 639200014.0, + "step": 24704 + }, + { + "epoch": 2.713046343070503, + "grad_norm": 2.722357988357544, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.758154034614563, + "num_tokens": 639219260.0, + "step": 24705 + }, + { + "epoch": 2.7131561607731167, + "grad_norm": 2.161864757537842, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7215888500213623, + "num_tokens": 639250650.0, + "step": 24706 + }, + { + "epoch": 2.7132659784757305, + "grad_norm": 2.4150071144104004, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7375611066818237, + "num_tokens": 639275641.0, + "step": 24707 + }, + { + "epoch": 2.713375796178344, + "grad_norm": 2.5289878845214844, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7354809641838074, + "num_tokens": 639297757.0, + "step": 24708 + }, + { + "epoch": 2.7134856138809575, + "grad_norm": 2.210482597351074, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.731377363204956, + "num_tokens": 639325714.0, + "step": 24709 + }, + { + "epoch": 2.7135954315835713, + "grad_norm": 2.6485848426818848, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7417539954185486, + "num_tokens": 639345003.0, + "step": 24710 + }, + { + "epoch": 2.713705249286185, + "grad_norm": 2.3936898708343506, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7308958768844604, + "num_tokens": 639369979.0, + "step": 24711 + }, + { + "epoch": 2.713815066988799, + "grad_norm": 2.262826919555664, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7267941832542419, + "num_tokens": 639398053.0, + "step": 24712 + }, + { + "epoch": 2.713924884691412, + "grad_norm": 2.180053234100342, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7125221490859985, + "num_tokens": 639428209.0, + "step": 24713 + }, + { + "epoch": 2.714034702394026, + "grad_norm": 2.1705713272094727, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7404419183731079, + "num_tokens": 639454434.0, + "step": 24714 + }, + { + "epoch": 2.7141445200966396, + "grad_norm": 2.428051233291626, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.765562891960144, + "num_tokens": 639476779.0, + "step": 24715 + }, + { + "epoch": 2.7142543377992534, + "grad_norm": 2.3073618412017822, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7380636930465698, + "num_tokens": 639503392.0, + "step": 24716 + }, + { + "epoch": 2.714364155501867, + "grad_norm": 2.368018865585327, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7583879232406616, + "num_tokens": 639527342.0, + "step": 24717 + }, + { + "epoch": 2.7144739732044805, + "grad_norm": 2.498326301574707, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7411487102508545, + "num_tokens": 639550694.0, + "step": 24718 + }, + { + "epoch": 2.7145837909070942, + "grad_norm": 1.9955976009368896, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7292875647544861, + "num_tokens": 639585039.0, + "step": 24719 + }, + { + "epoch": 2.714693608609708, + "grad_norm": 2.5987117290496826, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7461220026016235, + "num_tokens": 639607137.0, + "step": 24720 + }, + { + "epoch": 2.7148034263123213, + "grad_norm": 2.2016255855560303, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7319344282150269, + "num_tokens": 639635501.0, + "step": 24721 + }, + { + "epoch": 2.714913244014935, + "grad_norm": 2.2682480812072754, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7173465490341187, + "num_tokens": 639665461.0, + "step": 24722 + }, + { + "epoch": 2.715023061717549, + "grad_norm": 2.2806358337402344, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7164349555969238, + "num_tokens": 639693491.0, + "step": 24723 + }, + { + "epoch": 2.7151328794201626, + "grad_norm": 2.6470518112182617, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7350775003433228, + "num_tokens": 639715777.0, + "step": 24724 + }, + { + "epoch": 2.7152426971227763, + "grad_norm": 2.373847007751465, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7303417921066284, + "num_tokens": 639741348.0, + "step": 24725 + }, + { + "epoch": 2.7153525148253896, + "grad_norm": 2.5096945762634277, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7339850068092346, + "num_tokens": 639764709.0, + "step": 24726 + }, + { + "epoch": 2.7154623325280034, + "grad_norm": 2.126765727996826, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7445555925369263, + "num_tokens": 639793037.0, + "step": 24727 + }, + { + "epoch": 2.715572150230617, + "grad_norm": 2.3068020343780518, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7140278816223145, + "num_tokens": 639820215.0, + "step": 24728 + }, + { + "epoch": 2.715681967933231, + "grad_norm": 2.5582234859466553, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7265629768371582, + "num_tokens": 639842429.0, + "step": 24729 + }, + { + "epoch": 2.7157917856358447, + "grad_norm": 2.2752559185028076, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.727517306804657, + "num_tokens": 639873513.0, + "step": 24730 + }, + { + "epoch": 2.715901603338458, + "grad_norm": 2.3760499954223633, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7116047739982605, + "num_tokens": 639899481.0, + "step": 24731 + }, + { + "epoch": 2.7160114210410717, + "grad_norm": 2.4078245162963867, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7295135259628296, + "num_tokens": 639924799.0, + "step": 24732 + }, + { + "epoch": 2.7161212387436855, + "grad_norm": 2.2661850452423096, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7263508439064026, + "num_tokens": 639951117.0, + "step": 24733 + }, + { + "epoch": 2.7162310564462993, + "grad_norm": 2.3347113132476807, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7356473207473755, + "num_tokens": 639974759.0, + "step": 24734 + }, + { + "epoch": 2.716340874148913, + "grad_norm": 2.6967110633850098, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7260115146636963, + "num_tokens": 639997730.0, + "step": 24735 + }, + { + "epoch": 2.7164506918515263, + "grad_norm": 2.5709304809570312, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.737096905708313, + "num_tokens": 640020029.0, + "step": 24736 + }, + { + "epoch": 2.71656050955414, + "grad_norm": 2.303593158721924, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7353871464729309, + "num_tokens": 640048122.0, + "step": 24737 + }, + { + "epoch": 2.716670327256754, + "grad_norm": 2.1543362140655518, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7105387449264526, + "num_tokens": 640077578.0, + "step": 24738 + }, + { + "epoch": 2.7167801449593676, + "grad_norm": 2.344405174255371, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7639346718788147, + "num_tokens": 640102825.0, + "step": 24739 + }, + { + "epoch": 2.7168899626619814, + "grad_norm": 2.262387275695801, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7396013736724854, + "num_tokens": 640132048.0, + "step": 24740 + }, + { + "epoch": 2.7169997803645947, + "grad_norm": 2.630711555480957, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7311819195747375, + "num_tokens": 640156823.0, + "step": 24741 + }, + { + "epoch": 2.7171095980672084, + "grad_norm": 2.504127264022827, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7238447666168213, + "num_tokens": 640180505.0, + "step": 24742 + }, + { + "epoch": 2.717219415769822, + "grad_norm": 2.6870625019073486, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7587379813194275, + "num_tokens": 640199882.0, + "step": 24743 + }, + { + "epoch": 2.7173292334724355, + "grad_norm": 2.348529100418091, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7414908409118652, + "num_tokens": 640228809.0, + "step": 24744 + }, + { + "epoch": 2.7174390511750497, + "grad_norm": 2.3640613555908203, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7177668213844299, + "num_tokens": 640256072.0, + "step": 24745 + }, + { + "epoch": 2.717548868877663, + "grad_norm": 2.299501419067383, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7491001486778259, + "num_tokens": 640282519.0, + "step": 24746 + }, + { + "epoch": 2.7176586865802768, + "grad_norm": 2.3840508460998535, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7181296348571777, + "num_tokens": 640307450.0, + "step": 24747 + }, + { + "epoch": 2.7177685042828905, + "grad_norm": 2.5118772983551025, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7357276678085327, + "num_tokens": 640329891.0, + "step": 24748 + }, + { + "epoch": 2.717878321985504, + "grad_norm": 2.0417425632476807, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7019329071044922, + "num_tokens": 640364993.0, + "step": 24749 + }, + { + "epoch": 2.7179881396881176, + "grad_norm": 2.3067567348480225, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7310998439788818, + "num_tokens": 640392942.0, + "step": 24750 + }, + { + "epoch": 2.7180979573907313, + "grad_norm": 2.2792227268218994, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7237151861190796, + "num_tokens": 640419676.0, + "step": 24751 + }, + { + "epoch": 2.718207775093345, + "grad_norm": 2.2150092124938965, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7399923801422119, + "num_tokens": 640448091.0, + "step": 24752 + }, + { + "epoch": 2.718317592795959, + "grad_norm": 2.3282470703125, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7176913022994995, + "num_tokens": 640477778.0, + "step": 24753 + }, + { + "epoch": 2.718427410498572, + "grad_norm": 2.447050094604492, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.732561469078064, + "num_tokens": 640500398.0, + "step": 24754 + }, + { + "epoch": 2.718537228201186, + "grad_norm": 2.2542884349823, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7424412965774536, + "num_tokens": 640528757.0, + "step": 24755 + }, + { + "epoch": 2.7186470459037997, + "grad_norm": 2.357431650161743, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7067546844482422, + "num_tokens": 640554774.0, + "step": 24756 + }, + { + "epoch": 2.7187568636064134, + "grad_norm": 2.2669522762298584, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7222133874893188, + "num_tokens": 640582467.0, + "step": 24757 + }, + { + "epoch": 2.718866681309027, + "grad_norm": 2.215985059738159, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7214173078536987, + "num_tokens": 640614279.0, + "step": 24758 + }, + { + "epoch": 2.7189764990116405, + "grad_norm": 2.2380614280700684, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7227408289909363, + "num_tokens": 640642335.0, + "step": 24759 + }, + { + "epoch": 2.7190863167142543, + "grad_norm": 2.2994816303253174, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.71030592918396, + "num_tokens": 640669322.0, + "step": 24760 + }, + { + "epoch": 2.719196134416868, + "grad_norm": 2.920593738555908, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7329148054122925, + "num_tokens": 640687288.0, + "step": 24761 + }, + { + "epoch": 2.719305952119482, + "grad_norm": 2.4862148761749268, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7293676137924194, + "num_tokens": 640710975.0, + "step": 24762 + }, + { + "epoch": 2.7194157698220955, + "grad_norm": 1.9031494855880737, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7278842926025391, + "num_tokens": 640747982.0, + "step": 24763 + }, + { + "epoch": 2.719525587524709, + "grad_norm": 2.352177858352661, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7541464567184448, + "num_tokens": 640775056.0, + "step": 24764 + }, + { + "epoch": 2.7196354052273226, + "grad_norm": 2.2631962299346924, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.718944787979126, + "num_tokens": 640802794.0, + "step": 24765 + }, + { + "epoch": 2.7197452229299364, + "grad_norm": 2.394230842590332, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7188585996627808, + "num_tokens": 640828970.0, + "step": 24766 + }, + { + "epoch": 2.71985504063255, + "grad_norm": 2.343838691711426, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7361323237419128, + "num_tokens": 640856830.0, + "step": 24767 + }, + { + "epoch": 2.719964858335164, + "grad_norm": 2.2059710025787354, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7087939977645874, + "num_tokens": 640888052.0, + "step": 24768 + }, + { + "epoch": 2.720074676037777, + "grad_norm": 2.386906623840332, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7451550960540771, + "num_tokens": 640912422.0, + "step": 24769 + }, + { + "epoch": 2.720184493740391, + "grad_norm": 2.614853858947754, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7348595261573792, + "num_tokens": 640933340.0, + "step": 24770 + }, + { + "epoch": 2.7202943114430047, + "grad_norm": 2.6892600059509277, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7538951635360718, + "num_tokens": 640953858.0, + "step": 24771 + }, + { + "epoch": 2.720404129145618, + "grad_norm": 2.3647005558013916, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7328484058380127, + "num_tokens": 640979591.0, + "step": 24772 + }, + { + "epoch": 2.720513946848232, + "grad_norm": 2.237821578979492, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7292413711547852, + "num_tokens": 641005981.0, + "step": 24773 + }, + { + "epoch": 2.7206237645508455, + "grad_norm": 2.475654125213623, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7454298734664917, + "num_tokens": 641030596.0, + "step": 24774 + }, + { + "epoch": 2.7207335822534593, + "grad_norm": 2.3215696811676025, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7074116468429565, + "num_tokens": 641057299.0, + "step": 24775 + }, + { + "epoch": 2.720843399956073, + "grad_norm": 2.4999032020568848, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7293193936347961, + "num_tokens": 641083248.0, + "step": 24776 + }, + { + "epoch": 2.7209532176586864, + "grad_norm": 2.3847296237945557, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7042115926742554, + "num_tokens": 641110826.0, + "step": 24777 + }, + { + "epoch": 2.7210630353613, + "grad_norm": 2.360124349594116, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7477293610572815, + "num_tokens": 641138748.0, + "step": 24778 + }, + { + "epoch": 2.721172853063914, + "grad_norm": 2.319129467010498, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7187132835388184, + "num_tokens": 641167288.0, + "step": 24779 + }, + { + "epoch": 2.7212826707665276, + "grad_norm": 2.21248722076416, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.722095251083374, + "num_tokens": 641196385.0, + "step": 24780 + }, + { + "epoch": 2.7213924884691414, + "grad_norm": 2.3780462741851807, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7225854396820068, + "num_tokens": 641222366.0, + "step": 24781 + }, + { + "epoch": 2.7215023061717547, + "grad_norm": 2.0557031631469727, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7242792844772339, + "num_tokens": 641255955.0, + "step": 24782 + }, + { + "epoch": 2.7216121238743685, + "grad_norm": 2.4847488403320312, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7174592018127441, + "num_tokens": 641280874.0, + "step": 24783 + }, + { + "epoch": 2.721721941576982, + "grad_norm": 2.4823992252349854, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7387850284576416, + "num_tokens": 641303592.0, + "step": 24784 + }, + { + "epoch": 2.721831759279596, + "grad_norm": 2.030853748321533, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.6971287727355957, + "num_tokens": 641336893.0, + "step": 24785 + }, + { + "epoch": 2.7219415769822097, + "grad_norm": 2.492133378982544, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7375126481056213, + "num_tokens": 641359312.0, + "step": 24786 + }, + { + "epoch": 2.722051394684823, + "grad_norm": 2.484318733215332, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7503212690353394, + "num_tokens": 641383693.0, + "step": 24787 + }, + { + "epoch": 2.722161212387437, + "grad_norm": 2.548171281814575, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7337851524353027, + "num_tokens": 641407736.0, + "step": 24788 + }, + { + "epoch": 2.7222710300900506, + "grad_norm": 2.322244167327881, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7042471170425415, + "num_tokens": 641433221.0, + "step": 24789 + }, + { + "epoch": 2.7223808477926643, + "grad_norm": 2.5230565071105957, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7272648215293884, + "num_tokens": 641456588.0, + "step": 24790 + }, + { + "epoch": 2.722490665495278, + "grad_norm": 2.0338361263275146, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7118465900421143, + "num_tokens": 641491270.0, + "step": 24791 + }, + { + "epoch": 2.7226004831978914, + "grad_norm": 2.111001491546631, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7214483022689819, + "num_tokens": 641524657.0, + "step": 24792 + }, + { + "epoch": 2.722710300900505, + "grad_norm": 2.273515462875366, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7119815349578857, + "num_tokens": 641551827.0, + "step": 24793 + }, + { + "epoch": 2.722820118603119, + "grad_norm": 2.88301420211792, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.757564902305603, + "num_tokens": 641570975.0, + "step": 24794 + }, + { + "epoch": 2.722929936305732, + "grad_norm": 2.6345934867858887, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7370449304580688, + "num_tokens": 641591883.0, + "step": 24795 + }, + { + "epoch": 2.7230397540083464, + "grad_norm": 2.5055363178253174, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7304335832595825, + "num_tokens": 641615483.0, + "step": 24796 + }, + { + "epoch": 2.7231495717109597, + "grad_norm": 2.299678325653076, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7505912184715271, + "num_tokens": 641640463.0, + "step": 24797 + }, + { + "epoch": 2.7232593894135735, + "grad_norm": 2.25616717338562, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7185468673706055, + "num_tokens": 641669341.0, + "step": 24798 + }, + { + "epoch": 2.7233692071161872, + "grad_norm": 2.467831611633301, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7105336785316467, + "num_tokens": 641695953.0, + "step": 24799 + }, + { + "epoch": 2.7234790248188006, + "grad_norm": 2.5077247619628906, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7284673452377319, + "num_tokens": 641720964.0, + "step": 24800 + }, + { + "epoch": 2.7235888425214143, + "grad_norm": 2.1070172786712646, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7177879214286804, + "num_tokens": 641752724.0, + "step": 24801 + }, + { + "epoch": 2.723698660224028, + "grad_norm": 2.6045596599578857, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7525940537452698, + "num_tokens": 641774184.0, + "step": 24802 + }, + { + "epoch": 2.723808477926642, + "grad_norm": 2.2570464611053467, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.720576822757721, + "num_tokens": 641802550.0, + "step": 24803 + }, + { + "epoch": 2.7239182956292556, + "grad_norm": 2.665552854537964, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7568618059158325, + "num_tokens": 641822945.0, + "step": 24804 + }, + { + "epoch": 2.724028113331869, + "grad_norm": 2.3400514125823975, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7461868524551392, + "num_tokens": 641848120.0, + "step": 24805 + }, + { + "epoch": 2.7241379310344827, + "grad_norm": 2.270925521850586, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7479302883148193, + "num_tokens": 641874104.0, + "step": 24806 + }, + { + "epoch": 2.7242477487370964, + "grad_norm": 2.4219961166381836, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.740320086479187, + "num_tokens": 641898951.0, + "step": 24807 + }, + { + "epoch": 2.72435756643971, + "grad_norm": 2.4941587448120117, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.747552752494812, + "num_tokens": 641921362.0, + "step": 24808 + }, + { + "epoch": 2.724467384142324, + "grad_norm": 3.040940284729004, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7452196478843689, + "num_tokens": 641938341.0, + "step": 24809 + }, + { + "epoch": 2.7245772018449372, + "grad_norm": 2.099461793899536, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7102272510528564, + "num_tokens": 641970748.0, + "step": 24810 + }, + { + "epoch": 2.724687019547551, + "grad_norm": 2.2523586750030518, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7103570699691772, + "num_tokens": 642001075.0, + "step": 24811 + }, + { + "epoch": 2.7247968372501647, + "grad_norm": 2.30906081199646, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7512447237968445, + "num_tokens": 642026261.0, + "step": 24812 + }, + { + "epoch": 2.7249066549527785, + "grad_norm": 2.249725103378296, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7182594537734985, + "num_tokens": 642055981.0, + "step": 24813 + }, + { + "epoch": 2.7250164726553923, + "grad_norm": 2.2833735942840576, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7069151997566223, + "num_tokens": 642082195.0, + "step": 24814 + }, + { + "epoch": 2.7251262903580056, + "grad_norm": 2.1312155723571777, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7272456884384155, + "num_tokens": 642114405.0, + "step": 24815 + }, + { + "epoch": 2.7252361080606193, + "grad_norm": 2.336336851119995, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7200806140899658, + "num_tokens": 642139251.0, + "step": 24816 + }, + { + "epoch": 2.725345925763233, + "grad_norm": 2.513751268386841, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7417880892753601, + "num_tokens": 642163813.0, + "step": 24817 + }, + { + "epoch": 2.725455743465847, + "grad_norm": 2.261354923248291, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7366374731063843, + "num_tokens": 642189645.0, + "step": 24818 + }, + { + "epoch": 2.7255655611684606, + "grad_norm": 2.138356924057007, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.732764482498169, + "num_tokens": 642219610.0, + "step": 24819 + }, + { + "epoch": 2.725675378871074, + "grad_norm": 2.6029210090637207, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7379032969474792, + "num_tokens": 642240897.0, + "step": 24820 + }, + { + "epoch": 2.7257851965736877, + "grad_norm": 2.249423027038574, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7247823476791382, + "num_tokens": 642270152.0, + "step": 24821 + }, + { + "epoch": 2.7258950142763014, + "grad_norm": 2.204089641571045, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.730635404586792, + "num_tokens": 642303524.0, + "step": 24822 + }, + { + "epoch": 2.7260048319789147, + "grad_norm": 2.5201892852783203, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7280260324478149, + "num_tokens": 642326904.0, + "step": 24823 + }, + { + "epoch": 2.7261146496815285, + "grad_norm": 2.4238171577453613, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7290865182876587, + "num_tokens": 642352381.0, + "step": 24824 + }, + { + "epoch": 2.7262244673841423, + "grad_norm": 2.3783111572265625, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7280024290084839, + "num_tokens": 642378034.0, + "step": 24825 + }, + { + "epoch": 2.726334285086756, + "grad_norm": 2.3875508308410645, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7413203716278076, + "num_tokens": 642403044.0, + "step": 24826 + }, + { + "epoch": 2.7264441027893698, + "grad_norm": 2.4685142040252686, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7236949801445007, + "num_tokens": 642426545.0, + "step": 24827 + }, + { + "epoch": 2.726553920491983, + "grad_norm": 2.2944931983947754, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.755693793296814, + "num_tokens": 642452117.0, + "step": 24828 + }, + { + "epoch": 2.726663738194597, + "grad_norm": 2.482264518737793, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7191377282142639, + "num_tokens": 642478164.0, + "step": 24829 + }, + { + "epoch": 2.7267735558972106, + "grad_norm": 2.1907248497009277, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7082469463348389, + "num_tokens": 642506938.0, + "step": 24830 + }, + { + "epoch": 2.7268833735998244, + "grad_norm": 2.0918633937835693, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7124181985855103, + "num_tokens": 642542506.0, + "step": 24831 + }, + { + "epoch": 2.726993191302438, + "grad_norm": 2.554081678390503, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.738567590713501, + "num_tokens": 642563542.0, + "step": 24832 + }, + { + "epoch": 2.7271030090050514, + "grad_norm": 2.087613344192505, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7266713976860046, + "num_tokens": 642595227.0, + "step": 24833 + }, + { + "epoch": 2.727212826707665, + "grad_norm": 2.911323308944702, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7493347525596619, + "num_tokens": 642613641.0, + "step": 24834 + }, + { + "epoch": 2.727322644410279, + "grad_norm": 2.2211620807647705, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7254222631454468, + "num_tokens": 642640041.0, + "step": 24835 + }, + { + "epoch": 2.7274324621128927, + "grad_norm": 2.7522361278533936, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7627235651016235, + "num_tokens": 642658862.0, + "step": 24836 + }, + { + "epoch": 2.7275422798155065, + "grad_norm": 2.446157693862915, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7384452819824219, + "num_tokens": 642683284.0, + "step": 24837 + }, + { + "epoch": 2.7276520975181198, + "grad_norm": 2.1069390773773193, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7432072162628174, + "num_tokens": 642711859.0, + "step": 24838 + }, + { + "epoch": 2.7277619152207335, + "grad_norm": 1.9862639904022217, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7164443731307983, + "num_tokens": 642749230.0, + "step": 24839 + }, + { + "epoch": 2.7278717329233473, + "grad_norm": 2.3726749420166016, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7164666652679443, + "num_tokens": 642775552.0, + "step": 24840 + }, + { + "epoch": 2.727981550625961, + "grad_norm": 2.5258355140686035, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7545309662818909, + "num_tokens": 642797175.0, + "step": 24841 + }, + { + "epoch": 2.728091368328575, + "grad_norm": 2.353379964828491, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7275683879852295, + "num_tokens": 642823312.0, + "step": 24842 + }, + { + "epoch": 2.728201186031188, + "grad_norm": 2.266838550567627, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7250430583953857, + "num_tokens": 642851144.0, + "step": 24843 + }, + { + "epoch": 2.728311003733802, + "grad_norm": 2.321892738342285, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7369345426559448, + "num_tokens": 642878449.0, + "step": 24844 + }, + { + "epoch": 2.7284208214364156, + "grad_norm": 2.493748664855957, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7538601756095886, + "num_tokens": 642900043.0, + "step": 24845 + }, + { + "epoch": 2.728530639139029, + "grad_norm": 2.063352584838867, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7071243524551392, + "num_tokens": 642933975.0, + "step": 24846 + }, + { + "epoch": 2.728640456841643, + "grad_norm": 2.570526599884033, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.737255334854126, + "num_tokens": 642957037.0, + "step": 24847 + }, + { + "epoch": 2.7287502745442564, + "grad_norm": 2.279778242111206, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7384814620018005, + "num_tokens": 642983748.0, + "step": 24848 + }, + { + "epoch": 2.72886009224687, + "grad_norm": 2.024035930633545, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.73188716173172, + "num_tokens": 643015345.0, + "step": 24849 + }, + { + "epoch": 2.728969909949484, + "grad_norm": 2.640608310699463, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7286486029624939, + "num_tokens": 643037919.0, + "step": 24850 + }, + { + "epoch": 2.7290797276520973, + "grad_norm": 2.1548702716827393, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7432326078414917, + "num_tokens": 643066006.0, + "step": 24851 + }, + { + "epoch": 2.729189545354711, + "grad_norm": 2.230785846710205, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7471652626991272, + "num_tokens": 643092663.0, + "step": 24852 + }, + { + "epoch": 2.729299363057325, + "grad_norm": 2.224950075149536, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7288881540298462, + "num_tokens": 643122114.0, + "step": 24853 + }, + { + "epoch": 2.7294091807599385, + "grad_norm": 2.5314688682556152, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7407757043838501, + "num_tokens": 643143373.0, + "step": 24854 + }, + { + "epoch": 2.7295189984625523, + "grad_norm": 2.346092462539673, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7490803003311157, + "num_tokens": 643166868.0, + "step": 24855 + }, + { + "epoch": 2.7296288161651656, + "grad_norm": 2.3719558715820312, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7220616340637207, + "num_tokens": 643195898.0, + "step": 24856 + }, + { + "epoch": 2.7297386338677794, + "grad_norm": 2.0939908027648926, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.705736517906189, + "num_tokens": 643228559.0, + "step": 24857 + }, + { + "epoch": 2.729848451570393, + "grad_norm": 2.2616491317749023, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7488182783126831, + "num_tokens": 643254623.0, + "step": 24858 + }, + { + "epoch": 2.729958269273007, + "grad_norm": 2.3893508911132812, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7450869679450989, + "num_tokens": 643280531.0, + "step": 24859 + }, + { + "epoch": 2.7300680869756206, + "grad_norm": 2.6920156478881836, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7392476797103882, + "num_tokens": 643300152.0, + "step": 24860 + }, + { + "epoch": 2.730177904678234, + "grad_norm": 2.211043119430542, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7325631380081177, + "num_tokens": 643327754.0, + "step": 24861 + }, + { + "epoch": 2.7302877223808477, + "grad_norm": 2.5264217853546143, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7399998307228088, + "num_tokens": 643352108.0, + "step": 24862 + }, + { + "epoch": 2.7303975400834615, + "grad_norm": 2.2088003158569336, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.727977454662323, + "num_tokens": 643380193.0, + "step": 24863 + }, + { + "epoch": 2.7305073577860752, + "grad_norm": 2.2240419387817383, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7139429450035095, + "num_tokens": 643409522.0, + "step": 24864 + }, + { + "epoch": 2.730617175488689, + "grad_norm": 2.223827600479126, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7344365119934082, + "num_tokens": 643437342.0, + "step": 24865 + }, + { + "epoch": 2.7307269931913023, + "grad_norm": 2.3382019996643066, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7279421091079712, + "num_tokens": 643463951.0, + "step": 24866 + }, + { + "epoch": 2.730836810893916, + "grad_norm": 2.3863048553466797, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.754904568195343, + "num_tokens": 643487187.0, + "step": 24867 + }, + { + "epoch": 2.73094662859653, + "grad_norm": 2.589257001876831, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7251777648925781, + "num_tokens": 643508957.0, + "step": 24868 + }, + { + "epoch": 2.7310564462991436, + "grad_norm": 2.330275058746338, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7301761507987976, + "num_tokens": 643534542.0, + "step": 24869 + }, + { + "epoch": 2.7311662640017573, + "grad_norm": 2.3649392127990723, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7340713739395142, + "num_tokens": 643560555.0, + "step": 24870 + }, + { + "epoch": 2.7312760817043706, + "grad_norm": 2.022068738937378, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7251870036125183, + "num_tokens": 643594435.0, + "step": 24871 + }, + { + "epoch": 2.7313858994069844, + "grad_norm": 2.192040205001831, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7113101482391357, + "num_tokens": 643623464.0, + "step": 24872 + }, + { + "epoch": 2.731495717109598, + "grad_norm": 2.289853572845459, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7263944149017334, + "num_tokens": 643651590.0, + "step": 24873 + }, + { + "epoch": 2.7316055348122115, + "grad_norm": 2.5016188621520996, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7327892780303955, + "num_tokens": 643679139.0, + "step": 24874 + }, + { + "epoch": 2.7317153525148252, + "grad_norm": 2.4883968830108643, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7230606079101562, + "num_tokens": 643704449.0, + "step": 24875 + }, + { + "epoch": 2.731825170217439, + "grad_norm": 2.5377211570739746, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7392994165420532, + "num_tokens": 643727376.0, + "step": 24876 + }, + { + "epoch": 2.7319349879200527, + "grad_norm": 2.457965135574341, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7256309986114502, + "num_tokens": 643751934.0, + "step": 24877 + }, + { + "epoch": 2.7320448056226665, + "grad_norm": 2.333280324935913, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7308942079544067, + "num_tokens": 643778941.0, + "step": 24878 + }, + { + "epoch": 2.73215462332528, + "grad_norm": 2.3647844791412354, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7328924536705017, + "num_tokens": 643805336.0, + "step": 24879 + }, + { + "epoch": 2.7322644410278936, + "grad_norm": 2.1680309772491455, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7017481327056885, + "num_tokens": 643834946.0, + "step": 24880 + }, + { + "epoch": 2.7323742587305073, + "grad_norm": 2.1379826068878174, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7227393388748169, + "num_tokens": 643864791.0, + "step": 24881 + }, + { + "epoch": 2.732484076433121, + "grad_norm": 2.1706695556640625, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7215501666069031, + "num_tokens": 643896644.0, + "step": 24882 + }, + { + "epoch": 2.732593894135735, + "grad_norm": 2.3586976528167725, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7413438558578491, + "num_tokens": 643924896.0, + "step": 24883 + }, + { + "epoch": 2.732703711838348, + "grad_norm": 2.1597089767456055, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7420670986175537, + "num_tokens": 643955376.0, + "step": 24884 + }, + { + "epoch": 2.732813529540962, + "grad_norm": 2.10839581489563, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7457516193389893, + "num_tokens": 643984504.0, + "step": 24885 + }, + { + "epoch": 2.7329233472435757, + "grad_norm": 2.433907985687256, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7291516065597534, + "num_tokens": 644010345.0, + "step": 24886 + }, + { + "epoch": 2.7330331649461894, + "grad_norm": 2.4304070472717285, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7336549758911133, + "num_tokens": 644034153.0, + "step": 24887 + }, + { + "epoch": 2.733142982648803, + "grad_norm": 2.337247610092163, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.735032320022583, + "num_tokens": 644059787.0, + "step": 24888 + }, + { + "epoch": 2.7332528003514165, + "grad_norm": 2.4819419384002686, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7461129426956177, + "num_tokens": 644082650.0, + "step": 24889 + }, + { + "epoch": 2.7333626180540302, + "grad_norm": 2.4420011043548584, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7488963603973389, + "num_tokens": 644106614.0, + "step": 24890 + }, + { + "epoch": 2.733472435756644, + "grad_norm": 2.502251386642456, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7232908010482788, + "num_tokens": 644130761.0, + "step": 24891 + }, + { + "epoch": 2.7335822534592578, + "grad_norm": 2.462930917739868, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7480511665344238, + "num_tokens": 644154838.0, + "step": 24892 + }, + { + "epoch": 2.7336920711618715, + "grad_norm": 2.4737415313720703, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7438400983810425, + "num_tokens": 644179150.0, + "step": 24893 + }, + { + "epoch": 2.733801888864485, + "grad_norm": 2.2121667861938477, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7409566640853882, + "num_tokens": 644206990.0, + "step": 24894 + }, + { + "epoch": 2.7339117065670986, + "grad_norm": 2.4459340572357178, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7125878930091858, + "num_tokens": 644233162.0, + "step": 24895 + }, + { + "epoch": 2.7340215242697123, + "grad_norm": 2.510406017303467, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7100428342819214, + "num_tokens": 644256968.0, + "step": 24896 + }, + { + "epoch": 2.734131341972326, + "grad_norm": 2.2461016178131104, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7252714037895203, + "num_tokens": 644284581.0, + "step": 24897 + }, + { + "epoch": 2.73424115967494, + "grad_norm": 2.58583927154541, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7337454557418823, + "num_tokens": 644308226.0, + "step": 24898 + }, + { + "epoch": 2.734350977377553, + "grad_norm": 2.1622886657714844, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.71480792760849, + "num_tokens": 644340434.0, + "step": 24899 + }, + { + "epoch": 2.734460795080167, + "grad_norm": 2.2313385009765625, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7339625954627991, + "num_tokens": 644371815.0, + "step": 24900 + }, + { + "epoch": 2.7345706127827807, + "grad_norm": 2.337764263153076, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7299428582191467, + "num_tokens": 644396875.0, + "step": 24901 + }, + { + "epoch": 2.734680430485394, + "grad_norm": 2.2709639072418213, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7255847454071045, + "num_tokens": 644423665.0, + "step": 24902 + }, + { + "epoch": 2.7347902481880078, + "grad_norm": 2.135880947113037, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.716800332069397, + "num_tokens": 644455080.0, + "step": 24903 + }, + { + "epoch": 2.7349000658906215, + "grad_norm": 2.4889893531799316, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.74639892578125, + "num_tokens": 644478301.0, + "step": 24904 + }, + { + "epoch": 2.7350098835932353, + "grad_norm": 2.607789993286133, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7392703890800476, + "num_tokens": 644500203.0, + "step": 24905 + }, + { + "epoch": 2.735119701295849, + "grad_norm": 2.2686781883239746, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7208268642425537, + "num_tokens": 644528204.0, + "step": 24906 + }, + { + "epoch": 2.7352295189984623, + "grad_norm": 2.813692331314087, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7516732215881348, + "num_tokens": 644547240.0, + "step": 24907 + }, + { + "epoch": 2.735339336701076, + "grad_norm": 2.4760358333587646, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7414034008979797, + "num_tokens": 644570855.0, + "step": 24908 + }, + { + "epoch": 2.73544915440369, + "grad_norm": 2.443904399871826, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7107968926429749, + "num_tokens": 644595127.0, + "step": 24909 + }, + { + "epoch": 2.7355589721063036, + "grad_norm": 2.1632325649261475, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7349410653114319, + "num_tokens": 644625702.0, + "step": 24910 + }, + { + "epoch": 2.7356687898089174, + "grad_norm": 2.581071138381958, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.717720627784729, + "num_tokens": 644649169.0, + "step": 24911 + }, + { + "epoch": 2.7357786075115307, + "grad_norm": 2.6918578147888184, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7476441860198975, + "num_tokens": 644670190.0, + "step": 24912 + }, + { + "epoch": 2.7358884252141444, + "grad_norm": 2.337312698364258, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7291108965873718, + "num_tokens": 644694420.0, + "step": 24913 + }, + { + "epoch": 2.735998242916758, + "grad_norm": 2.484372615814209, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.723515510559082, + "num_tokens": 644717165.0, + "step": 24914 + }, + { + "epoch": 2.736108060619372, + "grad_norm": 2.247385263442993, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7424880266189575, + "num_tokens": 644745186.0, + "step": 24915 + }, + { + "epoch": 2.7362178783219857, + "grad_norm": 2.289982318878174, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7201534509658813, + "num_tokens": 644773772.0, + "step": 24916 + }, + { + "epoch": 2.736327696024599, + "grad_norm": 2.417240858078003, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7190014123916626, + "num_tokens": 644800001.0, + "step": 24917 + }, + { + "epoch": 2.736437513727213, + "grad_norm": 2.3825197219848633, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7231389284133911, + "num_tokens": 644825060.0, + "step": 24918 + }, + { + "epoch": 2.7365473314298265, + "grad_norm": 2.4919824600219727, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7345709800720215, + "num_tokens": 644848268.0, + "step": 24919 + }, + { + "epoch": 2.7366571491324403, + "grad_norm": 2.692012071609497, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.740410566329956, + "num_tokens": 644868479.0, + "step": 24920 + }, + { + "epoch": 2.736766966835054, + "grad_norm": 1.8415846824645996, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7285380363464355, + "num_tokens": 644904886.0, + "step": 24921 + }, + { + "epoch": 2.7368767845376674, + "grad_norm": 1.977891445159912, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7114976048469543, + "num_tokens": 644942507.0, + "step": 24922 + }, + { + "epoch": 2.736986602240281, + "grad_norm": 2.700650691986084, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7607213258743286, + "num_tokens": 644960996.0, + "step": 24923 + }, + { + "epoch": 2.737096419942895, + "grad_norm": 2.2037458419799805, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7332940101623535, + "num_tokens": 644991209.0, + "step": 24924 + }, + { + "epoch": 2.737206237645508, + "grad_norm": 2.2235846519470215, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7130497694015503, + "num_tokens": 645021253.0, + "step": 24925 + }, + { + "epoch": 2.7373160553481224, + "grad_norm": 2.437479019165039, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7483597993850708, + "num_tokens": 645045743.0, + "step": 24926 + }, + { + "epoch": 2.7374258730507357, + "grad_norm": 2.4240524768829346, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7306750416755676, + "num_tokens": 645070358.0, + "step": 24927 + }, + { + "epoch": 2.7375356907533495, + "grad_norm": 2.4324636459350586, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7509896159172058, + "num_tokens": 645093569.0, + "step": 24928 + }, + { + "epoch": 2.737645508455963, + "grad_norm": 2.1879258155822754, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7184553742408752, + "num_tokens": 645123189.0, + "step": 24929 + }, + { + "epoch": 2.7377553261585765, + "grad_norm": 2.3379406929016113, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7427706718444824, + "num_tokens": 645148340.0, + "step": 24930 + }, + { + "epoch": 2.7378651438611903, + "grad_norm": 2.6280713081359863, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7520935535430908, + "num_tokens": 645170493.0, + "step": 24931 + }, + { + "epoch": 2.737974961563804, + "grad_norm": 2.4606540203094482, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7210012674331665, + "num_tokens": 645196556.0, + "step": 24932 + }, + { + "epoch": 2.738084779266418, + "grad_norm": 2.412400960922241, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7277275323867798, + "num_tokens": 645222041.0, + "step": 24933 + }, + { + "epoch": 2.7381945969690316, + "grad_norm": 2.518465518951416, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7498904466629028, + "num_tokens": 645243118.0, + "step": 24934 + }, + { + "epoch": 2.738304414671645, + "grad_norm": 2.9717369079589844, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7372373938560486, + "num_tokens": 645261831.0, + "step": 24935 + }, + { + "epoch": 2.7384142323742586, + "grad_norm": 2.317352771759033, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7147025465965271, + "num_tokens": 645289555.0, + "step": 24936 + }, + { + "epoch": 2.7385240500768724, + "grad_norm": 2.5910913944244385, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7430477738380432, + "num_tokens": 645311314.0, + "step": 24937 + }, + { + "epoch": 2.738633867779486, + "grad_norm": 2.3039956092834473, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7126573324203491, + "num_tokens": 645341909.0, + "step": 24938 + }, + { + "epoch": 2.7387436854821, + "grad_norm": 2.588033676147461, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.71732497215271, + "num_tokens": 645366844.0, + "step": 24939 + }, + { + "epoch": 2.738853503184713, + "grad_norm": 2.615067481994629, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7262439131736755, + "num_tokens": 645389199.0, + "step": 24940 + }, + { + "epoch": 2.738963320887327, + "grad_norm": 2.1051290035247803, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7250820398330688, + "num_tokens": 645423038.0, + "step": 24941 + }, + { + "epoch": 2.7390731385899407, + "grad_norm": 2.252572774887085, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7451250553131104, + "num_tokens": 645451731.0, + "step": 24942 + }, + { + "epoch": 2.7391829562925545, + "grad_norm": 2.2537403106689453, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7296100854873657, + "num_tokens": 645478334.0, + "step": 24943 + }, + { + "epoch": 2.7392927739951682, + "grad_norm": 2.4531333446502686, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7289068698883057, + "num_tokens": 645502011.0, + "step": 24944 + }, + { + "epoch": 2.7394025916977816, + "grad_norm": 2.415764808654785, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.729365885257721, + "num_tokens": 645526537.0, + "step": 24945 + }, + { + "epoch": 2.7395124094003953, + "grad_norm": 2.1248536109924316, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7654119729995728, + "num_tokens": 645555576.0, + "step": 24946 + }, + { + "epoch": 2.739622227103009, + "grad_norm": 2.575594663619995, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7373432517051697, + "num_tokens": 645579414.0, + "step": 24947 + }, + { + "epoch": 2.739732044805623, + "grad_norm": 2.701688528060913, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7453183531761169, + "num_tokens": 645600487.0, + "step": 24948 + }, + { + "epoch": 2.7398418625082366, + "grad_norm": 2.2584574222564697, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7258665561676025, + "num_tokens": 645627786.0, + "step": 24949 + }, + { + "epoch": 2.73995168021085, + "grad_norm": 2.610914468765259, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7430806756019592, + "num_tokens": 645650574.0, + "step": 24950 + }, + { + "epoch": 2.7400614979134637, + "grad_norm": 2.04746675491333, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7051336765289307, + "num_tokens": 645687087.0, + "step": 24951 + }, + { + "epoch": 2.7401713156160774, + "grad_norm": 2.390360116958618, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7368717193603516, + "num_tokens": 645710254.0, + "step": 24952 + }, + { + "epoch": 2.7402811333186907, + "grad_norm": 2.177945613861084, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.721659779548645, + "num_tokens": 645740072.0, + "step": 24953 + }, + { + "epoch": 2.7403909510213045, + "grad_norm": 2.960510492324829, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7415069341659546, + "num_tokens": 645758717.0, + "step": 24954 + }, + { + "epoch": 2.7405007687239182, + "grad_norm": 2.3077199459075928, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7186245918273926, + "num_tokens": 645783376.0, + "step": 24955 + }, + { + "epoch": 2.740610586426532, + "grad_norm": 2.4258434772491455, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7200308442115784, + "num_tokens": 645810540.0, + "step": 24956 + }, + { + "epoch": 2.7407204041291457, + "grad_norm": 2.411583423614502, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7290840148925781, + "num_tokens": 645836400.0, + "step": 24957 + }, + { + "epoch": 2.740830221831759, + "grad_norm": 2.1406779289245605, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7126954793930054, + "num_tokens": 645865217.0, + "step": 24958 + }, + { + "epoch": 2.740940039534373, + "grad_norm": 2.0782277584075928, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7159656882286072, + "num_tokens": 645897124.0, + "step": 24959 + }, + { + "epoch": 2.7410498572369866, + "grad_norm": 2.5574209690093994, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7581127285957336, + "num_tokens": 645919008.0, + "step": 24960 + }, + { + "epoch": 2.7411596749396003, + "grad_norm": 2.3957901000976562, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7287176251411438, + "num_tokens": 645944119.0, + "step": 24961 + }, + { + "epoch": 2.741269492642214, + "grad_norm": 2.155614137649536, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7126271724700928, + "num_tokens": 645976846.0, + "step": 24962 + }, + { + "epoch": 2.7413793103448274, + "grad_norm": 2.4766287803649902, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7404993772506714, + "num_tokens": 646000486.0, + "step": 24963 + }, + { + "epoch": 2.741489128047441, + "grad_norm": 2.3615221977233887, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7114286422729492, + "num_tokens": 646028394.0, + "step": 24964 + }, + { + "epoch": 2.741598945750055, + "grad_norm": 2.111093521118164, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7167946696281433, + "num_tokens": 646058019.0, + "step": 24965 + }, + { + "epoch": 2.7417087634526687, + "grad_norm": 2.4594640731811523, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7193903923034668, + "num_tokens": 646082838.0, + "step": 24966 + }, + { + "epoch": 2.7418185811552824, + "grad_norm": 2.1915156841278076, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7413201332092285, + "num_tokens": 646112665.0, + "step": 24967 + }, + { + "epoch": 2.7419283988578957, + "grad_norm": 2.116791248321533, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7181035876274109, + "num_tokens": 646143820.0, + "step": 24968 + }, + { + "epoch": 2.7420382165605095, + "grad_norm": 2.3784422874450684, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7188829183578491, + "num_tokens": 646170636.0, + "step": 24969 + }, + { + "epoch": 2.7421480342631233, + "grad_norm": 2.59920597076416, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7224709987640381, + "num_tokens": 646192243.0, + "step": 24970 + }, + { + "epoch": 2.742257851965737, + "grad_norm": 2.14249849319458, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7103117108345032, + "num_tokens": 646224106.0, + "step": 24971 + }, + { + "epoch": 2.7423676696683508, + "grad_norm": 2.258556604385376, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7187734842300415, + "num_tokens": 646251003.0, + "step": 24972 + }, + { + "epoch": 2.742477487370964, + "grad_norm": 2.2990355491638184, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7102248668670654, + "num_tokens": 646278161.0, + "step": 24973 + }, + { + "epoch": 2.742587305073578, + "grad_norm": 2.544560670852661, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7587592601776123, + "num_tokens": 646300023.0, + "step": 24974 + }, + { + "epoch": 2.7426971227761916, + "grad_norm": 2.1177151203155518, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7178667783737183, + "num_tokens": 646330269.0, + "step": 24975 + }, + { + "epoch": 2.742806940478805, + "grad_norm": 2.2271604537963867, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7185701131820679, + "num_tokens": 646359177.0, + "step": 24976 + }, + { + "epoch": 2.742916758181419, + "grad_norm": 2.7771286964416504, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7376977205276489, + "num_tokens": 646378780.0, + "step": 24977 + }, + { + "epoch": 2.7430265758840324, + "grad_norm": 2.2150635719299316, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7386350035667419, + "num_tokens": 646408380.0, + "step": 24978 + }, + { + "epoch": 2.743136393586646, + "grad_norm": 2.636026620864868, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7273287773132324, + "num_tokens": 646436526.0, + "step": 24979 + }, + { + "epoch": 2.74324621128926, + "grad_norm": 2.663325309753418, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.747934877872467, + "num_tokens": 646457486.0, + "step": 24980 + }, + { + "epoch": 2.7433560289918733, + "grad_norm": 2.6148569583892822, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7352282404899597, + "num_tokens": 646478215.0, + "step": 24981 + }, + { + "epoch": 2.743465846694487, + "grad_norm": 2.3485889434814453, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7317915558815002, + "num_tokens": 646502529.0, + "step": 24982 + }, + { + "epoch": 2.7435756643971008, + "grad_norm": 2.145035982131958, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7215948104858398, + "num_tokens": 646532780.0, + "step": 24983 + }, + { + "epoch": 2.7436854820997145, + "grad_norm": 2.5035364627838135, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7447926998138428, + "num_tokens": 646555900.0, + "step": 24984 + }, + { + "epoch": 2.7437952998023283, + "grad_norm": 1.9422197341918945, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7058913707733154, + "num_tokens": 646590148.0, + "step": 24985 + }, + { + "epoch": 2.7439051175049416, + "grad_norm": 2.451064348220825, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7486111521720886, + "num_tokens": 646615531.0, + "step": 24986 + }, + { + "epoch": 2.7440149352075554, + "grad_norm": 2.341850519180298, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7436178922653198, + "num_tokens": 646640800.0, + "step": 24987 + }, + { + "epoch": 2.744124752910169, + "grad_norm": 2.672354221343994, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7315489053726196, + "num_tokens": 646664533.0, + "step": 24988 + }, + { + "epoch": 2.744234570612783, + "grad_norm": 2.4968674182891846, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7487321496009827, + "num_tokens": 646687570.0, + "step": 24989 + }, + { + "epoch": 2.7443443883153966, + "grad_norm": 2.3059027194976807, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7171050906181335, + "num_tokens": 646713050.0, + "step": 24990 + }, + { + "epoch": 2.74445420601801, + "grad_norm": 2.5000717639923096, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.746565580368042, + "num_tokens": 646736710.0, + "step": 24991 + }, + { + "epoch": 2.7445640237206237, + "grad_norm": 2.4952375888824463, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7368795871734619, + "num_tokens": 646760872.0, + "step": 24992 + }, + { + "epoch": 2.7446738414232374, + "grad_norm": 2.4749667644500732, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7509036064147949, + "num_tokens": 646783717.0, + "step": 24993 + }, + { + "epoch": 2.744783659125851, + "grad_norm": 2.294769763946533, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7225892543792725, + "num_tokens": 646810412.0, + "step": 24994 + }, + { + "epoch": 2.744893476828465, + "grad_norm": 2.6407623291015625, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7265772223472595, + "num_tokens": 646833771.0, + "step": 24995 + }, + { + "epoch": 2.7450032945310783, + "grad_norm": 2.4128470420837402, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7367996573448181, + "num_tokens": 646858231.0, + "step": 24996 + }, + { + "epoch": 2.745113112233692, + "grad_norm": 2.741069793701172, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7322577238082886, + "num_tokens": 646879355.0, + "step": 24997 + }, + { + "epoch": 2.745222929936306, + "grad_norm": 2.204892158508301, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.700452446937561, + "num_tokens": 646910891.0, + "step": 24998 + }, + { + "epoch": 2.7453327476389195, + "grad_norm": 2.2579073905944824, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7428602576255798, + "num_tokens": 646938572.0, + "step": 24999 + }, + { + "epoch": 2.7454425653415333, + "grad_norm": 2.2720677852630615, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7435343265533447, + "num_tokens": 646964803.0, + "step": 25000 + }, + { + "epoch": 2.7455523830441466, + "grad_norm": 2.072003126144409, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7263807058334351, + "num_tokens": 646997019.0, + "step": 25001 + }, + { + "epoch": 2.7456622007467604, + "grad_norm": 2.483042001724243, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7464462518692017, + "num_tokens": 647020212.0, + "step": 25002 + }, + { + "epoch": 2.745772018449374, + "grad_norm": 2.175020456314087, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7172983884811401, + "num_tokens": 647049845.0, + "step": 25003 + }, + { + "epoch": 2.7458818361519874, + "grad_norm": 2.271313190460205, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7148116827011108, + "num_tokens": 647078205.0, + "step": 25004 + }, + { + "epoch": 2.745991653854601, + "grad_norm": 1.9575603008270264, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7165535092353821, + "num_tokens": 647113634.0, + "step": 25005 + }, + { + "epoch": 2.746101471557215, + "grad_norm": 2.6297683715820312, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7298174500465393, + "num_tokens": 647136875.0, + "step": 25006 + }, + { + "epoch": 2.7462112892598287, + "grad_norm": 2.3355939388275146, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7362180352210999, + "num_tokens": 647162932.0, + "step": 25007 + }, + { + "epoch": 2.7463211069624425, + "grad_norm": 2.2570438385009766, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7158784866333008, + "num_tokens": 647190846.0, + "step": 25008 + }, + { + "epoch": 2.746430924665056, + "grad_norm": 2.201002597808838, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7250155210494995, + "num_tokens": 647219098.0, + "step": 25009 + }, + { + "epoch": 2.7465407423676695, + "grad_norm": 2.408086061477661, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7269097566604614, + "num_tokens": 647243920.0, + "step": 25010 + }, + { + "epoch": 2.7466505600702833, + "grad_norm": 2.443681478500366, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7299348711967468, + "num_tokens": 647268324.0, + "step": 25011 + }, + { + "epoch": 2.746760377772897, + "grad_norm": 2.469147205352783, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7352587580680847, + "num_tokens": 647291575.0, + "step": 25012 + }, + { + "epoch": 2.746870195475511, + "grad_norm": 2.420092821121216, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.756009042263031, + "num_tokens": 647314862.0, + "step": 25013 + }, + { + "epoch": 2.746980013178124, + "grad_norm": 2.3501713275909424, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7268011569976807, + "num_tokens": 647338860.0, + "step": 25014 + }, + { + "epoch": 2.747089830880738, + "grad_norm": 2.1672613620758057, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.727813720703125, + "num_tokens": 647367882.0, + "step": 25015 + }, + { + "epoch": 2.7471996485833516, + "grad_norm": 2.316093921661377, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7124911546707153, + "num_tokens": 647394841.0, + "step": 25016 + }, + { + "epoch": 2.7473094662859654, + "grad_norm": 2.323477268218994, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7312341928482056, + "num_tokens": 647419836.0, + "step": 25017 + }, + { + "epoch": 2.747419283988579, + "grad_norm": 2.4591805934906006, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7361438274383545, + "num_tokens": 647445445.0, + "step": 25018 + }, + { + "epoch": 2.7475291016911925, + "grad_norm": 2.14176607131958, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7388660311698914, + "num_tokens": 647476707.0, + "step": 25019 + }, + { + "epoch": 2.7476389193938062, + "grad_norm": 3.202596426010132, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7446577548980713, + "num_tokens": 647493810.0, + "step": 25020 + }, + { + "epoch": 2.74774873709642, + "grad_norm": 2.264577865600586, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7452919483184814, + "num_tokens": 647521427.0, + "step": 25021 + }, + { + "epoch": 2.7478585547990337, + "grad_norm": 2.2681682109832764, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7281436920166016, + "num_tokens": 647549213.0, + "step": 25022 + }, + { + "epoch": 2.7479683725016475, + "grad_norm": 2.0944879055023193, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6924837827682495, + "num_tokens": 647581008.0, + "step": 25023 + }, + { + "epoch": 2.748078190204261, + "grad_norm": 2.3232805728912354, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7399917840957642, + "num_tokens": 647605790.0, + "step": 25024 + }, + { + "epoch": 2.7481880079068746, + "grad_norm": 2.2428112030029297, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.733054518699646, + "num_tokens": 647631816.0, + "step": 25025 + }, + { + "epoch": 2.7482978256094883, + "grad_norm": 2.5035691261291504, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7450509667396545, + "num_tokens": 647656111.0, + "step": 25026 + }, + { + "epoch": 2.7484076433121016, + "grad_norm": 2.7012367248535156, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7349218130111694, + "num_tokens": 647677258.0, + "step": 25027 + }, + { + "epoch": 2.748517461014716, + "grad_norm": 2.8072569370269775, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7350056171417236, + "num_tokens": 647697016.0, + "step": 25028 + }, + { + "epoch": 2.748627278717329, + "grad_norm": 2.353955030441284, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.726402997970581, + "num_tokens": 647723095.0, + "step": 25029 + }, + { + "epoch": 2.748737096419943, + "grad_norm": 2.431410312652588, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7200420498847961, + "num_tokens": 647747672.0, + "step": 25030 + }, + { + "epoch": 2.7488469141225567, + "grad_norm": 2.3874781131744385, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.729682207107544, + "num_tokens": 647774007.0, + "step": 25031 + }, + { + "epoch": 2.74895673182517, + "grad_norm": 2.268799304962158, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7215075492858887, + "num_tokens": 647804272.0, + "step": 25032 + }, + { + "epoch": 2.7490665495277837, + "grad_norm": 2.515507459640503, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7418577671051025, + "num_tokens": 647827428.0, + "step": 25033 + }, + { + "epoch": 2.7491763672303975, + "grad_norm": 2.6579370498657227, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7344341278076172, + "num_tokens": 647849150.0, + "step": 25034 + }, + { + "epoch": 2.7492861849330112, + "grad_norm": 2.2582638263702393, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7470924854278564, + "num_tokens": 647876113.0, + "step": 25035 + }, + { + "epoch": 2.749396002635625, + "grad_norm": 2.2849278450012207, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7191107273101807, + "num_tokens": 647903846.0, + "step": 25036 + }, + { + "epoch": 2.7495058203382383, + "grad_norm": 2.004986047744751, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7074762582778931, + "num_tokens": 647938368.0, + "step": 25037 + }, + { + "epoch": 2.749615638040852, + "grad_norm": 2.074274778366089, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7204969525337219, + "num_tokens": 647970228.0, + "step": 25038 + }, + { + "epoch": 2.749725455743466, + "grad_norm": 2.15354323387146, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7229421138763428, + "num_tokens": 647999975.0, + "step": 25039 + }, + { + "epoch": 2.7498352734460796, + "grad_norm": 2.480302095413208, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7338537573814392, + "num_tokens": 648025289.0, + "step": 25040 + }, + { + "epoch": 2.7499450911486933, + "grad_norm": 2.1819756031036377, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7337306141853333, + "num_tokens": 648056141.0, + "step": 25041 + }, + { + "epoch": 2.7500549088513067, + "grad_norm": 2.1210031509399414, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7331379055976868, + "num_tokens": 648087984.0, + "step": 25042 + }, + { + "epoch": 2.7501647265539204, + "grad_norm": 2.1434059143066406, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7434900999069214, + "num_tokens": 648118362.0, + "step": 25043 + }, + { + "epoch": 2.750274544256534, + "grad_norm": 2.3123419284820557, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7561904191970825, + "num_tokens": 648142816.0, + "step": 25044 + }, + { + "epoch": 2.750384361959148, + "grad_norm": 2.2139668464660645, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7540227174758911, + "num_tokens": 648170513.0, + "step": 25045 + }, + { + "epoch": 2.7504941796617617, + "grad_norm": 2.394972562789917, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7388180494308472, + "num_tokens": 648195166.0, + "step": 25046 + }, + { + "epoch": 2.750603997364375, + "grad_norm": 2.2786190509796143, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7276505827903748, + "num_tokens": 648222767.0, + "step": 25047 + }, + { + "epoch": 2.7507138150669888, + "grad_norm": 2.007183313369751, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7272083759307861, + "num_tokens": 648255722.0, + "step": 25048 + }, + { + "epoch": 2.7508236327696025, + "grad_norm": 2.4362030029296875, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7456242442131042, + "num_tokens": 648280043.0, + "step": 25049 + }, + { + "epoch": 2.7509334504722163, + "grad_norm": 2.2954535484313965, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7058199048042297, + "num_tokens": 648307814.0, + "step": 25050 + }, + { + "epoch": 2.75104326817483, + "grad_norm": 2.2827987670898438, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7179086208343506, + "num_tokens": 648336217.0, + "step": 25051 + }, + { + "epoch": 2.7511530858774433, + "grad_norm": 2.2292885780334473, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7480302453041077, + "num_tokens": 648362643.0, + "step": 25052 + }, + { + "epoch": 2.751262903580057, + "grad_norm": 2.2807657718658447, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7182585000991821, + "num_tokens": 648389974.0, + "step": 25053 + }, + { + "epoch": 2.751372721282671, + "grad_norm": 2.2719967365264893, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7090750336647034, + "num_tokens": 648418638.0, + "step": 25054 + }, + { + "epoch": 2.751482538985284, + "grad_norm": 2.535221576690674, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7356213331222534, + "num_tokens": 648440188.0, + "step": 25055 + }, + { + "epoch": 2.7515923566878984, + "grad_norm": 2.3939082622528076, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7528215646743774, + "num_tokens": 648463041.0, + "step": 25056 + }, + { + "epoch": 2.7517021743905117, + "grad_norm": 2.2970023155212402, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7444140911102295, + "num_tokens": 648489706.0, + "step": 25057 + }, + { + "epoch": 2.7518119920931254, + "grad_norm": 2.4284298419952393, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.724903404712677, + "num_tokens": 648513222.0, + "step": 25058 + }, + { + "epoch": 2.751921809795739, + "grad_norm": 2.441762685775757, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.745022177696228, + "num_tokens": 648538625.0, + "step": 25059 + }, + { + "epoch": 2.7520316274983525, + "grad_norm": 2.4751105308532715, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7350447177886963, + "num_tokens": 648562289.0, + "step": 25060 + }, + { + "epoch": 2.7521414452009663, + "grad_norm": 2.0635366439819336, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7154581546783447, + "num_tokens": 648595217.0, + "step": 25061 + }, + { + "epoch": 2.75225126290358, + "grad_norm": 2.3817989826202393, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7300845384597778, + "num_tokens": 648620006.0, + "step": 25062 + }, + { + "epoch": 2.7523610806061938, + "grad_norm": 2.2261157035827637, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.694480836391449, + "num_tokens": 648648475.0, + "step": 25063 + }, + { + "epoch": 2.7524708983088075, + "grad_norm": 2.400883436203003, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7329423427581787, + "num_tokens": 648672852.0, + "step": 25064 + }, + { + "epoch": 2.752580716011421, + "grad_norm": 2.4095470905303955, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7554595470428467, + "num_tokens": 648698109.0, + "step": 25065 + }, + { + "epoch": 2.7526905337140346, + "grad_norm": 2.25229549407959, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7254056930541992, + "num_tokens": 648725701.0, + "step": 25066 + }, + { + "epoch": 2.7528003514166484, + "grad_norm": 2.467803478240967, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7579960823059082, + "num_tokens": 648749283.0, + "step": 25067 + }, + { + "epoch": 2.752910169119262, + "grad_norm": 2.422908067703247, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7625386118888855, + "num_tokens": 648771991.0, + "step": 25068 + }, + { + "epoch": 2.753019986821876, + "grad_norm": 2.42423152923584, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.743972659111023, + "num_tokens": 648795677.0, + "step": 25069 + }, + { + "epoch": 2.753129804524489, + "grad_norm": 2.4593465328216553, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7338801026344299, + "num_tokens": 648818931.0, + "step": 25070 + }, + { + "epoch": 2.753239622227103, + "grad_norm": 2.3292112350463867, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.715818464756012, + "num_tokens": 648847331.0, + "step": 25071 + }, + { + "epoch": 2.7533494399297167, + "grad_norm": 2.665318250656128, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7231691479682922, + "num_tokens": 648868817.0, + "step": 25072 + }, + { + "epoch": 2.7534592576323305, + "grad_norm": 2.3466734886169434, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7231142520904541, + "num_tokens": 648895614.0, + "step": 25073 + }, + { + "epoch": 2.753569075334944, + "grad_norm": 2.321286201477051, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7361485362052917, + "num_tokens": 648921781.0, + "step": 25074 + }, + { + "epoch": 2.7536788930375575, + "grad_norm": 2.1778621673583984, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7125808596611023, + "num_tokens": 648952615.0, + "step": 25075 + }, + { + "epoch": 2.7537887107401713, + "grad_norm": 2.4047813415527344, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.741674542427063, + "num_tokens": 648976681.0, + "step": 25076 + }, + { + "epoch": 2.753898528442785, + "grad_norm": 2.49751877784729, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7386112809181213, + "num_tokens": 649001451.0, + "step": 25077 + }, + { + "epoch": 2.754008346145399, + "grad_norm": 2.2129275798797607, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.710058331489563, + "num_tokens": 649030563.0, + "step": 25078 + }, + { + "epoch": 2.7541181638480126, + "grad_norm": 2.6323251724243164, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7304872274398804, + "num_tokens": 649052981.0, + "step": 25079 + }, + { + "epoch": 2.754227981550626, + "grad_norm": 2.2400007247924805, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.730492115020752, + "num_tokens": 649081091.0, + "step": 25080 + }, + { + "epoch": 2.7543377992532396, + "grad_norm": 2.4217722415924072, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7344973683357239, + "num_tokens": 649105045.0, + "step": 25081 + }, + { + "epoch": 2.7544476169558534, + "grad_norm": 2.2402114868164062, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7390127778053284, + "num_tokens": 649133027.0, + "step": 25082 + }, + { + "epoch": 2.7545574346584667, + "grad_norm": 2.7537612915039062, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7457934617996216, + "num_tokens": 649151761.0, + "step": 25083 + }, + { + "epoch": 2.7546672523610805, + "grad_norm": 2.0810134410858154, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7266827821731567, + "num_tokens": 649183703.0, + "step": 25084 + }, + { + "epoch": 2.754777070063694, + "grad_norm": 2.421682834625244, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7245244979858398, + "num_tokens": 649209334.0, + "step": 25085 + }, + { + "epoch": 2.754886887766308, + "grad_norm": 2.294696569442749, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7503716349601746, + "num_tokens": 649235028.0, + "step": 25086 + }, + { + "epoch": 2.7549967054689217, + "grad_norm": 2.465902805328369, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7330020666122437, + "num_tokens": 649258176.0, + "step": 25087 + }, + { + "epoch": 2.755106523171535, + "grad_norm": 2.47029447555542, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.736451268196106, + "num_tokens": 649281539.0, + "step": 25088 + }, + { + "epoch": 2.755216340874149, + "grad_norm": 2.1042897701263428, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7011877298355103, + "num_tokens": 649312403.0, + "step": 25089 + }, + { + "epoch": 2.7553261585767626, + "grad_norm": 2.266350507736206, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.711716890335083, + "num_tokens": 649341352.0, + "step": 25090 + }, + { + "epoch": 2.7554359762793763, + "grad_norm": 2.6455111503601074, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7392234206199646, + "num_tokens": 649363178.0, + "step": 25091 + }, + { + "epoch": 2.75554579398199, + "grad_norm": 2.8921258449554443, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7527955174446106, + "num_tokens": 649381050.0, + "step": 25092 + }, + { + "epoch": 2.7556556116846034, + "grad_norm": 2.492572069168091, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7351155281066895, + "num_tokens": 649405317.0, + "step": 25093 + }, + { + "epoch": 2.755765429387217, + "grad_norm": 2.3969709873199463, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7360393404960632, + "num_tokens": 649430816.0, + "step": 25094 + }, + { + "epoch": 2.755875247089831, + "grad_norm": 2.3725433349609375, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7560449242591858, + "num_tokens": 649456318.0, + "step": 25095 + }, + { + "epoch": 2.7559850647924446, + "grad_norm": 2.4991695880889893, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.732276201248169, + "num_tokens": 649480182.0, + "step": 25096 + }, + { + "epoch": 2.7560948824950584, + "grad_norm": 2.5264394283294678, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7381228804588318, + "num_tokens": 649504133.0, + "step": 25097 + }, + { + "epoch": 2.7562047001976717, + "grad_norm": 2.198577642440796, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.727191150188446, + "num_tokens": 649533342.0, + "step": 25098 + }, + { + "epoch": 2.7563145179002855, + "grad_norm": 2.088914632797241, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7092956900596619, + "num_tokens": 649567140.0, + "step": 25099 + }, + { + "epoch": 2.7564243356028992, + "grad_norm": 2.380803108215332, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7355901598930359, + "num_tokens": 649591643.0, + "step": 25100 + }, + { + "epoch": 2.756534153305513, + "grad_norm": 2.0724658966064453, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7122113108634949, + "num_tokens": 649624483.0, + "step": 25101 + }, + { + "epoch": 2.7566439710081267, + "grad_norm": 2.3066771030426025, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7733617424964905, + "num_tokens": 649650361.0, + "step": 25102 + }, + { + "epoch": 2.75675378871074, + "grad_norm": 2.649904251098633, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7213587760925293, + "num_tokens": 649670929.0, + "step": 25103 + }, + { + "epoch": 2.756863606413354, + "grad_norm": 1.9886608123779297, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7297966480255127, + "num_tokens": 649704762.0, + "step": 25104 + }, + { + "epoch": 2.7569734241159676, + "grad_norm": 2.52449631690979, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7124213576316833, + "num_tokens": 649728888.0, + "step": 25105 + }, + { + "epoch": 2.757083241818581, + "grad_norm": 2.217264175415039, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.6996515393257141, + "num_tokens": 649756537.0, + "step": 25106 + }, + { + "epoch": 2.757193059521195, + "grad_norm": 2.360668420791626, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7302011251449585, + "num_tokens": 649784537.0, + "step": 25107 + }, + { + "epoch": 2.7573028772238084, + "grad_norm": 2.2792670726776123, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7067952156066895, + "num_tokens": 649812101.0, + "step": 25108 + }, + { + "epoch": 2.757412694926422, + "grad_norm": 2.0620954036712646, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7420936822891235, + "num_tokens": 649840318.0, + "step": 25109 + }, + { + "epoch": 2.757522512629036, + "grad_norm": 2.4666261672973633, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7416566610336304, + "num_tokens": 649865001.0, + "step": 25110 + }, + { + "epoch": 2.7576323303316492, + "grad_norm": 2.467151165008545, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7484632134437561, + "num_tokens": 649889271.0, + "step": 25111 + }, + { + "epoch": 2.757742148034263, + "grad_norm": 2.3003623485565186, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7215981483459473, + "num_tokens": 649916816.0, + "step": 25112 + }, + { + "epoch": 2.7578519657368767, + "grad_norm": 2.4663267135620117, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7148013114929199, + "num_tokens": 649941801.0, + "step": 25113 + }, + { + "epoch": 2.7579617834394905, + "grad_norm": 2.5840296745300293, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7336286306381226, + "num_tokens": 649964154.0, + "step": 25114 + }, + { + "epoch": 2.7580716011421043, + "grad_norm": 2.172292947769165, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7217040061950684, + "num_tokens": 649994431.0, + "step": 25115 + }, + { + "epoch": 2.7581814188447176, + "grad_norm": 2.198852300643921, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7386876344680786, + "num_tokens": 650023704.0, + "step": 25116 + }, + { + "epoch": 2.7582912365473313, + "grad_norm": 2.587078809738159, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7285948991775513, + "num_tokens": 650046884.0, + "step": 25117 + }, + { + "epoch": 2.758401054249945, + "grad_norm": 2.383643388748169, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7490569353103638, + "num_tokens": 650071453.0, + "step": 25118 + }, + { + "epoch": 2.758510871952559, + "grad_norm": 2.307605028152466, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7488443851470947, + "num_tokens": 650096453.0, + "step": 25119 + }, + { + "epoch": 2.7586206896551726, + "grad_norm": 2.5002310276031494, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7276740074157715, + "num_tokens": 650120391.0, + "step": 25120 + }, + { + "epoch": 2.758730507357786, + "grad_norm": 2.4889636039733887, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7239028811454773, + "num_tokens": 650144385.0, + "step": 25121 + }, + { + "epoch": 2.7588403250603997, + "grad_norm": 2.450613498687744, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7119084596633911, + "num_tokens": 650169422.0, + "step": 25122 + }, + { + "epoch": 2.7589501427630134, + "grad_norm": 2.348534345626831, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7256971597671509, + "num_tokens": 650194483.0, + "step": 25123 + }, + { + "epoch": 2.759059960465627, + "grad_norm": 2.454322099685669, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7229977250099182, + "num_tokens": 650220021.0, + "step": 25124 + }, + { + "epoch": 2.759169778168241, + "grad_norm": 2.2680504322052, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7281848192214966, + "num_tokens": 650246245.0, + "step": 25125 + }, + { + "epoch": 2.7592795958708543, + "grad_norm": 2.370455503463745, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.719725489616394, + "num_tokens": 650272698.0, + "step": 25126 + }, + { + "epoch": 2.759389413573468, + "grad_norm": 2.3438477516174316, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7355316877365112, + "num_tokens": 650297928.0, + "step": 25127 + }, + { + "epoch": 2.7594992312760818, + "grad_norm": 2.4656307697296143, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7231991291046143, + "num_tokens": 650323221.0, + "step": 25128 + }, + { + "epoch": 2.7596090489786955, + "grad_norm": 2.101016044616699, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7427817583084106, + "num_tokens": 650354418.0, + "step": 25129 + }, + { + "epoch": 2.7597188666813093, + "grad_norm": 2.466978073120117, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7567448019981384, + "num_tokens": 650376365.0, + "step": 25130 + }, + { + "epoch": 2.7598286843839226, + "grad_norm": 2.433581590652466, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7154908776283264, + "num_tokens": 650400045.0, + "step": 25131 + }, + { + "epoch": 2.7599385020865363, + "grad_norm": 3.02335786819458, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7380260229110718, + "num_tokens": 650418615.0, + "step": 25132 + }, + { + "epoch": 2.76004831978915, + "grad_norm": 2.2413148880004883, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.753592848777771, + "num_tokens": 650443488.0, + "step": 25133 + }, + { + "epoch": 2.7601581374917634, + "grad_norm": 2.2476229667663574, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7239505052566528, + "num_tokens": 650473019.0, + "step": 25134 + }, + { + "epoch": 2.760267955194377, + "grad_norm": 2.2770214080810547, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7408163547515869, + "num_tokens": 650499245.0, + "step": 25135 + }, + { + "epoch": 2.760377772896991, + "grad_norm": 2.470726490020752, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7336430549621582, + "num_tokens": 650522784.0, + "step": 25136 + }, + { + "epoch": 2.7604875905996047, + "grad_norm": 2.5844576358795166, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7374358773231506, + "num_tokens": 650543810.0, + "step": 25137 + }, + { + "epoch": 2.7605974083022184, + "grad_norm": 2.4916062355041504, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.743425190448761, + "num_tokens": 650565417.0, + "step": 25138 + }, + { + "epoch": 2.7607072260048318, + "grad_norm": 2.1373674869537354, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7337751388549805, + "num_tokens": 650596977.0, + "step": 25139 + }, + { + "epoch": 2.7608170437074455, + "grad_norm": 2.448108196258545, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7222121357917786, + "num_tokens": 650621860.0, + "step": 25140 + }, + { + "epoch": 2.7609268614100593, + "grad_norm": 2.405298948287964, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7279503345489502, + "num_tokens": 650646498.0, + "step": 25141 + }, + { + "epoch": 2.761036679112673, + "grad_norm": 2.8353824615478516, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7537897229194641, + "num_tokens": 650664077.0, + "step": 25142 + }, + { + "epoch": 2.761146496815287, + "grad_norm": 2.7117156982421875, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7242006063461304, + "num_tokens": 650686151.0, + "step": 25143 + }, + { + "epoch": 2.7612563145179, + "grad_norm": 2.36879563331604, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7048429846763611, + "num_tokens": 650714203.0, + "step": 25144 + }, + { + "epoch": 2.761366132220514, + "grad_norm": 2.6874184608459473, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7503513693809509, + "num_tokens": 650734054.0, + "step": 25145 + }, + { + "epoch": 2.7614759499231276, + "grad_norm": 2.4179534912109375, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7121821641921997, + "num_tokens": 650761726.0, + "step": 25146 + }, + { + "epoch": 2.7615857676257414, + "grad_norm": 2.564116954803467, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7367877960205078, + "num_tokens": 650784206.0, + "step": 25147 + }, + { + "epoch": 2.761695585328355, + "grad_norm": 2.5405941009521484, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.721402645111084, + "num_tokens": 650807736.0, + "step": 25148 + }, + { + "epoch": 2.7618054030309684, + "grad_norm": 2.4435079097747803, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7394526600837708, + "num_tokens": 650830542.0, + "step": 25149 + }, + { + "epoch": 2.761915220733582, + "grad_norm": 2.2563846111297607, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7382901906967163, + "num_tokens": 650859543.0, + "step": 25150 + }, + { + "epoch": 2.762025038436196, + "grad_norm": 2.527423143386841, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7442396283149719, + "num_tokens": 650881841.0, + "step": 25151 + }, + { + "epoch": 2.7621348561388097, + "grad_norm": 2.414135217666626, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7430206537246704, + "num_tokens": 650906005.0, + "step": 25152 + }, + { + "epoch": 2.7622446738414235, + "grad_norm": 2.1661791801452637, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7340365052223206, + "num_tokens": 650933378.0, + "step": 25153 + }, + { + "epoch": 2.762354491544037, + "grad_norm": 2.4067413806915283, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.714768648147583, + "num_tokens": 650960849.0, + "step": 25154 + }, + { + "epoch": 2.7624643092466505, + "grad_norm": 2.004452705383301, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7277054190635681, + "num_tokens": 650990245.0, + "step": 25155 + }, + { + "epoch": 2.7625741269492643, + "grad_norm": 2.841745376586914, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7435207962989807, + "num_tokens": 651007513.0, + "step": 25156 + }, + { + "epoch": 2.7626839446518776, + "grad_norm": 2.4248199462890625, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7348060011863708, + "num_tokens": 651033184.0, + "step": 25157 + }, + { + "epoch": 2.762793762354492, + "grad_norm": 2.2338364124298096, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7302082180976868, + "num_tokens": 651061072.0, + "step": 25158 + }, + { + "epoch": 2.762903580057105, + "grad_norm": 2.4463677406311035, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7020987272262573, + "num_tokens": 651086402.0, + "step": 25159 + }, + { + "epoch": 2.763013397759719, + "grad_norm": 2.3876805305480957, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7279661893844604, + "num_tokens": 651110355.0, + "step": 25160 + }, + { + "epoch": 2.7631232154623326, + "grad_norm": 2.403362274169922, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7866351008415222, + "num_tokens": 651132937.0, + "step": 25161 + }, + { + "epoch": 2.763233033164946, + "grad_norm": 2.2924556732177734, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7312793731689453, + "num_tokens": 651159254.0, + "step": 25162 + }, + { + "epoch": 2.7633428508675597, + "grad_norm": 2.4163119792938232, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.71940016746521, + "num_tokens": 651183771.0, + "step": 25163 + }, + { + "epoch": 2.7634526685701735, + "grad_norm": 2.4214138984680176, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7324463129043579, + "num_tokens": 651210647.0, + "step": 25164 + }, + { + "epoch": 2.763562486272787, + "grad_norm": 2.5915775299072266, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7290977239608765, + "num_tokens": 651232450.0, + "step": 25165 + }, + { + "epoch": 2.763672303975401, + "grad_norm": 2.290024518966675, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7132316827774048, + "num_tokens": 651262363.0, + "step": 25166 + }, + { + "epoch": 2.7637821216780143, + "grad_norm": 2.192936897277832, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7254288196563721, + "num_tokens": 651294423.0, + "step": 25167 + }, + { + "epoch": 2.763891939380628, + "grad_norm": 2.4252898693084717, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.722406268119812, + "num_tokens": 651320244.0, + "step": 25168 + }, + { + "epoch": 2.764001757083242, + "grad_norm": 2.4373855590820312, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7300191521644592, + "num_tokens": 651345228.0, + "step": 25169 + }, + { + "epoch": 2.7641115747858556, + "grad_norm": 2.39127516746521, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7406086921691895, + "num_tokens": 651370252.0, + "step": 25170 + }, + { + "epoch": 2.7642213924884693, + "grad_norm": 2.125122547149658, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7385132312774658, + "num_tokens": 651399536.0, + "step": 25171 + }, + { + "epoch": 2.7643312101910826, + "grad_norm": 2.3148295879364014, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7111462354660034, + "num_tokens": 651427551.0, + "step": 25172 + }, + { + "epoch": 2.7644410278936964, + "grad_norm": 2.244025468826294, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7236698865890503, + "num_tokens": 651456709.0, + "step": 25173 + }, + { + "epoch": 2.76455084559631, + "grad_norm": 2.323864459991455, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7298274040222168, + "num_tokens": 651483211.0, + "step": 25174 + }, + { + "epoch": 2.764660663298924, + "grad_norm": 2.786454200744629, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7456361055374146, + "num_tokens": 651503020.0, + "step": 25175 + }, + { + "epoch": 2.7647704810015377, + "grad_norm": 2.371631145477295, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7298915386199951, + "num_tokens": 651527231.0, + "step": 25176 + }, + { + "epoch": 2.764880298704151, + "grad_norm": 2.293851375579834, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7461341619491577, + "num_tokens": 651554015.0, + "step": 25177 + }, + { + "epoch": 2.7649901164067647, + "grad_norm": 2.361233711242676, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7322307229042053, + "num_tokens": 651580250.0, + "step": 25178 + }, + { + "epoch": 2.7650999341093785, + "grad_norm": 2.2327516078948975, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7361838817596436, + "num_tokens": 651606831.0, + "step": 25179 + }, + { + "epoch": 2.7652097518119922, + "grad_norm": 2.4287328720092773, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7338212728500366, + "num_tokens": 651629974.0, + "step": 25180 + }, + { + "epoch": 2.765319569514606, + "grad_norm": 2.2376596927642822, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7336034178733826, + "num_tokens": 651657057.0, + "step": 25181 + }, + { + "epoch": 2.7654293872172193, + "grad_norm": 2.364144802093506, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7459985017776489, + "num_tokens": 651682385.0, + "step": 25182 + }, + { + "epoch": 2.765539204919833, + "grad_norm": 2.391026258468628, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7264752388000488, + "num_tokens": 651706546.0, + "step": 25183 + }, + { + "epoch": 2.765649022622447, + "grad_norm": 2.0951993465423584, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7465972304344177, + "num_tokens": 651734169.0, + "step": 25184 + }, + { + "epoch": 2.76575884032506, + "grad_norm": 2.842658281326294, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.733336329460144, + "num_tokens": 651760196.0, + "step": 25185 + }, + { + "epoch": 2.765868658027674, + "grad_norm": 2.27774715423584, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7190223336219788, + "num_tokens": 651791156.0, + "step": 25186 + }, + { + "epoch": 2.7659784757302877, + "grad_norm": 2.7778160572052, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7499706745147705, + "num_tokens": 651809139.0, + "step": 25187 + }, + { + "epoch": 2.7660882934329014, + "grad_norm": 2.6246070861816406, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7434587478637695, + "num_tokens": 651829919.0, + "step": 25188 + }, + { + "epoch": 2.766198111135515, + "grad_norm": 2.2470614910125732, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.713708221912384, + "num_tokens": 651859312.0, + "step": 25189 + }, + { + "epoch": 2.7663079288381285, + "grad_norm": 2.1549692153930664, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7280446887016296, + "num_tokens": 651889055.0, + "step": 25190 + }, + { + "epoch": 2.7664177465407422, + "grad_norm": 2.7868926525115967, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.766815185546875, + "num_tokens": 651906880.0, + "step": 25191 + }, + { + "epoch": 2.766527564243356, + "grad_norm": 2.482125759124756, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7434506416320801, + "num_tokens": 651929096.0, + "step": 25192 + }, + { + "epoch": 2.7666373819459698, + "grad_norm": 2.676745891571045, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7159041166305542, + "num_tokens": 651951758.0, + "step": 25193 + }, + { + "epoch": 2.7667471996485835, + "grad_norm": 2.317526340484619, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7268158197402954, + "num_tokens": 651976696.0, + "step": 25194 + }, + { + "epoch": 2.766857017351197, + "grad_norm": 2.2216410636901855, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7440614700317383, + "num_tokens": 652004851.0, + "step": 25195 + }, + { + "epoch": 2.7669668350538106, + "grad_norm": 2.5840418338775635, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7295459508895874, + "num_tokens": 652027235.0, + "step": 25196 + }, + { + "epoch": 2.7670766527564243, + "grad_norm": 2.24434232711792, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.724746584892273, + "num_tokens": 652053645.0, + "step": 25197 + }, + { + "epoch": 2.767186470459038, + "grad_norm": 2.2760894298553467, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7215006351470947, + "num_tokens": 652080875.0, + "step": 25198 + }, + { + "epoch": 2.767296288161652, + "grad_norm": 2.3894245624542236, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.719863772392273, + "num_tokens": 652105946.0, + "step": 25199 + }, + { + "epoch": 2.767406105864265, + "grad_norm": 2.386565923690796, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7316669225692749, + "num_tokens": 652132432.0, + "step": 25200 + }, + { + "epoch": 2.767515923566879, + "grad_norm": 2.0205111503601074, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7033336162567139, + "num_tokens": 652168571.0, + "step": 25201 + }, + { + "epoch": 2.7676257412694927, + "grad_norm": 2.505464553833008, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7346623539924622, + "num_tokens": 652192561.0, + "step": 25202 + }, + { + "epoch": 2.7677355589721064, + "grad_norm": 2.49387264251709, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7304707765579224, + "num_tokens": 652216199.0, + "step": 25203 + }, + { + "epoch": 2.76784537667472, + "grad_norm": 2.506801128387451, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7181283235549927, + "num_tokens": 652240440.0, + "step": 25204 + }, + { + "epoch": 2.7679551943773335, + "grad_norm": 2.5308525562286377, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7246596813201904, + "num_tokens": 652264582.0, + "step": 25205 + }, + { + "epoch": 2.7680650120799473, + "grad_norm": 2.267113447189331, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7523452043533325, + "num_tokens": 652291585.0, + "step": 25206 + }, + { + "epoch": 2.768174829782561, + "grad_norm": 2.1890311241149902, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7091253995895386, + "num_tokens": 652323196.0, + "step": 25207 + }, + { + "epoch": 2.7682846474851748, + "grad_norm": 2.1794981956481934, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7035581469535828, + "num_tokens": 652355899.0, + "step": 25208 + }, + { + "epoch": 2.7683944651877885, + "grad_norm": 2.391085624694824, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7053435444831848, + "num_tokens": 652381810.0, + "step": 25209 + }, + { + "epoch": 2.768504282890402, + "grad_norm": 2.634993314743042, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7499948740005493, + "num_tokens": 652403246.0, + "step": 25210 + }, + { + "epoch": 2.7686141005930156, + "grad_norm": 2.441519021987915, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7156552076339722, + "num_tokens": 652427886.0, + "step": 25211 + }, + { + "epoch": 2.7687239182956294, + "grad_norm": 2.290491819381714, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7365450263023376, + "num_tokens": 652454926.0, + "step": 25212 + }, + { + "epoch": 2.7688337359982427, + "grad_norm": 2.330562114715576, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7455071210861206, + "num_tokens": 652481219.0, + "step": 25213 + }, + { + "epoch": 2.7689435537008564, + "grad_norm": 2.2055702209472656, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7178436517715454, + "num_tokens": 652507642.0, + "step": 25214 + }, + { + "epoch": 2.76905337140347, + "grad_norm": 2.526543378829956, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7557637691497803, + "num_tokens": 652530668.0, + "step": 25215 + }, + { + "epoch": 2.769163189106084, + "grad_norm": 2.5813217163085938, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7386506199836731, + "num_tokens": 652552515.0, + "step": 25216 + }, + { + "epoch": 2.7692730068086977, + "grad_norm": 2.632347345352173, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7519650459289551, + "num_tokens": 652572944.0, + "step": 25217 + }, + { + "epoch": 2.769382824511311, + "grad_norm": 2.1582529544830322, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7236100435256958, + "num_tokens": 652601639.0, + "step": 25218 + }, + { + "epoch": 2.7694926422139248, + "grad_norm": 2.385552406311035, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7168034911155701, + "num_tokens": 652626592.0, + "step": 25219 + }, + { + "epoch": 2.7696024599165385, + "grad_norm": 2.244636297225952, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7033271789550781, + "num_tokens": 652655322.0, + "step": 25220 + }, + { + "epoch": 2.7697122776191523, + "grad_norm": 2.5497019290924072, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7487196922302246, + "num_tokens": 652679234.0, + "step": 25221 + }, + { + "epoch": 2.769822095321766, + "grad_norm": 2.7534420490264893, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7368872761726379, + "num_tokens": 652699212.0, + "step": 25222 + }, + { + "epoch": 2.7699319130243794, + "grad_norm": 2.2867326736450195, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7384203672409058, + "num_tokens": 652726972.0, + "step": 25223 + }, + { + "epoch": 2.770041730726993, + "grad_norm": 2.5004589557647705, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7518038153648376, + "num_tokens": 652748924.0, + "step": 25224 + }, + { + "epoch": 2.770151548429607, + "grad_norm": 2.5541844367980957, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7273188829421997, + "num_tokens": 652772025.0, + "step": 25225 + }, + { + "epoch": 2.7702613661322206, + "grad_norm": 2.3358042240142822, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7491083145141602, + "num_tokens": 652798692.0, + "step": 25226 + }, + { + "epoch": 2.7703711838348344, + "grad_norm": 2.1307051181793213, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7265685796737671, + "num_tokens": 652830187.0, + "step": 25227 + }, + { + "epoch": 2.7704810015374477, + "grad_norm": 2.4719996452331543, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7400397658348083, + "num_tokens": 652853772.0, + "step": 25228 + }, + { + "epoch": 2.7705908192400615, + "grad_norm": 2.477102041244507, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7284771203994751, + "num_tokens": 652879406.0, + "step": 25229 + }, + { + "epoch": 2.770700636942675, + "grad_norm": 2.367506504058838, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7482413053512573, + "num_tokens": 652904322.0, + "step": 25230 + }, + { + "epoch": 2.770810454645289, + "grad_norm": 2.2046620845794678, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7341001033782959, + "num_tokens": 652932569.0, + "step": 25231 + }, + { + "epoch": 2.7709202723479027, + "grad_norm": 2.3417322635650635, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7274542450904846, + "num_tokens": 652957935.0, + "step": 25232 + }, + { + "epoch": 2.771030090050516, + "grad_norm": 2.314439535140991, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7230842113494873, + "num_tokens": 652983410.0, + "step": 25233 + }, + { + "epoch": 2.77113990775313, + "grad_norm": 2.283546209335327, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7217962741851807, + "num_tokens": 653009975.0, + "step": 25234 + }, + { + "epoch": 2.7712497254557436, + "grad_norm": 2.189856767654419, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7455177903175354, + "num_tokens": 653036431.0, + "step": 25235 + }, + { + "epoch": 2.771359543158357, + "grad_norm": 2.5388975143432617, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7459473013877869, + "num_tokens": 653058546.0, + "step": 25236 + }, + { + "epoch": 2.771469360860971, + "grad_norm": 2.3109002113342285, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7594446539878845, + "num_tokens": 653083335.0, + "step": 25237 + }, + { + "epoch": 2.7715791785635844, + "grad_norm": 2.234193801879883, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7432755827903748, + "num_tokens": 653110027.0, + "step": 25238 + }, + { + "epoch": 2.771688996266198, + "grad_norm": 2.333345890045166, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7344695329666138, + "num_tokens": 653136265.0, + "step": 25239 + }, + { + "epoch": 2.771798813968812, + "grad_norm": 2.4815051555633545, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7275806069374084, + "num_tokens": 653159842.0, + "step": 25240 + }, + { + "epoch": 2.771908631671425, + "grad_norm": 2.418010950088501, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7304900884628296, + "num_tokens": 653186815.0, + "step": 25241 + }, + { + "epoch": 2.772018449374039, + "grad_norm": 2.4507083892822266, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7195810079574585, + "num_tokens": 653211266.0, + "step": 25242 + }, + { + "epoch": 2.7721282670766527, + "grad_norm": 2.2693533897399902, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7245174646377563, + "num_tokens": 653240510.0, + "step": 25243 + }, + { + "epoch": 2.7722380847792665, + "grad_norm": 2.6541340351104736, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.717244029045105, + "num_tokens": 653261176.0, + "step": 25244 + }, + { + "epoch": 2.7723479024818802, + "grad_norm": 2.2013437747955322, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7073605060577393, + "num_tokens": 653290730.0, + "step": 25245 + }, + { + "epoch": 2.7724577201844935, + "grad_norm": 2.3971974849700928, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7182713150978088, + "num_tokens": 653316082.0, + "step": 25246 + }, + { + "epoch": 2.7725675378871073, + "grad_norm": 2.3422207832336426, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7450293898582458, + "num_tokens": 653341875.0, + "step": 25247 + }, + { + "epoch": 2.772677355589721, + "grad_norm": 2.44097900390625, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7541904449462891, + "num_tokens": 653366545.0, + "step": 25248 + }, + { + "epoch": 2.772787173292335, + "grad_norm": 2.4858968257904053, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7252858281135559, + "num_tokens": 653391297.0, + "step": 25249 + }, + { + "epoch": 2.7728969909949486, + "grad_norm": 2.440760850906372, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7387781739234924, + "num_tokens": 653414622.0, + "step": 25250 + }, + { + "epoch": 2.773006808697562, + "grad_norm": 2.4281229972839355, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.751797616481781, + "num_tokens": 653438389.0, + "step": 25251 + }, + { + "epoch": 2.7731166264001756, + "grad_norm": 2.440171241760254, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7255958914756775, + "num_tokens": 653462007.0, + "step": 25252 + }, + { + "epoch": 2.7732264441027894, + "grad_norm": 2.130753993988037, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7230736017227173, + "num_tokens": 653491402.0, + "step": 25253 + }, + { + "epoch": 2.773336261805403, + "grad_norm": 2.24265456199646, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7119447588920593, + "num_tokens": 653521194.0, + "step": 25254 + }, + { + "epoch": 2.773446079508017, + "grad_norm": 2.4009673595428467, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7199629545211792, + "num_tokens": 653546587.0, + "step": 25255 + }, + { + "epoch": 2.7735558972106302, + "grad_norm": 2.3810737133026123, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7245222330093384, + "num_tokens": 653572844.0, + "step": 25256 + }, + { + "epoch": 2.773665714913244, + "grad_norm": 2.0976386070251465, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7178980112075806, + "num_tokens": 653604279.0, + "step": 25257 + }, + { + "epoch": 2.7737755326158577, + "grad_norm": 2.249952793121338, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7199815511703491, + "num_tokens": 653630499.0, + "step": 25258 + }, + { + "epoch": 2.7738853503184715, + "grad_norm": 2.466381549835205, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7331649661064148, + "num_tokens": 653652860.0, + "step": 25259 + }, + { + "epoch": 2.7739951680210853, + "grad_norm": 2.454063892364502, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7240274548530579, + "num_tokens": 653678067.0, + "step": 25260 + }, + { + "epoch": 2.7741049857236986, + "grad_norm": 2.244596004486084, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7278295755386353, + "num_tokens": 653705602.0, + "step": 25261 + }, + { + "epoch": 2.7742148034263123, + "grad_norm": 2.5468344688415527, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7535508871078491, + "num_tokens": 653728381.0, + "step": 25262 + }, + { + "epoch": 2.774324621128926, + "grad_norm": 2.498741626739502, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.727245569229126, + "num_tokens": 653752289.0, + "step": 25263 + }, + { + "epoch": 2.7744344388315394, + "grad_norm": 2.508122205734253, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7425249814987183, + "num_tokens": 653776572.0, + "step": 25264 + }, + { + "epoch": 2.774544256534153, + "grad_norm": 2.4402658939361572, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7404955625534058, + "num_tokens": 653801743.0, + "step": 25265 + }, + { + "epoch": 2.774654074236767, + "grad_norm": 2.3995285034179688, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7179350852966309, + "num_tokens": 653828938.0, + "step": 25266 + }, + { + "epoch": 2.7747638919393807, + "grad_norm": 2.2846243381500244, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7303495407104492, + "num_tokens": 653855618.0, + "step": 25267 + }, + { + "epoch": 2.7748737096419944, + "grad_norm": 2.551379680633545, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.746284544467926, + "num_tokens": 653878047.0, + "step": 25268 + }, + { + "epoch": 2.7749835273446077, + "grad_norm": 2.068012237548828, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7050470113754272, + "num_tokens": 653910661.0, + "step": 25269 + }, + { + "epoch": 2.7750933450472215, + "grad_norm": 2.6253340244293213, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7449843883514404, + "num_tokens": 653931833.0, + "step": 25270 + }, + { + "epoch": 2.7752031627498353, + "grad_norm": 2.3783295154571533, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7236259579658508, + "num_tokens": 653956526.0, + "step": 25271 + }, + { + "epoch": 2.775312980452449, + "grad_norm": 2.377990961074829, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7244412899017334, + "num_tokens": 653981615.0, + "step": 25272 + }, + { + "epoch": 2.7754227981550628, + "grad_norm": 2.2209126949310303, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7386691570281982, + "num_tokens": 654009960.0, + "step": 25273 + }, + { + "epoch": 2.775532615857676, + "grad_norm": 2.5632781982421875, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7320452928543091, + "num_tokens": 654032444.0, + "step": 25274 + }, + { + "epoch": 2.77564243356029, + "grad_norm": 2.8974714279174805, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7322939038276672, + "num_tokens": 654051166.0, + "step": 25275 + }, + { + "epoch": 2.7757522512629036, + "grad_norm": 2.4468064308166504, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7144263982772827, + "num_tokens": 654076902.0, + "step": 25276 + }, + { + "epoch": 2.7758620689655173, + "grad_norm": 2.6366636753082275, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7579107284545898, + "num_tokens": 654096938.0, + "step": 25277 + }, + { + "epoch": 2.775971886668131, + "grad_norm": 2.289259910583496, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7344439029693604, + "num_tokens": 654122767.0, + "step": 25278 + }, + { + "epoch": 2.7760817043707444, + "grad_norm": 2.5453343391418457, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7291807532310486, + "num_tokens": 654145677.0, + "step": 25279 + }, + { + "epoch": 2.776191522073358, + "grad_norm": 2.365541696548462, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7346206903457642, + "num_tokens": 654169535.0, + "step": 25280 + }, + { + "epoch": 2.776301339775972, + "grad_norm": 2.3939754962921143, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7615472078323364, + "num_tokens": 654192822.0, + "step": 25281 + }, + { + "epoch": 2.7764111574785857, + "grad_norm": 2.0082147121429443, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7230773568153381, + "num_tokens": 654225419.0, + "step": 25282 + }, + { + "epoch": 2.7765209751811994, + "grad_norm": 2.3391826152801514, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7334256768226624, + "num_tokens": 654249585.0, + "step": 25283 + }, + { + "epoch": 2.7766307928838128, + "grad_norm": 2.4647884368896484, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7312573194503784, + "num_tokens": 654273537.0, + "step": 25284 + }, + { + "epoch": 2.7767406105864265, + "grad_norm": 2.189997673034668, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7227479219436646, + "num_tokens": 654303510.0, + "step": 25285 + }, + { + "epoch": 2.7768504282890403, + "grad_norm": 2.0873665809631348, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7130135297775269, + "num_tokens": 654333327.0, + "step": 25286 + }, + { + "epoch": 2.7769602459916536, + "grad_norm": 2.683638334274292, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7352750301361084, + "num_tokens": 654353996.0, + "step": 25287 + }, + { + "epoch": 2.777070063694268, + "grad_norm": 2.725646495819092, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7495752573013306, + "num_tokens": 654374873.0, + "step": 25288 + }, + { + "epoch": 2.777179881396881, + "grad_norm": 2.526484489440918, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7444278597831726, + "num_tokens": 654397303.0, + "step": 25289 + }, + { + "epoch": 2.777289699099495, + "grad_norm": 2.4817097187042236, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7171362042427063, + "num_tokens": 654421715.0, + "step": 25290 + }, + { + "epoch": 2.7773995168021086, + "grad_norm": 2.3595902919769287, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7236889004707336, + "num_tokens": 654446136.0, + "step": 25291 + }, + { + "epoch": 2.777509334504722, + "grad_norm": 2.416698455810547, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.760277271270752, + "num_tokens": 654469152.0, + "step": 25292 + }, + { + "epoch": 2.7776191522073357, + "grad_norm": 2.2472901344299316, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7644180059432983, + "num_tokens": 654495377.0, + "step": 25293 + }, + { + "epoch": 2.7777289699099494, + "grad_norm": 2.611774444580078, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7303466796875, + "num_tokens": 654516281.0, + "step": 25294 + }, + { + "epoch": 2.777838787612563, + "grad_norm": 2.459476947784424, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7376195788383484, + "num_tokens": 654540311.0, + "step": 25295 + }, + { + "epoch": 2.777948605315177, + "grad_norm": 2.5871517658233643, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.731613039970398, + "num_tokens": 654563102.0, + "step": 25296 + }, + { + "epoch": 2.7780584230177903, + "grad_norm": 2.234151840209961, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7179872989654541, + "num_tokens": 654592510.0, + "step": 25297 + }, + { + "epoch": 2.778168240720404, + "grad_norm": 2.442129373550415, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7409918904304504, + "num_tokens": 654615826.0, + "step": 25298 + }, + { + "epoch": 2.778278058423018, + "grad_norm": 2.5155029296875, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7233117818832397, + "num_tokens": 654641014.0, + "step": 25299 + }, + { + "epoch": 2.7783878761256315, + "grad_norm": 2.2457079887390137, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7434150576591492, + "num_tokens": 654669484.0, + "step": 25300 + }, + { + "epoch": 2.7784976938282453, + "grad_norm": 2.5130224227905273, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7302103042602539, + "num_tokens": 654691943.0, + "step": 25301 + }, + { + "epoch": 2.7786075115308586, + "grad_norm": 2.0959036350250244, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7399902939796448, + "num_tokens": 654724098.0, + "step": 25302 + }, + { + "epoch": 2.7787173292334724, + "grad_norm": 2.318640947341919, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7260921001434326, + "num_tokens": 654751649.0, + "step": 25303 + }, + { + "epoch": 2.778827146936086, + "grad_norm": 2.6869592666625977, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7425272464752197, + "num_tokens": 654774086.0, + "step": 25304 + }, + { + "epoch": 2.7789369646387, + "grad_norm": 2.289982318878174, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7114111185073853, + "num_tokens": 654800126.0, + "step": 25305 + }, + { + "epoch": 2.7790467823413136, + "grad_norm": 2.252004384994507, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7196437120437622, + "num_tokens": 654828206.0, + "step": 25306 + }, + { + "epoch": 2.779156600043927, + "grad_norm": 2.6701316833496094, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.774074912071228, + "num_tokens": 654847164.0, + "step": 25307 + }, + { + "epoch": 2.7792664177465407, + "grad_norm": 2.582059621810913, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7404615879058838, + "num_tokens": 654869336.0, + "step": 25308 + }, + { + "epoch": 2.7793762354491545, + "grad_norm": 2.353247880935669, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7415418028831482, + "num_tokens": 654897779.0, + "step": 25309 + }, + { + "epoch": 2.779486053151768, + "grad_norm": 2.7443294525146484, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7350850701332092, + "num_tokens": 654917220.0, + "step": 25310 + }, + { + "epoch": 2.779595870854382, + "grad_norm": 2.653696298599243, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7281978726387024, + "num_tokens": 654939692.0, + "step": 25311 + }, + { + "epoch": 2.7797056885569953, + "grad_norm": 2.346813201904297, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7049345970153809, + "num_tokens": 654965842.0, + "step": 25312 + }, + { + "epoch": 2.779815506259609, + "grad_norm": 2.290191888809204, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7178528308868408, + "num_tokens": 654993830.0, + "step": 25313 + }, + { + "epoch": 2.779925323962223, + "grad_norm": 2.487321615219116, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7369112968444824, + "num_tokens": 655015958.0, + "step": 25314 + }, + { + "epoch": 2.780035141664836, + "grad_norm": 2.211980104446411, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.749314546585083, + "num_tokens": 655044444.0, + "step": 25315 + }, + { + "epoch": 2.78014495936745, + "grad_norm": 2.2568671703338623, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7406207323074341, + "num_tokens": 655071755.0, + "step": 25316 + }, + { + "epoch": 2.7802547770700636, + "grad_norm": 2.2958829402923584, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7029387950897217, + "num_tokens": 655099814.0, + "step": 25317 + }, + { + "epoch": 2.7803645947726774, + "grad_norm": 2.4280691146850586, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7444039583206177, + "num_tokens": 655129052.0, + "step": 25318 + }, + { + "epoch": 2.780474412475291, + "grad_norm": 2.367297649383545, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7477184534072876, + "num_tokens": 655153983.0, + "step": 25319 + }, + { + "epoch": 2.7805842301779045, + "grad_norm": 2.359661340713501, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7245982885360718, + "num_tokens": 655180614.0, + "step": 25320 + }, + { + "epoch": 2.780694047880518, + "grad_norm": 2.2370238304138184, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7192991971969604, + "num_tokens": 655209629.0, + "step": 25321 + }, + { + "epoch": 2.780803865583132, + "grad_norm": 2.4121646881103516, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7428832054138184, + "num_tokens": 655234356.0, + "step": 25322 + }, + { + "epoch": 2.7809136832857457, + "grad_norm": 2.2360520362854004, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7296780347824097, + "num_tokens": 655263774.0, + "step": 25323 + }, + { + "epoch": 2.7810235009883595, + "grad_norm": 2.3132123947143555, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7375766634941101, + "num_tokens": 655290867.0, + "step": 25324 + }, + { + "epoch": 2.781133318690973, + "grad_norm": 2.193533182144165, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7192821502685547, + "num_tokens": 655319273.0, + "step": 25325 + }, + { + "epoch": 2.7812431363935866, + "grad_norm": 2.30926513671875, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7476011514663696, + "num_tokens": 655344099.0, + "step": 25326 + }, + { + "epoch": 2.7813529540962003, + "grad_norm": 2.0997064113616943, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7068634629249573, + "num_tokens": 655376303.0, + "step": 25327 + }, + { + "epoch": 2.781462771798814, + "grad_norm": 2.3493728637695312, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7380239963531494, + "num_tokens": 655404148.0, + "step": 25328 + }, + { + "epoch": 2.781572589501428, + "grad_norm": 2.641500949859619, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7235643863677979, + "num_tokens": 655425556.0, + "step": 25329 + }, + { + "epoch": 2.781682407204041, + "grad_norm": 2.4666318893432617, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7550884485244751, + "num_tokens": 655448683.0, + "step": 25330 + }, + { + "epoch": 2.781792224906655, + "grad_norm": 2.2322912216186523, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7365018725395203, + "num_tokens": 655475223.0, + "step": 25331 + }, + { + "epoch": 2.7819020426092687, + "grad_norm": 2.175265312194824, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7397814989089966, + "num_tokens": 655503829.0, + "step": 25332 + }, + { + "epoch": 2.7820118603118824, + "grad_norm": 2.2831618785858154, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7319673299789429, + "num_tokens": 655530470.0, + "step": 25333 + }, + { + "epoch": 2.782121678014496, + "grad_norm": 2.6156904697418213, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7387219667434692, + "num_tokens": 655553107.0, + "step": 25334 + }, + { + "epoch": 2.7822314957171095, + "grad_norm": 2.3648269176483154, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7343701124191284, + "num_tokens": 655579577.0, + "step": 25335 + }, + { + "epoch": 2.7823413134197232, + "grad_norm": 2.4369893074035645, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7444160580635071, + "num_tokens": 655604063.0, + "step": 25336 + }, + { + "epoch": 2.782451131122337, + "grad_norm": 2.404797315597534, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7464630603790283, + "num_tokens": 655629811.0, + "step": 25337 + }, + { + "epoch": 2.7825609488249503, + "grad_norm": 2.3711514472961426, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7128006815910339, + "num_tokens": 655658096.0, + "step": 25338 + }, + { + "epoch": 2.7826707665275645, + "grad_norm": 2.3473927974700928, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7210593819618225, + "num_tokens": 655686339.0, + "step": 25339 + }, + { + "epoch": 2.782780584230178, + "grad_norm": 2.5720906257629395, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7192696332931519, + "num_tokens": 655710433.0, + "step": 25340 + }, + { + "epoch": 2.7828904019327916, + "grad_norm": 2.2513082027435303, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7286717891693115, + "num_tokens": 655738374.0, + "step": 25341 + }, + { + "epoch": 2.7830002196354053, + "grad_norm": 2.4185614585876465, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7312403917312622, + "num_tokens": 655763612.0, + "step": 25342 + }, + { + "epoch": 2.7831100373380186, + "grad_norm": 2.2049400806427, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7184392213821411, + "num_tokens": 655794128.0, + "step": 25343 + }, + { + "epoch": 2.7832198550406324, + "grad_norm": 2.5541164875030518, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7378353476524353, + "num_tokens": 655817587.0, + "step": 25344 + }, + { + "epoch": 2.783329672743246, + "grad_norm": 2.6140036582946777, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7339416742324829, + "num_tokens": 655840349.0, + "step": 25345 + }, + { + "epoch": 2.78343949044586, + "grad_norm": 2.5129683017730713, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7398369312286377, + "num_tokens": 655863924.0, + "step": 25346 + }, + { + "epoch": 2.7835493081484737, + "grad_norm": 2.6045148372650146, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7345640659332275, + "num_tokens": 655886611.0, + "step": 25347 + }, + { + "epoch": 2.783659125851087, + "grad_norm": 2.2639636993408203, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7333314418792725, + "num_tokens": 655914869.0, + "step": 25348 + }, + { + "epoch": 2.7837689435537007, + "grad_norm": 2.3278257846832275, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7423253059387207, + "num_tokens": 655941675.0, + "step": 25349 + }, + { + "epoch": 2.7838787612563145, + "grad_norm": 2.147221088409424, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.708669900894165, + "num_tokens": 655971798.0, + "step": 25350 + }, + { + "epoch": 2.7839885789589283, + "grad_norm": 2.3424172401428223, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7519689798355103, + "num_tokens": 655995942.0, + "step": 25351 + }, + { + "epoch": 2.784098396661542, + "grad_norm": 2.024977207183838, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.712443470954895, + "num_tokens": 656030686.0, + "step": 25352 + }, + { + "epoch": 2.7842082143641553, + "grad_norm": 2.262699604034424, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7288110256195068, + "num_tokens": 656060557.0, + "step": 25353 + }, + { + "epoch": 2.784318032066769, + "grad_norm": 2.2784616947174072, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.748988687992096, + "num_tokens": 656086004.0, + "step": 25354 + }, + { + "epoch": 2.784427849769383, + "grad_norm": 2.395599126815796, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7230717539787292, + "num_tokens": 656111463.0, + "step": 25355 + }, + { + "epoch": 2.7845376674719966, + "grad_norm": 2.512430191040039, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7320252656936646, + "num_tokens": 656136462.0, + "step": 25356 + }, + { + "epoch": 2.7846474851746104, + "grad_norm": 2.4432573318481445, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7336517572402954, + "num_tokens": 656160109.0, + "step": 25357 + }, + { + "epoch": 2.7847573028772237, + "grad_norm": 2.3446712493896484, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7263761758804321, + "num_tokens": 656184727.0, + "step": 25358 + }, + { + "epoch": 2.7848671205798374, + "grad_norm": 2.828428268432617, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7383676767349243, + "num_tokens": 656204424.0, + "step": 25359 + }, + { + "epoch": 2.784976938282451, + "grad_norm": 2.29382586479187, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7236040830612183, + "num_tokens": 656231785.0, + "step": 25360 + }, + { + "epoch": 2.785086755985065, + "grad_norm": 2.519564390182495, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7296997308731079, + "num_tokens": 656255960.0, + "step": 25361 + }, + { + "epoch": 2.7851965736876787, + "grad_norm": 1.9677324295043945, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7039827704429626, + "num_tokens": 656289877.0, + "step": 25362 + }, + { + "epoch": 2.785306391390292, + "grad_norm": 2.598111629486084, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7200995087623596, + "num_tokens": 656313410.0, + "step": 25363 + }, + { + "epoch": 2.7854162090929058, + "grad_norm": 2.424517869949341, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7132949233055115, + "num_tokens": 656338139.0, + "step": 25364 + }, + { + "epoch": 2.7855260267955195, + "grad_norm": 2.4129655361175537, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7200032472610474, + "num_tokens": 656365724.0, + "step": 25365 + }, + { + "epoch": 2.785635844498133, + "grad_norm": 2.248138904571533, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7315258979797363, + "num_tokens": 656397010.0, + "step": 25366 + }, + { + "epoch": 2.7857456622007466, + "grad_norm": 2.2271406650543213, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7009552121162415, + "num_tokens": 656426045.0, + "step": 25367 + }, + { + "epoch": 2.7858554799033604, + "grad_norm": 2.752955675125122, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7503451108932495, + "num_tokens": 656445907.0, + "step": 25368 + }, + { + "epoch": 2.785965297605974, + "grad_norm": 2.3678390979766846, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7218148708343506, + "num_tokens": 656471183.0, + "step": 25369 + }, + { + "epoch": 2.786075115308588, + "grad_norm": 2.23884916305542, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.728685736656189, + "num_tokens": 656499016.0, + "step": 25370 + }, + { + "epoch": 2.786184933011201, + "grad_norm": 2.525200605392456, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.728603720664978, + "num_tokens": 656523225.0, + "step": 25371 + }, + { + "epoch": 2.786294750713815, + "grad_norm": 2.4253292083740234, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.7094154953956604, + "num_tokens": 656548239.0, + "step": 25372 + }, + { + "epoch": 2.7864045684164287, + "grad_norm": 2.250598192214966, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7364732623100281, + "num_tokens": 656574403.0, + "step": 25373 + }, + { + "epoch": 2.7865143861190425, + "grad_norm": 2.5006604194641113, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7452189922332764, + "num_tokens": 656597325.0, + "step": 25374 + }, + { + "epoch": 2.786624203821656, + "grad_norm": 2.207716941833496, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7293426394462585, + "num_tokens": 656624767.0, + "step": 25375 + }, + { + "epoch": 2.7867340215242695, + "grad_norm": 2.4474270343780518, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7318745851516724, + "num_tokens": 656650865.0, + "step": 25376 + }, + { + "epoch": 2.7868438392268833, + "grad_norm": 2.3112635612487793, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7256482839584351, + "num_tokens": 656679434.0, + "step": 25377 + }, + { + "epoch": 2.786953656929497, + "grad_norm": 2.124185085296631, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7391852140426636, + "num_tokens": 656707461.0, + "step": 25378 + }, + { + "epoch": 2.787063474632111, + "grad_norm": 2.463909149169922, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7367351055145264, + "num_tokens": 656731067.0, + "step": 25379 + }, + { + "epoch": 2.7871732923347246, + "grad_norm": 2.4379940032958984, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7090607285499573, + "num_tokens": 656756016.0, + "step": 25380 + }, + { + "epoch": 2.787283110037338, + "grad_norm": 2.6019811630249023, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7485610246658325, + "num_tokens": 656776780.0, + "step": 25381 + }, + { + "epoch": 2.7873929277399516, + "grad_norm": 2.3832755088806152, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7245760560035706, + "num_tokens": 656802819.0, + "step": 25382 + }, + { + "epoch": 2.7875027454425654, + "grad_norm": 2.5738000869750977, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7329021692276001, + "num_tokens": 656825854.0, + "step": 25383 + }, + { + "epoch": 2.787612563145179, + "grad_norm": 2.7785935401916504, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.745746374130249, + "num_tokens": 656846330.0, + "step": 25384 + }, + { + "epoch": 2.787722380847793, + "grad_norm": 2.5691874027252197, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7443281412124634, + "num_tokens": 656870508.0, + "step": 25385 + }, + { + "epoch": 2.787832198550406, + "grad_norm": 2.4676108360290527, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7494781017303467, + "num_tokens": 656894746.0, + "step": 25386 + }, + { + "epoch": 2.78794201625302, + "grad_norm": 2.1723597049713135, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7413150072097778, + "num_tokens": 656924634.0, + "step": 25387 + }, + { + "epoch": 2.7880518339556337, + "grad_norm": 2.0630481243133545, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7417608499526978, + "num_tokens": 656954425.0, + "step": 25388 + }, + { + "epoch": 2.7881616516582475, + "grad_norm": 2.339531898498535, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7241603136062622, + "num_tokens": 656981932.0, + "step": 25389 + }, + { + "epoch": 2.7882714693608612, + "grad_norm": 2.2973899841308594, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6960715055465698, + "num_tokens": 657014208.0, + "step": 25390 + }, + { + "epoch": 2.7883812870634745, + "grad_norm": 2.0501670837402344, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7261431813240051, + "num_tokens": 657043290.0, + "step": 25391 + }, + { + "epoch": 2.7884911047660883, + "grad_norm": 2.6523821353912354, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7438316345214844, + "num_tokens": 657064427.0, + "step": 25392 + }, + { + "epoch": 2.788600922468702, + "grad_norm": 2.351836919784546, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7325122952461243, + "num_tokens": 657089835.0, + "step": 25393 + }, + { + "epoch": 2.7887107401713154, + "grad_norm": 2.2567691802978516, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.732405424118042, + "num_tokens": 657117918.0, + "step": 25394 + }, + { + "epoch": 2.788820557873929, + "grad_norm": 2.1279797554016113, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7001944780349731, + "num_tokens": 657150714.0, + "step": 25395 + }, + { + "epoch": 2.788930375576543, + "grad_norm": 2.365858554840088, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7311674952507019, + "num_tokens": 657175174.0, + "step": 25396 + }, + { + "epoch": 2.7890401932791566, + "grad_norm": 2.3803062438964844, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7208814024925232, + "num_tokens": 657202885.0, + "step": 25397 + }, + { + "epoch": 2.7891500109817704, + "grad_norm": 2.1708390712738037, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7271100282669067, + "num_tokens": 657231845.0, + "step": 25398 + }, + { + "epoch": 2.7892598286843837, + "grad_norm": 2.1701457500457764, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7292470932006836, + "num_tokens": 657261026.0, + "step": 25399 + }, + { + "epoch": 2.7893696463869975, + "grad_norm": 2.2340269088745117, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7165011167526245, + "num_tokens": 657291041.0, + "step": 25400 + }, + { + "epoch": 2.7894794640896112, + "grad_norm": 2.1836318969726562, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7322407960891724, + "num_tokens": 657319078.0, + "step": 25401 + }, + { + "epoch": 2.789589281792225, + "grad_norm": 2.783801794052124, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7629213333129883, + "num_tokens": 657338584.0, + "step": 25402 + }, + { + "epoch": 2.7896990994948387, + "grad_norm": 2.6135568618774414, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7264847159385681, + "num_tokens": 657361728.0, + "step": 25403 + }, + { + "epoch": 2.789808917197452, + "grad_norm": 2.5577919483184814, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7431222796440125, + "num_tokens": 657384977.0, + "step": 25404 + }, + { + "epoch": 2.789918734900066, + "grad_norm": 2.144247531890869, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7121623158454895, + "num_tokens": 657415883.0, + "step": 25405 + }, + { + "epoch": 2.7900285526026796, + "grad_norm": 2.491504430770874, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7678136229515076, + "num_tokens": 657438089.0, + "step": 25406 + }, + { + "epoch": 2.7901383703052933, + "grad_norm": 2.1594321727752686, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7209886312484741, + "num_tokens": 657467505.0, + "step": 25407 + }, + { + "epoch": 2.790248188007907, + "grad_norm": 2.2449326515197754, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7134419083595276, + "num_tokens": 657496368.0, + "step": 25408 + }, + { + "epoch": 2.7903580057105204, + "grad_norm": 2.0753366947174072, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7301415205001831, + "num_tokens": 657529212.0, + "step": 25409 + }, + { + "epoch": 2.790467823413134, + "grad_norm": 2.2230308055877686, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7326577305793762, + "num_tokens": 657556847.0, + "step": 25410 + }, + { + "epoch": 2.790577641115748, + "grad_norm": 2.1693451404571533, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.71268630027771, + "num_tokens": 657587255.0, + "step": 25411 + }, + { + "epoch": 2.7906874588183617, + "grad_norm": 2.7073450088500977, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7559763193130493, + "num_tokens": 657607807.0, + "step": 25412 + }, + { + "epoch": 2.7907972765209754, + "grad_norm": 2.307889461517334, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7583810091018677, + "num_tokens": 657632519.0, + "step": 25413 + }, + { + "epoch": 2.7909070942235887, + "grad_norm": 2.3042685985565186, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7028224468231201, + "num_tokens": 657660615.0, + "step": 25414 + }, + { + "epoch": 2.7910169119262025, + "grad_norm": 2.302288055419922, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7131606340408325, + "num_tokens": 657688699.0, + "step": 25415 + }, + { + "epoch": 2.7911267296288162, + "grad_norm": 2.2328784465789795, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7375027537345886, + "num_tokens": 657717203.0, + "step": 25416 + }, + { + "epoch": 2.7912365473314296, + "grad_norm": 2.391507387161255, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7274143099784851, + "num_tokens": 657743949.0, + "step": 25417 + }, + { + "epoch": 2.7913463650340438, + "grad_norm": 2.1085987091064453, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7224785685539246, + "num_tokens": 657774713.0, + "step": 25418 + }, + { + "epoch": 2.791456182736657, + "grad_norm": 2.371183156967163, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7559784650802612, + "num_tokens": 657799743.0, + "step": 25419 + }, + { + "epoch": 2.791566000439271, + "grad_norm": 2.1462221145629883, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7447890043258667, + "num_tokens": 657829342.0, + "step": 25420 + }, + { + "epoch": 2.7916758181418846, + "grad_norm": 2.3465397357940674, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7600053548812866, + "num_tokens": 657854566.0, + "step": 25421 + }, + { + "epoch": 2.791785635844498, + "grad_norm": 2.491004228591919, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7452781200408936, + "num_tokens": 657879366.0, + "step": 25422 + }, + { + "epoch": 2.7918954535471117, + "grad_norm": 2.351895332336426, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7317887544631958, + "num_tokens": 657905414.0, + "step": 25423 + }, + { + "epoch": 2.7920052712497254, + "grad_norm": 2.340961456298828, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7317245602607727, + "num_tokens": 657931013.0, + "step": 25424 + }, + { + "epoch": 2.792115088952339, + "grad_norm": 2.2356772422790527, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7084814310073853, + "num_tokens": 657959293.0, + "step": 25425 + }, + { + "epoch": 2.792224906654953, + "grad_norm": 2.2359373569488525, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7112192511558533, + "num_tokens": 657985708.0, + "step": 25426 + }, + { + "epoch": 2.7923347243575662, + "grad_norm": 2.2970595359802246, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7296017408370972, + "num_tokens": 658013079.0, + "step": 25427 + }, + { + "epoch": 2.79244454206018, + "grad_norm": 2.190155029296875, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7187628746032715, + "num_tokens": 658042567.0, + "step": 25428 + }, + { + "epoch": 2.7925543597627938, + "grad_norm": 2.1544766426086426, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7353262901306152, + "num_tokens": 658072626.0, + "step": 25429 + }, + { + "epoch": 2.7926641774654075, + "grad_norm": 2.105180501937866, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7151485085487366, + "num_tokens": 658102681.0, + "step": 25430 + }, + { + "epoch": 2.7927739951680213, + "grad_norm": 2.4727439880371094, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7526289224624634, + "num_tokens": 658125645.0, + "step": 25431 + }, + { + "epoch": 2.7928838128706346, + "grad_norm": 2.5053293704986572, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7391675114631653, + "num_tokens": 658148908.0, + "step": 25432 + }, + { + "epoch": 2.7929936305732483, + "grad_norm": 2.4918646812438965, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7236226201057434, + "num_tokens": 658171824.0, + "step": 25433 + }, + { + "epoch": 2.793103448275862, + "grad_norm": 2.4015536308288574, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7439287900924683, + "num_tokens": 658196709.0, + "step": 25434 + }, + { + "epoch": 2.793213265978476, + "grad_norm": 2.81268310546875, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7518771290779114, + "num_tokens": 658216649.0, + "step": 25435 + }, + { + "epoch": 2.7933230836810896, + "grad_norm": 2.374866485595703, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7498195171356201, + "num_tokens": 658241009.0, + "step": 25436 + }, + { + "epoch": 2.793432901383703, + "grad_norm": 2.4084482192993164, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7230855226516724, + "num_tokens": 658267432.0, + "step": 25437 + }, + { + "epoch": 2.7935427190863167, + "grad_norm": 2.3364129066467285, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7314373254776001, + "num_tokens": 658294948.0, + "step": 25438 + }, + { + "epoch": 2.7936525367889304, + "grad_norm": 2.425947427749634, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.748920202255249, + "num_tokens": 658319781.0, + "step": 25439 + }, + { + "epoch": 2.793762354491544, + "grad_norm": 2.293116331100464, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7240966558456421, + "num_tokens": 658347955.0, + "step": 25440 + }, + { + "epoch": 2.793872172194158, + "grad_norm": 2.6171841621398926, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7584526538848877, + "num_tokens": 658370227.0, + "step": 25441 + }, + { + "epoch": 2.7939819898967713, + "grad_norm": 2.236720085144043, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7335871458053589, + "num_tokens": 658399478.0, + "step": 25442 + }, + { + "epoch": 2.794091807599385, + "grad_norm": 2.7368013858795166, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7371503710746765, + "num_tokens": 658419334.0, + "step": 25443 + }, + { + "epoch": 2.794201625301999, + "grad_norm": 2.205439329147339, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7448376417160034, + "num_tokens": 658446821.0, + "step": 25444 + }, + { + "epoch": 2.794311443004612, + "grad_norm": 2.3117337226867676, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7598830461502075, + "num_tokens": 658472890.0, + "step": 25445 + }, + { + "epoch": 2.794421260707226, + "grad_norm": 2.1408350467681885, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7258971333503723, + "num_tokens": 658501453.0, + "step": 25446 + }, + { + "epoch": 2.7945310784098396, + "grad_norm": 2.3790926933288574, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7502284049987793, + "num_tokens": 658525789.0, + "step": 25447 + }, + { + "epoch": 2.7946408961124534, + "grad_norm": 2.6437017917633057, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7354332208633423, + "num_tokens": 658547999.0, + "step": 25448 + }, + { + "epoch": 2.794750713815067, + "grad_norm": 2.7410295009613037, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7541533708572388, + "num_tokens": 658567804.0, + "step": 25449 + }, + { + "epoch": 2.7948605315176804, + "grad_norm": 2.2544162273406982, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7256959676742554, + "num_tokens": 658595068.0, + "step": 25450 + }, + { + "epoch": 2.794970349220294, + "grad_norm": 2.5045876502990723, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7293564081192017, + "num_tokens": 658621032.0, + "step": 25451 + }, + { + "epoch": 2.795080166922908, + "grad_norm": 2.5585548877716064, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7388224005699158, + "num_tokens": 658641357.0, + "step": 25452 + }, + { + "epoch": 2.7951899846255217, + "grad_norm": 2.239983081817627, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7095764875411987, + "num_tokens": 658671439.0, + "step": 25453 + }, + { + "epoch": 2.7952998023281355, + "grad_norm": 2.3971736431121826, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7345064878463745, + "num_tokens": 658696053.0, + "step": 25454 + }, + { + "epoch": 2.7954096200307488, + "grad_norm": 2.34340500831604, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7452085018157959, + "num_tokens": 658721063.0, + "step": 25455 + }, + { + "epoch": 2.7955194377333625, + "grad_norm": 2.289044141769409, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.750691294670105, + "num_tokens": 658746630.0, + "step": 25456 + }, + { + "epoch": 2.7956292554359763, + "grad_norm": 2.306670904159546, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7404550313949585, + "num_tokens": 658771703.0, + "step": 25457 + }, + { + "epoch": 2.79573907313859, + "grad_norm": 2.1669719219207764, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7383323907852173, + "num_tokens": 658800801.0, + "step": 25458 + }, + { + "epoch": 2.795848890841204, + "grad_norm": 2.11525297164917, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7231079339981079, + "num_tokens": 658832820.0, + "step": 25459 + }, + { + "epoch": 2.795958708543817, + "grad_norm": 2.0769073963165283, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7011858224868774, + "num_tokens": 658868258.0, + "step": 25460 + }, + { + "epoch": 2.796068526246431, + "grad_norm": 2.454887866973877, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7320391535758972, + "num_tokens": 658891424.0, + "step": 25461 + }, + { + "epoch": 2.7961783439490446, + "grad_norm": 2.2112081050872803, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7069199085235596, + "num_tokens": 658919905.0, + "step": 25462 + }, + { + "epoch": 2.7962881616516584, + "grad_norm": 2.41664457321167, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7309249639511108, + "num_tokens": 658945164.0, + "step": 25463 + }, + { + "epoch": 2.796397979354272, + "grad_norm": 2.426353931427002, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7554286122322083, + "num_tokens": 658967595.0, + "step": 25464 + }, + { + "epoch": 2.7965077970568855, + "grad_norm": 2.080927610397339, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7310959696769714, + "num_tokens": 658999509.0, + "step": 25465 + }, + { + "epoch": 2.796617614759499, + "grad_norm": 2.244079113006592, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7273555994033813, + "num_tokens": 659028209.0, + "step": 25466 + }, + { + "epoch": 2.796727432462113, + "grad_norm": 2.2988975048065186, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7321621179580688, + "num_tokens": 659054733.0, + "step": 25467 + }, + { + "epoch": 2.7968372501647263, + "grad_norm": 2.251495838165283, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.696164071559906, + "num_tokens": 659086566.0, + "step": 25468 + }, + { + "epoch": 2.7969470678673405, + "grad_norm": 2.348947286605835, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7158765196800232, + "num_tokens": 659113076.0, + "step": 25469 + }, + { + "epoch": 2.797056885569954, + "grad_norm": 2.2550907135009766, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7456742525100708, + "num_tokens": 659140510.0, + "step": 25470 + }, + { + "epoch": 2.7971667032725676, + "grad_norm": 2.160104751586914, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7298311591148376, + "num_tokens": 659171065.0, + "step": 25471 + }, + { + "epoch": 2.7972765209751813, + "grad_norm": 2.315948247909546, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7149772644042969, + "num_tokens": 659198950.0, + "step": 25472 + }, + { + "epoch": 2.7973863386777946, + "grad_norm": 2.1641688346862793, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7137700319290161, + "num_tokens": 659228794.0, + "step": 25473 + }, + { + "epoch": 2.7974961563804084, + "grad_norm": 2.577172040939331, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7253220677375793, + "num_tokens": 659250785.0, + "step": 25474 + }, + { + "epoch": 2.797605974083022, + "grad_norm": 2.5391674041748047, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7679332494735718, + "num_tokens": 659273650.0, + "step": 25475 + }, + { + "epoch": 2.797715791785636, + "grad_norm": 2.353461742401123, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7526370882987976, + "num_tokens": 659297636.0, + "step": 25476 + }, + { + "epoch": 2.7978256094882497, + "grad_norm": 2.326483726501465, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7302093505859375, + "num_tokens": 659325961.0, + "step": 25477 + }, + { + "epoch": 2.797935427190863, + "grad_norm": 2.1985251903533936, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7134280204772949, + "num_tokens": 659355236.0, + "step": 25478 + }, + { + "epoch": 2.7980452448934767, + "grad_norm": 2.4382612705230713, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7475031614303589, + "num_tokens": 659380549.0, + "step": 25479 + }, + { + "epoch": 2.7981550625960905, + "grad_norm": 2.376310348510742, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7530205249786377, + "num_tokens": 659404170.0, + "step": 25480 + }, + { + "epoch": 2.7982648802987042, + "grad_norm": 2.2367115020751953, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7574777603149414, + "num_tokens": 659429358.0, + "step": 25481 + }, + { + "epoch": 2.798374698001318, + "grad_norm": 2.4571521282196045, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7359597682952881, + "num_tokens": 659452045.0, + "step": 25482 + }, + { + "epoch": 2.7984845157039313, + "grad_norm": 2.0838112831115723, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7244186401367188, + "num_tokens": 659482183.0, + "step": 25483 + }, + { + "epoch": 2.798594333406545, + "grad_norm": 2.4281601905822754, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7232575416564941, + "num_tokens": 659506878.0, + "step": 25484 + }, + { + "epoch": 2.798704151109159, + "grad_norm": 2.282628297805786, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7281534671783447, + "num_tokens": 659534891.0, + "step": 25485 + }, + { + "epoch": 2.7988139688117726, + "grad_norm": 2.291804790496826, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7382885813713074, + "num_tokens": 659563039.0, + "step": 25486 + }, + { + "epoch": 2.7989237865143863, + "grad_norm": 2.4160680770874023, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7193877100944519, + "num_tokens": 659587458.0, + "step": 25487 + }, + { + "epoch": 2.7990336042169996, + "grad_norm": 2.4366207122802734, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7371740341186523, + "num_tokens": 659610684.0, + "step": 25488 + }, + { + "epoch": 2.7991434219196134, + "grad_norm": 2.5593526363372803, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7554987668991089, + "num_tokens": 659631616.0, + "step": 25489 + }, + { + "epoch": 2.799253239622227, + "grad_norm": 2.322260856628418, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7226953506469727, + "num_tokens": 659659811.0, + "step": 25490 + }, + { + "epoch": 2.799363057324841, + "grad_norm": 2.619619369506836, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7510976195335388, + "num_tokens": 659681866.0, + "step": 25491 + }, + { + "epoch": 2.7994728750274547, + "grad_norm": 2.2518653869628906, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7289144396781921, + "num_tokens": 659708529.0, + "step": 25492 + }, + { + "epoch": 2.799582692730068, + "grad_norm": 2.2098851203918457, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7346497774124146, + "num_tokens": 659737239.0, + "step": 25493 + }, + { + "epoch": 2.7996925104326817, + "grad_norm": 2.6365420818328857, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7420412302017212, + "num_tokens": 659758319.0, + "step": 25494 + }, + { + "epoch": 2.7998023281352955, + "grad_norm": 2.4673068523406982, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.729504406452179, + "num_tokens": 659781992.0, + "step": 25495 + }, + { + "epoch": 2.799912145837909, + "grad_norm": 2.6734490394592285, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7300716638565063, + "num_tokens": 659803204.0, + "step": 25496 + }, + { + "epoch": 2.8000219635405226, + "grad_norm": 2.44639253616333, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7282640933990479, + "num_tokens": 659827306.0, + "step": 25497 + }, + { + "epoch": 2.8001317812431363, + "grad_norm": 2.654583692550659, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7353211641311646, + "num_tokens": 659847819.0, + "step": 25498 + }, + { + "epoch": 2.80024159894575, + "grad_norm": 2.5235695838928223, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7255652546882629, + "num_tokens": 659871600.0, + "step": 25499 + }, + { + "epoch": 2.800351416648364, + "grad_norm": 2.1655945777893066, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7398099303245544, + "num_tokens": 659900805.0, + "step": 25500 + }, + { + "epoch": 2.800461234350977, + "grad_norm": 2.1632771492004395, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7262107133865356, + "num_tokens": 659929084.0, + "step": 25501 + }, + { + "epoch": 2.800571052053591, + "grad_norm": 2.0897109508514404, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7168819904327393, + "num_tokens": 659961327.0, + "step": 25502 + }, + { + "epoch": 2.8006808697562047, + "grad_norm": 2.465684175491333, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7113704085350037, + "num_tokens": 659986817.0, + "step": 25503 + }, + { + "epoch": 2.8007906874588184, + "grad_norm": 2.758192300796509, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.721591055393219, + "num_tokens": 660008085.0, + "step": 25504 + }, + { + "epoch": 2.800900505161432, + "grad_norm": 2.2803561687469482, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7449572086334229, + "num_tokens": 660034970.0, + "step": 25505 + }, + { + "epoch": 2.8010103228640455, + "grad_norm": 2.653695821762085, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7430989742279053, + "num_tokens": 660056318.0, + "step": 25506 + }, + { + "epoch": 2.8011201405666593, + "grad_norm": 2.4359726905822754, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7506155967712402, + "num_tokens": 660079209.0, + "step": 25507 + }, + { + "epoch": 2.801229958269273, + "grad_norm": 2.282996892929077, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7090567350387573, + "num_tokens": 660107530.0, + "step": 25508 + }, + { + "epoch": 2.8013397759718868, + "grad_norm": 2.4959373474121094, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7384473085403442, + "num_tokens": 660131087.0, + "step": 25509 + }, + { + "epoch": 2.8014495936745005, + "grad_norm": 2.192079544067383, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7141167521476746, + "num_tokens": 660162306.0, + "step": 25510 + }, + { + "epoch": 2.801559411377114, + "grad_norm": 2.403949022293091, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7371723651885986, + "num_tokens": 660186485.0, + "step": 25511 + }, + { + "epoch": 2.8016692290797276, + "grad_norm": 2.3069512844085693, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.728142261505127, + "num_tokens": 660210493.0, + "step": 25512 + }, + { + "epoch": 2.8017790467823414, + "grad_norm": 2.4391350746154785, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7177391648292542, + "num_tokens": 660233809.0, + "step": 25513 + }, + { + "epoch": 2.801888864484955, + "grad_norm": 2.2305641174316406, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7322276830673218, + "num_tokens": 660262261.0, + "step": 25514 + }, + { + "epoch": 2.801998682187569, + "grad_norm": 2.514998435974121, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7424920797348022, + "num_tokens": 660285282.0, + "step": 25515 + }, + { + "epoch": 2.802108499890182, + "grad_norm": 2.381173610687256, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7379899621009827, + "num_tokens": 660312026.0, + "step": 25516 + }, + { + "epoch": 2.802218317592796, + "grad_norm": 2.526085138320923, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7265781164169312, + "num_tokens": 660336797.0, + "step": 25517 + }, + { + "epoch": 2.8023281352954097, + "grad_norm": 2.3673713207244873, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.735122561454773, + "num_tokens": 660363648.0, + "step": 25518 + }, + { + "epoch": 2.802437952998023, + "grad_norm": 2.568284749984741, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7310873866081238, + "num_tokens": 660388352.0, + "step": 25519 + }, + { + "epoch": 2.802547770700637, + "grad_norm": 2.5034189224243164, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7290623188018799, + "num_tokens": 660412677.0, + "step": 25520 + }, + { + "epoch": 2.8026575884032505, + "grad_norm": 2.22973895072937, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7051292657852173, + "num_tokens": 660442384.0, + "step": 25521 + }, + { + "epoch": 2.8027674061058643, + "grad_norm": 2.1422650814056396, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7467200756072998, + "num_tokens": 660471277.0, + "step": 25522 + }, + { + "epoch": 2.802877223808478, + "grad_norm": 2.2591707706451416, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7301095128059387, + "num_tokens": 660499776.0, + "step": 25523 + }, + { + "epoch": 2.8029870415110913, + "grad_norm": 2.1570844650268555, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7219924926757812, + "num_tokens": 660529542.0, + "step": 25524 + }, + { + "epoch": 2.803096859213705, + "grad_norm": 2.308529853820801, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7249009013175964, + "num_tokens": 660557255.0, + "step": 25525 + }, + { + "epoch": 2.803206676916319, + "grad_norm": 2.6253252029418945, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7415434122085571, + "num_tokens": 660577448.0, + "step": 25526 + }, + { + "epoch": 2.8033164946189326, + "grad_norm": 2.0640854835510254, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7216392755508423, + "num_tokens": 660611802.0, + "step": 25527 + }, + { + "epoch": 2.8034263123215464, + "grad_norm": 2.767693519592285, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7616247534751892, + "num_tokens": 660632030.0, + "step": 25528 + }, + { + "epoch": 2.8035361300241597, + "grad_norm": 2.420703649520874, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7171218395233154, + "num_tokens": 660656917.0, + "step": 25529 + }, + { + "epoch": 2.8036459477267734, + "grad_norm": 2.686248779296875, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7544503211975098, + "num_tokens": 660677591.0, + "step": 25530 + }, + { + "epoch": 2.803755765429387, + "grad_norm": 2.3096327781677246, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7484288215637207, + "num_tokens": 660704066.0, + "step": 25531 + }, + { + "epoch": 2.803865583132001, + "grad_norm": 2.016181230545044, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7194206714630127, + "num_tokens": 660736158.0, + "step": 25532 + }, + { + "epoch": 2.8039754008346147, + "grad_norm": 2.199732542037964, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7388525605201721, + "num_tokens": 660764931.0, + "step": 25533 + }, + { + "epoch": 2.804085218537228, + "grad_norm": 2.3386614322662354, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7207945585250854, + "num_tokens": 660791878.0, + "step": 25534 + }, + { + "epoch": 2.804195036239842, + "grad_norm": 2.3856875896453857, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.741142988204956, + "num_tokens": 660818303.0, + "step": 25535 + }, + { + "epoch": 2.8043048539424555, + "grad_norm": 2.132518768310547, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7263081073760986, + "num_tokens": 660847686.0, + "step": 25536 + }, + { + "epoch": 2.8044146716450693, + "grad_norm": 2.353048801422119, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.729613184928894, + "num_tokens": 660874618.0, + "step": 25537 + }, + { + "epoch": 2.804524489347683, + "grad_norm": 2.6066911220550537, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7524520754814148, + "num_tokens": 660895920.0, + "step": 25538 + }, + { + "epoch": 2.8046343070502964, + "grad_norm": 2.568366050720215, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7354916334152222, + "num_tokens": 660918930.0, + "step": 25539 + }, + { + "epoch": 2.80474412475291, + "grad_norm": 2.242729663848877, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7541273236274719, + "num_tokens": 660946257.0, + "step": 25540 + }, + { + "epoch": 2.804853942455524, + "grad_norm": 2.458164930343628, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7217128872871399, + "num_tokens": 660970851.0, + "step": 25541 + }, + { + "epoch": 2.8049637601581376, + "grad_norm": 2.1720588207244873, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7007956504821777, + "num_tokens": 661001842.0, + "step": 25542 + }, + { + "epoch": 2.8050735778607514, + "grad_norm": 2.1150195598602295, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7246536612510681, + "num_tokens": 661033093.0, + "step": 25543 + }, + { + "epoch": 2.8051833955633647, + "grad_norm": 2.5724172592163086, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7420421242713928, + "num_tokens": 661055272.0, + "step": 25544 + }, + { + "epoch": 2.8052932132659785, + "grad_norm": 2.343766927719116, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7331249713897705, + "num_tokens": 661080432.0, + "step": 25545 + }, + { + "epoch": 2.8054030309685922, + "grad_norm": 2.5104236602783203, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7432672381401062, + "num_tokens": 661105508.0, + "step": 25546 + }, + { + "epoch": 2.8055128486712055, + "grad_norm": 2.280275821685791, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7289918661117554, + "num_tokens": 661135946.0, + "step": 25547 + }, + { + "epoch": 2.8056226663738193, + "grad_norm": 2.5136163234710693, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7279151082038879, + "num_tokens": 661159454.0, + "step": 25548 + }, + { + "epoch": 2.805732484076433, + "grad_norm": 2.465754270553589, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7661105394363403, + "num_tokens": 661182410.0, + "step": 25549 + }, + { + "epoch": 2.805842301779047, + "grad_norm": 2.8872430324554443, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7469805479049683, + "num_tokens": 661201229.0, + "step": 25550 + }, + { + "epoch": 2.8059521194816606, + "grad_norm": 2.282451868057251, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7112452983856201, + "num_tokens": 661229079.0, + "step": 25551 + }, + { + "epoch": 2.806061937184274, + "grad_norm": 2.4797744750976562, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7360896468162537, + "num_tokens": 661252597.0, + "step": 25552 + }, + { + "epoch": 2.8061717548868876, + "grad_norm": 2.3386571407318115, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7078896760940552, + "num_tokens": 661280899.0, + "step": 25553 + }, + { + "epoch": 2.8062815725895014, + "grad_norm": 2.324101686477661, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7309324741363525, + "num_tokens": 661309258.0, + "step": 25554 + }, + { + "epoch": 2.806391390292115, + "grad_norm": 2.4674723148345947, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7133639454841614, + "num_tokens": 661334630.0, + "step": 25555 + }, + { + "epoch": 2.806501207994729, + "grad_norm": 2.314042568206787, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7375971078872681, + "num_tokens": 661361697.0, + "step": 25556 + }, + { + "epoch": 2.806611025697342, + "grad_norm": 2.400620698928833, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7359815835952759, + "num_tokens": 661387248.0, + "step": 25557 + }, + { + "epoch": 2.806720843399956, + "grad_norm": 2.4014687538146973, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.727333128452301, + "num_tokens": 661412647.0, + "step": 25558 + }, + { + "epoch": 2.8068306611025697, + "grad_norm": 2.321681261062622, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7277380228042603, + "num_tokens": 661438448.0, + "step": 25559 + }, + { + "epoch": 2.8069404788051835, + "grad_norm": 2.4435293674468994, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7354451417922974, + "num_tokens": 661463863.0, + "step": 25560 + }, + { + "epoch": 2.8070502965077972, + "grad_norm": 2.255068302154541, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7246417999267578, + "num_tokens": 661492452.0, + "step": 25561 + }, + { + "epoch": 2.8071601142104106, + "grad_norm": 2.657421827316284, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7409595251083374, + "num_tokens": 661514392.0, + "step": 25562 + }, + { + "epoch": 2.8072699319130243, + "grad_norm": 2.1881954669952393, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7442505955696106, + "num_tokens": 661540750.0, + "step": 25563 + }, + { + "epoch": 2.807379749615638, + "grad_norm": 2.4072272777557373, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7125269174575806, + "num_tokens": 661568038.0, + "step": 25564 + }, + { + "epoch": 2.807489567318252, + "grad_norm": 2.2475273609161377, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.736483633518219, + "num_tokens": 661594930.0, + "step": 25565 + }, + { + "epoch": 2.8075993850208656, + "grad_norm": 2.2456247806549072, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7210235595703125, + "num_tokens": 661623558.0, + "step": 25566 + }, + { + "epoch": 2.807709202723479, + "grad_norm": 2.19632887840271, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7323883771896362, + "num_tokens": 661652206.0, + "step": 25567 + }, + { + "epoch": 2.8078190204260927, + "grad_norm": 2.1546316146850586, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.719124972820282, + "num_tokens": 661683089.0, + "step": 25568 + }, + { + "epoch": 2.8079288381287064, + "grad_norm": 2.379890203475952, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7535757422447205, + "num_tokens": 661707272.0, + "step": 25569 + }, + { + "epoch": 2.80803865583132, + "grad_norm": 2.4837520122528076, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7434277534484863, + "num_tokens": 661730598.0, + "step": 25570 + }, + { + "epoch": 2.808148473533934, + "grad_norm": 2.5211448669433594, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7471719980239868, + "num_tokens": 661753754.0, + "step": 25571 + }, + { + "epoch": 2.8082582912365472, + "grad_norm": 2.3625385761260986, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7376947402954102, + "num_tokens": 661779277.0, + "step": 25572 + }, + { + "epoch": 2.808368108939161, + "grad_norm": 2.482583522796631, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7393755316734314, + "num_tokens": 661804152.0, + "step": 25573 + }, + { + "epoch": 2.8084779266417748, + "grad_norm": 2.584307909011841, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7186405062675476, + "num_tokens": 661826690.0, + "step": 25574 + }, + { + "epoch": 2.808587744344388, + "grad_norm": 2.1154515743255615, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.716701090335846, + "num_tokens": 661858275.0, + "step": 25575 + }, + { + "epoch": 2.808697562047002, + "grad_norm": 2.4100422859191895, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7206733226776123, + "num_tokens": 661885693.0, + "step": 25576 + }, + { + "epoch": 2.8088073797496156, + "grad_norm": 2.402217388153076, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7265552282333374, + "num_tokens": 661910365.0, + "step": 25577 + }, + { + "epoch": 2.8089171974522293, + "grad_norm": 2.666828155517578, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7479602098464966, + "num_tokens": 661930340.0, + "step": 25578 + }, + { + "epoch": 2.809027015154843, + "grad_norm": 2.69403338432312, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7349384427070618, + "num_tokens": 661950143.0, + "step": 25579 + }, + { + "epoch": 2.8091368328574564, + "grad_norm": 2.6832761764526367, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7476945519447327, + "num_tokens": 661972186.0, + "step": 25580 + }, + { + "epoch": 2.80924665056007, + "grad_norm": 2.263627767562866, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7384021878242493, + "num_tokens": 661998612.0, + "step": 25581 + }, + { + "epoch": 2.809356468262684, + "grad_norm": 2.322995662689209, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7372061610221863, + "num_tokens": 662027212.0, + "step": 25582 + }, + { + "epoch": 2.8094662859652977, + "grad_norm": 2.488116502761841, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7399005889892578, + "num_tokens": 662052504.0, + "step": 25583 + }, + { + "epoch": 2.8095761036679114, + "grad_norm": 2.540128231048584, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7517269253730774, + "num_tokens": 662075237.0, + "step": 25584 + }, + { + "epoch": 2.8096859213705248, + "grad_norm": 2.36517596244812, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7264972925186157, + "num_tokens": 662100237.0, + "step": 25585 + }, + { + "epoch": 2.8097957390731385, + "grad_norm": 2.384888172149658, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7595982551574707, + "num_tokens": 662125680.0, + "step": 25586 + }, + { + "epoch": 2.8099055567757523, + "grad_norm": 2.71012020111084, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7268739938735962, + "num_tokens": 662146244.0, + "step": 25587 + }, + { + "epoch": 2.810015374478366, + "grad_norm": 2.1693882942199707, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7353929877281189, + "num_tokens": 662177330.0, + "step": 25588 + }, + { + "epoch": 2.81012519218098, + "grad_norm": 2.6885952949523926, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7534915804862976, + "num_tokens": 662197208.0, + "step": 25589 + }, + { + "epoch": 2.810235009883593, + "grad_norm": 2.7462844848632812, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7667988538742065, + "num_tokens": 662216227.0, + "step": 25590 + }, + { + "epoch": 2.810344827586207, + "grad_norm": 2.433159828186035, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7256976962089539, + "num_tokens": 662241684.0, + "step": 25591 + }, + { + "epoch": 2.8104546452888206, + "grad_norm": 2.6873397827148438, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7395778298377991, + "num_tokens": 662262461.0, + "step": 25592 + }, + { + "epoch": 2.8105644629914344, + "grad_norm": 2.3514225482940674, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7394641041755676, + "num_tokens": 662287571.0, + "step": 25593 + }, + { + "epoch": 2.810674280694048, + "grad_norm": 2.1446876525878906, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7272757291793823, + "num_tokens": 662317521.0, + "step": 25594 + }, + { + "epoch": 2.8107840983966614, + "grad_norm": 2.1878578662872314, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7104157209396362, + "num_tokens": 662346423.0, + "step": 25595 + }, + { + "epoch": 2.810893916099275, + "grad_norm": 2.6764910221099854, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7265372276306152, + "num_tokens": 662367305.0, + "step": 25596 + }, + { + "epoch": 2.811003733801889, + "grad_norm": 2.3798346519470215, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7322183847427368, + "num_tokens": 662392026.0, + "step": 25597 + }, + { + "epoch": 2.8111135515045023, + "grad_norm": 2.4398226737976074, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7479130029678345, + "num_tokens": 662417973.0, + "step": 25598 + }, + { + "epoch": 2.8112233692071165, + "grad_norm": 2.0859055519104004, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7340520620346069, + "num_tokens": 662449961.0, + "step": 25599 + }, + { + "epoch": 2.8113331869097298, + "grad_norm": 2.331707000732422, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7280774116516113, + "num_tokens": 662474543.0, + "step": 25600 + }, + { + "epoch": 2.8114430046123435, + "grad_norm": 2.321763753890991, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7138590216636658, + "num_tokens": 662503803.0, + "step": 25601 + }, + { + "epoch": 2.8115528223149573, + "grad_norm": 2.0222504138946533, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7348490953445435, + "num_tokens": 662538210.0, + "step": 25602 + }, + { + "epoch": 2.8116626400175706, + "grad_norm": 2.5898048877716064, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7593640685081482, + "num_tokens": 662559687.0, + "step": 25603 + }, + { + "epoch": 2.8117724577201844, + "grad_norm": 2.4001994132995605, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7502648234367371, + "num_tokens": 662583448.0, + "step": 25604 + }, + { + "epoch": 2.811882275422798, + "grad_norm": 2.3876123428344727, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7253580093383789, + "num_tokens": 662608705.0, + "step": 25605 + }, + { + "epoch": 2.811992093125412, + "grad_norm": 2.6302578449249268, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.71117103099823, + "num_tokens": 662630715.0, + "step": 25606 + }, + { + "epoch": 2.8121019108280256, + "grad_norm": 2.2797436714172363, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7214590907096863, + "num_tokens": 662657962.0, + "step": 25607 + }, + { + "epoch": 2.812211728530639, + "grad_norm": 2.391709566116333, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7354463934898376, + "num_tokens": 662684548.0, + "step": 25608 + }, + { + "epoch": 2.8123215462332527, + "grad_norm": 2.337773323059082, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.744888186454773, + "num_tokens": 662710245.0, + "step": 25609 + }, + { + "epoch": 2.8124313639358665, + "grad_norm": 2.4594266414642334, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7384430170059204, + "num_tokens": 662735937.0, + "step": 25610 + }, + { + "epoch": 2.81254118163848, + "grad_norm": 2.7540063858032227, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7376033067703247, + "num_tokens": 662755845.0, + "step": 25611 + }, + { + "epoch": 2.812650999341094, + "grad_norm": 2.4598443508148193, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7259385585784912, + "num_tokens": 662781038.0, + "step": 25612 + }, + { + "epoch": 2.8127608170437073, + "grad_norm": 2.50272536277771, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7320413589477539, + "num_tokens": 662805811.0, + "step": 25613 + }, + { + "epoch": 2.812870634746321, + "grad_norm": 2.6315836906433105, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7479791045188904, + "num_tokens": 662826943.0, + "step": 25614 + }, + { + "epoch": 2.812980452448935, + "grad_norm": 2.2001595497131348, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7253325581550598, + "num_tokens": 662854053.0, + "step": 25615 + }, + { + "epoch": 2.8130902701515486, + "grad_norm": 2.259782075881958, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7175329923629761, + "num_tokens": 662882986.0, + "step": 25616 + }, + { + "epoch": 2.8132000878541623, + "grad_norm": 2.23101544380188, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7321202754974365, + "num_tokens": 662913476.0, + "step": 25617 + }, + { + "epoch": 2.8133099055567756, + "grad_norm": 2.3875341415405273, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7421287298202515, + "num_tokens": 662940969.0, + "step": 25618 + }, + { + "epoch": 2.8134197232593894, + "grad_norm": 2.1584062576293945, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7494020462036133, + "num_tokens": 662969481.0, + "step": 25619 + }, + { + "epoch": 2.813529540962003, + "grad_norm": 2.747779369354248, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7248748540878296, + "num_tokens": 662990900.0, + "step": 25620 + }, + { + "epoch": 2.813639358664617, + "grad_norm": 2.077765941619873, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7003029584884644, + "num_tokens": 663025348.0, + "step": 25621 + }, + { + "epoch": 2.8137491763672307, + "grad_norm": 2.143810749053955, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7283554673194885, + "num_tokens": 663055529.0, + "step": 25622 + }, + { + "epoch": 2.813858994069844, + "grad_norm": 2.2783825397491455, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7437133193016052, + "num_tokens": 663082171.0, + "step": 25623 + }, + { + "epoch": 2.8139688117724577, + "grad_norm": 2.683729410171509, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7436184883117676, + "num_tokens": 663102817.0, + "step": 25624 + }, + { + "epoch": 2.8140786294750715, + "grad_norm": 2.385998010635376, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7232122421264648, + "num_tokens": 663126762.0, + "step": 25625 + }, + { + "epoch": 2.814188447177685, + "grad_norm": 2.581232786178589, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7339358329772949, + "num_tokens": 663149265.0, + "step": 25626 + }, + { + "epoch": 2.8142982648802986, + "grad_norm": 2.6341423988342285, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7483196258544922, + "num_tokens": 663171640.0, + "step": 25627 + }, + { + "epoch": 2.8144080825829123, + "grad_norm": 2.0412724018096924, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.742388904094696, + "num_tokens": 663205859.0, + "step": 25628 + }, + { + "epoch": 2.814517900285526, + "grad_norm": 2.4616246223449707, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7537577748298645, + "num_tokens": 663229362.0, + "step": 25629 + }, + { + "epoch": 2.81462771798814, + "grad_norm": 2.4899003505706787, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7508960962295532, + "num_tokens": 663251463.0, + "step": 25630 + }, + { + "epoch": 2.814737535690753, + "grad_norm": 2.48496150970459, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7249428629875183, + "num_tokens": 663275502.0, + "step": 25631 + }, + { + "epoch": 2.814847353393367, + "grad_norm": 2.126258373260498, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7264047265052795, + "num_tokens": 663305986.0, + "step": 25632 + }, + { + "epoch": 2.8149571710959806, + "grad_norm": 2.3759765625, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7545732855796814, + "num_tokens": 663330428.0, + "step": 25633 + }, + { + "epoch": 2.8150669887985944, + "grad_norm": 2.3850908279418945, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.730451762676239, + "num_tokens": 663356398.0, + "step": 25634 + }, + { + "epoch": 2.815176806501208, + "grad_norm": 2.7641844749450684, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7320247888565063, + "num_tokens": 663376647.0, + "step": 25635 + }, + { + "epoch": 2.8152866242038215, + "grad_norm": 2.5608322620391846, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7412673234939575, + "num_tokens": 663399079.0, + "step": 25636 + }, + { + "epoch": 2.8153964419064352, + "grad_norm": 2.3386075496673584, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7418010234832764, + "num_tokens": 663424783.0, + "step": 25637 + }, + { + "epoch": 2.815506259609049, + "grad_norm": 2.102947473526001, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7232689261436462, + "num_tokens": 663456579.0, + "step": 25638 + }, + { + "epoch": 2.8156160773116627, + "grad_norm": 2.391782283782959, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7219187021255493, + "num_tokens": 663481581.0, + "step": 25639 + }, + { + "epoch": 2.8157258950142765, + "grad_norm": 2.4866862297058105, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7365779876708984, + "num_tokens": 663504593.0, + "step": 25640 + }, + { + "epoch": 2.81583571271689, + "grad_norm": 2.326270580291748, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7404793500900269, + "num_tokens": 663529251.0, + "step": 25641 + }, + { + "epoch": 2.8159455304195036, + "grad_norm": 2.420668840408325, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7521709203720093, + "num_tokens": 663553956.0, + "step": 25642 + }, + { + "epoch": 2.8160553481221173, + "grad_norm": 2.5593292713165283, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7400535941123962, + "num_tokens": 663575866.0, + "step": 25643 + }, + { + "epoch": 2.816165165824731, + "grad_norm": 2.0530433654785156, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7241295576095581, + "num_tokens": 663609903.0, + "step": 25644 + }, + { + "epoch": 2.816274983527345, + "grad_norm": 2.306032180786133, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7301933169364929, + "num_tokens": 663637331.0, + "step": 25645 + }, + { + "epoch": 2.816384801229958, + "grad_norm": 2.301365613937378, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.739662230014801, + "num_tokens": 663667723.0, + "step": 25646 + }, + { + "epoch": 2.816494618932572, + "grad_norm": 2.428536891937256, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7238531708717346, + "num_tokens": 663691961.0, + "step": 25647 + }, + { + "epoch": 2.8166044366351857, + "grad_norm": 2.8168587684631348, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7591735124588013, + "num_tokens": 663712363.0, + "step": 25648 + }, + { + "epoch": 2.816714254337799, + "grad_norm": 2.073014974594116, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7174922227859497, + "num_tokens": 663744232.0, + "step": 25649 + }, + { + "epoch": 2.816824072040413, + "grad_norm": 2.2495031356811523, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7295854091644287, + "num_tokens": 663771402.0, + "step": 25650 + }, + { + "epoch": 2.8169338897430265, + "grad_norm": 2.2725582122802734, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7314413785934448, + "num_tokens": 663798841.0, + "step": 25651 + }, + { + "epoch": 2.8170437074456403, + "grad_norm": 2.2737793922424316, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7255374789237976, + "num_tokens": 663825270.0, + "step": 25652 + }, + { + "epoch": 2.817153525148254, + "grad_norm": 2.0258796215057373, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.716092050075531, + "num_tokens": 663856581.0, + "step": 25653 + }, + { + "epoch": 2.8172633428508673, + "grad_norm": 2.3010144233703613, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.754427969455719, + "num_tokens": 663882578.0, + "step": 25654 + }, + { + "epoch": 2.817373160553481, + "grad_norm": 2.2957255840301514, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7259179353713989, + "num_tokens": 663908507.0, + "step": 25655 + }, + { + "epoch": 2.817482978256095, + "grad_norm": 2.3210341930389404, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7445058226585388, + "num_tokens": 663934012.0, + "step": 25656 + }, + { + "epoch": 2.8175927959587086, + "grad_norm": 1.9465603828430176, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7188494801521301, + "num_tokens": 663971194.0, + "step": 25657 + }, + { + "epoch": 2.8177026136613224, + "grad_norm": 2.31722092628479, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7454618215560913, + "num_tokens": 663996352.0, + "step": 25658 + }, + { + "epoch": 2.8178124313639357, + "grad_norm": 2.6709322929382324, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7251067757606506, + "num_tokens": 664016776.0, + "step": 25659 + }, + { + "epoch": 2.8179222490665494, + "grad_norm": 2.3311405181884766, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7565211057662964, + "num_tokens": 664042536.0, + "step": 25660 + }, + { + "epoch": 2.818032066769163, + "grad_norm": 2.212102174758911, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7566804885864258, + "num_tokens": 664070228.0, + "step": 25661 + }, + { + "epoch": 2.818141884471777, + "grad_norm": 2.490784168243408, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7456201314926147, + "num_tokens": 664093475.0, + "step": 25662 + }, + { + "epoch": 2.8182517021743907, + "grad_norm": 2.6237049102783203, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7323688268661499, + "num_tokens": 664114673.0, + "step": 25663 + }, + { + "epoch": 2.818361519877004, + "grad_norm": 2.452789068222046, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7335289716720581, + "num_tokens": 664139000.0, + "step": 25664 + }, + { + "epoch": 2.8184713375796178, + "grad_norm": 2.236845016479492, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7204862236976624, + "num_tokens": 664169452.0, + "step": 25665 + }, + { + "epoch": 2.8185811552822315, + "grad_norm": 3.0669283866882324, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7420870065689087, + "num_tokens": 664190664.0, + "step": 25666 + }, + { + "epoch": 2.8186909729848453, + "grad_norm": 2.462683916091919, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7328574061393738, + "num_tokens": 664214414.0, + "step": 25667 + }, + { + "epoch": 2.818800790687459, + "grad_norm": 2.172879934310913, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7324820756912231, + "num_tokens": 664243671.0, + "step": 25668 + }, + { + "epoch": 2.8189106083900723, + "grad_norm": 2.282320261001587, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.719640851020813, + "num_tokens": 664272682.0, + "step": 25669 + }, + { + "epoch": 2.819020426092686, + "grad_norm": 2.4107232093811035, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7326862812042236, + "num_tokens": 664297167.0, + "step": 25670 + }, + { + "epoch": 2.8191302437953, + "grad_norm": 2.1715824604034424, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7196294069290161, + "num_tokens": 664327436.0, + "step": 25671 + }, + { + "epoch": 2.8192400614979136, + "grad_norm": 2.3209757804870605, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7303779125213623, + "num_tokens": 664355794.0, + "step": 25672 + }, + { + "epoch": 2.8193498792005274, + "grad_norm": 2.24379301071167, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7077928781509399, + "num_tokens": 664385123.0, + "step": 25673 + }, + { + "epoch": 2.8194596969031407, + "grad_norm": 3.3356189727783203, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7690433859825134, + "num_tokens": 664399166.0, + "step": 25674 + }, + { + "epoch": 2.8195695146057544, + "grad_norm": 2.258521556854248, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7210990190505981, + "num_tokens": 664426734.0, + "step": 25675 + }, + { + "epoch": 2.819679332308368, + "grad_norm": 2.461794137954712, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7502461671829224, + "num_tokens": 664449383.0, + "step": 25676 + }, + { + "epoch": 2.8197891500109815, + "grad_norm": 2.185835361480713, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7382667660713196, + "num_tokens": 664478943.0, + "step": 25677 + }, + { + "epoch": 2.8198989677135953, + "grad_norm": 2.48580265045166, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7253886461257935, + "num_tokens": 664504458.0, + "step": 25678 + }, + { + "epoch": 2.820008785416209, + "grad_norm": 2.0776607990264893, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7195855379104614, + "num_tokens": 664535890.0, + "step": 25679 + }, + { + "epoch": 2.820118603118823, + "grad_norm": 2.5611510276794434, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7390909194946289, + "num_tokens": 664558345.0, + "step": 25680 + }, + { + "epoch": 2.8202284208214365, + "grad_norm": 2.498035192489624, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7356387972831726, + "num_tokens": 664580299.0, + "step": 25681 + }, + { + "epoch": 2.82033823852405, + "grad_norm": 2.8301830291748047, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7412897348403931, + "num_tokens": 664599077.0, + "step": 25682 + }, + { + "epoch": 2.8204480562266636, + "grad_norm": 2.4784791469573975, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7330272197723389, + "num_tokens": 664624121.0, + "step": 25683 + }, + { + "epoch": 2.8205578739292774, + "grad_norm": 1.9632023572921753, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7314087748527527, + "num_tokens": 664660951.0, + "step": 25684 + }, + { + "epoch": 2.820667691631891, + "grad_norm": 2.5164787769317627, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7556416988372803, + "num_tokens": 664684410.0, + "step": 25685 + }, + { + "epoch": 2.820777509334505, + "grad_norm": 2.5395400524139404, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7331873774528503, + "num_tokens": 664706851.0, + "step": 25686 + }, + { + "epoch": 2.820887327037118, + "grad_norm": 2.427612781524658, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.728496789932251, + "num_tokens": 664731863.0, + "step": 25687 + }, + { + "epoch": 2.820997144739732, + "grad_norm": 2.7264153957366943, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.738981306552887, + "num_tokens": 664752228.0, + "step": 25688 + }, + { + "epoch": 2.8211069624423457, + "grad_norm": 2.507338285446167, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7620171904563904, + "num_tokens": 664773933.0, + "step": 25689 + }, + { + "epoch": 2.8212167801449595, + "grad_norm": 2.189025640487671, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7299408912658691, + "num_tokens": 664802609.0, + "step": 25690 + }, + { + "epoch": 2.8213265978475732, + "grad_norm": 2.880929708480835, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7310291528701782, + "num_tokens": 664821200.0, + "step": 25691 + }, + { + "epoch": 2.8214364155501865, + "grad_norm": 2.3514695167541504, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7189176082611084, + "num_tokens": 664848426.0, + "step": 25692 + }, + { + "epoch": 2.8215462332528003, + "grad_norm": 2.218549966812134, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7341263294219971, + "num_tokens": 664874057.0, + "step": 25693 + }, + { + "epoch": 2.821656050955414, + "grad_norm": 2.6173696517944336, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7361299991607666, + "num_tokens": 664897031.0, + "step": 25694 + }, + { + "epoch": 2.821765868658028, + "grad_norm": 2.429603099822998, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7025125622749329, + "num_tokens": 664922382.0, + "step": 25695 + }, + { + "epoch": 2.8218756863606416, + "grad_norm": 2.398561477661133, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7323747873306274, + "num_tokens": 664946332.0, + "step": 25696 + }, + { + "epoch": 2.821985504063255, + "grad_norm": 2.4163010120391846, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7333012819290161, + "num_tokens": 664970614.0, + "step": 25697 + }, + { + "epoch": 2.8220953217658686, + "grad_norm": 2.3444972038269043, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7260594367980957, + "num_tokens": 664997642.0, + "step": 25698 + }, + { + "epoch": 2.8222051394684824, + "grad_norm": 2.172978401184082, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7450113296508789, + "num_tokens": 665025423.0, + "step": 25699 + }, + { + "epoch": 2.8223149571710957, + "grad_norm": 2.2872345447540283, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7268397808074951, + "num_tokens": 665053357.0, + "step": 25700 + }, + { + "epoch": 2.82242477487371, + "grad_norm": 2.6367673873901367, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7363834381103516, + "num_tokens": 665078191.0, + "step": 25701 + }, + { + "epoch": 2.822534592576323, + "grad_norm": 2.594320297241211, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7575473785400391, + "num_tokens": 665100123.0, + "step": 25702 + }, + { + "epoch": 2.822644410278937, + "grad_norm": 2.16170597076416, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.6966595649719238, + "num_tokens": 665130091.0, + "step": 25703 + }, + { + "epoch": 2.8227542279815507, + "grad_norm": 2.929936408996582, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7393950819969177, + "num_tokens": 665148247.0, + "step": 25704 + }, + { + "epoch": 2.822864045684164, + "grad_norm": 2.0781328678131104, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.71233069896698, + "num_tokens": 665178671.0, + "step": 25705 + }, + { + "epoch": 2.822973863386778, + "grad_norm": 2.1676530838012695, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7345081567764282, + "num_tokens": 665206942.0, + "step": 25706 + }, + { + "epoch": 2.8230836810893916, + "grad_norm": 2.7869486808776855, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7286961078643799, + "num_tokens": 665228016.0, + "step": 25707 + }, + { + "epoch": 2.8231934987920053, + "grad_norm": 2.4245598316192627, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7271358966827393, + "num_tokens": 665252564.0, + "step": 25708 + }, + { + "epoch": 2.823303316494619, + "grad_norm": 2.3656063079833984, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7153942584991455, + "num_tokens": 665279318.0, + "step": 25709 + }, + { + "epoch": 2.8234131341972324, + "grad_norm": 2.247124671936035, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.74196457862854, + "num_tokens": 665305386.0, + "step": 25710 + }, + { + "epoch": 2.823522951899846, + "grad_norm": 2.6161577701568604, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7739224433898926, + "num_tokens": 665327383.0, + "step": 25711 + }, + { + "epoch": 2.82363276960246, + "grad_norm": 2.152952194213867, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7170717716217041, + "num_tokens": 665356505.0, + "step": 25712 + }, + { + "epoch": 2.8237425873050737, + "grad_norm": 2.4342200756073, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7176365852355957, + "num_tokens": 665382325.0, + "step": 25713 + }, + { + "epoch": 2.8238524050076874, + "grad_norm": 2.274118423461914, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7438845634460449, + "num_tokens": 665408629.0, + "step": 25714 + }, + { + "epoch": 2.8239622227103007, + "grad_norm": 2.2669970989227295, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7243452072143555, + "num_tokens": 665436323.0, + "step": 25715 + }, + { + "epoch": 2.8240720404129145, + "grad_norm": 2.2283523082733154, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7720555067062378, + "num_tokens": 665463777.0, + "step": 25716 + }, + { + "epoch": 2.8241818581155282, + "grad_norm": 2.605025053024292, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7370826005935669, + "num_tokens": 665485836.0, + "step": 25717 + }, + { + "epoch": 2.824291675818142, + "grad_norm": 2.4298911094665527, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7249000072479248, + "num_tokens": 665510241.0, + "step": 25718 + }, + { + "epoch": 2.8244014935207558, + "grad_norm": 2.236868143081665, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7119995355606079, + "num_tokens": 665539111.0, + "step": 25719 + }, + { + "epoch": 2.824511311223369, + "grad_norm": 2.1280272006988525, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7112056016921997, + "num_tokens": 665572159.0, + "step": 25720 + }, + { + "epoch": 2.824621128925983, + "grad_norm": 2.259158134460449, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7194656133651733, + "num_tokens": 665600417.0, + "step": 25721 + }, + { + "epoch": 2.8247309466285966, + "grad_norm": 2.5126938819885254, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.724990963935852, + "num_tokens": 665625111.0, + "step": 25722 + }, + { + "epoch": 2.8248407643312103, + "grad_norm": 2.7159526348114014, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7357733249664307, + "num_tokens": 665645082.0, + "step": 25723 + }, + { + "epoch": 2.824950582033824, + "grad_norm": 2.1410231590270996, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7173059582710266, + "num_tokens": 665675922.0, + "step": 25724 + }, + { + "epoch": 2.8250603997364374, + "grad_norm": 2.408400297164917, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7468820810317993, + "num_tokens": 665701672.0, + "step": 25725 + }, + { + "epoch": 2.825170217439051, + "grad_norm": 2.416314125061035, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7411679625511169, + "num_tokens": 665726620.0, + "step": 25726 + }, + { + "epoch": 2.825280035141665, + "grad_norm": 2.3343985080718994, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7122172117233276, + "num_tokens": 665752886.0, + "step": 25727 + }, + { + "epoch": 2.8253898528442782, + "grad_norm": 2.4411075115203857, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7277637720108032, + "num_tokens": 665776049.0, + "step": 25728 + }, + { + "epoch": 2.825499670546892, + "grad_norm": 2.3706274032592773, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7371866703033447, + "num_tokens": 665801166.0, + "step": 25729 + }, + { + "epoch": 2.8256094882495058, + "grad_norm": 2.227084159851074, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7087529897689819, + "num_tokens": 665829909.0, + "step": 25730 + }, + { + "epoch": 2.8257193059521195, + "grad_norm": 2.7169079780578613, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.738996684551239, + "num_tokens": 665851708.0, + "step": 25731 + }, + { + "epoch": 2.8258291236547333, + "grad_norm": 2.4509215354919434, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.725213348865509, + "num_tokens": 665875189.0, + "step": 25732 + }, + { + "epoch": 2.8259389413573466, + "grad_norm": 2.4714107513427734, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7249427437782288, + "num_tokens": 665897904.0, + "step": 25733 + }, + { + "epoch": 2.8260487590599603, + "grad_norm": 2.4961836338043213, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7525399923324585, + "num_tokens": 665921091.0, + "step": 25734 + }, + { + "epoch": 2.826158576762574, + "grad_norm": 2.180497646331787, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7494467496871948, + "num_tokens": 665948538.0, + "step": 25735 + }, + { + "epoch": 2.826268394465188, + "grad_norm": 2.4374756813049316, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7200877070426941, + "num_tokens": 665974966.0, + "step": 25736 + }, + { + "epoch": 2.8263782121678016, + "grad_norm": 2.7537600994110107, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7396751046180725, + "num_tokens": 665994760.0, + "step": 25737 + }, + { + "epoch": 2.826488029870415, + "grad_norm": 2.2991342544555664, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7274779081344604, + "num_tokens": 666021963.0, + "step": 25738 + }, + { + "epoch": 2.8265978475730287, + "grad_norm": 2.137462615966797, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7341912388801575, + "num_tokens": 666050709.0, + "step": 25739 + }, + { + "epoch": 2.8267076652756424, + "grad_norm": 2.1045024394989014, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.719009280204773, + "num_tokens": 666080470.0, + "step": 25740 + }, + { + "epoch": 2.826817482978256, + "grad_norm": 2.317903757095337, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7243776321411133, + "num_tokens": 666105655.0, + "step": 25741 + }, + { + "epoch": 2.82692730068087, + "grad_norm": 2.531607151031494, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7353441119194031, + "num_tokens": 666129224.0, + "step": 25742 + }, + { + "epoch": 2.8270371183834833, + "grad_norm": 2.3466925621032715, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7343953251838684, + "num_tokens": 666153247.0, + "step": 25743 + }, + { + "epoch": 2.827146936086097, + "grad_norm": 2.6212410926818848, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7316836714744568, + "num_tokens": 666175056.0, + "step": 25744 + }, + { + "epoch": 2.8272567537887108, + "grad_norm": 2.167714834213257, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7238268852233887, + "num_tokens": 666207477.0, + "step": 25745 + }, + { + "epoch": 2.8273665714913245, + "grad_norm": 2.422255039215088, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7246284484863281, + "num_tokens": 666233368.0, + "step": 25746 + }, + { + "epoch": 2.8274763891939383, + "grad_norm": 2.464472532272339, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7300600409507751, + "num_tokens": 666258159.0, + "step": 25747 + }, + { + "epoch": 2.8275862068965516, + "grad_norm": 2.41705060005188, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7273036241531372, + "num_tokens": 666284219.0, + "step": 25748 + }, + { + "epoch": 2.8276960245991654, + "grad_norm": 2.6214637756347656, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7380982637405396, + "num_tokens": 666305589.0, + "step": 25749 + }, + { + "epoch": 2.827805842301779, + "grad_norm": 2.476191759109497, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7153114080429077, + "num_tokens": 666330272.0, + "step": 25750 + }, + { + "epoch": 2.827915660004393, + "grad_norm": 2.361365795135498, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7350555658340454, + "num_tokens": 666356032.0, + "step": 25751 + }, + { + "epoch": 2.8280254777070066, + "grad_norm": 3.0828983783721924, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7538868188858032, + "num_tokens": 666372562.0, + "step": 25752 + }, + { + "epoch": 2.82813529540962, + "grad_norm": 2.7019248008728027, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7354550957679749, + "num_tokens": 666395419.0, + "step": 25753 + }, + { + "epoch": 2.8282451131122337, + "grad_norm": 2.2950284481048584, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7648118734359741, + "num_tokens": 666420506.0, + "step": 25754 + }, + { + "epoch": 2.8283549308148475, + "grad_norm": 2.1975479125976562, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7107134461402893, + "num_tokens": 666450363.0, + "step": 25755 + }, + { + "epoch": 2.8284647485174608, + "grad_norm": 2.5440635681152344, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7419388294219971, + "num_tokens": 666472731.0, + "step": 25756 + }, + { + "epoch": 2.8285745662200745, + "grad_norm": 2.3987584114074707, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.723197340965271, + "num_tokens": 666498792.0, + "step": 25757 + }, + { + "epoch": 2.8286843839226883, + "grad_norm": 2.296600103378296, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7190722823143005, + "num_tokens": 666526752.0, + "step": 25758 + }, + { + "epoch": 2.828794201625302, + "grad_norm": 2.5579240322113037, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7589805126190186, + "num_tokens": 666548945.0, + "step": 25759 + }, + { + "epoch": 2.828904019327916, + "grad_norm": 2.659680128097534, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7233823537826538, + "num_tokens": 666569835.0, + "step": 25760 + }, + { + "epoch": 2.829013837030529, + "grad_norm": 2.496217966079712, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7346433997154236, + "num_tokens": 666593958.0, + "step": 25761 + }, + { + "epoch": 2.829123654733143, + "grad_norm": 2.4726459980010986, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7246478796005249, + "num_tokens": 666618422.0, + "step": 25762 + }, + { + "epoch": 2.8292334724357566, + "grad_norm": 2.72475004196167, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7534587383270264, + "num_tokens": 666639019.0, + "step": 25763 + }, + { + "epoch": 2.8293432901383704, + "grad_norm": 2.505540609359741, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7315754294395447, + "num_tokens": 666662132.0, + "step": 25764 + }, + { + "epoch": 2.829453107840984, + "grad_norm": 2.802196741104126, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7428719997406006, + "num_tokens": 666680745.0, + "step": 25765 + }, + { + "epoch": 2.8295629255435975, + "grad_norm": 2.2827608585357666, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.745945394039154, + "num_tokens": 666708013.0, + "step": 25766 + }, + { + "epoch": 2.829672743246211, + "grad_norm": 2.2362990379333496, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6947228908538818, + "num_tokens": 666738869.0, + "step": 25767 + }, + { + "epoch": 2.829782560948825, + "grad_norm": 2.3045642375946045, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.735817015171051, + "num_tokens": 666764646.0, + "step": 25768 + }, + { + "epoch": 2.8298923786514387, + "grad_norm": 2.291525363922119, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7141858339309692, + "num_tokens": 666792890.0, + "step": 25769 + }, + { + "epoch": 2.8300021963540525, + "grad_norm": 2.3704495429992676, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.741241455078125, + "num_tokens": 666817196.0, + "step": 25770 + }, + { + "epoch": 2.830112014056666, + "grad_norm": 2.663032054901123, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7729086875915527, + "num_tokens": 666837965.0, + "step": 25771 + }, + { + "epoch": 2.8302218317592795, + "grad_norm": 2.4660239219665527, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7255356311798096, + "num_tokens": 666861412.0, + "step": 25772 + }, + { + "epoch": 2.8303316494618933, + "grad_norm": 2.65057110786438, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7260798215866089, + "num_tokens": 666883947.0, + "step": 25773 + }, + { + "epoch": 2.830441467164507, + "grad_norm": 2.762298107147217, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7515125274658203, + "num_tokens": 666904341.0, + "step": 25774 + }, + { + "epoch": 2.830551284867121, + "grad_norm": 2.4298360347747803, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7386369705200195, + "num_tokens": 666926188.0, + "step": 25775 + }, + { + "epoch": 2.830661102569734, + "grad_norm": 2.5069050788879395, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7330979704856873, + "num_tokens": 666950234.0, + "step": 25776 + }, + { + "epoch": 2.830770920272348, + "grad_norm": 2.3246965408325195, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7244029641151428, + "num_tokens": 666976412.0, + "step": 25777 + }, + { + "epoch": 2.8308807379749616, + "grad_norm": 2.2129452228546143, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7061273455619812, + "num_tokens": 667005457.0, + "step": 25778 + }, + { + "epoch": 2.830990555677575, + "grad_norm": 2.6075644493103027, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7379378080368042, + "num_tokens": 667028380.0, + "step": 25779 + }, + { + "epoch": 2.831100373380189, + "grad_norm": 2.042983293533325, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7114431262016296, + "num_tokens": 667059785.0, + "step": 25780 + }, + { + "epoch": 2.8312101910828025, + "grad_norm": 2.339535713195801, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7426326274871826, + "num_tokens": 667083512.0, + "step": 25781 + }, + { + "epoch": 2.8313200087854162, + "grad_norm": 2.510606527328491, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7636445164680481, + "num_tokens": 667104526.0, + "step": 25782 + }, + { + "epoch": 2.83142982648803, + "grad_norm": 2.736414909362793, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7213466167449951, + "num_tokens": 667125859.0, + "step": 25783 + }, + { + "epoch": 2.8315396441906433, + "grad_norm": 2.3630125522613525, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7350915670394897, + "num_tokens": 667151027.0, + "step": 25784 + }, + { + "epoch": 2.831649461893257, + "grad_norm": 1.8913112878799438, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7491156458854675, + "num_tokens": 667186120.0, + "step": 25785 + }, + { + "epoch": 2.831759279595871, + "grad_norm": 2.6756045818328857, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7250332832336426, + "num_tokens": 667209178.0, + "step": 25786 + }, + { + "epoch": 2.8318690972984846, + "grad_norm": 2.4307053089141846, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7440499663352966, + "num_tokens": 667233824.0, + "step": 25787 + }, + { + "epoch": 2.8319789150010983, + "grad_norm": 2.679708480834961, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7459039688110352, + "num_tokens": 667255284.0, + "step": 25788 + }, + { + "epoch": 2.8320887327037116, + "grad_norm": 2.24686861038208, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.71324622631073, + "num_tokens": 667284434.0, + "step": 25789 + }, + { + "epoch": 2.8321985504063254, + "grad_norm": 2.477538585662842, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.729473888874054, + "num_tokens": 667307963.0, + "step": 25790 + }, + { + "epoch": 2.832308368108939, + "grad_norm": 2.6697194576263428, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7350386381149292, + "num_tokens": 667330400.0, + "step": 25791 + }, + { + "epoch": 2.832418185811553, + "grad_norm": 2.446413278579712, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.717519998550415, + "num_tokens": 667353480.0, + "step": 25792 + }, + { + "epoch": 2.8325280035141667, + "grad_norm": 2.436789035797119, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.743205726146698, + "num_tokens": 667378113.0, + "step": 25793 + }, + { + "epoch": 2.83263782121678, + "grad_norm": 2.236645221710205, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7553484439849854, + "num_tokens": 667405448.0, + "step": 25794 + }, + { + "epoch": 2.8327476389193937, + "grad_norm": 2.3724539279937744, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7468146085739136, + "num_tokens": 667431979.0, + "step": 25795 + }, + { + "epoch": 2.8328574566220075, + "grad_norm": 2.5479750633239746, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7468550205230713, + "num_tokens": 667454702.0, + "step": 25796 + }, + { + "epoch": 2.8329672743246213, + "grad_norm": 2.7472643852233887, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7261242866516113, + "num_tokens": 667476501.0, + "step": 25797 + }, + { + "epoch": 2.833077092027235, + "grad_norm": 2.3810951709747314, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7301959991455078, + "num_tokens": 667502412.0, + "step": 25798 + }, + { + "epoch": 2.8331869097298483, + "grad_norm": 2.7681920528411865, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7563339471817017, + "num_tokens": 667521563.0, + "step": 25799 + }, + { + "epoch": 2.833296727432462, + "grad_norm": 2.363736629486084, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7302349209785461, + "num_tokens": 667548691.0, + "step": 25800 + }, + { + "epoch": 2.833406545135076, + "grad_norm": 2.3051233291625977, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7221847772598267, + "num_tokens": 667575700.0, + "step": 25801 + }, + { + "epoch": 2.8335163628376896, + "grad_norm": 2.4570729732513428, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7342585325241089, + "num_tokens": 667599917.0, + "step": 25802 + }, + { + "epoch": 2.8336261805403034, + "grad_norm": 2.194737434387207, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7151023149490356, + "num_tokens": 667630484.0, + "step": 25803 + }, + { + "epoch": 2.8337359982429167, + "grad_norm": 2.0465846061706543, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7071868777275085, + "num_tokens": 667662177.0, + "step": 25804 + }, + { + "epoch": 2.8338458159455304, + "grad_norm": 2.37099289894104, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7166353464126587, + "num_tokens": 667687948.0, + "step": 25805 + }, + { + "epoch": 2.833955633648144, + "grad_norm": 2.461137056350708, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7317787408828735, + "num_tokens": 667711620.0, + "step": 25806 + }, + { + "epoch": 2.8340654513507575, + "grad_norm": 2.547327756881714, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7561028599739075, + "num_tokens": 667733656.0, + "step": 25807 + }, + { + "epoch": 2.8341752690533712, + "grad_norm": 2.0319983959198, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7239487171173096, + "num_tokens": 667766584.0, + "step": 25808 + }, + { + "epoch": 2.834285086755985, + "grad_norm": 2.2969207763671875, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7156492471694946, + "num_tokens": 667795665.0, + "step": 25809 + }, + { + "epoch": 2.8343949044585988, + "grad_norm": 2.4970791339874268, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7348383665084839, + "num_tokens": 667820599.0, + "step": 25810 + }, + { + "epoch": 2.8345047221612125, + "grad_norm": 2.2132365703582764, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7384524941444397, + "num_tokens": 667849042.0, + "step": 25811 + }, + { + "epoch": 2.834614539863826, + "grad_norm": 2.1697938442230225, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7521251440048218, + "num_tokens": 667877786.0, + "step": 25812 + }, + { + "epoch": 2.8347243575664396, + "grad_norm": 2.7148547172546387, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7503846883773804, + "num_tokens": 667900474.0, + "step": 25813 + }, + { + "epoch": 2.8348341752690533, + "grad_norm": 2.095388650894165, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.739702582359314, + "num_tokens": 667930778.0, + "step": 25814 + }, + { + "epoch": 2.834943992971667, + "grad_norm": 2.3333091735839844, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.753791332244873, + "num_tokens": 667956985.0, + "step": 25815 + }, + { + "epoch": 2.835053810674281, + "grad_norm": 2.6804184913635254, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7600652575492859, + "num_tokens": 667976498.0, + "step": 25816 + }, + { + "epoch": 2.835163628376894, + "grad_norm": 2.567223310470581, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7030841112136841, + "num_tokens": 667999562.0, + "step": 25817 + }, + { + "epoch": 2.835273446079508, + "grad_norm": 2.422990083694458, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7277611494064331, + "num_tokens": 668024673.0, + "step": 25818 + }, + { + "epoch": 2.8353832637821217, + "grad_norm": 2.732513189315796, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7503407001495361, + "num_tokens": 668044984.0, + "step": 25819 + }, + { + "epoch": 2.8354930814847354, + "grad_norm": 2.879716396331787, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.749485433101654, + "num_tokens": 668062676.0, + "step": 25820 + }, + { + "epoch": 2.835602899187349, + "grad_norm": 2.0413265228271484, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.715955376625061, + "num_tokens": 668095964.0, + "step": 25821 + }, + { + "epoch": 2.8357127168899625, + "grad_norm": 2.3772666454315186, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7295109033584595, + "num_tokens": 668120610.0, + "step": 25822 + }, + { + "epoch": 2.8358225345925763, + "grad_norm": 2.139162540435791, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.70804762840271, + "num_tokens": 668154666.0, + "step": 25823 + }, + { + "epoch": 2.83593235229519, + "grad_norm": 2.5688652992248535, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7166432738304138, + "num_tokens": 668179455.0, + "step": 25824 + }, + { + "epoch": 2.836042169997804, + "grad_norm": 2.3820111751556396, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.746063232421875, + "num_tokens": 668204743.0, + "step": 25825 + }, + { + "epoch": 2.8361519877004175, + "grad_norm": 2.405557870864868, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7321897745132446, + "num_tokens": 668229429.0, + "step": 25826 + }, + { + "epoch": 2.836261805403031, + "grad_norm": 2.4754326343536377, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7455970644950867, + "num_tokens": 668252642.0, + "step": 25827 + }, + { + "epoch": 2.8363716231056446, + "grad_norm": 2.493213415145874, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.733285665512085, + "num_tokens": 668274927.0, + "step": 25828 + }, + { + "epoch": 2.8364814408082584, + "grad_norm": 2.5321998596191406, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7320233583450317, + "num_tokens": 668300529.0, + "step": 25829 + }, + { + "epoch": 2.8365912585108717, + "grad_norm": 2.4921863079071045, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7462965846061707, + "num_tokens": 668323105.0, + "step": 25830 + }, + { + "epoch": 2.836701076213486, + "grad_norm": 2.7616019248962402, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7349819540977478, + "num_tokens": 668343240.0, + "step": 25831 + }, + { + "epoch": 2.836810893916099, + "grad_norm": 2.436337947845459, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7242185473442078, + "num_tokens": 668366235.0, + "step": 25832 + }, + { + "epoch": 2.836920711618713, + "grad_norm": 2.451174020767212, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7113006711006165, + "num_tokens": 668390790.0, + "step": 25833 + }, + { + "epoch": 2.8370305293213267, + "grad_norm": 2.368696928024292, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.72941654920578, + "num_tokens": 668417367.0, + "step": 25834 + }, + { + "epoch": 2.83714034702394, + "grad_norm": 2.9741227626800537, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7433381080627441, + "num_tokens": 668436434.0, + "step": 25835 + }, + { + "epoch": 2.837250164726554, + "grad_norm": 2.3411102294921875, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7464321851730347, + "num_tokens": 668462158.0, + "step": 25836 + }, + { + "epoch": 2.8373599824291675, + "grad_norm": 2.4902760982513428, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7272329330444336, + "num_tokens": 668484016.0, + "step": 25837 + }, + { + "epoch": 2.8374698001317813, + "grad_norm": 2.392254590988159, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7317491173744202, + "num_tokens": 668509697.0, + "step": 25838 + }, + { + "epoch": 2.837579617834395, + "grad_norm": 2.4822795391082764, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7318286895751953, + "num_tokens": 668534839.0, + "step": 25839 + }, + { + "epoch": 2.8376894355370084, + "grad_norm": 2.544687032699585, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7308530807495117, + "num_tokens": 668558902.0, + "step": 25840 + }, + { + "epoch": 2.837799253239622, + "grad_norm": 2.258523464202881, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7168310880661011, + "num_tokens": 668589644.0, + "step": 25841 + }, + { + "epoch": 2.837909070942236, + "grad_norm": 2.019676923751831, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7193024754524231, + "num_tokens": 668623296.0, + "step": 25842 + }, + { + "epoch": 2.8380188886448496, + "grad_norm": 2.1303672790527344, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7134701013565063, + "num_tokens": 668652015.0, + "step": 25843 + }, + { + "epoch": 2.8381287063474634, + "grad_norm": 2.244487762451172, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7274784445762634, + "num_tokens": 668681762.0, + "step": 25844 + }, + { + "epoch": 2.8382385240500767, + "grad_norm": 2.2757694721221924, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7538739442825317, + "num_tokens": 668706177.0, + "step": 25845 + }, + { + "epoch": 2.8383483417526905, + "grad_norm": 2.8284735679626465, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7624391317367554, + "num_tokens": 668724696.0, + "step": 25846 + }, + { + "epoch": 2.838458159455304, + "grad_norm": 2.424206018447876, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.743216872215271, + "num_tokens": 668748726.0, + "step": 25847 + }, + { + "epoch": 2.838567977157918, + "grad_norm": 2.5499353408813477, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7493937015533447, + "num_tokens": 668771087.0, + "step": 25848 + }, + { + "epoch": 2.8386777948605317, + "grad_norm": 2.482548475265503, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7230144739151001, + "num_tokens": 668796119.0, + "step": 25849 + }, + { + "epoch": 2.838787612563145, + "grad_norm": 2.3639211654663086, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7322115898132324, + "num_tokens": 668821536.0, + "step": 25850 + }, + { + "epoch": 2.838897430265759, + "grad_norm": 2.368849039077759, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7320558428764343, + "num_tokens": 668849118.0, + "step": 25851 + }, + { + "epoch": 2.8390072479683726, + "grad_norm": 2.2916343212127686, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7105458378791809, + "num_tokens": 668878656.0, + "step": 25852 + }, + { + "epoch": 2.8391170656709863, + "grad_norm": 2.6825993061065674, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7663260698318481, + "num_tokens": 668897345.0, + "step": 25853 + }, + { + "epoch": 2.8392268833736, + "grad_norm": 2.4760239124298096, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7281705141067505, + "num_tokens": 668921684.0, + "step": 25854 + }, + { + "epoch": 2.8393367010762134, + "grad_norm": 2.2675719261169434, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7305089235305786, + "num_tokens": 668948081.0, + "step": 25855 + }, + { + "epoch": 2.839446518778827, + "grad_norm": 2.103862762451172, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7052294015884399, + "num_tokens": 668978585.0, + "step": 25856 + }, + { + "epoch": 2.839556336481441, + "grad_norm": 2.1256704330444336, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7121811509132385, + "num_tokens": 669009970.0, + "step": 25857 + }, + { + "epoch": 2.839666154184054, + "grad_norm": 2.3629627227783203, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7240278124809265, + "num_tokens": 669036112.0, + "step": 25858 + }, + { + "epoch": 2.839775971886668, + "grad_norm": 2.3179280757904053, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7391256093978882, + "num_tokens": 669062067.0, + "step": 25859 + }, + { + "epoch": 2.8398857895892817, + "grad_norm": 2.3072357177734375, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7467398047447205, + "num_tokens": 669089372.0, + "step": 25860 + }, + { + "epoch": 2.8399956072918955, + "grad_norm": 2.291597604751587, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7374301552772522, + "num_tokens": 669115387.0, + "step": 25861 + }, + { + "epoch": 2.8401054249945092, + "grad_norm": 2.452155590057373, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7235066294670105, + "num_tokens": 669139887.0, + "step": 25862 + }, + { + "epoch": 2.8402152426971226, + "grad_norm": 2.2595884799957275, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7095215320587158, + "num_tokens": 669169332.0, + "step": 25863 + }, + { + "epoch": 2.8403250603997363, + "grad_norm": 2.6503939628601074, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7271592617034912, + "num_tokens": 669193611.0, + "step": 25864 + }, + { + "epoch": 2.84043487810235, + "grad_norm": 2.443004608154297, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7504431009292603, + "num_tokens": 669217802.0, + "step": 25865 + }, + { + "epoch": 2.840544695804964, + "grad_norm": 2.519214391708374, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7498241662979126, + "num_tokens": 669241755.0, + "step": 25866 + }, + { + "epoch": 2.8406545135075776, + "grad_norm": 2.763927459716797, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7081411480903625, + "num_tokens": 669263832.0, + "step": 25867 + }, + { + "epoch": 2.840764331210191, + "grad_norm": 2.4612629413604736, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7430119514465332, + "num_tokens": 669286704.0, + "step": 25868 + }, + { + "epoch": 2.8408741489128047, + "grad_norm": 2.3531038761138916, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7326470613479614, + "num_tokens": 669313161.0, + "step": 25869 + }, + { + "epoch": 2.8409839666154184, + "grad_norm": 2.7673230171203613, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7195643186569214, + "num_tokens": 669334300.0, + "step": 25870 + }, + { + "epoch": 2.841093784318032, + "grad_norm": 2.232966423034668, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.732943058013916, + "num_tokens": 669362793.0, + "step": 25871 + }, + { + "epoch": 2.841203602020646, + "grad_norm": 2.219411611557007, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7304286956787109, + "num_tokens": 669391541.0, + "step": 25872 + }, + { + "epoch": 2.8413134197232592, + "grad_norm": 2.138784646987915, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7382459044456482, + "num_tokens": 669418882.0, + "step": 25873 + }, + { + "epoch": 2.841423237425873, + "grad_norm": 2.6252822875976562, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7196098566055298, + "num_tokens": 669441562.0, + "step": 25874 + }, + { + "epoch": 2.8415330551284868, + "grad_norm": 2.28424334526062, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7058824896812439, + "num_tokens": 669468471.0, + "step": 25875 + }, + { + "epoch": 2.8416428728311005, + "grad_norm": 2.482255458831787, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7175213098526001, + "num_tokens": 669493183.0, + "step": 25876 + }, + { + "epoch": 2.8417526905337143, + "grad_norm": 2.5927987098693848, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7287365198135376, + "num_tokens": 669516414.0, + "step": 25877 + }, + { + "epoch": 2.8418625082363276, + "grad_norm": 2.3082616329193115, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7141503691673279, + "num_tokens": 669542733.0, + "step": 25878 + }, + { + "epoch": 2.8419723259389413, + "grad_norm": 2.500563859939575, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7445361614227295, + "num_tokens": 669566008.0, + "step": 25879 + }, + { + "epoch": 2.842082143641555, + "grad_norm": 2.4904332160949707, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7329731583595276, + "num_tokens": 669593634.0, + "step": 25880 + }, + { + "epoch": 2.8421919613441684, + "grad_norm": 2.2578344345092773, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7374323010444641, + "num_tokens": 669621603.0, + "step": 25881 + }, + { + "epoch": 2.8423017790467826, + "grad_norm": 2.4334824085235596, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7390854954719543, + "num_tokens": 669644770.0, + "step": 25882 + }, + { + "epoch": 2.842411596749396, + "grad_norm": 2.419856548309326, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.735472559928894, + "num_tokens": 669669975.0, + "step": 25883 + }, + { + "epoch": 2.8425214144520097, + "grad_norm": 2.4919025897979736, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7072954177856445, + "num_tokens": 669693620.0, + "step": 25884 + }, + { + "epoch": 2.8426312321546234, + "grad_norm": 2.2601234912872314, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7263450622558594, + "num_tokens": 669720800.0, + "step": 25885 + }, + { + "epoch": 2.8427410498572367, + "grad_norm": 2.562488317489624, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7519887685775757, + "num_tokens": 669743147.0, + "step": 25886 + }, + { + "epoch": 2.8428508675598505, + "grad_norm": 2.248410224914551, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7243733406066895, + "num_tokens": 669773579.0, + "step": 25887 + }, + { + "epoch": 2.8429606852624643, + "grad_norm": 2.294243335723877, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7229527831077576, + "num_tokens": 669801002.0, + "step": 25888 + }, + { + "epoch": 2.843070502965078, + "grad_norm": 2.401505947113037, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7517528533935547, + "num_tokens": 669823367.0, + "step": 25889 + }, + { + "epoch": 2.8431803206676918, + "grad_norm": 2.108839988708496, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7306579351425171, + "num_tokens": 669853913.0, + "step": 25890 + }, + { + "epoch": 2.843290138370305, + "grad_norm": 2.1242504119873047, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7181931734085083, + "num_tokens": 669885033.0, + "step": 25891 + }, + { + "epoch": 2.843399956072919, + "grad_norm": 2.225290060043335, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7233319282531738, + "num_tokens": 669915585.0, + "step": 25892 + }, + { + "epoch": 2.8435097737755326, + "grad_norm": 2.415205240249634, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7246636152267456, + "num_tokens": 669940756.0, + "step": 25893 + }, + { + "epoch": 2.8436195914781464, + "grad_norm": 2.68721342086792, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7428116798400879, + "num_tokens": 669961158.0, + "step": 25894 + }, + { + "epoch": 2.84372940918076, + "grad_norm": 2.458542585372925, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7355964183807373, + "num_tokens": 669984561.0, + "step": 25895 + }, + { + "epoch": 2.8438392268833734, + "grad_norm": 2.4896974563598633, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7325549721717834, + "num_tokens": 670008493.0, + "step": 25896 + }, + { + "epoch": 2.843949044585987, + "grad_norm": 2.337451934814453, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7176989316940308, + "num_tokens": 670033800.0, + "step": 25897 + }, + { + "epoch": 2.844058862288601, + "grad_norm": 2.2909750938415527, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7278923988342285, + "num_tokens": 670060141.0, + "step": 25898 + }, + { + "epoch": 2.8441686799912147, + "grad_norm": 2.3167057037353516, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7415562868118286, + "num_tokens": 670087456.0, + "step": 25899 + }, + { + "epoch": 2.8442784976938285, + "grad_norm": 2.091615676879883, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7327682375907898, + "num_tokens": 670118073.0, + "step": 25900 + }, + { + "epoch": 2.8443883153964418, + "grad_norm": 2.4249327182769775, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7339533567428589, + "num_tokens": 670142476.0, + "step": 25901 + }, + { + "epoch": 2.8444981330990555, + "grad_norm": 2.3016061782836914, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7331796884536743, + "num_tokens": 670168641.0, + "step": 25902 + }, + { + "epoch": 2.8446079508016693, + "grad_norm": 2.1871447563171387, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7233359217643738, + "num_tokens": 670201852.0, + "step": 25903 + }, + { + "epoch": 2.844717768504283, + "grad_norm": 2.6062092781066895, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7473973631858826, + "num_tokens": 670224372.0, + "step": 25904 + }, + { + "epoch": 2.844827586206897, + "grad_norm": 2.327793598175049, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7310324907302856, + "num_tokens": 670252213.0, + "step": 25905 + }, + { + "epoch": 2.84493740390951, + "grad_norm": 2.4401185512542725, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.715034008026123, + "num_tokens": 670280173.0, + "step": 25906 + }, + { + "epoch": 2.845047221612124, + "grad_norm": 2.1720683574676514, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7488967776298523, + "num_tokens": 670310822.0, + "step": 25907 + }, + { + "epoch": 2.8451570393147376, + "grad_norm": 2.0832431316375732, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7377454042434692, + "num_tokens": 670338498.0, + "step": 25908 + }, + { + "epoch": 2.845266857017351, + "grad_norm": 2.5909175872802734, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7596107721328735, + "num_tokens": 670360003.0, + "step": 25909 + }, + { + "epoch": 2.8453766747199647, + "grad_norm": 2.1586098670959473, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.723482072353363, + "num_tokens": 670392819.0, + "step": 25910 + }, + { + "epoch": 2.8454864924225785, + "grad_norm": 2.2780590057373047, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7222023606300354, + "num_tokens": 670420464.0, + "step": 25911 + }, + { + "epoch": 2.845596310125192, + "grad_norm": 2.159945249557495, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.707808256149292, + "num_tokens": 670451696.0, + "step": 25912 + }, + { + "epoch": 2.845706127827806, + "grad_norm": 2.453247308731079, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7266706228256226, + "num_tokens": 670480806.0, + "step": 25913 + }, + { + "epoch": 2.8458159455304193, + "grad_norm": 2.740562915802002, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7442071437835693, + "num_tokens": 670502025.0, + "step": 25914 + }, + { + "epoch": 2.845925763233033, + "grad_norm": 2.153446912765503, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7283737659454346, + "num_tokens": 670533258.0, + "step": 25915 + }, + { + "epoch": 2.846035580935647, + "grad_norm": 2.359145402908325, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7324630618095398, + "num_tokens": 670560018.0, + "step": 25916 + }, + { + "epoch": 2.8461453986382605, + "grad_norm": 2.2225286960601807, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.6998416185379028, + "num_tokens": 670590028.0, + "step": 25917 + }, + { + "epoch": 2.8462552163408743, + "grad_norm": 2.377708911895752, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.748214840888977, + "num_tokens": 670615101.0, + "step": 25918 + }, + { + "epoch": 2.8463650340434876, + "grad_norm": 2.150541067123413, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7479680776596069, + "num_tokens": 670646292.0, + "step": 25919 + }, + { + "epoch": 2.8464748517461014, + "grad_norm": 2.5288984775543213, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7367016673088074, + "num_tokens": 670667824.0, + "step": 25920 + }, + { + "epoch": 2.846584669448715, + "grad_norm": 2.628229856491089, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7295970320701599, + "num_tokens": 670691045.0, + "step": 25921 + }, + { + "epoch": 2.846694487151329, + "grad_norm": 2.103834867477417, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7259595394134521, + "num_tokens": 670722992.0, + "step": 25922 + }, + { + "epoch": 2.8468043048539426, + "grad_norm": 2.1702568531036377, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7277942895889282, + "num_tokens": 670751400.0, + "step": 25923 + }, + { + "epoch": 2.846914122556556, + "grad_norm": 2.195066213607788, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7097835540771484, + "num_tokens": 670780341.0, + "step": 25924 + }, + { + "epoch": 2.8470239402591697, + "grad_norm": 2.1539695262908936, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.723839282989502, + "num_tokens": 670809210.0, + "step": 25925 + }, + { + "epoch": 2.8471337579617835, + "grad_norm": 2.215932846069336, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7231053709983826, + "num_tokens": 670837346.0, + "step": 25926 + }, + { + "epoch": 2.8472435756643972, + "grad_norm": 2.078362226486206, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7442702651023865, + "num_tokens": 670866039.0, + "step": 25927 + }, + { + "epoch": 2.847353393367011, + "grad_norm": 2.5011849403381348, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7297683358192444, + "num_tokens": 670889048.0, + "step": 25928 + }, + { + "epoch": 2.8474632110696243, + "grad_norm": 2.2161929607391357, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7061201930046082, + "num_tokens": 670918102.0, + "step": 25929 + }, + { + "epoch": 2.847573028772238, + "grad_norm": 2.354379415512085, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7323333024978638, + "num_tokens": 670944479.0, + "step": 25930 + }, + { + "epoch": 2.847682846474852, + "grad_norm": 2.3515350818634033, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7346829175949097, + "num_tokens": 670972970.0, + "step": 25931 + }, + { + "epoch": 2.8477926641774656, + "grad_norm": 2.5700559616088867, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7206858396530151, + "num_tokens": 670997133.0, + "step": 25932 + }, + { + "epoch": 2.8479024818800793, + "grad_norm": 2.491637945175171, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7275898456573486, + "num_tokens": 671023043.0, + "step": 25933 + }, + { + "epoch": 2.8480122995826926, + "grad_norm": 2.6191792488098145, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7278192043304443, + "num_tokens": 671046145.0, + "step": 25934 + }, + { + "epoch": 2.8481221172853064, + "grad_norm": 2.249368906021118, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7100151777267456, + "num_tokens": 671073990.0, + "step": 25935 + }, + { + "epoch": 2.84823193498792, + "grad_norm": 2.470719814300537, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7464562058448792, + "num_tokens": 671096700.0, + "step": 25936 + }, + { + "epoch": 2.8483417526905335, + "grad_norm": 2.328589916229248, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7226863503456116, + "num_tokens": 671121925.0, + "step": 25937 + }, + { + "epoch": 2.8484515703931472, + "grad_norm": 2.2263636589050293, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7324304580688477, + "num_tokens": 671149683.0, + "step": 25938 + }, + { + "epoch": 2.848561388095761, + "grad_norm": 2.404332399368286, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.730587363243103, + "num_tokens": 671174455.0, + "step": 25939 + }, + { + "epoch": 2.8486712057983747, + "grad_norm": 2.474984645843506, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7541991472244263, + "num_tokens": 671197544.0, + "step": 25940 + }, + { + "epoch": 2.8487810235009885, + "grad_norm": 2.2033050060272217, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7138848304748535, + "num_tokens": 671227099.0, + "step": 25941 + }, + { + "epoch": 2.848890841203602, + "grad_norm": 2.7578206062316895, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7518391609191895, + "num_tokens": 671247278.0, + "step": 25942 + }, + { + "epoch": 2.8490006589062156, + "grad_norm": 2.7255373001098633, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7259358167648315, + "num_tokens": 671270021.0, + "step": 25943 + }, + { + "epoch": 2.8491104766088293, + "grad_norm": 2.3186001777648926, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7457330822944641, + "num_tokens": 671297070.0, + "step": 25944 + }, + { + "epoch": 2.849220294311443, + "grad_norm": 2.176652193069458, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7205716967582703, + "num_tokens": 671327388.0, + "step": 25945 + }, + { + "epoch": 2.849330112014057, + "grad_norm": 2.290858268737793, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7268654108047485, + "num_tokens": 671357515.0, + "step": 25946 + }, + { + "epoch": 2.84943992971667, + "grad_norm": 2.309148073196411, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7435191869735718, + "num_tokens": 671383917.0, + "step": 25947 + }, + { + "epoch": 2.849549747419284, + "grad_norm": 2.608041286468506, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.737700879573822, + "num_tokens": 671405249.0, + "step": 25948 + }, + { + "epoch": 2.8496595651218977, + "grad_norm": 2.3637800216674805, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7184426784515381, + "num_tokens": 671434360.0, + "step": 25949 + }, + { + "epoch": 2.8497693828245114, + "grad_norm": 2.2017877101898193, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7288721799850464, + "num_tokens": 671463945.0, + "step": 25950 + }, + { + "epoch": 2.849879200527125, + "grad_norm": 2.3478891849517822, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7474612593650818, + "num_tokens": 671488614.0, + "step": 25951 + }, + { + "epoch": 2.8499890182297385, + "grad_norm": 2.3881208896636963, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7281873226165771, + "num_tokens": 671516910.0, + "step": 25952 + }, + { + "epoch": 2.8500988359323522, + "grad_norm": 2.392853260040283, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7370712757110596, + "num_tokens": 671542002.0, + "step": 25953 + }, + { + "epoch": 2.850208653634966, + "grad_norm": 2.657078981399536, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7296543717384338, + "num_tokens": 671564639.0, + "step": 25954 + }, + { + "epoch": 2.8503184713375798, + "grad_norm": 2.396664619445801, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7251418232917786, + "num_tokens": 671591875.0, + "step": 25955 + }, + { + "epoch": 2.8504282890401935, + "grad_norm": 2.3710484504699707, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7174952030181885, + "num_tokens": 671618230.0, + "step": 25956 + }, + { + "epoch": 2.850538106742807, + "grad_norm": 2.137425422668457, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7325335741043091, + "num_tokens": 671650632.0, + "step": 25957 + }, + { + "epoch": 2.8506479244454206, + "grad_norm": 2.485731840133667, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7219128608703613, + "num_tokens": 671674132.0, + "step": 25958 + }, + { + "epoch": 2.8507577421480343, + "grad_norm": 2.3415913581848145, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7185600996017456, + "num_tokens": 671701303.0, + "step": 25959 + }, + { + "epoch": 2.8508675598506477, + "grad_norm": 2.202716827392578, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7163019180297852, + "num_tokens": 671731503.0, + "step": 25960 + }, + { + "epoch": 2.850977377553262, + "grad_norm": 2.30066180229187, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.7572211027145386, + "num_tokens": 671758936.0, + "step": 25961 + }, + { + "epoch": 2.851087195255875, + "grad_norm": 2.1935036182403564, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.720588207244873, + "num_tokens": 671789371.0, + "step": 25962 + }, + { + "epoch": 2.851197012958489, + "grad_norm": 2.3864524364471436, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7405687570571899, + "num_tokens": 671815168.0, + "step": 25963 + }, + { + "epoch": 2.8513068306611027, + "grad_norm": 2.5974347591400146, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7607443332672119, + "num_tokens": 671835725.0, + "step": 25964 + }, + { + "epoch": 2.851416648363716, + "grad_norm": 2.1841814517974854, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7136887907981873, + "num_tokens": 671864361.0, + "step": 25965 + }, + { + "epoch": 2.8515264660663298, + "grad_norm": 2.2711594104766846, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7173734903335571, + "num_tokens": 671893204.0, + "step": 25966 + }, + { + "epoch": 2.8516362837689435, + "grad_norm": 2.4946799278259277, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7318190932273865, + "num_tokens": 671917250.0, + "step": 25967 + }, + { + "epoch": 2.8517461014715573, + "grad_norm": 2.152456521987915, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.716960072517395, + "num_tokens": 671948757.0, + "step": 25968 + }, + { + "epoch": 2.851855919174171, + "grad_norm": 2.6230568885803223, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7337849140167236, + "num_tokens": 671971315.0, + "step": 25969 + }, + { + "epoch": 2.8519657368767843, + "grad_norm": 2.3302297592163086, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7049375176429749, + "num_tokens": 672001667.0, + "step": 25970 + }, + { + "epoch": 2.852075554579398, + "grad_norm": 2.6554741859436035, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7532839775085449, + "num_tokens": 672021785.0, + "step": 25971 + }, + { + "epoch": 2.852185372282012, + "grad_norm": 2.290133237838745, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7355892658233643, + "num_tokens": 672047617.0, + "step": 25972 + }, + { + "epoch": 2.8522951899846256, + "grad_norm": 2.6688621044158936, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.725955605506897, + "num_tokens": 672068906.0, + "step": 25973 + }, + { + "epoch": 2.8524050076872394, + "grad_norm": 2.227597713470459, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7233368754386902, + "num_tokens": 672096120.0, + "step": 25974 + }, + { + "epoch": 2.8525148253898527, + "grad_norm": 2.332202911376953, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7420382499694824, + "num_tokens": 672123519.0, + "step": 25975 + }, + { + "epoch": 2.8526246430924664, + "grad_norm": 2.421292304992676, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7331043481826782, + "num_tokens": 672148463.0, + "step": 25976 + }, + { + "epoch": 2.85273446079508, + "grad_norm": 2.3088345527648926, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7625990509986877, + "num_tokens": 672175014.0, + "step": 25977 + }, + { + "epoch": 2.852844278497694, + "grad_norm": 2.611315965652466, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7342395782470703, + "num_tokens": 672196646.0, + "step": 25978 + }, + { + "epoch": 2.8529540962003077, + "grad_norm": 2.4153599739074707, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7317668795585632, + "num_tokens": 672223961.0, + "step": 25979 + }, + { + "epoch": 2.853063913902921, + "grad_norm": 2.2139692306518555, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7334173917770386, + "num_tokens": 672252712.0, + "step": 25980 + }, + { + "epoch": 2.853173731605535, + "grad_norm": 2.425684690475464, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7313108444213867, + "num_tokens": 672276394.0, + "step": 25981 + }, + { + "epoch": 2.8532835493081485, + "grad_norm": 2.3837645053863525, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7517650723457336, + "num_tokens": 672300503.0, + "step": 25982 + }, + { + "epoch": 2.8533933670107623, + "grad_norm": 2.6211369037628174, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7351549863815308, + "num_tokens": 672322330.0, + "step": 25983 + }, + { + "epoch": 2.853503184713376, + "grad_norm": 2.511178493499756, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7341736555099487, + "num_tokens": 672345273.0, + "step": 25984 + }, + { + "epoch": 2.8536130024159894, + "grad_norm": 2.3268682956695557, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7214099168777466, + "num_tokens": 672372616.0, + "step": 25985 + }, + { + "epoch": 2.853722820118603, + "grad_norm": 2.383729934692383, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7787104249000549, + "num_tokens": 672394953.0, + "step": 25986 + }, + { + "epoch": 2.853832637821217, + "grad_norm": 2.216449022293091, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7239387035369873, + "num_tokens": 672423046.0, + "step": 25987 + }, + { + "epoch": 2.85394245552383, + "grad_norm": 2.7386958599090576, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7459028959274292, + "num_tokens": 672443008.0, + "step": 25988 + }, + { + "epoch": 2.854052273226444, + "grad_norm": 2.435328483581543, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7330908179283142, + "num_tokens": 672467994.0, + "step": 25989 + }, + { + "epoch": 2.8541620909290577, + "grad_norm": 2.203688859939575, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7085007429122925, + "num_tokens": 672497637.0, + "step": 25990 + }, + { + "epoch": 2.8542719086316715, + "grad_norm": 2.282499074935913, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7259048223495483, + "num_tokens": 672522773.0, + "step": 25991 + }, + { + "epoch": 2.854381726334285, + "grad_norm": 2.1461219787597656, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7308632731437683, + "num_tokens": 672550357.0, + "step": 25992 + }, + { + "epoch": 2.8544915440368985, + "grad_norm": 2.5320956707000732, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7580849528312683, + "num_tokens": 672572533.0, + "step": 25993 + }, + { + "epoch": 2.8546013617395123, + "grad_norm": 2.252108097076416, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7310798764228821, + "num_tokens": 672599089.0, + "step": 25994 + }, + { + "epoch": 2.854711179442126, + "grad_norm": 2.314568042755127, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7421775460243225, + "num_tokens": 672626113.0, + "step": 25995 + }, + { + "epoch": 2.85482099714474, + "grad_norm": 2.4459009170532227, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7272582650184631, + "num_tokens": 672651555.0, + "step": 25996 + }, + { + "epoch": 2.8549308148473536, + "grad_norm": 2.5889089107513428, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7240224480628967, + "num_tokens": 672674340.0, + "step": 25997 + }, + { + "epoch": 2.855040632549967, + "grad_norm": 2.1678354740142822, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7516201734542847, + "num_tokens": 672704680.0, + "step": 25998 + }, + { + "epoch": 2.8551504502525806, + "grad_norm": 2.4162917137145996, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7111278176307678, + "num_tokens": 672731569.0, + "step": 25999 + }, + { + "epoch": 2.8552602679551944, + "grad_norm": 2.07141375541687, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7085859775543213, + "num_tokens": 672763196.0, + "step": 26000 + }, + { + "epoch": 2.855370085657808, + "grad_norm": 2.5761823654174805, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7586205005645752, + "num_tokens": 672784965.0, + "step": 26001 + }, + { + "epoch": 2.855479903360422, + "grad_norm": 2.168850898742676, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7032061815261841, + "num_tokens": 672816672.0, + "step": 26002 + }, + { + "epoch": 2.855589721063035, + "grad_norm": 2.4213337898254395, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7541863918304443, + "num_tokens": 672840746.0, + "step": 26003 + }, + { + "epoch": 2.855699538765649, + "grad_norm": 2.3173580169677734, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7467247247695923, + "num_tokens": 672866472.0, + "step": 26004 + }, + { + "epoch": 2.8558093564682627, + "grad_norm": 2.2815935611724854, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.727765679359436, + "num_tokens": 672894541.0, + "step": 26005 + }, + { + "epoch": 2.8559191741708765, + "grad_norm": 2.4314184188842773, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7376130223274231, + "num_tokens": 672919064.0, + "step": 26006 + }, + { + "epoch": 2.8560289918734902, + "grad_norm": 2.2720835208892822, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7274394035339355, + "num_tokens": 672946396.0, + "step": 26007 + }, + { + "epoch": 2.8561388095761036, + "grad_norm": 2.1828482151031494, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7260891199111938, + "num_tokens": 672975469.0, + "step": 26008 + }, + { + "epoch": 2.8562486272787173, + "grad_norm": 2.0966274738311768, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7487075328826904, + "num_tokens": 673005393.0, + "step": 26009 + }, + { + "epoch": 2.856358444981331, + "grad_norm": 2.7855746746063232, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7552766799926758, + "num_tokens": 673023441.0, + "step": 26010 + }, + { + "epoch": 2.8564682626839444, + "grad_norm": 2.359259843826294, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7583030462265015, + "num_tokens": 673049411.0, + "step": 26011 + }, + { + "epoch": 2.8565780803865586, + "grad_norm": 2.5827994346618652, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7344681620597839, + "num_tokens": 673071869.0, + "step": 26012 + }, + { + "epoch": 2.856687898089172, + "grad_norm": 2.321255922317505, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7132076621055603, + "num_tokens": 673098439.0, + "step": 26013 + }, + { + "epoch": 2.8567977157917857, + "grad_norm": 2.621631622314453, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.736630916595459, + "num_tokens": 673120897.0, + "step": 26014 + }, + { + "epoch": 2.8569075334943994, + "grad_norm": 2.641618251800537, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7359521389007568, + "num_tokens": 673141537.0, + "step": 26015 + }, + { + "epoch": 2.8570173511970127, + "grad_norm": 2.3307740688323975, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7305746674537659, + "num_tokens": 673169449.0, + "step": 26016 + }, + { + "epoch": 2.8571271688996265, + "grad_norm": 2.15978741645813, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7360882759094238, + "num_tokens": 673197705.0, + "step": 26017 + }, + { + "epoch": 2.8572369866022402, + "grad_norm": 2.1384100914001465, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7138987183570862, + "num_tokens": 673227243.0, + "step": 26018 + }, + { + "epoch": 2.857346804304854, + "grad_norm": 2.4887921810150146, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7235065698623657, + "num_tokens": 673252030.0, + "step": 26019 + }, + { + "epoch": 2.8574566220074678, + "grad_norm": 2.189324378967285, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7392985820770264, + "num_tokens": 673280208.0, + "step": 26020 + }, + { + "epoch": 2.857566439710081, + "grad_norm": 2.0471324920654297, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.71800297498703, + "num_tokens": 673311114.0, + "step": 26021 + }, + { + "epoch": 2.857676257412695, + "grad_norm": 2.271862268447876, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7348171472549438, + "num_tokens": 673336523.0, + "step": 26022 + }, + { + "epoch": 2.8577860751153086, + "grad_norm": 2.326575756072998, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7199498414993286, + "num_tokens": 673363109.0, + "step": 26023 + }, + { + "epoch": 2.8578958928179223, + "grad_norm": 2.336613655090332, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7484067678451538, + "num_tokens": 673388866.0, + "step": 26024 + }, + { + "epoch": 2.858005710520536, + "grad_norm": 2.537804365158081, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7330340147018433, + "num_tokens": 673413250.0, + "step": 26025 + }, + { + "epoch": 2.8581155282231494, + "grad_norm": 2.707249641418457, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7408236265182495, + "num_tokens": 673433658.0, + "step": 26026 + }, + { + "epoch": 2.858225345925763, + "grad_norm": 2.539193868637085, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7484515309333801, + "num_tokens": 673453861.0, + "step": 26027 + }, + { + "epoch": 2.858335163628377, + "grad_norm": 2.161278009414673, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6908791065216064, + "num_tokens": 673484844.0, + "step": 26028 + }, + { + "epoch": 2.8584449813309907, + "grad_norm": 2.4412825107574463, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7503259181976318, + "num_tokens": 673510058.0, + "step": 26029 + }, + { + "epoch": 2.8585547990336044, + "grad_norm": 2.2600467205047607, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7333425283432007, + "num_tokens": 673537680.0, + "step": 26030 + }, + { + "epoch": 2.8586646167362177, + "grad_norm": 2.4689810276031494, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7338601350784302, + "num_tokens": 673562118.0, + "step": 26031 + }, + { + "epoch": 2.8587744344388315, + "grad_norm": 2.3666625022888184, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7506060004234314, + "num_tokens": 673587339.0, + "step": 26032 + }, + { + "epoch": 2.8588842521414453, + "grad_norm": 2.4564645290374756, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.733616828918457, + "num_tokens": 673612062.0, + "step": 26033 + }, + { + "epoch": 2.858994069844059, + "grad_norm": 2.170238494873047, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7378480434417725, + "num_tokens": 673639397.0, + "step": 26034 + }, + { + "epoch": 2.8591038875466728, + "grad_norm": 2.1057136058807373, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.738874077796936, + "num_tokens": 673672041.0, + "step": 26035 + }, + { + "epoch": 2.859213705249286, + "grad_norm": 2.565124750137329, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7240226864814758, + "num_tokens": 673696084.0, + "step": 26036 + }, + { + "epoch": 2.8593235229519, + "grad_norm": 2.1541860103607178, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7235585451126099, + "num_tokens": 673725454.0, + "step": 26037 + }, + { + "epoch": 2.8594333406545136, + "grad_norm": 2.6148879528045654, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7393674254417419, + "num_tokens": 673746551.0, + "step": 26038 + }, + { + "epoch": 2.859543158357127, + "grad_norm": 2.37392258644104, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7435147166252136, + "num_tokens": 673772122.0, + "step": 26039 + }, + { + "epoch": 2.8596529760597407, + "grad_norm": 2.335585355758667, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7303740382194519, + "num_tokens": 673797623.0, + "step": 26040 + }, + { + "epoch": 2.8597627937623544, + "grad_norm": 2.0838534832000732, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7222718000411987, + "num_tokens": 673828902.0, + "step": 26041 + }, + { + "epoch": 2.859872611464968, + "grad_norm": 2.1557376384735107, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.729154109954834, + "num_tokens": 673859008.0, + "step": 26042 + }, + { + "epoch": 2.859982429167582, + "grad_norm": 2.466921329498291, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7454081177711487, + "num_tokens": 673881917.0, + "step": 26043 + }, + { + "epoch": 2.8600922468701953, + "grad_norm": 2.1471095085144043, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.735016942024231, + "num_tokens": 673912750.0, + "step": 26044 + }, + { + "epoch": 2.860202064572809, + "grad_norm": 2.521805763244629, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7297808527946472, + "num_tokens": 673934835.0, + "step": 26045 + }, + { + "epoch": 2.8603118822754228, + "grad_norm": 2.207022190093994, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7298760414123535, + "num_tokens": 673963165.0, + "step": 26046 + }, + { + "epoch": 2.8604216999780365, + "grad_norm": 2.542116403579712, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7426578998565674, + "num_tokens": 673984440.0, + "step": 26047 + }, + { + "epoch": 2.8605315176806503, + "grad_norm": 2.2965290546417236, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7457028031349182, + "num_tokens": 674009921.0, + "step": 26048 + }, + { + "epoch": 2.8606413353832636, + "grad_norm": 2.487238883972168, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7226334810256958, + "num_tokens": 674034168.0, + "step": 26049 + }, + { + "epoch": 2.8607511530858774, + "grad_norm": 2.7629923820495605, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7489025592803955, + "num_tokens": 674054595.0, + "step": 26050 + }, + { + "epoch": 2.860860970788491, + "grad_norm": 2.869940996170044, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7346878051757812, + "num_tokens": 674073090.0, + "step": 26051 + }, + { + "epoch": 2.860970788491105, + "grad_norm": 2.1755788326263428, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7280145287513733, + "num_tokens": 674102260.0, + "step": 26052 + }, + { + "epoch": 2.8610806061937186, + "grad_norm": 2.621778964996338, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7224787473678589, + "num_tokens": 674123703.0, + "step": 26053 + }, + { + "epoch": 2.861190423896332, + "grad_norm": 2.2954318523406982, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7155081033706665, + "num_tokens": 674151101.0, + "step": 26054 + }, + { + "epoch": 2.8613002415989457, + "grad_norm": 2.4539971351623535, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7200920581817627, + "num_tokens": 674176981.0, + "step": 26055 + }, + { + "epoch": 2.8614100593015594, + "grad_norm": 2.361583948135376, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7161208391189575, + "num_tokens": 674202164.0, + "step": 26056 + }, + { + "epoch": 2.861519877004173, + "grad_norm": 2.492140769958496, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7415369749069214, + "num_tokens": 674227273.0, + "step": 26057 + }, + { + "epoch": 2.861629694706787, + "grad_norm": 2.3881185054779053, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7433515191078186, + "num_tokens": 674251889.0, + "step": 26058 + }, + { + "epoch": 2.8617395124094003, + "grad_norm": 2.1590960025787354, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7079675793647766, + "num_tokens": 674282868.0, + "step": 26059 + }, + { + "epoch": 2.861849330112014, + "grad_norm": 2.3136494159698486, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.728240966796875, + "num_tokens": 674310603.0, + "step": 26060 + }, + { + "epoch": 2.861959147814628, + "grad_norm": 2.508218288421631, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7344368100166321, + "num_tokens": 674333678.0, + "step": 26061 + }, + { + "epoch": 2.862068965517241, + "grad_norm": 2.2778451442718506, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7519779205322266, + "num_tokens": 674361458.0, + "step": 26062 + }, + { + "epoch": 2.8621787832198553, + "grad_norm": 2.5569651126861572, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7465084791183472, + "num_tokens": 674384351.0, + "step": 26063 + }, + { + "epoch": 2.8622886009224686, + "grad_norm": 2.2259066104888916, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7112159729003906, + "num_tokens": 674411895.0, + "step": 26064 + }, + { + "epoch": 2.8623984186250824, + "grad_norm": 2.3037538528442383, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7052062153816223, + "num_tokens": 674441370.0, + "step": 26065 + }, + { + "epoch": 2.862508236327696, + "grad_norm": 2.064920425415039, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.709897518157959, + "num_tokens": 674473057.0, + "step": 26066 + }, + { + "epoch": 2.8626180540303094, + "grad_norm": 2.3102309703826904, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7367669939994812, + "num_tokens": 674500205.0, + "step": 26067 + }, + { + "epoch": 2.862727871732923, + "grad_norm": 2.5433924198150635, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7344985008239746, + "num_tokens": 674522500.0, + "step": 26068 + }, + { + "epoch": 2.862837689435537, + "grad_norm": 2.1925790309906006, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7137369513511658, + "num_tokens": 674553971.0, + "step": 26069 + }, + { + "epoch": 2.8629475071381507, + "grad_norm": 2.3401315212249756, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7157326936721802, + "num_tokens": 674583947.0, + "step": 26070 + }, + { + "epoch": 2.8630573248407645, + "grad_norm": 2.45676851272583, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7227885723114014, + "num_tokens": 674609286.0, + "step": 26071 + }, + { + "epoch": 2.863167142543378, + "grad_norm": 2.9377834796905518, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7360000610351562, + "num_tokens": 674627776.0, + "step": 26072 + }, + { + "epoch": 2.8632769602459915, + "grad_norm": 2.3341894149780273, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7356652021408081, + "num_tokens": 674652947.0, + "step": 26073 + }, + { + "epoch": 2.8633867779486053, + "grad_norm": 2.247997999191284, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7229117155075073, + "num_tokens": 674680767.0, + "step": 26074 + }, + { + "epoch": 2.863496595651219, + "grad_norm": 2.2997307777404785, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7193018794059753, + "num_tokens": 674706645.0, + "step": 26075 + }, + { + "epoch": 2.863606413353833, + "grad_norm": 2.3382086753845215, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7278238534927368, + "num_tokens": 674735371.0, + "step": 26076 + }, + { + "epoch": 2.863716231056446, + "grad_norm": 2.3955819606781006, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7219119071960449, + "num_tokens": 674761652.0, + "step": 26077 + }, + { + "epoch": 2.86382604875906, + "grad_norm": 2.224229574203491, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7410974502563477, + "num_tokens": 674790968.0, + "step": 26078 + }, + { + "epoch": 2.8639358664616736, + "grad_norm": 2.2104482650756836, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.7029114961624146, + "num_tokens": 674820329.0, + "step": 26079 + }, + { + "epoch": 2.8640456841642874, + "grad_norm": 2.91898775100708, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7406156063079834, + "num_tokens": 674843544.0, + "step": 26080 + }, + { + "epoch": 2.864155501866901, + "grad_norm": 2.576582670211792, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.755128800868988, + "num_tokens": 674866950.0, + "step": 26081 + }, + { + "epoch": 2.8642653195695145, + "grad_norm": 2.4171693325042725, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.702378511428833, + "num_tokens": 674892458.0, + "step": 26082 + }, + { + "epoch": 2.8643751372721282, + "grad_norm": 2.737813711166382, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7341934442520142, + "num_tokens": 674912625.0, + "step": 26083 + }, + { + "epoch": 2.864484954974742, + "grad_norm": 2.32753586769104, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7253928184509277, + "num_tokens": 674937917.0, + "step": 26084 + }, + { + "epoch": 2.8645947726773557, + "grad_norm": 2.3394315242767334, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7502462863922119, + "num_tokens": 674963788.0, + "step": 26085 + }, + { + "epoch": 2.8647045903799695, + "grad_norm": 2.6050543785095215, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7464016079902649, + "num_tokens": 674985862.0, + "step": 26086 + }, + { + "epoch": 2.864814408082583, + "grad_norm": 2.661987543106079, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7334893941879272, + "num_tokens": 675006808.0, + "step": 26087 + }, + { + "epoch": 2.8649242257851966, + "grad_norm": 2.3506197929382324, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7317095994949341, + "num_tokens": 675033623.0, + "step": 26088 + }, + { + "epoch": 2.8650340434878103, + "grad_norm": 2.5669498443603516, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7477585673332214, + "num_tokens": 675053083.0, + "step": 26089 + }, + { + "epoch": 2.8651438611904236, + "grad_norm": 2.211822032928467, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7231296300888062, + "num_tokens": 675082485.0, + "step": 26090 + }, + { + "epoch": 2.8652536788930374, + "grad_norm": 2.7503671646118164, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7146047353744507, + "num_tokens": 675105349.0, + "step": 26091 + }, + { + "epoch": 2.865363496595651, + "grad_norm": 2.431550979614258, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7320960760116577, + "num_tokens": 675129468.0, + "step": 26092 + }, + { + "epoch": 2.865473314298265, + "grad_norm": 2.2127974033355713, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7318171262741089, + "num_tokens": 675158711.0, + "step": 26093 + }, + { + "epoch": 2.8655831320008787, + "grad_norm": 2.234315872192383, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7561231851577759, + "num_tokens": 675186157.0, + "step": 26094 + }, + { + "epoch": 2.865692949703492, + "grad_norm": 2.199598550796509, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.727146327495575, + "num_tokens": 675213161.0, + "step": 26095 + }, + { + "epoch": 2.8658027674061057, + "grad_norm": 2.2722132205963135, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7257177829742432, + "num_tokens": 675243061.0, + "step": 26096 + }, + { + "epoch": 2.8659125851087195, + "grad_norm": 2.4646499156951904, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7449392676353455, + "num_tokens": 675267805.0, + "step": 26097 + }, + { + "epoch": 2.8660224028113332, + "grad_norm": 2.6646130084991455, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7518017888069153, + "num_tokens": 675288863.0, + "step": 26098 + }, + { + "epoch": 2.866132220513947, + "grad_norm": 2.3499624729156494, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7618499994277954, + "num_tokens": 675312823.0, + "step": 26099 + }, + { + "epoch": 2.8662420382165603, + "grad_norm": 2.2580103874206543, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7272421717643738, + "num_tokens": 675343825.0, + "step": 26100 + }, + { + "epoch": 2.866351855919174, + "grad_norm": 2.716646194458008, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7234218716621399, + "num_tokens": 675364877.0, + "step": 26101 + }, + { + "epoch": 2.866461673621788, + "grad_norm": 2.2688207626342773, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7333601713180542, + "num_tokens": 675390248.0, + "step": 26102 + }, + { + "epoch": 2.8665714913244016, + "grad_norm": 2.147129774093628, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7294948697090149, + "num_tokens": 675418776.0, + "step": 26103 + }, + { + "epoch": 2.8666813090270153, + "grad_norm": 1.9637190103530884, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7466251850128174, + "num_tokens": 675452449.0, + "step": 26104 + }, + { + "epoch": 2.8667911267296287, + "grad_norm": 2.590373992919922, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7294464111328125, + "num_tokens": 675472964.0, + "step": 26105 + }, + { + "epoch": 2.8669009444322424, + "grad_norm": 2.379490852355957, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7317508459091187, + "num_tokens": 675499433.0, + "step": 26106 + }, + { + "epoch": 2.867010762134856, + "grad_norm": 2.1289780139923096, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7092477083206177, + "num_tokens": 675531158.0, + "step": 26107 + }, + { + "epoch": 2.86712057983747, + "grad_norm": 2.7233798503875732, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7466641664505005, + "num_tokens": 675552905.0, + "step": 26108 + }, + { + "epoch": 2.8672303975400837, + "grad_norm": 2.437614917755127, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7369483113288879, + "num_tokens": 675577409.0, + "step": 26109 + }, + { + "epoch": 2.867340215242697, + "grad_norm": 2.4810702800750732, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7476343512535095, + "num_tokens": 675600582.0, + "step": 26110 + }, + { + "epoch": 2.8674500329453108, + "grad_norm": 2.5017971992492676, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7507959008216858, + "num_tokens": 675623921.0, + "step": 26111 + }, + { + "epoch": 2.8675598506479245, + "grad_norm": 2.6608591079711914, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7607781887054443, + "num_tokens": 675644555.0, + "step": 26112 + }, + { + "epoch": 2.8676696683505383, + "grad_norm": 2.6022212505340576, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7356772422790527, + "num_tokens": 675667029.0, + "step": 26113 + }, + { + "epoch": 2.867779486053152, + "grad_norm": 2.037449359893799, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7092716693878174, + "num_tokens": 675702160.0, + "step": 26114 + }, + { + "epoch": 2.8678893037557653, + "grad_norm": 2.4540905952453613, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7504037618637085, + "num_tokens": 675723715.0, + "step": 26115 + }, + { + "epoch": 2.867999121458379, + "grad_norm": 2.4152679443359375, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7255935072898865, + "num_tokens": 675749148.0, + "step": 26116 + }, + { + "epoch": 2.868108939160993, + "grad_norm": 2.2832062244415283, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7210936546325684, + "num_tokens": 675776758.0, + "step": 26117 + }, + { + "epoch": 2.868218756863606, + "grad_norm": 2.4675378799438477, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7259271740913391, + "num_tokens": 675800674.0, + "step": 26118 + }, + { + "epoch": 2.86832857456622, + "grad_norm": 2.256422519683838, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7113494873046875, + "num_tokens": 675827345.0, + "step": 26119 + }, + { + "epoch": 2.8684383922688337, + "grad_norm": 1.944210410118103, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7266241908073425, + "num_tokens": 675861466.0, + "step": 26120 + }, + { + "epoch": 2.8685482099714474, + "grad_norm": 2.4725635051727295, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7393964529037476, + "num_tokens": 675884135.0, + "step": 26121 + }, + { + "epoch": 2.868658027674061, + "grad_norm": 2.4348044395446777, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7339212894439697, + "num_tokens": 675908730.0, + "step": 26122 + }, + { + "epoch": 2.8687678453766745, + "grad_norm": 2.4387459754943848, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.754191517829895, + "num_tokens": 675932288.0, + "step": 26123 + }, + { + "epoch": 2.8688776630792883, + "grad_norm": 2.5877633094787598, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7467256188392639, + "num_tokens": 675954517.0, + "step": 26124 + }, + { + "epoch": 2.868987480781902, + "grad_norm": 2.2558791637420654, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7367814779281616, + "num_tokens": 675980211.0, + "step": 26125 + }, + { + "epoch": 2.869097298484516, + "grad_norm": 2.3225314617156982, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7507908940315247, + "num_tokens": 676005009.0, + "step": 26126 + }, + { + "epoch": 2.8692071161871295, + "grad_norm": 2.3781440258026123, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7323378920555115, + "num_tokens": 676030630.0, + "step": 26127 + }, + { + "epoch": 2.869316933889743, + "grad_norm": 2.3498926162719727, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7436615228652954, + "num_tokens": 676055394.0, + "step": 26128 + }, + { + "epoch": 2.8694267515923566, + "grad_norm": 2.3270087242126465, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7436249852180481, + "num_tokens": 676081325.0, + "step": 26129 + }, + { + "epoch": 2.8695365692949704, + "grad_norm": 2.3261148929595947, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7130658626556396, + "num_tokens": 676107605.0, + "step": 26130 + }, + { + "epoch": 2.869646386997584, + "grad_norm": 2.5241384506225586, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7342376708984375, + "num_tokens": 676130392.0, + "step": 26131 + }, + { + "epoch": 2.869756204700198, + "grad_norm": 2.4858593940734863, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7265874147415161, + "num_tokens": 676155618.0, + "step": 26132 + }, + { + "epoch": 2.869866022402811, + "grad_norm": 2.478811264038086, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7374082207679749, + "num_tokens": 676180193.0, + "step": 26133 + }, + { + "epoch": 2.869975840105425, + "grad_norm": 2.370349168777466, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7025119066238403, + "num_tokens": 676207552.0, + "step": 26134 + }, + { + "epoch": 2.8700856578080387, + "grad_norm": 2.543881893157959, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7305110692977905, + "num_tokens": 676231269.0, + "step": 26135 + }, + { + "epoch": 2.8701954755106525, + "grad_norm": 2.7019965648651123, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7525022029876709, + "num_tokens": 676251030.0, + "step": 26136 + }, + { + "epoch": 2.870305293213266, + "grad_norm": 2.3843672275543213, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7508929967880249, + "num_tokens": 676277159.0, + "step": 26137 + }, + { + "epoch": 2.8704151109158795, + "grad_norm": 2.4893505573272705, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7350600361824036, + "num_tokens": 676302300.0, + "step": 26138 + }, + { + "epoch": 2.8705249286184933, + "grad_norm": 2.5981364250183105, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7474493384361267, + "num_tokens": 676322622.0, + "step": 26139 + }, + { + "epoch": 2.870634746321107, + "grad_norm": 2.406719207763672, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7198715209960938, + "num_tokens": 676347935.0, + "step": 26140 + }, + { + "epoch": 2.8707445640237204, + "grad_norm": 2.119452476501465, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7119343280792236, + "num_tokens": 676379047.0, + "step": 26141 + }, + { + "epoch": 2.8708543817263346, + "grad_norm": 2.1877996921539307, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.6928259134292603, + "num_tokens": 676408422.0, + "step": 26142 + }, + { + "epoch": 2.870964199428948, + "grad_norm": 2.3824679851531982, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7266907691955566, + "num_tokens": 676435121.0, + "step": 26143 + }, + { + "epoch": 2.8710740171315616, + "grad_norm": 2.3871374130249023, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7211506366729736, + "num_tokens": 676460750.0, + "step": 26144 + }, + { + "epoch": 2.8711838348341754, + "grad_norm": 2.4034652709960938, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7479248046875, + "num_tokens": 676482900.0, + "step": 26145 + }, + { + "epoch": 2.8712936525367887, + "grad_norm": 2.3482205867767334, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7307521104812622, + "num_tokens": 676510402.0, + "step": 26146 + }, + { + "epoch": 2.8714034702394025, + "grad_norm": 2.137873649597168, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7384151220321655, + "num_tokens": 676541557.0, + "step": 26147 + }, + { + "epoch": 2.871513287942016, + "grad_norm": 2.2688357830047607, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7500889301300049, + "num_tokens": 676567699.0, + "step": 26148 + }, + { + "epoch": 2.87162310564463, + "grad_norm": 2.5290098190307617, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7383993268013, + "num_tokens": 676590430.0, + "step": 26149 + }, + { + "epoch": 2.8717329233472437, + "grad_norm": 2.384063243865967, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7270551919937134, + "num_tokens": 676616324.0, + "step": 26150 + }, + { + "epoch": 2.871842741049857, + "grad_norm": 2.195810317993164, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7345470190048218, + "num_tokens": 676645348.0, + "step": 26151 + }, + { + "epoch": 2.871952558752471, + "grad_norm": 2.760153293609619, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7489530444145203, + "num_tokens": 676664263.0, + "step": 26152 + }, + { + "epoch": 2.8720623764550846, + "grad_norm": 2.2616374492645264, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7365515828132629, + "num_tokens": 676689852.0, + "step": 26153 + }, + { + "epoch": 2.8721721941576983, + "grad_norm": 2.427013397216797, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7120306491851807, + "num_tokens": 676716524.0, + "step": 26154 + }, + { + "epoch": 2.872282011860312, + "grad_norm": 2.295386791229248, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7436360120773315, + "num_tokens": 676741984.0, + "step": 26155 + }, + { + "epoch": 2.8723918295629254, + "grad_norm": 2.123973846435547, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7099959254264832, + "num_tokens": 676774168.0, + "step": 26156 + }, + { + "epoch": 2.872501647265539, + "grad_norm": 2.077890157699585, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7178739309310913, + "num_tokens": 676806321.0, + "step": 26157 + }, + { + "epoch": 2.872611464968153, + "grad_norm": 2.4328513145446777, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7437248826026917, + "num_tokens": 676830146.0, + "step": 26158 + }, + { + "epoch": 2.8727212826707667, + "grad_norm": 2.26668381690979, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7465667128562927, + "num_tokens": 676856918.0, + "step": 26159 + }, + { + "epoch": 2.8728311003733804, + "grad_norm": 2.7397515773773193, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7416986227035522, + "num_tokens": 676877334.0, + "step": 26160 + }, + { + "epoch": 2.8729409180759937, + "grad_norm": 2.315197467803955, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7376258373260498, + "num_tokens": 676902466.0, + "step": 26161 + }, + { + "epoch": 2.8730507357786075, + "grad_norm": 2.5417299270629883, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7366876006126404, + "num_tokens": 676923813.0, + "step": 26162 + }, + { + "epoch": 2.8731605534812212, + "grad_norm": 2.479495048522949, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7355863451957703, + "num_tokens": 676946898.0, + "step": 26163 + }, + { + "epoch": 2.873270371183835, + "grad_norm": 2.4389734268188477, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7260369062423706, + "num_tokens": 676972284.0, + "step": 26164 + }, + { + "epoch": 2.8733801888864487, + "grad_norm": 2.4747440814971924, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7300031185150146, + "num_tokens": 676997213.0, + "step": 26165 + }, + { + "epoch": 2.873490006589062, + "grad_norm": 2.5449163913726807, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7256336808204651, + "num_tokens": 677020832.0, + "step": 26166 + }, + { + "epoch": 2.873599824291676, + "grad_norm": 2.535780191421509, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7217196226119995, + "num_tokens": 677044474.0, + "step": 26167 + }, + { + "epoch": 2.8737096419942896, + "grad_norm": 2.245002031326294, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7396478652954102, + "num_tokens": 677071226.0, + "step": 26168 + }, + { + "epoch": 2.873819459696903, + "grad_norm": 2.4440085887908936, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7531775832176208, + "num_tokens": 677094413.0, + "step": 26169 + }, + { + "epoch": 2.8739292773995166, + "grad_norm": 1.9684038162231445, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7196190357208252, + "num_tokens": 677128378.0, + "step": 26170 + }, + { + "epoch": 2.8740390951021304, + "grad_norm": 2.5871739387512207, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7233754396438599, + "num_tokens": 677151911.0, + "step": 26171 + }, + { + "epoch": 2.874148912804744, + "grad_norm": 2.1714537143707275, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7423961162567139, + "num_tokens": 677180932.0, + "step": 26172 + }, + { + "epoch": 2.874258730507358, + "grad_norm": 2.3833577632904053, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7524327039718628, + "num_tokens": 677204864.0, + "step": 26173 + }, + { + "epoch": 2.8743685482099712, + "grad_norm": 2.3442225456237793, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7269874215126038, + "num_tokens": 677229322.0, + "step": 26174 + }, + { + "epoch": 2.874478365912585, + "grad_norm": 2.6416802406311035, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7325122356414795, + "num_tokens": 677251165.0, + "step": 26175 + }, + { + "epoch": 2.8745881836151987, + "grad_norm": 2.399223566055298, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7314037680625916, + "num_tokens": 677276506.0, + "step": 26176 + }, + { + "epoch": 2.8746980013178125, + "grad_norm": 2.377908945083618, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7178906202316284, + "num_tokens": 677303683.0, + "step": 26177 + }, + { + "epoch": 2.8748078190204263, + "grad_norm": 2.262446403503418, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7486633658409119, + "num_tokens": 677329909.0, + "step": 26178 + }, + { + "epoch": 2.8749176367230396, + "grad_norm": 2.55283784866333, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7303644418716431, + "num_tokens": 677352625.0, + "step": 26179 + }, + { + "epoch": 2.8750274544256533, + "grad_norm": 2.341200351715088, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7356511354446411, + "num_tokens": 677377575.0, + "step": 26180 + }, + { + "epoch": 2.875137272128267, + "grad_norm": 2.0470900535583496, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.718993604183197, + "num_tokens": 677413143.0, + "step": 26181 + }, + { + "epoch": 2.875247089830881, + "grad_norm": 2.5280673503875732, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7270161509513855, + "num_tokens": 677437420.0, + "step": 26182 + }, + { + "epoch": 2.8753569075334946, + "grad_norm": 2.4560537338256836, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7061388492584229, + "num_tokens": 677460817.0, + "step": 26183 + }, + { + "epoch": 2.875466725236108, + "grad_norm": 2.2524192333221436, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7009720206260681, + "num_tokens": 677490990.0, + "step": 26184 + }, + { + "epoch": 2.8755765429387217, + "grad_norm": 2.313016891479492, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.6993207931518555, + "num_tokens": 677519083.0, + "step": 26185 + }, + { + "epoch": 2.8756863606413354, + "grad_norm": 2.6893036365509033, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.724539041519165, + "num_tokens": 677541676.0, + "step": 26186 + }, + { + "epoch": 2.875796178343949, + "grad_norm": 2.1410627365112305, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7081063389778137, + "num_tokens": 677574898.0, + "step": 26187 + }, + { + "epoch": 2.875905996046563, + "grad_norm": 2.525646686553955, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7255637645721436, + "num_tokens": 677597511.0, + "step": 26188 + }, + { + "epoch": 2.8760158137491763, + "grad_norm": 2.4525644779205322, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7363426089286804, + "num_tokens": 677621779.0, + "step": 26189 + }, + { + "epoch": 2.87612563145179, + "grad_norm": 2.563823938369751, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7204048037528992, + "num_tokens": 677645561.0, + "step": 26190 + }, + { + "epoch": 2.8762354491544038, + "grad_norm": 2.323984146118164, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.738201379776001, + "num_tokens": 677669700.0, + "step": 26191 + }, + { + "epoch": 2.876345266857017, + "grad_norm": 2.2892043590545654, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7139412760734558, + "num_tokens": 677698827.0, + "step": 26192 + }, + { + "epoch": 2.8764550845596313, + "grad_norm": 2.1351473331451416, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7274273633956909, + "num_tokens": 677729257.0, + "step": 26193 + }, + { + "epoch": 2.8765649022622446, + "grad_norm": 2.3081142902374268, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7178987264633179, + "num_tokens": 677755695.0, + "step": 26194 + }, + { + "epoch": 2.8766747199648584, + "grad_norm": 2.2908101081848145, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7229307889938354, + "num_tokens": 677783088.0, + "step": 26195 + }, + { + "epoch": 2.876784537667472, + "grad_norm": 2.2127318382263184, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7316231727600098, + "num_tokens": 677814594.0, + "step": 26196 + }, + { + "epoch": 2.8768943553700854, + "grad_norm": 2.4727773666381836, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7526023983955383, + "num_tokens": 677836354.0, + "step": 26197 + }, + { + "epoch": 2.877004173072699, + "grad_norm": 2.228646755218506, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7446768879890442, + "num_tokens": 677865061.0, + "step": 26198 + }, + { + "epoch": 2.877113990775313, + "grad_norm": 2.3633370399475098, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7055018544197083, + "num_tokens": 677893485.0, + "step": 26199 + }, + { + "epoch": 2.8772238084779267, + "grad_norm": 3.0146846771240234, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7245960235595703, + "num_tokens": 677917057.0, + "step": 26200 + }, + { + "epoch": 2.8773336261805404, + "grad_norm": 2.1148619651794434, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7425165176391602, + "num_tokens": 677946716.0, + "step": 26201 + }, + { + "epoch": 2.8774434438831538, + "grad_norm": 2.379995822906494, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7122139930725098, + "num_tokens": 677974654.0, + "step": 26202 + }, + { + "epoch": 2.8775532615857675, + "grad_norm": 2.404270887374878, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7215895056724548, + "num_tokens": 678000009.0, + "step": 26203 + }, + { + "epoch": 2.8776630792883813, + "grad_norm": 2.195399761199951, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7367876768112183, + "num_tokens": 678027105.0, + "step": 26204 + }, + { + "epoch": 2.877772896990995, + "grad_norm": 2.2704622745513916, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7323908805847168, + "num_tokens": 678057760.0, + "step": 26205 + }, + { + "epoch": 2.877882714693609, + "grad_norm": 2.4108211994171143, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.735015869140625, + "num_tokens": 678084386.0, + "step": 26206 + }, + { + "epoch": 2.877992532396222, + "grad_norm": 2.282165288925171, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7234851717948914, + "num_tokens": 678110250.0, + "step": 26207 + }, + { + "epoch": 2.878102350098836, + "grad_norm": 2.287454843521118, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7112797498703003, + "num_tokens": 678138519.0, + "step": 26208 + }, + { + "epoch": 2.8782121678014496, + "grad_norm": 2.5968706607818604, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7248753309249878, + "num_tokens": 678160459.0, + "step": 26209 + }, + { + "epoch": 2.8783219855040634, + "grad_norm": 2.5558764934539795, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7399678826332092, + "num_tokens": 678183192.0, + "step": 26210 + }, + { + "epoch": 2.878431803206677, + "grad_norm": 2.190950870513916, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7345512509346008, + "num_tokens": 678212127.0, + "step": 26211 + }, + { + "epoch": 2.8785416209092904, + "grad_norm": 2.237236261367798, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7368566989898682, + "num_tokens": 678240686.0, + "step": 26212 + }, + { + "epoch": 2.878651438611904, + "grad_norm": 2.1780734062194824, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7600082159042358, + "num_tokens": 678270121.0, + "step": 26213 + }, + { + "epoch": 2.878761256314518, + "grad_norm": 2.1312880516052246, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7028642892837524, + "num_tokens": 678300259.0, + "step": 26214 + }, + { + "epoch": 2.8788710740171317, + "grad_norm": 2.258708953857422, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7226409912109375, + "num_tokens": 678325595.0, + "step": 26215 + }, + { + "epoch": 2.8789808917197455, + "grad_norm": 2.4146463871002197, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.725575864315033, + "num_tokens": 678351866.0, + "step": 26216 + }, + { + "epoch": 2.879090709422359, + "grad_norm": 2.2726328372955322, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7205511331558228, + "num_tokens": 678380016.0, + "step": 26217 + }, + { + "epoch": 2.8792005271249725, + "grad_norm": 2.381664276123047, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7441354393959045, + "num_tokens": 678406271.0, + "step": 26218 + }, + { + "epoch": 2.8793103448275863, + "grad_norm": 2.2621283531188965, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7401172518730164, + "num_tokens": 678433377.0, + "step": 26219 + }, + { + "epoch": 2.8794201625301996, + "grad_norm": 2.4045565128326416, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7381173372268677, + "num_tokens": 678457691.0, + "step": 26220 + }, + { + "epoch": 2.8795299802328134, + "grad_norm": 2.2080719470977783, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7181092500686646, + "num_tokens": 678486304.0, + "step": 26221 + }, + { + "epoch": 2.879639797935427, + "grad_norm": 2.3069984912872314, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7191804647445679, + "num_tokens": 678512215.0, + "step": 26222 + }, + { + "epoch": 2.879749615638041, + "grad_norm": 2.282541513442993, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7347648739814758, + "num_tokens": 678537965.0, + "step": 26223 + }, + { + "epoch": 2.8798594333406546, + "grad_norm": 2.149322509765625, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7367019057273865, + "num_tokens": 678567305.0, + "step": 26224 + }, + { + "epoch": 2.879969251043268, + "grad_norm": 2.64266300201416, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7378735542297363, + "num_tokens": 678588741.0, + "step": 26225 + }, + { + "epoch": 2.8800790687458817, + "grad_norm": 2.4138693809509277, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7251973748207092, + "num_tokens": 678614744.0, + "step": 26226 + }, + { + "epoch": 2.8801888864484955, + "grad_norm": 2.8325769901275635, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7516745328903198, + "num_tokens": 678633366.0, + "step": 26227 + }, + { + "epoch": 2.8802987041511092, + "grad_norm": 2.6503474712371826, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7447682619094849, + "num_tokens": 678655254.0, + "step": 26228 + }, + { + "epoch": 2.880408521853723, + "grad_norm": 2.433185338973999, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7537016272544861, + "num_tokens": 678678121.0, + "step": 26229 + }, + { + "epoch": 2.8805183395563363, + "grad_norm": 2.088700294494629, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7279558181762695, + "num_tokens": 678708706.0, + "step": 26230 + }, + { + "epoch": 2.88062815725895, + "grad_norm": 2.260592222213745, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7168852090835571, + "num_tokens": 678736911.0, + "step": 26231 + }, + { + "epoch": 2.880737974961564, + "grad_norm": 2.4015462398529053, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7300374507904053, + "num_tokens": 678760187.0, + "step": 26232 + }, + { + "epoch": 2.8808477926641776, + "grad_norm": 2.5611259937286377, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.735298216342926, + "num_tokens": 678783427.0, + "step": 26233 + }, + { + "epoch": 2.8809576103667913, + "grad_norm": 2.5506904125213623, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7373957633972168, + "num_tokens": 678806767.0, + "step": 26234 + }, + { + "epoch": 2.8810674280694046, + "grad_norm": 2.571021556854248, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7274684906005859, + "num_tokens": 678831097.0, + "step": 26235 + }, + { + "epoch": 2.8811772457720184, + "grad_norm": 2.244168996810913, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7184418439865112, + "num_tokens": 678859565.0, + "step": 26236 + }, + { + "epoch": 2.881287063474632, + "grad_norm": 2.228411912918091, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.732788622379303, + "num_tokens": 678887636.0, + "step": 26237 + }, + { + "epoch": 2.881396881177246, + "grad_norm": 2.3211829662323, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7304509878158569, + "num_tokens": 678915873.0, + "step": 26238 + }, + { + "epoch": 2.8815066988798597, + "grad_norm": 2.2845258712768555, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7274272441864014, + "num_tokens": 678941675.0, + "step": 26239 + }, + { + "epoch": 2.881616516582473, + "grad_norm": 2.502244472503662, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7399542927742004, + "num_tokens": 678965197.0, + "step": 26240 + }, + { + "epoch": 2.8817263342850867, + "grad_norm": 2.4487264156341553, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7099971771240234, + "num_tokens": 678990892.0, + "step": 26241 + }, + { + "epoch": 2.8818361519877005, + "grad_norm": 2.499160051345825, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7459467053413391, + "num_tokens": 679012837.0, + "step": 26242 + }, + { + "epoch": 2.881945969690314, + "grad_norm": 2.1524343490600586, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7052186131477356, + "num_tokens": 679045032.0, + "step": 26243 + }, + { + "epoch": 2.882055787392928, + "grad_norm": 2.7906410694122314, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7396534085273743, + "num_tokens": 679064203.0, + "step": 26244 + }, + { + "epoch": 2.8821656050955413, + "grad_norm": 2.2693543434143066, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7272236943244934, + "num_tokens": 679091818.0, + "step": 26245 + }, + { + "epoch": 2.882275422798155, + "grad_norm": 2.4437179565429688, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7606638073921204, + "num_tokens": 679117480.0, + "step": 26246 + }, + { + "epoch": 2.882385240500769, + "grad_norm": 2.2170488834381104, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7226132154464722, + "num_tokens": 679147118.0, + "step": 26247 + }, + { + "epoch": 2.882495058203382, + "grad_norm": 2.4465293884277344, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7212592959403992, + "num_tokens": 679171613.0, + "step": 26248 + }, + { + "epoch": 2.882604875905996, + "grad_norm": 2.258948564529419, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7317878007888794, + "num_tokens": 679200715.0, + "step": 26249 + }, + { + "epoch": 2.8827146936086097, + "grad_norm": 2.1272144317626953, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7411679029464722, + "num_tokens": 679229449.0, + "step": 26250 + }, + { + "epoch": 2.8828245113112234, + "grad_norm": 2.362872838973999, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.707736611366272, + "num_tokens": 679255059.0, + "step": 26251 + }, + { + "epoch": 2.882934329013837, + "grad_norm": 2.7348060607910156, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7385272979736328, + "num_tokens": 679274261.0, + "step": 26252 + }, + { + "epoch": 2.8830441467164505, + "grad_norm": 2.0644102096557617, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7562189102172852, + "num_tokens": 679302410.0, + "step": 26253 + }, + { + "epoch": 2.8831539644190642, + "grad_norm": 2.3129403591156006, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7551623582839966, + "num_tokens": 679327577.0, + "step": 26254 + }, + { + "epoch": 2.883263782121678, + "grad_norm": 2.355541706085205, + "learning_rate": 1e-06, + "loss": 0.781, + "mean_token_accuracy": 0.7443494200706482, + "num_tokens": 679353635.0, + "step": 26255 + }, + { + "epoch": 2.8833735998242918, + "grad_norm": 2.3606491088867188, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7245780229568481, + "num_tokens": 679382008.0, + "step": 26256 + }, + { + "epoch": 2.8834834175269055, + "grad_norm": 2.2676308155059814, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.721531331539154, + "num_tokens": 679410828.0, + "step": 26257 + }, + { + "epoch": 2.883593235229519, + "grad_norm": 2.0201289653778076, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7276177406311035, + "num_tokens": 679443533.0, + "step": 26258 + }, + { + "epoch": 2.8837030529321326, + "grad_norm": 2.774583101272583, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7463948726654053, + "num_tokens": 679463114.0, + "step": 26259 + }, + { + "epoch": 2.8838128706347463, + "grad_norm": 2.2289462089538574, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7568237781524658, + "num_tokens": 679490246.0, + "step": 26260 + }, + { + "epoch": 2.88392268833736, + "grad_norm": 2.1925253868103027, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7375262975692749, + "num_tokens": 679517195.0, + "step": 26261 + }, + { + "epoch": 2.884032506039974, + "grad_norm": 2.5888922214508057, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7425982356071472, + "num_tokens": 679540984.0, + "step": 26262 + }, + { + "epoch": 2.884142323742587, + "grad_norm": 2.6263139247894287, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7234017252922058, + "num_tokens": 679564476.0, + "step": 26263 + }, + { + "epoch": 2.884252141445201, + "grad_norm": 2.2442855834960938, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7305848598480225, + "num_tokens": 679591224.0, + "step": 26264 + }, + { + "epoch": 2.8843619591478147, + "grad_norm": 2.228501081466675, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7550901174545288, + "num_tokens": 679618867.0, + "step": 26265 + }, + { + "epoch": 2.8844717768504284, + "grad_norm": 2.2290589809417725, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.718049168586731, + "num_tokens": 679647701.0, + "step": 26266 + }, + { + "epoch": 2.884581594553042, + "grad_norm": 2.821516752243042, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7515387535095215, + "num_tokens": 679666694.0, + "step": 26267 + }, + { + "epoch": 2.8846914122556555, + "grad_norm": 2.4200503826141357, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7449929118156433, + "num_tokens": 679690306.0, + "step": 26268 + }, + { + "epoch": 2.8848012299582693, + "grad_norm": 2.788438558578491, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7283564805984497, + "num_tokens": 679710856.0, + "step": 26269 + }, + { + "epoch": 2.884911047660883, + "grad_norm": 2.1611077785491943, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.714741051197052, + "num_tokens": 679740890.0, + "step": 26270 + }, + { + "epoch": 2.8850208653634963, + "grad_norm": 2.1040616035461426, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7321511507034302, + "num_tokens": 679772541.0, + "step": 26271 + }, + { + "epoch": 2.88513068306611, + "grad_norm": 2.664867401123047, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7418819665908813, + "num_tokens": 679792942.0, + "step": 26272 + }, + { + "epoch": 2.885240500768724, + "grad_norm": 2.2276813983917236, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7195686101913452, + "num_tokens": 679821926.0, + "step": 26273 + }, + { + "epoch": 2.8853503184713376, + "grad_norm": 2.319814920425415, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7442914247512817, + "num_tokens": 679846469.0, + "step": 26274 + }, + { + "epoch": 2.8854601361739514, + "grad_norm": 2.108583927154541, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7166738510131836, + "num_tokens": 679876950.0, + "step": 26275 + }, + { + "epoch": 2.8855699538765647, + "grad_norm": 2.153409242630005, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7142268419265747, + "num_tokens": 679907480.0, + "step": 26276 + }, + { + "epoch": 2.8856797715791784, + "grad_norm": 2.3853979110717773, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7401353716850281, + "num_tokens": 679932416.0, + "step": 26277 + }, + { + "epoch": 2.885789589281792, + "grad_norm": 2.3649725914001465, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7296006083488464, + "num_tokens": 679956498.0, + "step": 26278 + }, + { + "epoch": 2.885899406984406, + "grad_norm": 2.194330930709839, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7207301259040833, + "num_tokens": 679983183.0, + "step": 26279 + }, + { + "epoch": 2.8860092246870197, + "grad_norm": 2.1235721111297607, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7573229670524597, + "num_tokens": 680013070.0, + "step": 26280 + }, + { + "epoch": 2.886119042389633, + "grad_norm": 2.491328716278076, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7183574438095093, + "num_tokens": 680038849.0, + "step": 26281 + }, + { + "epoch": 2.8862288600922468, + "grad_norm": 2.326154947280884, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7312666177749634, + "num_tokens": 680064328.0, + "step": 26282 + }, + { + "epoch": 2.8863386777948605, + "grad_norm": 2.3922080993652344, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7113099694252014, + "num_tokens": 680091408.0, + "step": 26283 + }, + { + "epoch": 2.8864484954974743, + "grad_norm": 2.105257272720337, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7377043962478638, + "num_tokens": 680123767.0, + "step": 26284 + }, + { + "epoch": 2.886558313200088, + "grad_norm": 2.8247883319854736, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.726304292678833, + "num_tokens": 680143812.0, + "step": 26285 + }, + { + "epoch": 2.8866681309027014, + "grad_norm": 2.3081796169281006, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7546758651733398, + "num_tokens": 680169260.0, + "step": 26286 + }, + { + "epoch": 2.886777948605315, + "grad_norm": 2.3446390628814697, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.745667576789856, + "num_tokens": 680193558.0, + "step": 26287 + }, + { + "epoch": 2.886887766307929, + "grad_norm": 2.3280935287475586, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7246472239494324, + "num_tokens": 680220597.0, + "step": 26288 + }, + { + "epoch": 2.8869975840105426, + "grad_norm": 2.734440803527832, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7515494227409363, + "num_tokens": 680240737.0, + "step": 26289 + }, + { + "epoch": 2.8871074017131564, + "grad_norm": 2.3285105228424072, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7417671084403992, + "num_tokens": 680266014.0, + "step": 26290 + }, + { + "epoch": 2.8872172194157697, + "grad_norm": 2.426842451095581, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7363346815109253, + "num_tokens": 680290115.0, + "step": 26291 + }, + { + "epoch": 2.8873270371183835, + "grad_norm": 2.3014180660247803, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7250670194625854, + "num_tokens": 680318250.0, + "step": 26292 + }, + { + "epoch": 2.887436854820997, + "grad_norm": 2.397196054458618, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7308735847473145, + "num_tokens": 680342706.0, + "step": 26293 + }, + { + "epoch": 2.887546672523611, + "grad_norm": 2.2217445373535156, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7199392318725586, + "num_tokens": 680373243.0, + "step": 26294 + }, + { + "epoch": 2.8876564902262247, + "grad_norm": 2.0069997310638428, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7263767719268799, + "num_tokens": 680404986.0, + "step": 26295 + }, + { + "epoch": 2.887766307928838, + "grad_norm": 2.3546361923217773, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7274889945983887, + "num_tokens": 680430597.0, + "step": 26296 + }, + { + "epoch": 2.887876125631452, + "grad_norm": 2.218684434890747, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.745515763759613, + "num_tokens": 680458734.0, + "step": 26297 + }, + { + "epoch": 2.8879859433340656, + "grad_norm": 2.2075512409210205, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7058041095733643, + "num_tokens": 680490034.0, + "step": 26298 + }, + { + "epoch": 2.888095761036679, + "grad_norm": 2.335512638092041, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7462233304977417, + "num_tokens": 680515547.0, + "step": 26299 + }, + { + "epoch": 2.8882055787392926, + "grad_norm": 2.4607675075531006, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7383993864059448, + "num_tokens": 680538587.0, + "step": 26300 + }, + { + "epoch": 2.8883153964419064, + "grad_norm": 2.332122802734375, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7252563238143921, + "num_tokens": 680565636.0, + "step": 26301 + }, + { + "epoch": 2.88842521414452, + "grad_norm": 2.3998982906341553, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7389856576919556, + "num_tokens": 680589704.0, + "step": 26302 + }, + { + "epoch": 2.888535031847134, + "grad_norm": 2.5665245056152344, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7244888544082642, + "num_tokens": 680611431.0, + "step": 26303 + }, + { + "epoch": 2.888644849549747, + "grad_norm": 2.610043525695801, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7230088114738464, + "num_tokens": 680634011.0, + "step": 26304 + }, + { + "epoch": 2.888754667252361, + "grad_norm": 2.3981869220733643, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7493854761123657, + "num_tokens": 680657458.0, + "step": 26305 + }, + { + "epoch": 2.8888644849549747, + "grad_norm": 2.1840176582336426, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7589952945709229, + "num_tokens": 680683301.0, + "step": 26306 + }, + { + "epoch": 2.8889743026575885, + "grad_norm": 2.4309840202331543, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7144399881362915, + "num_tokens": 680709705.0, + "step": 26307 + }, + { + "epoch": 2.8890841203602022, + "grad_norm": 2.3440632820129395, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7235455513000488, + "num_tokens": 680736867.0, + "step": 26308 + }, + { + "epoch": 2.8891939380628155, + "grad_norm": 2.466400623321533, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7318257689476013, + "num_tokens": 680761365.0, + "step": 26309 + }, + { + "epoch": 2.8893037557654293, + "grad_norm": 2.434830665588379, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7151015400886536, + "num_tokens": 680785840.0, + "step": 26310 + }, + { + "epoch": 2.889413573468043, + "grad_norm": 2.4618728160858154, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7326459884643555, + "num_tokens": 680811875.0, + "step": 26311 + }, + { + "epoch": 2.889523391170657, + "grad_norm": 2.4295716285705566, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7660066485404968, + "num_tokens": 680834686.0, + "step": 26312 + }, + { + "epoch": 2.8896332088732706, + "grad_norm": 2.1973907947540283, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7238122224807739, + "num_tokens": 680863684.0, + "step": 26313 + }, + { + "epoch": 2.889743026575884, + "grad_norm": 2.4775094985961914, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7345819473266602, + "num_tokens": 680886106.0, + "step": 26314 + }, + { + "epoch": 2.8898528442784976, + "grad_norm": 2.226147174835205, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7247980237007141, + "num_tokens": 680914405.0, + "step": 26315 + }, + { + "epoch": 2.8899626619811114, + "grad_norm": 2.188567638397217, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7299536466598511, + "num_tokens": 680943760.0, + "step": 26316 + }, + { + "epoch": 2.890072479683725, + "grad_norm": 2.275393009185791, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7420990467071533, + "num_tokens": 680971377.0, + "step": 26317 + }, + { + "epoch": 2.890182297386339, + "grad_norm": 2.0765128135681152, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7265045642852783, + "num_tokens": 681002931.0, + "step": 26318 + }, + { + "epoch": 2.8902921150889522, + "grad_norm": 2.26865816116333, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7288254499435425, + "num_tokens": 681030516.0, + "step": 26319 + }, + { + "epoch": 2.890401932791566, + "grad_norm": 2.165602207183838, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7062305212020874, + "num_tokens": 681059702.0, + "step": 26320 + }, + { + "epoch": 2.8905117504941797, + "grad_norm": 2.1987380981445312, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6881870031356812, + "num_tokens": 681092970.0, + "step": 26321 + }, + { + "epoch": 2.890621568196793, + "grad_norm": 2.549100637435913, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7489530444145203, + "num_tokens": 681116302.0, + "step": 26322 + }, + { + "epoch": 2.8907313858994073, + "grad_norm": 2.3969109058380127, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.723568320274353, + "num_tokens": 681140960.0, + "step": 26323 + }, + { + "epoch": 2.8908412036020206, + "grad_norm": 2.4838900566101074, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7228132486343384, + "num_tokens": 681166078.0, + "step": 26324 + }, + { + "epoch": 2.8909510213046343, + "grad_norm": 2.4559438228607178, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7124754786491394, + "num_tokens": 681194378.0, + "step": 26325 + }, + { + "epoch": 2.891060839007248, + "grad_norm": 2.336630344390869, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7290415167808533, + "num_tokens": 681220387.0, + "step": 26326 + }, + { + "epoch": 2.8911706567098614, + "grad_norm": 2.3623993396759033, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7241056561470032, + "num_tokens": 681246376.0, + "step": 26327 + }, + { + "epoch": 2.891280474412475, + "grad_norm": 2.19858717918396, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7151713967323303, + "num_tokens": 681277320.0, + "step": 26328 + }, + { + "epoch": 2.891390292115089, + "grad_norm": 2.4774153232574463, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7457534074783325, + "num_tokens": 681301160.0, + "step": 26329 + }, + { + "epoch": 2.8915001098177027, + "grad_norm": 2.587228536605835, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7416452169418335, + "num_tokens": 681322226.0, + "step": 26330 + }, + { + "epoch": 2.8916099275203164, + "grad_norm": 2.6118953227996826, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.743657648563385, + "num_tokens": 681345895.0, + "step": 26331 + }, + { + "epoch": 2.8917197452229297, + "grad_norm": 2.563438892364502, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7440786957740784, + "num_tokens": 681369058.0, + "step": 26332 + }, + { + "epoch": 2.8918295629255435, + "grad_norm": 2.458247661590576, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7232604026794434, + "num_tokens": 681393988.0, + "step": 26333 + }, + { + "epoch": 2.8919393806281573, + "grad_norm": 2.4386327266693115, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7443906664848328, + "num_tokens": 681419271.0, + "step": 26334 + }, + { + "epoch": 2.892049198330771, + "grad_norm": 2.4541409015655518, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7154693603515625, + "num_tokens": 681445181.0, + "step": 26335 + }, + { + "epoch": 2.8921590160333848, + "grad_norm": 2.029036521911621, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7160075902938843, + "num_tokens": 681478406.0, + "step": 26336 + }, + { + "epoch": 2.892268833735998, + "grad_norm": 2.439177989959717, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7281807661056519, + "num_tokens": 681502338.0, + "step": 26337 + }, + { + "epoch": 2.892378651438612, + "grad_norm": 2.4555673599243164, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7133402228355408, + "num_tokens": 681531407.0, + "step": 26338 + }, + { + "epoch": 2.8924884691412256, + "grad_norm": 2.1150829792022705, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7146569490432739, + "num_tokens": 681563462.0, + "step": 26339 + }, + { + "epoch": 2.8925982868438394, + "grad_norm": 2.7302417755126953, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7503483295440674, + "num_tokens": 681584214.0, + "step": 26340 + }, + { + "epoch": 2.892708104546453, + "grad_norm": 2.215440273284912, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7435382604598999, + "num_tokens": 681613791.0, + "step": 26341 + }, + { + "epoch": 2.8928179222490664, + "grad_norm": 2.2085955142974854, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7204151153564453, + "num_tokens": 681646841.0, + "step": 26342 + }, + { + "epoch": 2.89292773995168, + "grad_norm": 2.316970109939575, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7155371308326721, + "num_tokens": 681672239.0, + "step": 26343 + }, + { + "epoch": 2.893037557654294, + "grad_norm": 2.7058184146881104, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7271572351455688, + "num_tokens": 681692934.0, + "step": 26344 + }, + { + "epoch": 2.8931473753569077, + "grad_norm": 2.384493827819824, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7526124715805054, + "num_tokens": 681716216.0, + "step": 26345 + }, + { + "epoch": 2.8932571930595214, + "grad_norm": 2.1992807388305664, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7262520790100098, + "num_tokens": 681745947.0, + "step": 26346 + }, + { + "epoch": 2.8933670107621348, + "grad_norm": 2.7414567470550537, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7463840842247009, + "num_tokens": 681764836.0, + "step": 26347 + }, + { + "epoch": 2.8934768284647485, + "grad_norm": 2.4341180324554443, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7213171720504761, + "num_tokens": 681790826.0, + "step": 26348 + }, + { + "epoch": 2.8935866461673623, + "grad_norm": 2.256014108657837, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7383005023002625, + "num_tokens": 681818482.0, + "step": 26349 + }, + { + "epoch": 2.8936964638699756, + "grad_norm": 2.496448278427124, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7159954309463501, + "num_tokens": 681843663.0, + "step": 26350 + }, + { + "epoch": 2.8938062815725893, + "grad_norm": 2.31913685798645, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7303429245948792, + "num_tokens": 681870011.0, + "step": 26351 + }, + { + "epoch": 2.893916099275203, + "grad_norm": 2.398580312728882, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7475782632827759, + "num_tokens": 681894347.0, + "step": 26352 + }, + { + "epoch": 2.894025916977817, + "grad_norm": 2.184804916381836, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7439631819725037, + "num_tokens": 681921147.0, + "step": 26353 + }, + { + "epoch": 2.8941357346804306, + "grad_norm": 2.340114116668701, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7423080801963806, + "num_tokens": 681946221.0, + "step": 26354 + }, + { + "epoch": 2.894245552383044, + "grad_norm": 2.21777606010437, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7019614577293396, + "num_tokens": 681977249.0, + "step": 26355 + }, + { + "epoch": 2.8943553700856577, + "grad_norm": 2.2594945430755615, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.729988157749176, + "num_tokens": 682004527.0, + "step": 26356 + }, + { + "epoch": 2.8944651877882714, + "grad_norm": 2.407104253768921, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7563353776931763, + "num_tokens": 682028877.0, + "step": 26357 + }, + { + "epoch": 2.894575005490885, + "grad_norm": 2.295823574066162, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7432487607002258, + "num_tokens": 682055359.0, + "step": 26358 + }, + { + "epoch": 2.894684823193499, + "grad_norm": 2.6710257530212402, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7394965291023254, + "num_tokens": 682077733.0, + "step": 26359 + }, + { + "epoch": 2.8947946408961123, + "grad_norm": 2.635396718978882, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7233980894088745, + "num_tokens": 682099772.0, + "step": 26360 + }, + { + "epoch": 2.894904458598726, + "grad_norm": 2.3203654289245605, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7409635782241821, + "num_tokens": 682127088.0, + "step": 26361 + }, + { + "epoch": 2.89501427630134, + "grad_norm": 2.610605478286743, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7361326813697815, + "num_tokens": 682149027.0, + "step": 26362 + }, + { + "epoch": 2.8951240940039535, + "grad_norm": 2.550147533416748, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7242748737335205, + "num_tokens": 682172011.0, + "step": 26363 + }, + { + "epoch": 2.8952339117065673, + "grad_norm": 2.3860936164855957, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7179349660873413, + "num_tokens": 682199704.0, + "step": 26364 + }, + { + "epoch": 2.8953437294091806, + "grad_norm": 2.6388614177703857, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7718885540962219, + "num_tokens": 682219918.0, + "step": 26365 + }, + { + "epoch": 2.8954535471117944, + "grad_norm": 2.7227180004119873, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7454631328582764, + "num_tokens": 682241287.0, + "step": 26366 + }, + { + "epoch": 2.895563364814408, + "grad_norm": 2.383418321609497, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.726902961730957, + "num_tokens": 682266925.0, + "step": 26367 + }, + { + "epoch": 2.895673182517022, + "grad_norm": 2.8251914978027344, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7434543371200562, + "num_tokens": 682287503.0, + "step": 26368 + }, + { + "epoch": 2.8957830002196356, + "grad_norm": 2.367819309234619, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.75701904296875, + "num_tokens": 682311142.0, + "step": 26369 + }, + { + "epoch": 2.895892817922249, + "grad_norm": 2.0015170574188232, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7170486450195312, + "num_tokens": 682346342.0, + "step": 26370 + }, + { + "epoch": 2.8960026356248627, + "grad_norm": 2.467181921005249, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7294671535491943, + "num_tokens": 682370628.0, + "step": 26371 + }, + { + "epoch": 2.8961124533274765, + "grad_norm": 2.613461971282959, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7344232201576233, + "num_tokens": 682393547.0, + "step": 26372 + }, + { + "epoch": 2.89622227103009, + "grad_norm": 2.270012140274048, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7416347861289978, + "num_tokens": 682421741.0, + "step": 26373 + }, + { + "epoch": 2.896332088732704, + "grad_norm": 2.258551597595215, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7269847393035889, + "num_tokens": 682449528.0, + "step": 26374 + }, + { + "epoch": 2.8964419064353173, + "grad_norm": 2.4772374629974365, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7193218469619751, + "num_tokens": 682473028.0, + "step": 26375 + }, + { + "epoch": 2.896551724137931, + "grad_norm": 2.358569860458374, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7232990264892578, + "num_tokens": 682500082.0, + "step": 26376 + }, + { + "epoch": 2.896661541840545, + "grad_norm": 2.2577292919158936, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.721611738204956, + "num_tokens": 682530425.0, + "step": 26377 + }, + { + "epoch": 2.896771359543158, + "grad_norm": 2.581756353378296, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7231228947639465, + "num_tokens": 682554091.0, + "step": 26378 + }, + { + "epoch": 2.896881177245772, + "grad_norm": 2.3646907806396484, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.726481556892395, + "num_tokens": 682579492.0, + "step": 26379 + }, + { + "epoch": 2.8969909949483856, + "grad_norm": 2.4874963760375977, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7537999153137207, + "num_tokens": 682603660.0, + "step": 26380 + }, + { + "epoch": 2.8971008126509994, + "grad_norm": 2.3577969074249268, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7207005023956299, + "num_tokens": 682631859.0, + "step": 26381 + }, + { + "epoch": 2.897210630353613, + "grad_norm": 2.4809398651123047, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.744910478591919, + "num_tokens": 682656306.0, + "step": 26382 + }, + { + "epoch": 2.8973204480562265, + "grad_norm": 2.5118424892425537, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.734785258769989, + "num_tokens": 682679307.0, + "step": 26383 + }, + { + "epoch": 2.89743026575884, + "grad_norm": 2.339132070541382, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7380131483078003, + "num_tokens": 682704005.0, + "step": 26384 + }, + { + "epoch": 2.897540083461454, + "grad_norm": 2.291254758834839, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7444248795509338, + "num_tokens": 682730369.0, + "step": 26385 + }, + { + "epoch": 2.8976499011640677, + "grad_norm": 2.118598699569702, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7202510833740234, + "num_tokens": 682761701.0, + "step": 26386 + }, + { + "epoch": 2.8977597188666815, + "grad_norm": 2.236107587814331, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7385984063148499, + "num_tokens": 682787398.0, + "step": 26387 + }, + { + "epoch": 2.897869536569295, + "grad_norm": 2.3865411281585693, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7235749363899231, + "num_tokens": 682813490.0, + "step": 26388 + }, + { + "epoch": 2.8979793542719086, + "grad_norm": 2.5105323791503906, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.743554949760437, + "num_tokens": 682839551.0, + "step": 26389 + }, + { + "epoch": 2.8980891719745223, + "grad_norm": 2.380683660507202, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7265936136245728, + "num_tokens": 682864892.0, + "step": 26390 + }, + { + "epoch": 2.898198989677136, + "grad_norm": 2.351040840148926, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7134681940078735, + "num_tokens": 682892896.0, + "step": 26391 + }, + { + "epoch": 2.89830880737975, + "grad_norm": 2.59724497795105, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7299870848655701, + "num_tokens": 682915798.0, + "step": 26392 + }, + { + "epoch": 2.898418625082363, + "grad_norm": 2.339320182800293, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7316338419914246, + "num_tokens": 682940538.0, + "step": 26393 + }, + { + "epoch": 2.898528442784977, + "grad_norm": 2.2560324668884277, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7074216604232788, + "num_tokens": 682972109.0, + "step": 26394 + }, + { + "epoch": 2.8986382604875907, + "grad_norm": 2.252100944519043, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7239791750907898, + "num_tokens": 682998786.0, + "step": 26395 + }, + { + "epoch": 2.8987480781902044, + "grad_norm": 2.415160894393921, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7098333835601807, + "num_tokens": 683025291.0, + "step": 26396 + }, + { + "epoch": 2.898857895892818, + "grad_norm": 2.1492443084716797, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7357999086380005, + "num_tokens": 683056370.0, + "step": 26397 + }, + { + "epoch": 2.8989677135954315, + "grad_norm": 2.300731897354126, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7307378649711609, + "num_tokens": 683084465.0, + "step": 26398 + }, + { + "epoch": 2.8990775312980452, + "grad_norm": 2.5830159187316895, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7315016984939575, + "num_tokens": 683108337.0, + "step": 26399 + }, + { + "epoch": 2.899187349000659, + "grad_norm": 2.392293691635132, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7266120314598083, + "num_tokens": 683134858.0, + "step": 26400 + }, + { + "epoch": 2.8992971667032723, + "grad_norm": 2.5510010719299316, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7301294803619385, + "num_tokens": 683158821.0, + "step": 26401 + }, + { + "epoch": 2.899406984405886, + "grad_norm": 2.2951691150665283, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.722330629825592, + "num_tokens": 683185864.0, + "step": 26402 + }, + { + "epoch": 2.8995168021085, + "grad_norm": 2.374673843383789, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7261747121810913, + "num_tokens": 683211009.0, + "step": 26403 + }, + { + "epoch": 2.8996266198111136, + "grad_norm": 2.1719110012054443, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7442858219146729, + "num_tokens": 683240253.0, + "step": 26404 + }, + { + "epoch": 2.8997364375137273, + "grad_norm": 2.5175416469573975, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7204465866088867, + "num_tokens": 683264149.0, + "step": 26405 + }, + { + "epoch": 2.8998462552163407, + "grad_norm": 2.345224142074585, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7208187580108643, + "num_tokens": 683292588.0, + "step": 26406 + }, + { + "epoch": 2.8999560729189544, + "grad_norm": 2.543657064437866, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7446160316467285, + "num_tokens": 683314964.0, + "step": 26407 + }, + { + "epoch": 2.900065890621568, + "grad_norm": 2.2220921516418457, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.769597053527832, + "num_tokens": 683340361.0, + "step": 26408 + }, + { + "epoch": 2.900175708324182, + "grad_norm": 2.601766347885132, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7385344505310059, + "num_tokens": 683362288.0, + "step": 26409 + }, + { + "epoch": 2.9002855260267957, + "grad_norm": 2.333050012588501, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7287411093711853, + "num_tokens": 683387158.0, + "step": 26410 + }, + { + "epoch": 2.900395343729409, + "grad_norm": 2.7373111248016357, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7234311103820801, + "num_tokens": 683408163.0, + "step": 26411 + }, + { + "epoch": 2.9005051614320227, + "grad_norm": 2.4293878078460693, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.725629985332489, + "num_tokens": 683434116.0, + "step": 26412 + }, + { + "epoch": 2.9006149791346365, + "grad_norm": 2.221436023712158, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7299439907073975, + "num_tokens": 683463104.0, + "step": 26413 + }, + { + "epoch": 2.9007247968372503, + "grad_norm": 2.3395938873291016, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7248796820640564, + "num_tokens": 683488989.0, + "step": 26414 + }, + { + "epoch": 2.900834614539864, + "grad_norm": 2.09926438331604, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.724223256111145, + "num_tokens": 683520293.0, + "step": 26415 + }, + { + "epoch": 2.9009444322424773, + "grad_norm": 2.67453932762146, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7560833096504211, + "num_tokens": 683542115.0, + "step": 26416 + }, + { + "epoch": 2.901054249945091, + "grad_norm": 2.232314348220825, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7331894040107727, + "num_tokens": 683572409.0, + "step": 26417 + }, + { + "epoch": 2.901164067647705, + "grad_norm": 2.466435432434082, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7345463037490845, + "num_tokens": 683597498.0, + "step": 26418 + }, + { + "epoch": 2.9012738853503186, + "grad_norm": 1.9892386198043823, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7444679141044617, + "num_tokens": 683632653.0, + "step": 26419 + }, + { + "epoch": 2.9013837030529324, + "grad_norm": 2.166069269180298, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7319080829620361, + "num_tokens": 683662090.0, + "step": 26420 + }, + { + "epoch": 2.9014935207555457, + "grad_norm": 2.614471673965454, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7369571328163147, + "num_tokens": 683684165.0, + "step": 26421 + }, + { + "epoch": 2.9016033384581594, + "grad_norm": 2.394601583480835, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7159687280654907, + "num_tokens": 683710140.0, + "step": 26422 + }, + { + "epoch": 2.901713156160773, + "grad_norm": 2.4819159507751465, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7140302658081055, + "num_tokens": 683738357.0, + "step": 26423 + }, + { + "epoch": 2.9018229738633865, + "grad_norm": 2.183547258377075, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7552890777587891, + "num_tokens": 683765150.0, + "step": 26424 + }, + { + "epoch": 2.9019327915660007, + "grad_norm": 2.438584327697754, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7102273106575012, + "num_tokens": 683791127.0, + "step": 26425 + }, + { + "epoch": 2.902042609268614, + "grad_norm": 2.4447999000549316, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7376465797424316, + "num_tokens": 683814724.0, + "step": 26426 + }, + { + "epoch": 2.9021524269712278, + "grad_norm": 2.730436325073242, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7641215324401855, + "num_tokens": 683834075.0, + "step": 26427 + }, + { + "epoch": 2.9022622446738415, + "grad_norm": 2.044635534286499, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7446107864379883, + "num_tokens": 683865794.0, + "step": 26428 + }, + { + "epoch": 2.902372062376455, + "grad_norm": 2.427354335784912, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7360644340515137, + "num_tokens": 683891188.0, + "step": 26429 + }, + { + "epoch": 2.9024818800790686, + "grad_norm": 2.0290260314941406, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6974058151245117, + "num_tokens": 683926150.0, + "step": 26430 + }, + { + "epoch": 2.9025916977816824, + "grad_norm": 2.658327579498291, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7651911377906799, + "num_tokens": 683946763.0, + "step": 26431 + }, + { + "epoch": 2.902701515484296, + "grad_norm": 2.3304836750030518, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7360970973968506, + "num_tokens": 683972539.0, + "step": 26432 + }, + { + "epoch": 2.90281133318691, + "grad_norm": 2.538961410522461, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7463622689247131, + "num_tokens": 683995041.0, + "step": 26433 + }, + { + "epoch": 2.902921150889523, + "grad_norm": 2.0451900959014893, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7095948457717896, + "num_tokens": 684028733.0, + "step": 26434 + }, + { + "epoch": 2.903030968592137, + "grad_norm": 2.3171327114105225, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.725354015827179, + "num_tokens": 684056832.0, + "step": 26435 + }, + { + "epoch": 2.9031407862947507, + "grad_norm": 2.1609461307525635, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7574236392974854, + "num_tokens": 684085015.0, + "step": 26436 + }, + { + "epoch": 2.9032506039973645, + "grad_norm": 2.566770553588867, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7483593225479126, + "num_tokens": 684109006.0, + "step": 26437 + }, + { + "epoch": 2.903360421699978, + "grad_norm": 2.5599050521850586, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.7486332654953003, + "num_tokens": 684132450.0, + "step": 26438 + }, + { + "epoch": 2.9034702394025915, + "grad_norm": 2.053046703338623, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7315258979797363, + "num_tokens": 684164024.0, + "step": 26439 + }, + { + "epoch": 2.9035800571052053, + "grad_norm": 2.1795456409454346, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7199488878250122, + "num_tokens": 684193418.0, + "step": 26440 + }, + { + "epoch": 2.903689874807819, + "grad_norm": 2.58467698097229, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7248489260673523, + "num_tokens": 684219420.0, + "step": 26441 + }, + { + "epoch": 2.903799692510433, + "grad_norm": 2.382816791534424, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7840788960456848, + "num_tokens": 684241687.0, + "step": 26442 + }, + { + "epoch": 2.9039095102130466, + "grad_norm": 2.259256601333618, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7380002737045288, + "num_tokens": 684270672.0, + "step": 26443 + }, + { + "epoch": 2.90401932791566, + "grad_norm": 2.4535915851593018, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7232187390327454, + "num_tokens": 684296671.0, + "step": 26444 + }, + { + "epoch": 2.9041291456182736, + "grad_norm": 2.382889747619629, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7250924706459045, + "num_tokens": 684322906.0, + "step": 26445 + }, + { + "epoch": 2.9042389633208874, + "grad_norm": 2.208678722381592, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7224290370941162, + "num_tokens": 684352381.0, + "step": 26446 + }, + { + "epoch": 2.904348781023501, + "grad_norm": 2.4823646545410156, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7360258102416992, + "num_tokens": 684376296.0, + "step": 26447 + }, + { + "epoch": 2.904458598726115, + "grad_norm": 2.483401298522949, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7344979047775269, + "num_tokens": 684400919.0, + "step": 26448 + }, + { + "epoch": 2.904568416428728, + "grad_norm": 2.383718967437744, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7241619825363159, + "num_tokens": 684428314.0, + "step": 26449 + }, + { + "epoch": 2.904678234131342, + "grad_norm": 2.4962403774261475, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7168781757354736, + "num_tokens": 684452540.0, + "step": 26450 + }, + { + "epoch": 2.9047880518339557, + "grad_norm": 2.4832417964935303, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7223571538925171, + "num_tokens": 684477807.0, + "step": 26451 + }, + { + "epoch": 2.904897869536569, + "grad_norm": 2.846247434616089, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7573599815368652, + "num_tokens": 684496994.0, + "step": 26452 + }, + { + "epoch": 2.905007687239183, + "grad_norm": 2.099796772003174, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7189090251922607, + "num_tokens": 684529066.0, + "step": 26453 + }, + { + "epoch": 2.9051175049417965, + "grad_norm": 2.5192489624023438, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7521841526031494, + "num_tokens": 684551104.0, + "step": 26454 + }, + { + "epoch": 2.9052273226444103, + "grad_norm": 2.34440016746521, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7333126664161682, + "num_tokens": 684577229.0, + "step": 26455 + }, + { + "epoch": 2.905337140347024, + "grad_norm": 2.2149157524108887, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7367754578590393, + "num_tokens": 684605572.0, + "step": 26456 + }, + { + "epoch": 2.9054469580496374, + "grad_norm": 2.1451029777526855, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7373542189598083, + "num_tokens": 684636023.0, + "step": 26457 + }, + { + "epoch": 2.905556775752251, + "grad_norm": 2.418710708618164, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7247332334518433, + "num_tokens": 684663762.0, + "step": 26458 + }, + { + "epoch": 2.905666593454865, + "grad_norm": 2.5864858627319336, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.747114896774292, + "num_tokens": 684685399.0, + "step": 26459 + }, + { + "epoch": 2.9057764111574786, + "grad_norm": 2.5624754428863525, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7572238445281982, + "num_tokens": 684706935.0, + "step": 26460 + }, + { + "epoch": 2.9058862288600924, + "grad_norm": 2.2866122722625732, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.737104058265686, + "num_tokens": 684735676.0, + "step": 26461 + }, + { + "epoch": 2.9059960465627057, + "grad_norm": 2.1525914669036865, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7254621982574463, + "num_tokens": 684767002.0, + "step": 26462 + }, + { + "epoch": 2.9061058642653195, + "grad_norm": 2.335435390472412, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.718095600605011, + "num_tokens": 684794623.0, + "step": 26463 + }, + { + "epoch": 2.9062156819679332, + "grad_norm": 2.415222644805908, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7318018674850464, + "num_tokens": 684820573.0, + "step": 26464 + }, + { + "epoch": 2.906325499670547, + "grad_norm": 2.1408469676971436, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7279086709022522, + "num_tokens": 684848905.0, + "step": 26465 + }, + { + "epoch": 2.9064353173731607, + "grad_norm": 2.4618866443634033, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7596874833106995, + "num_tokens": 684872597.0, + "step": 26466 + }, + { + "epoch": 2.906545135075774, + "grad_norm": 2.3983242511749268, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7052791118621826, + "num_tokens": 684899741.0, + "step": 26467 + }, + { + "epoch": 2.906654952778388, + "grad_norm": 2.5271615982055664, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7456557154655457, + "num_tokens": 684922939.0, + "step": 26468 + }, + { + "epoch": 2.9067647704810016, + "grad_norm": 2.317707061767578, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.729975700378418, + "num_tokens": 684948105.0, + "step": 26469 + }, + { + "epoch": 2.9068745881836153, + "grad_norm": 2.09769344329834, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7175348997116089, + "num_tokens": 684979234.0, + "step": 26470 + }, + { + "epoch": 2.906984405886229, + "grad_norm": 2.5982096195220947, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7435808181762695, + "num_tokens": 685001535.0, + "step": 26471 + }, + { + "epoch": 2.9070942235888424, + "grad_norm": 2.531477928161621, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7337600588798523, + "num_tokens": 685024754.0, + "step": 26472 + }, + { + "epoch": 2.907204041291456, + "grad_norm": 2.208186626434326, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7343745231628418, + "num_tokens": 685054084.0, + "step": 26473 + }, + { + "epoch": 2.90731385899407, + "grad_norm": 2.1692581176757812, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.720708429813385, + "num_tokens": 685084931.0, + "step": 26474 + }, + { + "epoch": 2.9074236766966837, + "grad_norm": 2.319582462310791, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7203094363212585, + "num_tokens": 685112701.0, + "step": 26475 + }, + { + "epoch": 2.9075334943992974, + "grad_norm": 2.3187334537506104, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7265166640281677, + "num_tokens": 685139446.0, + "step": 26476 + }, + { + "epoch": 2.9076433121019107, + "grad_norm": 2.733548879623413, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7528973817825317, + "num_tokens": 685160733.0, + "step": 26477 + }, + { + "epoch": 2.9077531298045245, + "grad_norm": 2.153503656387329, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7251052856445312, + "num_tokens": 685189134.0, + "step": 26478 + }, + { + "epoch": 2.9078629475071383, + "grad_norm": 2.5749690532684326, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7559853792190552, + "num_tokens": 685211228.0, + "step": 26479 + }, + { + "epoch": 2.9079727652097516, + "grad_norm": 2.4199140071868896, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7286627292633057, + "num_tokens": 685236995.0, + "step": 26480 + }, + { + "epoch": 2.9080825829123653, + "grad_norm": 2.435600757598877, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7253928184509277, + "num_tokens": 685262502.0, + "step": 26481 + }, + { + "epoch": 2.908192400614979, + "grad_norm": 2.5070345401763916, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7423725128173828, + "num_tokens": 685283923.0, + "step": 26482 + }, + { + "epoch": 2.908302218317593, + "grad_norm": 2.2116127014160156, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7313697338104248, + "num_tokens": 685311142.0, + "step": 26483 + }, + { + "epoch": 2.9084120360202066, + "grad_norm": 2.503726005554199, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7451943159103394, + "num_tokens": 685334840.0, + "step": 26484 + }, + { + "epoch": 2.90852185372282, + "grad_norm": 2.3273208141326904, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7295803427696228, + "num_tokens": 685362203.0, + "step": 26485 + }, + { + "epoch": 2.9086316714254337, + "grad_norm": 2.512800455093384, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7262346744537354, + "num_tokens": 685385678.0, + "step": 26486 + }, + { + "epoch": 2.9087414891280474, + "grad_norm": 2.183121919631958, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.716775119304657, + "num_tokens": 685416591.0, + "step": 26487 + }, + { + "epoch": 2.908851306830661, + "grad_norm": 2.5848467350006104, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7304783463478088, + "num_tokens": 685439488.0, + "step": 26488 + }, + { + "epoch": 2.908961124533275, + "grad_norm": 2.2166714668273926, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7219272255897522, + "num_tokens": 685467009.0, + "step": 26489 + }, + { + "epoch": 2.9090709422358882, + "grad_norm": 2.272800922393799, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.737891674041748, + "num_tokens": 685492831.0, + "step": 26490 + }, + { + "epoch": 2.909180759938502, + "grad_norm": 2.364795446395874, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7224972248077393, + "num_tokens": 685520269.0, + "step": 26491 + }, + { + "epoch": 2.9092905776411158, + "grad_norm": 2.722593069076538, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.743344783782959, + "num_tokens": 685539928.0, + "step": 26492 + }, + { + "epoch": 2.9094003953437295, + "grad_norm": 2.650937080383301, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7146812677383423, + "num_tokens": 685564126.0, + "step": 26493 + }, + { + "epoch": 2.9095102130463433, + "grad_norm": 2.2538836002349854, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7317714691162109, + "num_tokens": 685590910.0, + "step": 26494 + }, + { + "epoch": 2.9096200307489566, + "grad_norm": 2.366006374359131, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7219057679176331, + "num_tokens": 685620084.0, + "step": 26495 + }, + { + "epoch": 2.9097298484515703, + "grad_norm": 2.2510018348693848, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7486234307289124, + "num_tokens": 685647562.0, + "step": 26496 + }, + { + "epoch": 2.909839666154184, + "grad_norm": 2.193889856338501, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7240880727767944, + "num_tokens": 685675349.0, + "step": 26497 + }, + { + "epoch": 2.909949483856798, + "grad_norm": 2.436054229736328, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7468666434288025, + "num_tokens": 685698449.0, + "step": 26498 + }, + { + "epoch": 2.9100593015594116, + "grad_norm": 2.292421340942383, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.705903172492981, + "num_tokens": 685727170.0, + "step": 26499 + }, + { + "epoch": 2.910169119262025, + "grad_norm": 2.1332309246063232, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.71869957447052, + "num_tokens": 685759830.0, + "step": 26500 + }, + { + "epoch": 2.9102789369646387, + "grad_norm": 2.056732177734375, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7341945171356201, + "num_tokens": 685790878.0, + "step": 26501 + }, + { + "epoch": 2.9103887546672524, + "grad_norm": 2.1093811988830566, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7371770739555359, + "num_tokens": 685822032.0, + "step": 26502 + }, + { + "epoch": 2.9104985723698658, + "grad_norm": 2.4319210052490234, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7259305119514465, + "num_tokens": 685849320.0, + "step": 26503 + }, + { + "epoch": 2.91060839007248, + "grad_norm": 2.441218852996826, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7181582450866699, + "num_tokens": 685873771.0, + "step": 26504 + }, + { + "epoch": 2.9107182077750933, + "grad_norm": 2.1713428497314453, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7132641077041626, + "num_tokens": 685904227.0, + "step": 26505 + }, + { + "epoch": 2.910828025477707, + "grad_norm": 2.3178956508636475, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7379769086837769, + "num_tokens": 685931888.0, + "step": 26506 + }, + { + "epoch": 2.910937843180321, + "grad_norm": 2.479628801345825, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7374539971351624, + "num_tokens": 685955824.0, + "step": 26507 + }, + { + "epoch": 2.911047660882934, + "grad_norm": 2.4187142848968506, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7279823422431946, + "num_tokens": 685980405.0, + "step": 26508 + }, + { + "epoch": 2.911157478585548, + "grad_norm": 2.725205659866333, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7378216981887817, + "num_tokens": 686000283.0, + "step": 26509 + }, + { + "epoch": 2.9112672962881616, + "grad_norm": 2.4127867221832275, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7446175813674927, + "num_tokens": 686024486.0, + "step": 26510 + }, + { + "epoch": 2.9113771139907754, + "grad_norm": 2.2245640754699707, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.75727778673172, + "num_tokens": 686049863.0, + "step": 26511 + }, + { + "epoch": 2.911486931693389, + "grad_norm": 2.3541553020477295, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7384867072105408, + "num_tokens": 686074858.0, + "step": 26512 + }, + { + "epoch": 2.9115967493960024, + "grad_norm": 2.4426050186157227, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7269608378410339, + "num_tokens": 686098909.0, + "step": 26513 + }, + { + "epoch": 2.911706567098616, + "grad_norm": 2.3958184719085693, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.724244236946106, + "num_tokens": 686122999.0, + "step": 26514 + }, + { + "epoch": 2.91181638480123, + "grad_norm": 2.6044764518737793, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7408022880554199, + "num_tokens": 686146006.0, + "step": 26515 + }, + { + "epoch": 2.9119262025038437, + "grad_norm": 2.060264825820923, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7203688025474548, + "num_tokens": 686179381.0, + "step": 26516 + }, + { + "epoch": 2.9120360202064575, + "grad_norm": 2.2804737091064453, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7265537977218628, + "num_tokens": 686207753.0, + "step": 26517 + }, + { + "epoch": 2.912145837909071, + "grad_norm": 2.4025444984436035, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7217671275138855, + "num_tokens": 686233578.0, + "step": 26518 + }, + { + "epoch": 2.9122556556116845, + "grad_norm": 2.4091780185699463, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7441086173057556, + "num_tokens": 686258649.0, + "step": 26519 + }, + { + "epoch": 2.9123654733142983, + "grad_norm": 2.320533275604248, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7738409042358398, + "num_tokens": 686282824.0, + "step": 26520 + }, + { + "epoch": 2.912475291016912, + "grad_norm": 2.0012056827545166, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7217680215835571, + "num_tokens": 686315799.0, + "step": 26521 + }, + { + "epoch": 2.912585108719526, + "grad_norm": 2.4768002033233643, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.74509596824646, + "num_tokens": 686339348.0, + "step": 26522 + }, + { + "epoch": 2.912694926422139, + "grad_norm": 2.7124202251434326, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7563717365264893, + "num_tokens": 686360029.0, + "step": 26523 + }, + { + "epoch": 2.912804744124753, + "grad_norm": 2.365832567214966, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7111719846725464, + "num_tokens": 686387739.0, + "step": 26524 + }, + { + "epoch": 2.9129145618273666, + "grad_norm": 2.526930332183838, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7358478903770447, + "num_tokens": 686410635.0, + "step": 26525 + }, + { + "epoch": 2.9130243795299804, + "grad_norm": 2.425642967224121, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7431150674819946, + "num_tokens": 686436426.0, + "step": 26526 + }, + { + "epoch": 2.913134197232594, + "grad_norm": 2.0671169757843018, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7170225977897644, + "num_tokens": 686466815.0, + "step": 26527 + }, + { + "epoch": 2.9132440149352075, + "grad_norm": 2.3595595359802246, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7181546092033386, + "num_tokens": 686491711.0, + "step": 26528 + }, + { + "epoch": 2.913353832637821, + "grad_norm": 2.391885757446289, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.716794490814209, + "num_tokens": 686515784.0, + "step": 26529 + }, + { + "epoch": 2.913463650340435, + "grad_norm": 2.6510355472564697, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7252743244171143, + "num_tokens": 686538227.0, + "step": 26530 + }, + { + "epoch": 2.9135734680430483, + "grad_norm": 2.194089651107788, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7090235948562622, + "num_tokens": 686568077.0, + "step": 26531 + }, + { + "epoch": 2.913683285745662, + "grad_norm": 2.2124505043029785, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.730396032333374, + "num_tokens": 686597018.0, + "step": 26532 + }, + { + "epoch": 2.913793103448276, + "grad_norm": 2.7024004459381104, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7367744445800781, + "num_tokens": 686618219.0, + "step": 26533 + }, + { + "epoch": 2.9139029211508896, + "grad_norm": 2.409641981124878, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.730478048324585, + "num_tokens": 686642533.0, + "step": 26534 + }, + { + "epoch": 2.9140127388535033, + "grad_norm": 2.49186110496521, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7154504060745239, + "num_tokens": 686665114.0, + "step": 26535 + }, + { + "epoch": 2.9141225565561166, + "grad_norm": 2.276822328567505, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.709947943687439, + "num_tokens": 686695160.0, + "step": 26536 + }, + { + "epoch": 2.9142323742587304, + "grad_norm": 2.593994379043579, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7162042856216431, + "num_tokens": 686717975.0, + "step": 26537 + }, + { + "epoch": 2.914342191961344, + "grad_norm": 2.4031617641448975, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7323489785194397, + "num_tokens": 686744794.0, + "step": 26538 + }, + { + "epoch": 2.914452009663958, + "grad_norm": 2.3625283241271973, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7112571001052856, + "num_tokens": 686770704.0, + "step": 26539 + }, + { + "epoch": 2.9145618273665717, + "grad_norm": 2.093304395675659, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7396500706672668, + "num_tokens": 686800460.0, + "step": 26540 + }, + { + "epoch": 2.914671645069185, + "grad_norm": 2.0101704597473145, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6909430623054504, + "num_tokens": 686835550.0, + "step": 26541 + }, + { + "epoch": 2.9147814627717987, + "grad_norm": 2.296487331390381, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7358196973800659, + "num_tokens": 686860874.0, + "step": 26542 + }, + { + "epoch": 2.9148912804744125, + "grad_norm": 2.284010171890259, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7210842370986938, + "num_tokens": 686889331.0, + "step": 26543 + }, + { + "epoch": 2.9150010981770262, + "grad_norm": 2.2268686294555664, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7247669100761414, + "num_tokens": 686919202.0, + "step": 26544 + }, + { + "epoch": 2.91511091587964, + "grad_norm": 2.4964723587036133, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7242387533187866, + "num_tokens": 686942904.0, + "step": 26545 + }, + { + "epoch": 2.9152207335822533, + "grad_norm": 2.1482300758361816, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7316635847091675, + "num_tokens": 686973126.0, + "step": 26546 + }, + { + "epoch": 2.915330551284867, + "grad_norm": 2.4799253940582275, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7364848852157593, + "num_tokens": 686998311.0, + "step": 26547 + }, + { + "epoch": 2.915440368987481, + "grad_norm": 2.2488999366760254, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7253152132034302, + "num_tokens": 687025114.0, + "step": 26548 + }, + { + "epoch": 2.9155501866900946, + "grad_norm": 2.298042058944702, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7642959356307983, + "num_tokens": 687050089.0, + "step": 26549 + }, + { + "epoch": 2.9156600043927083, + "grad_norm": 2.1830697059631348, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7167532444000244, + "num_tokens": 687079672.0, + "step": 26550 + }, + { + "epoch": 2.9157698220953217, + "grad_norm": 2.4192984104156494, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7409958243370056, + "num_tokens": 687104830.0, + "step": 26551 + }, + { + "epoch": 2.9158796397979354, + "grad_norm": 2.8383524417877197, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7534661293029785, + "num_tokens": 687124592.0, + "step": 26552 + }, + { + "epoch": 2.915989457500549, + "grad_norm": 2.4121787548065186, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7507843971252441, + "num_tokens": 687149660.0, + "step": 26553 + }, + { + "epoch": 2.9160992752031625, + "grad_norm": 2.5656399726867676, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7131919264793396, + "num_tokens": 687173940.0, + "step": 26554 + }, + { + "epoch": 2.9162090929057767, + "grad_norm": 2.174393653869629, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7240235209465027, + "num_tokens": 687206727.0, + "step": 26555 + }, + { + "epoch": 2.91631891060839, + "grad_norm": 2.5899417400360107, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7426251173019409, + "num_tokens": 687227811.0, + "step": 26556 + }, + { + "epoch": 2.9164287283110037, + "grad_norm": 2.177867889404297, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7309807538986206, + "num_tokens": 687256730.0, + "step": 26557 + }, + { + "epoch": 2.9165385460136175, + "grad_norm": 2.267066717147827, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7487396001815796, + "num_tokens": 687281628.0, + "step": 26558 + }, + { + "epoch": 2.916648363716231, + "grad_norm": 2.1291117668151855, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7386825680732727, + "num_tokens": 687310682.0, + "step": 26559 + }, + { + "epoch": 2.9167581814188446, + "grad_norm": 2.1724820137023926, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7138004899024963, + "num_tokens": 687340045.0, + "step": 26560 + }, + { + "epoch": 2.9168679991214583, + "grad_norm": 2.276419162750244, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7329046130180359, + "num_tokens": 687365898.0, + "step": 26561 + }, + { + "epoch": 2.916977816824072, + "grad_norm": 2.2712368965148926, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7566290497779846, + "num_tokens": 687390425.0, + "step": 26562 + }, + { + "epoch": 2.917087634526686, + "grad_norm": 2.6121459007263184, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7382830381393433, + "num_tokens": 687412248.0, + "step": 26563 + }, + { + "epoch": 2.917197452229299, + "grad_norm": 2.41768217086792, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7292556762695312, + "num_tokens": 687438042.0, + "step": 26564 + }, + { + "epoch": 2.917307269931913, + "grad_norm": 2.3408517837524414, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7181373834609985, + "num_tokens": 687464699.0, + "step": 26565 + }, + { + "epoch": 2.9174170876345267, + "grad_norm": 2.2700068950653076, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7392398118972778, + "num_tokens": 687490154.0, + "step": 26566 + }, + { + "epoch": 2.9175269053371404, + "grad_norm": 2.5479989051818848, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7538259029388428, + "num_tokens": 687510151.0, + "step": 26567 + }, + { + "epoch": 2.917636723039754, + "grad_norm": 2.396397829055786, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7242426872253418, + "num_tokens": 687534651.0, + "step": 26568 + }, + { + "epoch": 2.9177465407423675, + "grad_norm": 2.511866569519043, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7187954783439636, + "num_tokens": 687557368.0, + "step": 26569 + }, + { + "epoch": 2.9178563584449813, + "grad_norm": 2.334207773208618, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7371054887771606, + "num_tokens": 687581983.0, + "step": 26570 + }, + { + "epoch": 2.917966176147595, + "grad_norm": 2.2061541080474854, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7184467315673828, + "num_tokens": 687610019.0, + "step": 26571 + }, + { + "epoch": 2.9180759938502088, + "grad_norm": 2.068040609359741, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7091996073722839, + "num_tokens": 687644128.0, + "step": 26572 + }, + { + "epoch": 2.9181858115528225, + "grad_norm": 2.6985020637512207, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7143734693527222, + "num_tokens": 687664735.0, + "step": 26573 + }, + { + "epoch": 2.918295629255436, + "grad_norm": 2.3552353382110596, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7630314826965332, + "num_tokens": 687687794.0, + "step": 26574 + }, + { + "epoch": 2.9184054469580496, + "grad_norm": 2.480945110321045, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7218778729438782, + "num_tokens": 687711311.0, + "step": 26575 + }, + { + "epoch": 2.9185152646606634, + "grad_norm": 2.644075870513916, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7415025234222412, + "num_tokens": 687733156.0, + "step": 26576 + }, + { + "epoch": 2.918625082363277, + "grad_norm": 2.6118922233581543, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.738273024559021, + "num_tokens": 687755265.0, + "step": 26577 + }, + { + "epoch": 2.918734900065891, + "grad_norm": 2.313786029815674, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.738528311252594, + "num_tokens": 687780821.0, + "step": 26578 + }, + { + "epoch": 2.918844717768504, + "grad_norm": 2.582686185836792, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7511934638023376, + "num_tokens": 687803688.0, + "step": 26579 + }, + { + "epoch": 2.918954535471118, + "grad_norm": 2.375598430633545, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7428532838821411, + "num_tokens": 687829995.0, + "step": 26580 + }, + { + "epoch": 2.9190643531737317, + "grad_norm": 2.2108051776885986, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7249898910522461, + "num_tokens": 687859682.0, + "step": 26581 + }, + { + "epoch": 2.919174170876345, + "grad_norm": 2.056666612625122, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7134525179862976, + "num_tokens": 687893976.0, + "step": 26582 + }, + { + "epoch": 2.9192839885789588, + "grad_norm": 2.1812939643859863, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7500977516174316, + "num_tokens": 687922005.0, + "step": 26583 + }, + { + "epoch": 2.9193938062815725, + "grad_norm": 2.5417850017547607, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7354823350906372, + "num_tokens": 687946333.0, + "step": 26584 + }, + { + "epoch": 2.9195036239841863, + "grad_norm": 2.6144776344299316, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7600592374801636, + "num_tokens": 687966792.0, + "step": 26585 + }, + { + "epoch": 2.9196134416868, + "grad_norm": 2.2949938774108887, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.73597252368927, + "num_tokens": 687993284.0, + "step": 26586 + }, + { + "epoch": 2.9197232593894134, + "grad_norm": 2.5785443782806396, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7369483709335327, + "num_tokens": 688016354.0, + "step": 26587 + }, + { + "epoch": 2.919833077092027, + "grad_norm": 2.3074986934661865, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7486151456832886, + "num_tokens": 688042962.0, + "step": 26588 + }, + { + "epoch": 2.919942894794641, + "grad_norm": 2.601541519165039, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7253103852272034, + "num_tokens": 688065254.0, + "step": 26589 + }, + { + "epoch": 2.9200527124972546, + "grad_norm": 2.2199888229370117, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7017234563827515, + "num_tokens": 688094167.0, + "step": 26590 + }, + { + "epoch": 2.9201625301998684, + "grad_norm": 2.3117594718933105, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7358658909797668, + "num_tokens": 688121059.0, + "step": 26591 + }, + { + "epoch": 2.9202723479024817, + "grad_norm": 2.3495872020721436, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7292553186416626, + "num_tokens": 688148228.0, + "step": 26592 + }, + { + "epoch": 2.9203821656050954, + "grad_norm": 2.719465732574463, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.742285966873169, + "num_tokens": 688168188.0, + "step": 26593 + }, + { + "epoch": 2.920491983307709, + "grad_norm": 2.363252639770508, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7272492051124573, + "num_tokens": 688193376.0, + "step": 26594 + }, + { + "epoch": 2.920601801010323, + "grad_norm": 2.168191909790039, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7246717214584351, + "num_tokens": 688223449.0, + "step": 26595 + }, + { + "epoch": 2.9207116187129367, + "grad_norm": 2.3641505241394043, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7362505793571472, + "num_tokens": 688248470.0, + "step": 26596 + }, + { + "epoch": 2.92082143641555, + "grad_norm": 2.214050054550171, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7582638263702393, + "num_tokens": 688275671.0, + "step": 26597 + }, + { + "epoch": 2.920931254118164, + "grad_norm": 2.301772117614746, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7325005531311035, + "num_tokens": 688305394.0, + "step": 26598 + }, + { + "epoch": 2.9210410718207775, + "grad_norm": 2.6151676177978516, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7302831411361694, + "num_tokens": 688327261.0, + "step": 26599 + }, + { + "epoch": 2.9211508895233913, + "grad_norm": 2.2764875888824463, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7512332201004028, + "num_tokens": 688354440.0, + "step": 26600 + }, + { + "epoch": 2.921260707226005, + "grad_norm": 2.668140411376953, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7343586683273315, + "num_tokens": 688376412.0, + "step": 26601 + }, + { + "epoch": 2.9213705249286184, + "grad_norm": 2.1823551654815674, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7045356035232544, + "num_tokens": 688408083.0, + "step": 26602 + }, + { + "epoch": 2.921480342631232, + "grad_norm": 2.4699931144714355, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.753584623336792, + "num_tokens": 688430832.0, + "step": 26603 + }, + { + "epoch": 2.921590160333846, + "grad_norm": 2.4183335304260254, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7288392782211304, + "num_tokens": 688455208.0, + "step": 26604 + }, + { + "epoch": 2.921699978036459, + "grad_norm": 2.302682638168335, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7387104630470276, + "num_tokens": 688480867.0, + "step": 26605 + }, + { + "epoch": 2.9218097957390734, + "grad_norm": 2.3393678665161133, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7202304601669312, + "num_tokens": 688508797.0, + "step": 26606 + }, + { + "epoch": 2.9219196134416867, + "grad_norm": 2.209534168243408, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7093191146850586, + "num_tokens": 688539193.0, + "step": 26607 + }, + { + "epoch": 2.9220294311443005, + "grad_norm": 2.124241352081299, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7274127006530762, + "num_tokens": 688568712.0, + "step": 26608 + }, + { + "epoch": 2.9221392488469142, + "grad_norm": 2.566438913345337, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7278259992599487, + "num_tokens": 688592866.0, + "step": 26609 + }, + { + "epoch": 2.9222490665495275, + "grad_norm": 2.547276258468628, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7395415902137756, + "num_tokens": 688617035.0, + "step": 26610 + }, + { + "epoch": 2.9223588842521413, + "grad_norm": 2.312669038772583, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7413899898529053, + "num_tokens": 688642534.0, + "step": 26611 + }, + { + "epoch": 2.922468701954755, + "grad_norm": 2.469686985015869, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7316327691078186, + "num_tokens": 688666128.0, + "step": 26612 + }, + { + "epoch": 2.922578519657369, + "grad_norm": 2.522510051727295, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.76108717918396, + "num_tokens": 688688505.0, + "step": 26613 + }, + { + "epoch": 2.9226883373599826, + "grad_norm": 2.1030616760253906, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7182969450950623, + "num_tokens": 688719595.0, + "step": 26614 + }, + { + "epoch": 2.922798155062596, + "grad_norm": 2.6519100666046143, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7162126302719116, + "num_tokens": 688741262.0, + "step": 26615 + }, + { + "epoch": 2.9229079727652096, + "grad_norm": 2.5303287506103516, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7335894107818604, + "num_tokens": 688764920.0, + "step": 26616 + }, + { + "epoch": 2.9230177904678234, + "grad_norm": 2.2287487983703613, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7086299061775208, + "num_tokens": 688796330.0, + "step": 26617 + }, + { + "epoch": 2.923127608170437, + "grad_norm": 2.5298616886138916, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7550445795059204, + "num_tokens": 688817560.0, + "step": 26618 + }, + { + "epoch": 2.923237425873051, + "grad_norm": 2.473787784576416, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7410099506378174, + "num_tokens": 688840841.0, + "step": 26619 + }, + { + "epoch": 2.9233472435756642, + "grad_norm": 2.6134402751922607, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7356206178665161, + "num_tokens": 688863396.0, + "step": 26620 + }, + { + "epoch": 2.923457061278278, + "grad_norm": 2.6706795692443848, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7512566447257996, + "num_tokens": 688884373.0, + "step": 26621 + }, + { + "epoch": 2.9235668789808917, + "grad_norm": 2.5639941692352295, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7627249360084534, + "num_tokens": 688906250.0, + "step": 26622 + }, + { + "epoch": 2.9236766966835055, + "grad_norm": 2.592203140258789, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7390028238296509, + "num_tokens": 688927619.0, + "step": 26623 + }, + { + "epoch": 2.9237865143861193, + "grad_norm": 2.18121600151062, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7522118091583252, + "num_tokens": 688955302.0, + "step": 26624 + }, + { + "epoch": 2.9238963320887326, + "grad_norm": 2.2110702991485596, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7178685665130615, + "num_tokens": 688985863.0, + "step": 26625 + }, + { + "epoch": 2.9240061497913463, + "grad_norm": 2.3169052600860596, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7467806339263916, + "num_tokens": 689013871.0, + "step": 26626 + }, + { + "epoch": 2.92411596749396, + "grad_norm": 2.605954647064209, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7309044599533081, + "num_tokens": 689036331.0, + "step": 26627 + }, + { + "epoch": 2.924225785196574, + "grad_norm": 2.30442476272583, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7297869920730591, + "num_tokens": 689064330.0, + "step": 26628 + }, + { + "epoch": 2.9243356028991876, + "grad_norm": 2.4867165088653564, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7192149758338928, + "num_tokens": 689089193.0, + "step": 26629 + }, + { + "epoch": 2.924445420601801, + "grad_norm": 2.247227191925049, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7269819378852844, + "num_tokens": 689117563.0, + "step": 26630 + }, + { + "epoch": 2.9245552383044147, + "grad_norm": 2.487645387649536, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7372732162475586, + "num_tokens": 689141597.0, + "step": 26631 + }, + { + "epoch": 2.9246650560070284, + "grad_norm": 2.2971694469451904, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7498725652694702, + "num_tokens": 689167430.0, + "step": 26632 + }, + { + "epoch": 2.9247748737096417, + "grad_norm": 2.1113271713256836, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.742760419845581, + "num_tokens": 689199668.0, + "step": 26633 + }, + { + "epoch": 2.924884691412256, + "grad_norm": 2.169140100479126, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7310941219329834, + "num_tokens": 689232059.0, + "step": 26634 + }, + { + "epoch": 2.9249945091148692, + "grad_norm": 2.3487019538879395, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7174057960510254, + "num_tokens": 689259070.0, + "step": 26635 + }, + { + "epoch": 2.925104326817483, + "grad_norm": 2.3203938007354736, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7062528133392334, + "num_tokens": 689284076.0, + "step": 26636 + }, + { + "epoch": 2.9252141445200968, + "grad_norm": 2.5249252319335938, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7204858064651489, + "num_tokens": 689308431.0, + "step": 26637 + }, + { + "epoch": 2.92532396222271, + "grad_norm": 2.0537688732147217, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7307782173156738, + "num_tokens": 689340239.0, + "step": 26638 + }, + { + "epoch": 2.925433779925324, + "grad_norm": 2.2202157974243164, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7349093556404114, + "num_tokens": 689369272.0, + "step": 26639 + }, + { + "epoch": 2.9255435976279376, + "grad_norm": 2.2048308849334717, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.703317403793335, + "num_tokens": 689400323.0, + "step": 26640 + }, + { + "epoch": 2.9256534153305513, + "grad_norm": 2.67073392868042, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7379190325737, + "num_tokens": 689422422.0, + "step": 26641 + }, + { + "epoch": 2.925763233033165, + "grad_norm": 2.22615909576416, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7080377340316772, + "num_tokens": 689451044.0, + "step": 26642 + }, + { + "epoch": 2.9258730507357784, + "grad_norm": 2.6419174671173096, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7455933094024658, + "num_tokens": 689473795.0, + "step": 26643 + }, + { + "epoch": 2.925982868438392, + "grad_norm": 2.2721426486968994, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7240259647369385, + "num_tokens": 689503380.0, + "step": 26644 + }, + { + "epoch": 2.926092686141006, + "grad_norm": 2.5099844932556152, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7272073030471802, + "num_tokens": 689526467.0, + "step": 26645 + }, + { + "epoch": 2.9262025038436197, + "grad_norm": 2.2620084285736084, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7163965106010437, + "num_tokens": 689555347.0, + "step": 26646 + }, + { + "epoch": 2.9263123215462334, + "grad_norm": 1.9911044836044312, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6968181729316711, + "num_tokens": 689592022.0, + "step": 26647 + }, + { + "epoch": 2.9264221392488468, + "grad_norm": 2.5877685546875, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.733637273311615, + "num_tokens": 689612055.0, + "step": 26648 + }, + { + "epoch": 2.9265319569514605, + "grad_norm": 2.444319725036621, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7137641906738281, + "num_tokens": 689638227.0, + "step": 26649 + }, + { + "epoch": 2.9266417746540743, + "grad_norm": 2.2151708602905273, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7309063673019409, + "num_tokens": 689669711.0, + "step": 26650 + }, + { + "epoch": 2.926751592356688, + "grad_norm": 2.233203172683716, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7179452180862427, + "num_tokens": 689699272.0, + "step": 26651 + }, + { + "epoch": 2.926861410059302, + "grad_norm": 2.2280125617980957, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7140530347824097, + "num_tokens": 689729717.0, + "step": 26652 + }, + { + "epoch": 2.926971227761915, + "grad_norm": 2.262470006942749, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7265295386314392, + "num_tokens": 689755552.0, + "step": 26653 + }, + { + "epoch": 2.927081045464529, + "grad_norm": 2.5292277336120605, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7019904851913452, + "num_tokens": 689780275.0, + "step": 26654 + }, + { + "epoch": 2.9271908631671426, + "grad_norm": 2.2756540775299072, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7270911335945129, + "num_tokens": 689806057.0, + "step": 26655 + }, + { + "epoch": 2.9273006808697564, + "grad_norm": 2.4879395961761475, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7372873425483704, + "num_tokens": 689831959.0, + "step": 26656 + }, + { + "epoch": 2.92741049857237, + "grad_norm": 2.1045124530792236, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7342008948326111, + "num_tokens": 689863640.0, + "step": 26657 + }, + { + "epoch": 2.9275203162749834, + "grad_norm": 2.3711555004119873, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7220098376274109, + "num_tokens": 689892249.0, + "step": 26658 + }, + { + "epoch": 2.927630133977597, + "grad_norm": 2.395836114883423, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.73443603515625, + "num_tokens": 689916482.0, + "step": 26659 + }, + { + "epoch": 2.927739951680211, + "grad_norm": 2.363306760787964, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7418136596679688, + "num_tokens": 689941566.0, + "step": 26660 + }, + { + "epoch": 2.9278497693828243, + "grad_norm": 2.2337284088134766, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.735100269317627, + "num_tokens": 689968884.0, + "step": 26661 + }, + { + "epoch": 2.927959587085438, + "grad_norm": 2.254342555999756, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7205158472061157, + "num_tokens": 689997689.0, + "step": 26662 + }, + { + "epoch": 2.9280694047880518, + "grad_norm": 2.5020530223846436, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.744623601436615, + "num_tokens": 690020953.0, + "step": 26663 + }, + { + "epoch": 2.9281792224906655, + "grad_norm": 2.281703233718872, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7214957475662231, + "num_tokens": 690049019.0, + "step": 26664 + }, + { + "epoch": 2.9282890401932793, + "grad_norm": 2.362344264984131, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7390532493591309, + "num_tokens": 690075642.0, + "step": 26665 + }, + { + "epoch": 2.9283988578958926, + "grad_norm": 2.1534478664398193, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.727617621421814, + "num_tokens": 690107687.0, + "step": 26666 + }, + { + "epoch": 2.9285086755985064, + "grad_norm": 2.97796893119812, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7276121377944946, + "num_tokens": 690128154.0, + "step": 26667 + }, + { + "epoch": 2.92861849330112, + "grad_norm": 2.621630907058716, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7606607675552368, + "num_tokens": 690149424.0, + "step": 26668 + }, + { + "epoch": 2.928728311003734, + "grad_norm": 2.400404214859009, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7398658394813538, + "num_tokens": 690174474.0, + "step": 26669 + }, + { + "epoch": 2.9288381287063476, + "grad_norm": 2.569150447845459, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7385663390159607, + "num_tokens": 690198756.0, + "step": 26670 + }, + { + "epoch": 2.928947946408961, + "grad_norm": 2.656377077102661, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7656052112579346, + "num_tokens": 690219581.0, + "step": 26671 + }, + { + "epoch": 2.9290577641115747, + "grad_norm": 2.153930187225342, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7190364599227905, + "num_tokens": 690249638.0, + "step": 26672 + }, + { + "epoch": 2.9291675818141885, + "grad_norm": 2.3698792457580566, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7167904376983643, + "num_tokens": 690275987.0, + "step": 26673 + }, + { + "epoch": 2.929277399516802, + "grad_norm": 2.7275233268737793, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7548657655715942, + "num_tokens": 690295474.0, + "step": 26674 + }, + { + "epoch": 2.929387217219416, + "grad_norm": 2.705674409866333, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7402480840682983, + "num_tokens": 690315944.0, + "step": 26675 + }, + { + "epoch": 2.9294970349220293, + "grad_norm": 2.3928966522216797, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7356724739074707, + "num_tokens": 690340715.0, + "step": 26676 + }, + { + "epoch": 2.929606852624643, + "grad_norm": 2.605846643447876, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7357305884361267, + "num_tokens": 690361929.0, + "step": 26677 + }, + { + "epoch": 2.929716670327257, + "grad_norm": 2.1537256240844727, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.718639075756073, + "num_tokens": 690391351.0, + "step": 26678 + }, + { + "epoch": 2.9298264880298706, + "grad_norm": 2.1645476818084717, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7249721884727478, + "num_tokens": 690421413.0, + "step": 26679 + }, + { + "epoch": 2.9299363057324843, + "grad_norm": 2.5078935623168945, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7084606885910034, + "num_tokens": 690446312.0, + "step": 26680 + }, + { + "epoch": 2.9300461234350976, + "grad_norm": 2.516392469406128, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7043362855911255, + "num_tokens": 690471830.0, + "step": 26681 + }, + { + "epoch": 2.9301559411377114, + "grad_norm": 2.3810863494873047, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7122751474380493, + "num_tokens": 690501404.0, + "step": 26682 + }, + { + "epoch": 2.930265758840325, + "grad_norm": 2.48539662361145, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7280218601226807, + "num_tokens": 690524742.0, + "step": 26683 + }, + { + "epoch": 2.9303755765429385, + "grad_norm": 2.6075501441955566, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7730600833892822, + "num_tokens": 690546333.0, + "step": 26684 + }, + { + "epoch": 2.9304853942455527, + "grad_norm": 2.514162540435791, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7459023594856262, + "num_tokens": 690569697.0, + "step": 26685 + }, + { + "epoch": 2.930595211948166, + "grad_norm": 2.372187852859497, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7733727693557739, + "num_tokens": 690593988.0, + "step": 26686 + }, + { + "epoch": 2.9307050296507797, + "grad_norm": 2.4543356895446777, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7238948345184326, + "num_tokens": 690618230.0, + "step": 26687 + }, + { + "epoch": 2.9308148473533935, + "grad_norm": 2.6759912967681885, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7397271394729614, + "num_tokens": 690639264.0, + "step": 26688 + }, + { + "epoch": 2.930924665056007, + "grad_norm": 2.4854137897491455, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7533526420593262, + "num_tokens": 690661980.0, + "step": 26689 + }, + { + "epoch": 2.9310344827586206, + "grad_norm": 2.5608227252960205, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7406835556030273, + "num_tokens": 690685725.0, + "step": 26690 + }, + { + "epoch": 2.9311443004612343, + "grad_norm": 2.5264360904693604, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7293406128883362, + "num_tokens": 690710122.0, + "step": 26691 + }, + { + "epoch": 2.931254118163848, + "grad_norm": 2.553524971008301, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7329307794570923, + "num_tokens": 690737360.0, + "step": 26692 + }, + { + "epoch": 2.931363935866462, + "grad_norm": 2.6455938816070557, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7290163040161133, + "num_tokens": 690758151.0, + "step": 26693 + }, + { + "epoch": 2.931473753569075, + "grad_norm": 2.2651515007019043, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7642929553985596, + "num_tokens": 690782626.0, + "step": 26694 + }, + { + "epoch": 2.931583571271689, + "grad_norm": 2.681504249572754, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7464102506637573, + "num_tokens": 690804268.0, + "step": 26695 + }, + { + "epoch": 2.9316933889743026, + "grad_norm": 2.116495370864868, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.725355863571167, + "num_tokens": 690833758.0, + "step": 26696 + }, + { + "epoch": 2.9318032066769164, + "grad_norm": 2.1324522495269775, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7060028314590454, + "num_tokens": 690867639.0, + "step": 26697 + }, + { + "epoch": 2.93191302437953, + "grad_norm": 2.0094430446624756, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7150636911392212, + "num_tokens": 690902197.0, + "step": 26698 + }, + { + "epoch": 2.9320228420821435, + "grad_norm": 2.2044153213500977, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7134772539138794, + "num_tokens": 690932496.0, + "step": 26699 + }, + { + "epoch": 2.9321326597847572, + "grad_norm": 2.6687607765197754, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7382440567016602, + "num_tokens": 690956064.0, + "step": 26700 + }, + { + "epoch": 2.932242477487371, + "grad_norm": 2.3711163997650146, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7215360403060913, + "num_tokens": 690980200.0, + "step": 26701 + }, + { + "epoch": 2.9323522951899847, + "grad_norm": 2.4184234142303467, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7376004457473755, + "num_tokens": 691003199.0, + "step": 26702 + }, + { + "epoch": 2.9324621128925985, + "grad_norm": 2.5500218868255615, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7353935241699219, + "num_tokens": 691025235.0, + "step": 26703 + }, + { + "epoch": 2.932571930595212, + "grad_norm": 2.3424108028411865, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7584655284881592, + "num_tokens": 691049426.0, + "step": 26704 + }, + { + "epoch": 2.9326817482978256, + "grad_norm": 2.1549015045166016, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.731036365032196, + "num_tokens": 691080489.0, + "step": 26705 + }, + { + "epoch": 2.9327915660004393, + "grad_norm": 2.245224714279175, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.724723756313324, + "num_tokens": 691109836.0, + "step": 26706 + }, + { + "epoch": 2.932901383703053, + "grad_norm": 2.5482096672058105, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.734187126159668, + "num_tokens": 691133883.0, + "step": 26707 + }, + { + "epoch": 2.933011201405667, + "grad_norm": 2.332491159439087, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7335116267204285, + "num_tokens": 691159481.0, + "step": 26708 + }, + { + "epoch": 2.93312101910828, + "grad_norm": 3.0181968212127686, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7391782402992249, + "num_tokens": 691178012.0, + "step": 26709 + }, + { + "epoch": 2.933230836810894, + "grad_norm": 2.685840606689453, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.730478823184967, + "num_tokens": 691200489.0, + "step": 26710 + }, + { + "epoch": 2.9333406545135077, + "grad_norm": 2.6620240211486816, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7454484701156616, + "num_tokens": 691220284.0, + "step": 26711 + }, + { + "epoch": 2.933450472216121, + "grad_norm": 2.248142957687378, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7582160830497742, + "num_tokens": 691247468.0, + "step": 26712 + }, + { + "epoch": 2.9335602899187347, + "grad_norm": 2.369551181793213, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7285902500152588, + "num_tokens": 691274301.0, + "step": 26713 + }, + { + "epoch": 2.9336701076213485, + "grad_norm": 2.7728121280670166, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7467563152313232, + "num_tokens": 691294714.0, + "step": 26714 + }, + { + "epoch": 2.9337799253239623, + "grad_norm": 2.3061697483062744, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7142578959465027, + "num_tokens": 691320042.0, + "step": 26715 + }, + { + "epoch": 2.933889743026576, + "grad_norm": 3.0719785690307617, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.755875825881958, + "num_tokens": 691336829.0, + "step": 26716 + }, + { + "epoch": 2.9339995607291893, + "grad_norm": 2.0782887935638428, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7499650120735168, + "num_tokens": 691366283.0, + "step": 26717 + }, + { + "epoch": 2.934109378431803, + "grad_norm": 2.508998155593872, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7453061938285828, + "num_tokens": 691390445.0, + "step": 26718 + }, + { + "epoch": 2.934219196134417, + "grad_norm": 2.382600784301758, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7402814626693726, + "num_tokens": 691416398.0, + "step": 26719 + }, + { + "epoch": 2.9343290138370306, + "grad_norm": 1.9582290649414062, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7123665809631348, + "num_tokens": 691451529.0, + "step": 26720 + }, + { + "epoch": 2.9344388315396444, + "grad_norm": 2.449629783630371, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7406914830207825, + "num_tokens": 691476310.0, + "step": 26721 + }, + { + "epoch": 2.9345486492422577, + "grad_norm": 2.6030619144439697, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7438573241233826, + "num_tokens": 691499514.0, + "step": 26722 + }, + { + "epoch": 2.9346584669448714, + "grad_norm": 2.63960862159729, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.729429304599762, + "num_tokens": 691521563.0, + "step": 26723 + }, + { + "epoch": 2.934768284647485, + "grad_norm": 2.376081705093384, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7257117033004761, + "num_tokens": 691545425.0, + "step": 26724 + }, + { + "epoch": 2.934878102350099, + "grad_norm": 2.3128926753997803, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.707092821598053, + "num_tokens": 691571772.0, + "step": 26725 + }, + { + "epoch": 2.9349879200527127, + "grad_norm": 2.4527087211608887, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7175344228744507, + "num_tokens": 691597828.0, + "step": 26726 + }, + { + "epoch": 2.935097737755326, + "grad_norm": 2.2390964031219482, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7040667533874512, + "num_tokens": 691627954.0, + "step": 26727 + }, + { + "epoch": 2.9352075554579398, + "grad_norm": 2.344874620437622, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7040418386459351, + "num_tokens": 691655678.0, + "step": 26728 + }, + { + "epoch": 2.9353173731605535, + "grad_norm": 2.4577438831329346, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7260961532592773, + "num_tokens": 691681956.0, + "step": 26729 + }, + { + "epoch": 2.9354271908631673, + "grad_norm": 2.4430439472198486, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7324075102806091, + "num_tokens": 691708417.0, + "step": 26730 + }, + { + "epoch": 2.935537008565781, + "grad_norm": 2.650813341140747, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7308586835861206, + "num_tokens": 691728792.0, + "step": 26731 + }, + { + "epoch": 2.9356468262683943, + "grad_norm": 2.5138232707977295, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7447037696838379, + "num_tokens": 691751611.0, + "step": 26732 + }, + { + "epoch": 2.935756643971008, + "grad_norm": 2.6494650840759277, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7286134958267212, + "num_tokens": 691773539.0, + "step": 26733 + }, + { + "epoch": 2.935866461673622, + "grad_norm": 2.5090982913970947, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7351590394973755, + "num_tokens": 691796543.0, + "step": 26734 + }, + { + "epoch": 2.935976279376235, + "grad_norm": 2.278536081314087, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7377902269363403, + "num_tokens": 691824616.0, + "step": 26735 + }, + { + "epoch": 2.9360860970788494, + "grad_norm": 2.5114002227783203, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7191600799560547, + "num_tokens": 691847901.0, + "step": 26736 + }, + { + "epoch": 2.9361959147814627, + "grad_norm": 2.129002571105957, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7432354092597961, + "num_tokens": 691875459.0, + "step": 26737 + }, + { + "epoch": 2.9363057324840764, + "grad_norm": 2.2432501316070557, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7247282862663269, + "num_tokens": 691905817.0, + "step": 26738 + }, + { + "epoch": 2.93641555018669, + "grad_norm": 2.280726909637451, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.739733099937439, + "num_tokens": 691932060.0, + "step": 26739 + }, + { + "epoch": 2.9365253678893035, + "grad_norm": 2.222818613052368, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7134571075439453, + "num_tokens": 691959423.0, + "step": 26740 + }, + { + "epoch": 2.9366351855919173, + "grad_norm": 2.1822307109832764, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7347617149353027, + "num_tokens": 691989589.0, + "step": 26741 + }, + { + "epoch": 2.936745003294531, + "grad_norm": 2.273381471633911, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7352749109268188, + "num_tokens": 692018265.0, + "step": 26742 + }, + { + "epoch": 2.936854820997145, + "grad_norm": 2.3117218017578125, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7315110564231873, + "num_tokens": 692044060.0, + "step": 26743 + }, + { + "epoch": 2.9369646386997585, + "grad_norm": 2.3365840911865234, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7310073375701904, + "num_tokens": 692071309.0, + "step": 26744 + }, + { + "epoch": 2.937074456402372, + "grad_norm": 2.3782782554626465, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7481390833854675, + "num_tokens": 692095403.0, + "step": 26745 + }, + { + "epoch": 2.9371842741049856, + "grad_norm": 2.4321470260620117, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7210806608200073, + "num_tokens": 692120477.0, + "step": 26746 + }, + { + "epoch": 2.9372940918075994, + "grad_norm": 2.2492802143096924, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7235174179077148, + "num_tokens": 692147261.0, + "step": 26747 + }, + { + "epoch": 2.937403909510213, + "grad_norm": 2.5333478450775146, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7174922823905945, + "num_tokens": 692170175.0, + "step": 26748 + }, + { + "epoch": 2.937513727212827, + "grad_norm": 1.9311589002609253, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.719305157661438, + "num_tokens": 692207159.0, + "step": 26749 + }, + { + "epoch": 2.93762354491544, + "grad_norm": 2.707158088684082, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7490840554237366, + "num_tokens": 692228771.0, + "step": 26750 + }, + { + "epoch": 2.937733362618054, + "grad_norm": 2.4916248321533203, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7286214828491211, + "num_tokens": 692252062.0, + "step": 26751 + }, + { + "epoch": 2.9378431803206677, + "grad_norm": 2.493309736251831, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.741907000541687, + "num_tokens": 692276146.0, + "step": 26752 + }, + { + "epoch": 2.9379529980232815, + "grad_norm": 2.3511555194854736, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7376924753189087, + "num_tokens": 692302206.0, + "step": 26753 + }, + { + "epoch": 2.9380628157258952, + "grad_norm": 2.503474235534668, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.7369470000267029, + "num_tokens": 692325652.0, + "step": 26754 + }, + { + "epoch": 2.9381726334285085, + "grad_norm": 2.1641852855682373, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7200939655303955, + "num_tokens": 692355280.0, + "step": 26755 + }, + { + "epoch": 2.9382824511311223, + "grad_norm": 2.392252206802368, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7277575731277466, + "num_tokens": 692379783.0, + "step": 26756 + }, + { + "epoch": 2.938392268833736, + "grad_norm": 2.546257734298706, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7350952625274658, + "num_tokens": 692402919.0, + "step": 26757 + }, + { + "epoch": 2.93850208653635, + "grad_norm": 2.3541367053985596, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7392218112945557, + "num_tokens": 692429074.0, + "step": 26758 + }, + { + "epoch": 2.9386119042389636, + "grad_norm": 2.589203357696533, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7309855222702026, + "num_tokens": 692452534.0, + "step": 26759 + }, + { + "epoch": 2.938721721941577, + "grad_norm": 2.464076280593872, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7378344535827637, + "num_tokens": 692477274.0, + "step": 26760 + }, + { + "epoch": 2.9388315396441906, + "grad_norm": 2.5551390647888184, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7405946254730225, + "num_tokens": 692500243.0, + "step": 26761 + }, + { + "epoch": 2.9389413573468044, + "grad_norm": 2.4904918670654297, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7288413643836975, + "num_tokens": 692523281.0, + "step": 26762 + }, + { + "epoch": 2.9390511750494177, + "grad_norm": 2.305295944213867, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7382575273513794, + "num_tokens": 692550813.0, + "step": 26763 + }, + { + "epoch": 2.9391609927520315, + "grad_norm": 2.762411594390869, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.741490364074707, + "num_tokens": 692571744.0, + "step": 26764 + }, + { + "epoch": 2.939270810454645, + "grad_norm": 2.5880541801452637, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.739979088306427, + "num_tokens": 692594231.0, + "step": 26765 + }, + { + "epoch": 2.939380628157259, + "grad_norm": 2.658125877380371, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7422805428504944, + "num_tokens": 692614966.0, + "step": 26766 + }, + { + "epoch": 2.9394904458598727, + "grad_norm": 2.489586114883423, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7365260720252991, + "num_tokens": 692638923.0, + "step": 26767 + }, + { + "epoch": 2.939600263562486, + "grad_norm": 2.391573429107666, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7205656170845032, + "num_tokens": 692664177.0, + "step": 26768 + }, + { + "epoch": 2.9397100812651, + "grad_norm": 2.4881622791290283, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.742154061794281, + "num_tokens": 692686427.0, + "step": 26769 + }, + { + "epoch": 2.9398198989677136, + "grad_norm": 2.282651901245117, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7267152070999146, + "num_tokens": 692714060.0, + "step": 26770 + }, + { + "epoch": 2.9399297166703273, + "grad_norm": 2.5815556049346924, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7735079526901245, + "num_tokens": 692735925.0, + "step": 26771 + }, + { + "epoch": 2.940039534372941, + "grad_norm": 2.7564449310302734, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7315630912780762, + "num_tokens": 692755877.0, + "step": 26772 + }, + { + "epoch": 2.9401493520755544, + "grad_norm": 2.556349515914917, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7143329381942749, + "num_tokens": 692779742.0, + "step": 26773 + }, + { + "epoch": 2.940259169778168, + "grad_norm": 2.316997528076172, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7205222845077515, + "num_tokens": 692807891.0, + "step": 26774 + }, + { + "epoch": 2.940368987480782, + "grad_norm": 2.7470195293426514, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7353660464286804, + "num_tokens": 692827979.0, + "step": 26775 + }, + { + "epoch": 2.9404788051833957, + "grad_norm": 2.291609287261963, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7060482501983643, + "num_tokens": 692857900.0, + "step": 26776 + }, + { + "epoch": 2.9405886228860094, + "grad_norm": 2.2112503051757812, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7148485779762268, + "num_tokens": 692887200.0, + "step": 26777 + }, + { + "epoch": 2.9406984405886227, + "grad_norm": 2.14595365524292, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7383304238319397, + "num_tokens": 692916475.0, + "step": 26778 + }, + { + "epoch": 2.9408082582912365, + "grad_norm": 2.3869667053222656, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.726604700088501, + "num_tokens": 692941391.0, + "step": 26779 + }, + { + "epoch": 2.9409180759938502, + "grad_norm": 2.3464841842651367, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7394136190414429, + "num_tokens": 692965948.0, + "step": 26780 + }, + { + "epoch": 2.941027893696464, + "grad_norm": 2.5313470363616943, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7334699630737305, + "num_tokens": 692988724.0, + "step": 26781 + }, + { + "epoch": 2.9411377113990778, + "grad_norm": 2.6653358936309814, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7329643964767456, + "num_tokens": 693010367.0, + "step": 26782 + }, + { + "epoch": 2.941247529101691, + "grad_norm": 2.3193228244781494, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7382395267486572, + "num_tokens": 693036425.0, + "step": 26783 + }, + { + "epoch": 2.941357346804305, + "grad_norm": 2.9847185611724854, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7429986000061035, + "num_tokens": 693055801.0, + "step": 26784 + }, + { + "epoch": 2.9414671645069186, + "grad_norm": 2.738800287246704, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7331199645996094, + "num_tokens": 693076977.0, + "step": 26785 + }, + { + "epoch": 2.9415769822095323, + "grad_norm": 2.2677218914031982, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7144478559494019, + "num_tokens": 693105603.0, + "step": 26786 + }, + { + "epoch": 2.941686799912146, + "grad_norm": 2.3988590240478516, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7374889850616455, + "num_tokens": 693132176.0, + "step": 26787 + }, + { + "epoch": 2.9417966176147594, + "grad_norm": 2.271876573562622, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7435439229011536, + "num_tokens": 693158898.0, + "step": 26788 + }, + { + "epoch": 2.941906435317373, + "grad_norm": 2.497483491897583, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7452503442764282, + "num_tokens": 693181428.0, + "step": 26789 + }, + { + "epoch": 2.942016253019987, + "grad_norm": 2.2860934734344482, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7296042442321777, + "num_tokens": 693209373.0, + "step": 26790 + }, + { + "epoch": 2.9421260707226002, + "grad_norm": 2.6088175773620605, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7473452091217041, + "num_tokens": 693233344.0, + "step": 26791 + }, + { + "epoch": 2.942235888425214, + "grad_norm": 2.404852867126465, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7367738485336304, + "num_tokens": 693256521.0, + "step": 26792 + }, + { + "epoch": 2.9423457061278278, + "grad_norm": 2.4826605319976807, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7360141277313232, + "num_tokens": 693282044.0, + "step": 26793 + }, + { + "epoch": 2.9424555238304415, + "grad_norm": 2.3001692295074463, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7301380634307861, + "num_tokens": 693306197.0, + "step": 26794 + }, + { + "epoch": 2.9425653415330553, + "grad_norm": 2.2536206245422363, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7174688577651978, + "num_tokens": 693333976.0, + "step": 26795 + }, + { + "epoch": 2.9426751592356686, + "grad_norm": 2.5185158252716064, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7390434145927429, + "num_tokens": 693355923.0, + "step": 26796 + }, + { + "epoch": 2.9427849769382823, + "grad_norm": 2.2230286598205566, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7215798497200012, + "num_tokens": 693383526.0, + "step": 26797 + }, + { + "epoch": 2.942894794640896, + "grad_norm": 2.355315923690796, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.723546028137207, + "num_tokens": 693408827.0, + "step": 26798 + }, + { + "epoch": 2.94300461234351, + "grad_norm": 2.4265847206115723, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7362138628959656, + "num_tokens": 693434950.0, + "step": 26799 + }, + { + "epoch": 2.9431144300461236, + "grad_norm": 2.3150246143341064, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7358580827713013, + "num_tokens": 693461097.0, + "step": 26800 + }, + { + "epoch": 2.943224247748737, + "grad_norm": 2.4453015327453613, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7241513133049011, + "num_tokens": 693486772.0, + "step": 26801 + }, + { + "epoch": 2.9433340654513507, + "grad_norm": 2.1944234371185303, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7267380952835083, + "num_tokens": 693514765.0, + "step": 26802 + }, + { + "epoch": 2.9434438831539644, + "grad_norm": 2.5037527084350586, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7203353643417358, + "num_tokens": 693538919.0, + "step": 26803 + }, + { + "epoch": 2.943553700856578, + "grad_norm": 2.5294206142425537, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7543812990188599, + "num_tokens": 693560016.0, + "step": 26804 + }, + { + "epoch": 2.943663518559192, + "grad_norm": 2.174427032470703, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7237426042556763, + "num_tokens": 693590642.0, + "step": 26805 + }, + { + "epoch": 2.9437733362618053, + "grad_norm": 2.542675018310547, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7267521619796753, + "num_tokens": 693614369.0, + "step": 26806 + }, + { + "epoch": 2.943883153964419, + "grad_norm": 2.2513656616210938, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7273972630500793, + "num_tokens": 693643798.0, + "step": 26807 + }, + { + "epoch": 2.9439929716670328, + "grad_norm": 2.0452523231506348, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7085357904434204, + "num_tokens": 693676593.0, + "step": 26808 + }, + { + "epoch": 2.9441027893696465, + "grad_norm": 2.6417925357818604, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7441809773445129, + "num_tokens": 693698578.0, + "step": 26809 + }, + { + "epoch": 2.9442126070722603, + "grad_norm": 2.1290037631988525, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7332014441490173, + "num_tokens": 693728855.0, + "step": 26810 + }, + { + "epoch": 2.9443224247748736, + "grad_norm": 2.2455122470855713, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7026790380477905, + "num_tokens": 693757497.0, + "step": 26811 + }, + { + "epoch": 2.9444322424774874, + "grad_norm": 2.517486095428467, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7422538995742798, + "num_tokens": 693780563.0, + "step": 26812 + }, + { + "epoch": 2.944542060180101, + "grad_norm": 2.192981243133545, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7190530300140381, + "num_tokens": 693807523.0, + "step": 26813 + }, + { + "epoch": 2.9446518778827144, + "grad_norm": 2.476841449737549, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7248088121414185, + "num_tokens": 693831195.0, + "step": 26814 + }, + { + "epoch": 2.9447616955853286, + "grad_norm": 2.20589017868042, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7106854915618896, + "num_tokens": 693860382.0, + "step": 26815 + }, + { + "epoch": 2.944871513287942, + "grad_norm": 2.3331823348999023, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7231695652008057, + "num_tokens": 693887384.0, + "step": 26816 + }, + { + "epoch": 2.9449813309905557, + "grad_norm": 2.455612897872925, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7395125031471252, + "num_tokens": 693910703.0, + "step": 26817 + }, + { + "epoch": 2.9450911486931695, + "grad_norm": 2.4091649055480957, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7133784294128418, + "num_tokens": 693935462.0, + "step": 26818 + }, + { + "epoch": 2.9452009663957828, + "grad_norm": 2.535123109817505, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7485988736152649, + "num_tokens": 693956186.0, + "step": 26819 + }, + { + "epoch": 2.9453107840983965, + "grad_norm": 2.164019823074341, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7288900017738342, + "num_tokens": 693985511.0, + "step": 26820 + }, + { + "epoch": 2.9454206018010103, + "grad_norm": 2.4263694286346436, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.753791093826294, + "num_tokens": 694008428.0, + "step": 26821 + }, + { + "epoch": 2.945530419503624, + "grad_norm": 2.2543253898620605, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7328612804412842, + "num_tokens": 694035928.0, + "step": 26822 + }, + { + "epoch": 2.945640237206238, + "grad_norm": 2.356077194213867, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7112419605255127, + "num_tokens": 694061531.0, + "step": 26823 + }, + { + "epoch": 2.945750054908851, + "grad_norm": 2.439232110977173, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7286491990089417, + "num_tokens": 694084998.0, + "step": 26824 + }, + { + "epoch": 2.945859872611465, + "grad_norm": 2.5639073848724365, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7411396503448486, + "num_tokens": 694106285.0, + "step": 26825 + }, + { + "epoch": 2.9459696903140786, + "grad_norm": 2.24455189704895, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.723631739616394, + "num_tokens": 694135134.0, + "step": 26826 + }, + { + "epoch": 2.9460795080166924, + "grad_norm": 2.437748670578003, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7426265478134155, + "num_tokens": 694160220.0, + "step": 26827 + }, + { + "epoch": 2.946189325719306, + "grad_norm": 2.6291391849517822, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7266361713409424, + "num_tokens": 694181966.0, + "step": 26828 + }, + { + "epoch": 2.9462991434219195, + "grad_norm": 2.4481778144836426, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7324182987213135, + "num_tokens": 694207810.0, + "step": 26829 + }, + { + "epoch": 2.946408961124533, + "grad_norm": 2.1558573246002197, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7269632816314697, + "num_tokens": 694238163.0, + "step": 26830 + }, + { + "epoch": 2.946518778827147, + "grad_norm": 2.5791027545928955, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7523260116577148, + "num_tokens": 694260657.0, + "step": 26831 + }, + { + "epoch": 2.9466285965297607, + "grad_norm": 2.1273415088653564, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7056609392166138, + "num_tokens": 694290003.0, + "step": 26832 + }, + { + "epoch": 2.9467384142323745, + "grad_norm": 2.368140697479248, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7319711446762085, + "num_tokens": 694315162.0, + "step": 26833 + }, + { + "epoch": 2.946848231934988, + "grad_norm": 2.2527964115142822, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7308005690574646, + "num_tokens": 694342414.0, + "step": 26834 + }, + { + "epoch": 2.9469580496376016, + "grad_norm": 2.373486280441284, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7501866817474365, + "num_tokens": 694367724.0, + "step": 26835 + }, + { + "epoch": 2.9470678673402153, + "grad_norm": 2.63330078125, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7549410462379456, + "num_tokens": 694388463.0, + "step": 26836 + }, + { + "epoch": 2.947177685042829, + "grad_norm": 2.3179421424865723, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.74216228723526, + "num_tokens": 694415520.0, + "step": 26837 + }, + { + "epoch": 2.947287502745443, + "grad_norm": 2.5334391593933105, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7201430201530457, + "num_tokens": 694438833.0, + "step": 26838 + }, + { + "epoch": 2.947397320448056, + "grad_norm": 2.2114932537078857, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7484618425369263, + "num_tokens": 694466746.0, + "step": 26839 + }, + { + "epoch": 2.94750713815067, + "grad_norm": 2.6480560302734375, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7549716830253601, + "num_tokens": 694488620.0, + "step": 26840 + }, + { + "epoch": 2.9476169558532836, + "grad_norm": 2.4813730716705322, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7246131896972656, + "num_tokens": 694513493.0, + "step": 26841 + }, + { + "epoch": 2.947726773555897, + "grad_norm": 2.60400128364563, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7414542436599731, + "num_tokens": 694535953.0, + "step": 26842 + }, + { + "epoch": 2.9478365912585107, + "grad_norm": 2.2677226066589355, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7408088445663452, + "num_tokens": 694563545.0, + "step": 26843 + }, + { + "epoch": 2.9479464089611245, + "grad_norm": 2.226604461669922, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7384485006332397, + "num_tokens": 694592406.0, + "step": 26844 + }, + { + "epoch": 2.9480562266637382, + "grad_norm": 2.5449936389923096, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7203431129455566, + "num_tokens": 694615286.0, + "step": 26845 + }, + { + "epoch": 2.948166044366352, + "grad_norm": 2.280022144317627, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7081125974655151, + "num_tokens": 694643591.0, + "step": 26846 + }, + { + "epoch": 2.9482758620689653, + "grad_norm": 2.715632915496826, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7432510256767273, + "num_tokens": 694664116.0, + "step": 26847 + }, + { + "epoch": 2.948385679771579, + "grad_norm": 2.433321475982666, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.732269287109375, + "num_tokens": 694688798.0, + "step": 26848 + }, + { + "epoch": 2.948495497474193, + "grad_norm": 2.486419200897217, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.757315993309021, + "num_tokens": 694710847.0, + "step": 26849 + }, + { + "epoch": 2.9486053151768066, + "grad_norm": 2.7427310943603516, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7423402070999146, + "num_tokens": 694729400.0, + "step": 26850 + }, + { + "epoch": 2.9487151328794203, + "grad_norm": 2.1126887798309326, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7084712982177734, + "num_tokens": 694760140.0, + "step": 26851 + }, + { + "epoch": 2.9488249505820336, + "grad_norm": 2.5759875774383545, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7227113246917725, + "num_tokens": 694782721.0, + "step": 26852 + }, + { + "epoch": 2.9489347682846474, + "grad_norm": 2.330460548400879, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7357336282730103, + "num_tokens": 694807509.0, + "step": 26853 + }, + { + "epoch": 2.949044585987261, + "grad_norm": 2.4463350772857666, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7325688600540161, + "num_tokens": 694833524.0, + "step": 26854 + }, + { + "epoch": 2.949154403689875, + "grad_norm": 2.5634727478027344, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7611992955207825, + "num_tokens": 694855875.0, + "step": 26855 + }, + { + "epoch": 2.9492642213924887, + "grad_norm": 2.6400012969970703, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7310847043991089, + "num_tokens": 694877608.0, + "step": 26856 + }, + { + "epoch": 2.949374039095102, + "grad_norm": 2.355942726135254, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7335553765296936, + "num_tokens": 694901942.0, + "step": 26857 + }, + { + "epoch": 2.9494838567977157, + "grad_norm": 2.5044591426849365, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7426367998123169, + "num_tokens": 694925693.0, + "step": 26858 + }, + { + "epoch": 2.9495936745003295, + "grad_norm": 2.2233874797821045, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7306649088859558, + "num_tokens": 694957187.0, + "step": 26859 + }, + { + "epoch": 2.9497034922029433, + "grad_norm": 2.445309638977051, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7307213544845581, + "num_tokens": 694982495.0, + "step": 26860 + }, + { + "epoch": 2.949813309905557, + "grad_norm": 2.4434971809387207, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.739022970199585, + "num_tokens": 695006353.0, + "step": 26861 + }, + { + "epoch": 2.9499231276081703, + "grad_norm": 2.4104509353637695, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.743131160736084, + "num_tokens": 695030589.0, + "step": 26862 + }, + { + "epoch": 2.950032945310784, + "grad_norm": 2.491220712661743, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7227556705474854, + "num_tokens": 695055152.0, + "step": 26863 + }, + { + "epoch": 2.950142763013398, + "grad_norm": 2.549516201019287, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7396470308303833, + "num_tokens": 695077887.0, + "step": 26864 + }, + { + "epoch": 2.950252580716011, + "grad_norm": 2.142369031906128, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7180299758911133, + "num_tokens": 695108433.0, + "step": 26865 + }, + { + "epoch": 2.9503623984186254, + "grad_norm": 2.3754818439483643, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7247956991195679, + "num_tokens": 695134430.0, + "step": 26866 + }, + { + "epoch": 2.9504722161212387, + "grad_norm": 2.572493553161621, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.732578694820404, + "num_tokens": 695157626.0, + "step": 26867 + }, + { + "epoch": 2.9505820338238524, + "grad_norm": 2.28480863571167, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7177639007568359, + "num_tokens": 695185852.0, + "step": 26868 + }, + { + "epoch": 2.950691851526466, + "grad_norm": 2.374194622039795, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7245848178863525, + "num_tokens": 695213099.0, + "step": 26869 + }, + { + "epoch": 2.9508016692290795, + "grad_norm": 2.4913320541381836, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7146847248077393, + "num_tokens": 695239514.0, + "step": 26870 + }, + { + "epoch": 2.9509114869316933, + "grad_norm": 2.167130708694458, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7229927182197571, + "num_tokens": 695267943.0, + "step": 26871 + }, + { + "epoch": 2.951021304634307, + "grad_norm": 2.646092653274536, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7663824558258057, + "num_tokens": 695287393.0, + "step": 26872 + }, + { + "epoch": 2.9511311223369208, + "grad_norm": 2.428786516189575, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7328689694404602, + "num_tokens": 695310829.0, + "step": 26873 + }, + { + "epoch": 2.9512409400395345, + "grad_norm": 2.3714115619659424, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7451034784317017, + "num_tokens": 695335851.0, + "step": 26874 + }, + { + "epoch": 2.951350757742148, + "grad_norm": 2.3979909420013428, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7331103086471558, + "num_tokens": 695359724.0, + "step": 26875 + }, + { + "epoch": 2.9514605754447616, + "grad_norm": 2.4595792293548584, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.725046694278717, + "num_tokens": 695383927.0, + "step": 26876 + }, + { + "epoch": 2.9515703931473753, + "grad_norm": 2.0716195106506348, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7325184345245361, + "num_tokens": 695415377.0, + "step": 26877 + }, + { + "epoch": 2.951680210849989, + "grad_norm": 2.368441343307495, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7098497748374939, + "num_tokens": 695444063.0, + "step": 26878 + }, + { + "epoch": 2.951790028552603, + "grad_norm": 2.4874660968780518, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7413047552108765, + "num_tokens": 695468197.0, + "step": 26879 + }, + { + "epoch": 2.951899846255216, + "grad_norm": 2.142515182495117, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7041624784469604, + "num_tokens": 695498017.0, + "step": 26880 + }, + { + "epoch": 2.95200966395783, + "grad_norm": 2.3050625324249268, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7056248188018799, + "num_tokens": 695523901.0, + "step": 26881 + }, + { + "epoch": 2.9521194816604437, + "grad_norm": 2.2822530269622803, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7314379215240479, + "num_tokens": 695552061.0, + "step": 26882 + }, + { + "epoch": 2.9522292993630574, + "grad_norm": 2.209202766418457, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7023377418518066, + "num_tokens": 695582133.0, + "step": 26883 + }, + { + "epoch": 2.952339117065671, + "grad_norm": 2.2094290256500244, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7272381782531738, + "num_tokens": 695609439.0, + "step": 26884 + }, + { + "epoch": 2.9524489347682845, + "grad_norm": 2.5490832328796387, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7296494841575623, + "num_tokens": 695634479.0, + "step": 26885 + }, + { + "epoch": 2.9525587524708983, + "grad_norm": 2.40671706199646, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.721574068069458, + "num_tokens": 695659018.0, + "step": 26886 + }, + { + "epoch": 2.952668570173512, + "grad_norm": 2.210376024246216, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7297496795654297, + "num_tokens": 695685941.0, + "step": 26887 + }, + { + "epoch": 2.952778387876126, + "grad_norm": 2.7431278228759766, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7242069244384766, + "num_tokens": 695706360.0, + "step": 26888 + }, + { + "epoch": 2.9528882055787395, + "grad_norm": 2.200712203979492, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.741233229637146, + "num_tokens": 695736570.0, + "step": 26889 + }, + { + "epoch": 2.952998023281353, + "grad_norm": 2.5569794178009033, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7375938892364502, + "num_tokens": 695758780.0, + "step": 26890 + }, + { + "epoch": 2.9531078409839666, + "grad_norm": 2.611457109451294, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7290480136871338, + "num_tokens": 695781814.0, + "step": 26891 + }, + { + "epoch": 2.9532176586865804, + "grad_norm": 2.0114316940307617, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7372432947158813, + "num_tokens": 695816254.0, + "step": 26892 + }, + { + "epoch": 2.9533274763891937, + "grad_norm": 2.4270317554473877, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7268509268760681, + "num_tokens": 695841352.0, + "step": 26893 + }, + { + "epoch": 2.9534372940918074, + "grad_norm": 2.239783525466919, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7218834757804871, + "num_tokens": 695869778.0, + "step": 26894 + }, + { + "epoch": 2.953547111794421, + "grad_norm": 2.46254563331604, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.723955512046814, + "num_tokens": 695895169.0, + "step": 26895 + }, + { + "epoch": 2.953656929497035, + "grad_norm": 2.0367205142974854, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7405648827552795, + "num_tokens": 695929771.0, + "step": 26896 + }, + { + "epoch": 2.9537667471996487, + "grad_norm": 2.509666919708252, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7394883632659912, + "num_tokens": 695952223.0, + "step": 26897 + }, + { + "epoch": 2.953876564902262, + "grad_norm": 2.0509328842163086, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.715881884098053, + "num_tokens": 695983239.0, + "step": 26898 + }, + { + "epoch": 2.953986382604876, + "grad_norm": 2.4818062782287598, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.730717658996582, + "num_tokens": 696007787.0, + "step": 26899 + }, + { + "epoch": 2.9540962003074895, + "grad_norm": 2.4928314685821533, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7334155440330505, + "num_tokens": 696032528.0, + "step": 26900 + }, + { + "epoch": 2.9542060180101033, + "grad_norm": 2.296644926071167, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.709473729133606, + "num_tokens": 696061706.0, + "step": 26901 + }, + { + "epoch": 2.954315835712717, + "grad_norm": 2.2427525520324707, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7317463755607605, + "num_tokens": 696090386.0, + "step": 26902 + }, + { + "epoch": 2.9544256534153304, + "grad_norm": 2.3228001594543457, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7406590580940247, + "num_tokens": 696114610.0, + "step": 26903 + }, + { + "epoch": 2.954535471117944, + "grad_norm": 2.052825689315796, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7380216121673584, + "num_tokens": 696147123.0, + "step": 26904 + }, + { + "epoch": 2.954645288820558, + "grad_norm": 2.175126791000366, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7334833145141602, + "num_tokens": 696176382.0, + "step": 26905 + }, + { + "epoch": 2.9547551065231716, + "grad_norm": 2.3697755336761475, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7270136475563049, + "num_tokens": 696207007.0, + "step": 26906 + }, + { + "epoch": 2.9548649242257854, + "grad_norm": 2.37602162361145, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7331916093826294, + "num_tokens": 696233648.0, + "step": 26907 + }, + { + "epoch": 2.9549747419283987, + "grad_norm": 2.2601518630981445, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.732090175151825, + "num_tokens": 696261850.0, + "step": 26908 + }, + { + "epoch": 2.9550845596310125, + "grad_norm": 2.6711666584014893, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.751605212688446, + "num_tokens": 696283083.0, + "step": 26909 + }, + { + "epoch": 2.955194377333626, + "grad_norm": 2.0475645065307617, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.6996630430221558, + "num_tokens": 696318250.0, + "step": 26910 + }, + { + "epoch": 2.95530419503624, + "grad_norm": 2.597843647003174, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7431048154830933, + "num_tokens": 696338696.0, + "step": 26911 + }, + { + "epoch": 2.9554140127388537, + "grad_norm": 2.7466931343078613, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.744042694568634, + "num_tokens": 696359347.0, + "step": 26912 + }, + { + "epoch": 2.955523830441467, + "grad_norm": 2.6866297721862793, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7501607537269592, + "num_tokens": 696379979.0, + "step": 26913 + }, + { + "epoch": 2.955633648144081, + "grad_norm": 2.2317380905151367, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7156459093093872, + "num_tokens": 696410744.0, + "step": 26914 + }, + { + "epoch": 2.9557434658466946, + "grad_norm": 2.1370420455932617, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7236155867576599, + "num_tokens": 696442078.0, + "step": 26915 + }, + { + "epoch": 2.955853283549308, + "grad_norm": 2.5634872913360596, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7375504970550537, + "num_tokens": 696464422.0, + "step": 26916 + }, + { + "epoch": 2.955963101251922, + "grad_norm": 2.4785776138305664, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.740418553352356, + "num_tokens": 696488774.0, + "step": 26917 + }, + { + "epoch": 2.9560729189545354, + "grad_norm": 2.2694053649902344, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7145604491233826, + "num_tokens": 696516657.0, + "step": 26918 + }, + { + "epoch": 2.956182736657149, + "grad_norm": 2.7063231468200684, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.730766773223877, + "num_tokens": 696537618.0, + "step": 26919 + }, + { + "epoch": 2.956292554359763, + "grad_norm": 2.336271047592163, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7521325349807739, + "num_tokens": 696562197.0, + "step": 26920 + }, + { + "epoch": 2.956402372062376, + "grad_norm": 2.378974437713623, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.722698986530304, + "num_tokens": 696588380.0, + "step": 26921 + }, + { + "epoch": 2.95651218976499, + "grad_norm": 2.43066668510437, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.726534903049469, + "num_tokens": 696612837.0, + "step": 26922 + }, + { + "epoch": 2.9566220074676037, + "grad_norm": 2.5872621536254883, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7347043752670288, + "num_tokens": 696633842.0, + "step": 26923 + }, + { + "epoch": 2.9567318251702175, + "grad_norm": 2.3107333183288574, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7296277284622192, + "num_tokens": 696660609.0, + "step": 26924 + }, + { + "epoch": 2.9568416428728312, + "grad_norm": 2.272620439529419, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7215177416801453, + "num_tokens": 696688257.0, + "step": 26925 + }, + { + "epoch": 2.9569514605754446, + "grad_norm": 2.59841251373291, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7464306354522705, + "num_tokens": 696712726.0, + "step": 26926 + }, + { + "epoch": 2.9570612782780583, + "grad_norm": 2.1961312294006348, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7367086410522461, + "num_tokens": 696743551.0, + "step": 26927 + }, + { + "epoch": 2.957171095980672, + "grad_norm": 2.1243553161621094, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7452139258384705, + "num_tokens": 696772162.0, + "step": 26928 + }, + { + "epoch": 2.957280913683286, + "grad_norm": 2.3383595943450928, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7393780946731567, + "num_tokens": 696796890.0, + "step": 26929 + }, + { + "epoch": 2.9573907313858996, + "grad_norm": 2.5049304962158203, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7288800477981567, + "num_tokens": 696821425.0, + "step": 26930 + }, + { + "epoch": 2.957500549088513, + "grad_norm": 2.357576370239258, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7291726469993591, + "num_tokens": 696849118.0, + "step": 26931 + }, + { + "epoch": 2.9576103667911267, + "grad_norm": 2.3491854667663574, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7190442085266113, + "num_tokens": 696875342.0, + "step": 26932 + }, + { + "epoch": 2.9577201844937404, + "grad_norm": 2.4213273525238037, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7284356951713562, + "num_tokens": 696900139.0, + "step": 26933 + }, + { + "epoch": 2.957830002196354, + "grad_norm": 2.438739538192749, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7176340818405151, + "num_tokens": 696925544.0, + "step": 26934 + }, + { + "epoch": 2.957939819898968, + "grad_norm": 2.1662368774414062, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7302893400192261, + "num_tokens": 696957200.0, + "step": 26935 + }, + { + "epoch": 2.9580496376015812, + "grad_norm": 2.2837188243865967, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.71892249584198, + "num_tokens": 696984092.0, + "step": 26936 + }, + { + "epoch": 2.958159455304195, + "grad_norm": 2.536139965057373, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7442319989204407, + "num_tokens": 697009435.0, + "step": 26937 + }, + { + "epoch": 2.9582692730068088, + "grad_norm": 2.0996429920196533, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7384450435638428, + "num_tokens": 697040127.0, + "step": 26938 + }, + { + "epoch": 2.9583790907094225, + "grad_norm": 2.1349761486053467, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7292892336845398, + "num_tokens": 697068990.0, + "step": 26939 + }, + { + "epoch": 2.9584889084120363, + "grad_norm": 2.414569616317749, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7255587577819824, + "num_tokens": 697093253.0, + "step": 26940 + }, + { + "epoch": 2.9585987261146496, + "grad_norm": 2.040574073791504, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7168399691581726, + "num_tokens": 697127762.0, + "step": 26941 + }, + { + "epoch": 2.9587085438172633, + "grad_norm": 2.4249768257141113, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7193602323532104, + "num_tokens": 697152355.0, + "step": 26942 + }, + { + "epoch": 2.958818361519877, + "grad_norm": 2.331164836883545, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7521783113479614, + "num_tokens": 697177570.0, + "step": 26943 + }, + { + "epoch": 2.9589281792224904, + "grad_norm": 2.2573060989379883, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7622534036636353, + "num_tokens": 697203614.0, + "step": 26944 + }, + { + "epoch": 2.959037996925104, + "grad_norm": 2.5139265060424805, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7480204105377197, + "num_tokens": 697227800.0, + "step": 26945 + }, + { + "epoch": 2.959147814627718, + "grad_norm": 2.6098146438598633, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7329976558685303, + "num_tokens": 697249865.0, + "step": 26946 + }, + { + "epoch": 2.9592576323303317, + "grad_norm": 2.5771830081939697, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7329391241073608, + "num_tokens": 697272909.0, + "step": 26947 + }, + { + "epoch": 2.9593674500329454, + "grad_norm": 2.1886861324310303, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7483096718788147, + "num_tokens": 697300674.0, + "step": 26948 + }, + { + "epoch": 2.9594772677355587, + "grad_norm": 2.348724603652954, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7651707530021667, + "num_tokens": 697327265.0, + "step": 26949 + }, + { + "epoch": 2.9595870854381725, + "grad_norm": 2.332719326019287, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7547841668128967, + "num_tokens": 697352568.0, + "step": 26950 + }, + { + "epoch": 2.9596969031407863, + "grad_norm": 2.056102752685547, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7235190868377686, + "num_tokens": 697383626.0, + "step": 26951 + }, + { + "epoch": 2.9598067208434, + "grad_norm": 2.586784839630127, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7309491038322449, + "num_tokens": 697406230.0, + "step": 26952 + }, + { + "epoch": 2.9599165385460138, + "grad_norm": 2.550213098526001, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7430540323257446, + "num_tokens": 697428583.0, + "step": 26953 + }, + { + "epoch": 2.960026356248627, + "grad_norm": 2.6236519813537598, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7309632301330566, + "num_tokens": 697450490.0, + "step": 26954 + }, + { + "epoch": 2.960136173951241, + "grad_norm": 2.5039732456207275, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7190956473350525, + "num_tokens": 697474413.0, + "step": 26955 + }, + { + "epoch": 2.9602459916538546, + "grad_norm": 2.3876724243164062, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7121745347976685, + "num_tokens": 697500328.0, + "step": 26956 + }, + { + "epoch": 2.9603558093564684, + "grad_norm": 2.090691089630127, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7460612058639526, + "num_tokens": 697531539.0, + "step": 26957 + }, + { + "epoch": 2.960465627059082, + "grad_norm": 2.2785446643829346, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7128636837005615, + "num_tokens": 697557611.0, + "step": 26958 + }, + { + "epoch": 2.9605754447616954, + "grad_norm": 2.315984010696411, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7263880968093872, + "num_tokens": 697584171.0, + "step": 26959 + }, + { + "epoch": 2.960685262464309, + "grad_norm": 2.309333086013794, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7332226037979126, + "num_tokens": 697610562.0, + "step": 26960 + }, + { + "epoch": 2.960795080166923, + "grad_norm": 2.598212480545044, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7382920980453491, + "num_tokens": 697632479.0, + "step": 26961 + }, + { + "epoch": 2.9609048978695367, + "grad_norm": 2.2830874919891357, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7249132394790649, + "num_tokens": 697659337.0, + "step": 26962 + }, + { + "epoch": 2.9610147155721505, + "grad_norm": 2.062161445617676, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7340718507766724, + "num_tokens": 697692917.0, + "step": 26963 + }, + { + "epoch": 2.9611245332747638, + "grad_norm": 2.142548084259033, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7205640077590942, + "num_tokens": 697722169.0, + "step": 26964 + }, + { + "epoch": 2.9612343509773775, + "grad_norm": 2.2680981159210205, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7293626666069031, + "num_tokens": 697750091.0, + "step": 26965 + }, + { + "epoch": 2.9613441686799913, + "grad_norm": 2.412102699279785, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7204679846763611, + "num_tokens": 697776446.0, + "step": 26966 + }, + { + "epoch": 2.961453986382605, + "grad_norm": 2.7241127490997314, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7277867794036865, + "num_tokens": 697796895.0, + "step": 26967 + }, + { + "epoch": 2.961563804085219, + "grad_norm": 2.1955864429473877, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7075894474983215, + "num_tokens": 697827015.0, + "step": 26968 + }, + { + "epoch": 2.961673621787832, + "grad_norm": 2.66308856010437, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.759404182434082, + "num_tokens": 697849556.0, + "step": 26969 + }, + { + "epoch": 2.961783439490446, + "grad_norm": 2.3945603370666504, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7503583431243896, + "num_tokens": 697873031.0, + "step": 26970 + }, + { + "epoch": 2.9618932571930596, + "grad_norm": 2.2331066131591797, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7254944443702698, + "num_tokens": 697902202.0, + "step": 26971 + }, + { + "epoch": 2.962003074895673, + "grad_norm": 2.246889352798462, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7395975589752197, + "num_tokens": 697929995.0, + "step": 26972 + }, + { + "epoch": 2.9621128925982867, + "grad_norm": 2.3920443058013916, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7399474382400513, + "num_tokens": 697954809.0, + "step": 26973 + }, + { + "epoch": 2.9622227103009005, + "grad_norm": 2.3851377964019775, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.705109715461731, + "num_tokens": 697982219.0, + "step": 26974 + }, + { + "epoch": 2.962332528003514, + "grad_norm": 2.2716331481933594, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7226516008377075, + "num_tokens": 698010780.0, + "step": 26975 + }, + { + "epoch": 2.962442345706128, + "grad_norm": 2.25290584564209, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7400611042976379, + "num_tokens": 698036413.0, + "step": 26976 + }, + { + "epoch": 2.9625521634087413, + "grad_norm": 2.2689759731292725, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7441186904907227, + "num_tokens": 698062900.0, + "step": 26977 + }, + { + "epoch": 2.962661981111355, + "grad_norm": 2.5233511924743652, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7229098081588745, + "num_tokens": 698087121.0, + "step": 26978 + }, + { + "epoch": 2.962771798813969, + "grad_norm": 2.374901056289673, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7130321264266968, + "num_tokens": 698113106.0, + "step": 26979 + }, + { + "epoch": 2.9628816165165826, + "grad_norm": 2.363591432571411, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7439308166503906, + "num_tokens": 698136953.0, + "step": 26980 + }, + { + "epoch": 2.9629914342191963, + "grad_norm": 2.814612627029419, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7107354402542114, + "num_tokens": 698160688.0, + "step": 26981 + }, + { + "epoch": 2.9631012519218096, + "grad_norm": 2.3508479595184326, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7406536340713501, + "num_tokens": 698186106.0, + "step": 26982 + }, + { + "epoch": 2.9632110696244234, + "grad_norm": 2.3647398948669434, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7331795692443848, + "num_tokens": 698213086.0, + "step": 26983 + }, + { + "epoch": 2.963320887327037, + "grad_norm": 2.085867166519165, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7465249300003052, + "num_tokens": 698243181.0, + "step": 26984 + }, + { + "epoch": 2.963430705029651, + "grad_norm": 2.515357732772827, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7428468465805054, + "num_tokens": 698267595.0, + "step": 26985 + }, + { + "epoch": 2.9635405227322646, + "grad_norm": 2.5155837535858154, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7485407590866089, + "num_tokens": 698291108.0, + "step": 26986 + }, + { + "epoch": 2.963650340434878, + "grad_norm": 2.173163890838623, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7347773313522339, + "num_tokens": 698321436.0, + "step": 26987 + }, + { + "epoch": 2.9637601581374917, + "grad_norm": 2.3409743309020996, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.735141396522522, + "num_tokens": 698348173.0, + "step": 26988 + }, + { + "epoch": 2.9638699758401055, + "grad_norm": 2.0399434566497803, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7051094770431519, + "num_tokens": 698382618.0, + "step": 26989 + }, + { + "epoch": 2.9639797935427192, + "grad_norm": 1.9501500129699707, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7219807505607605, + "num_tokens": 698419430.0, + "step": 26990 + }, + { + "epoch": 2.964089611245333, + "grad_norm": 2.6168594360351562, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7500296831130981, + "num_tokens": 698440289.0, + "step": 26991 + }, + { + "epoch": 2.9641994289479463, + "grad_norm": 2.4234063625335693, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7310901284217834, + "num_tokens": 698468329.0, + "step": 26992 + }, + { + "epoch": 2.96430924665056, + "grad_norm": 2.264822006225586, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7339563965797424, + "num_tokens": 698493506.0, + "step": 26993 + }, + { + "epoch": 2.964419064353174, + "grad_norm": 2.71024227142334, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7508337497711182, + "num_tokens": 698512949.0, + "step": 26994 + }, + { + "epoch": 2.964528882055787, + "grad_norm": 2.3384571075439453, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7215216159820557, + "num_tokens": 698540485.0, + "step": 26995 + }, + { + "epoch": 2.9646386997584013, + "grad_norm": 2.2606561183929443, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7073082327842712, + "num_tokens": 698568590.0, + "step": 26996 + }, + { + "epoch": 2.9647485174610146, + "grad_norm": 2.3748371601104736, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7313220500946045, + "num_tokens": 698595857.0, + "step": 26997 + }, + { + "epoch": 2.9648583351636284, + "grad_norm": 2.0395009517669678, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7114110589027405, + "num_tokens": 698630604.0, + "step": 26998 + }, + { + "epoch": 2.964968152866242, + "grad_norm": 2.4752357006073, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7472206354141235, + "num_tokens": 698655461.0, + "step": 26999 + }, + { + "epoch": 2.9650779705688555, + "grad_norm": 2.05893611907959, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7090935111045837, + "num_tokens": 698690583.0, + "step": 27000 + }, + { + "epoch": 2.9651877882714692, + "grad_norm": 2.2254862785339355, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7151707410812378, + "num_tokens": 698717539.0, + "step": 27001 + }, + { + "epoch": 2.965297605974083, + "grad_norm": 2.5373404026031494, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7473552227020264, + "num_tokens": 698740933.0, + "step": 27002 + }, + { + "epoch": 2.9654074236766967, + "grad_norm": 2.4004199504852295, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7343835830688477, + "num_tokens": 698767324.0, + "step": 27003 + }, + { + "epoch": 2.9655172413793105, + "grad_norm": 2.588890314102173, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7257207632064819, + "num_tokens": 698795914.0, + "step": 27004 + }, + { + "epoch": 2.965627059081924, + "grad_norm": 2.485386848449707, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7278798818588257, + "num_tokens": 698821613.0, + "step": 27005 + }, + { + "epoch": 2.9657368767845376, + "grad_norm": 2.634791374206543, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7278204560279846, + "num_tokens": 698842556.0, + "step": 27006 + }, + { + "epoch": 2.9658466944871513, + "grad_norm": 2.3156678676605225, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7288039326667786, + "num_tokens": 698871271.0, + "step": 27007 + }, + { + "epoch": 2.965956512189765, + "grad_norm": 2.2976138591766357, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7124026417732239, + "num_tokens": 698897929.0, + "step": 27008 + }, + { + "epoch": 2.966066329892379, + "grad_norm": 2.454404354095459, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7133979797363281, + "num_tokens": 698922956.0, + "step": 27009 + }, + { + "epoch": 2.966176147594992, + "grad_norm": 2.093451976776123, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.743668794631958, + "num_tokens": 698952848.0, + "step": 27010 + }, + { + "epoch": 2.966285965297606, + "grad_norm": 2.3991541862487793, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.735808253288269, + "num_tokens": 698977153.0, + "step": 27011 + }, + { + "epoch": 2.9663957830002197, + "grad_norm": 2.748934030532837, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7675957679748535, + "num_tokens": 698996246.0, + "step": 27012 + }, + { + "epoch": 2.9665056007028334, + "grad_norm": 2.586141347885132, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7371211647987366, + "num_tokens": 699018202.0, + "step": 27013 + }, + { + "epoch": 2.966615418405447, + "grad_norm": 2.6721649169921875, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7324640154838562, + "num_tokens": 699039269.0, + "step": 27014 + }, + { + "epoch": 2.9667252361080605, + "grad_norm": 2.210578441619873, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7219085693359375, + "num_tokens": 699067947.0, + "step": 27015 + }, + { + "epoch": 2.9668350538106742, + "grad_norm": 2.358447551727295, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7261339426040649, + "num_tokens": 699093164.0, + "step": 27016 + }, + { + "epoch": 2.966944871513288, + "grad_norm": 2.4065425395965576, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7507739067077637, + "num_tokens": 699116466.0, + "step": 27017 + }, + { + "epoch": 2.9670546892159018, + "grad_norm": 2.5456855297088623, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7598973512649536, + "num_tokens": 699138646.0, + "step": 27018 + }, + { + "epoch": 2.9671645069185155, + "grad_norm": 2.39968204498291, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7356465458869934, + "num_tokens": 699163571.0, + "step": 27019 + }, + { + "epoch": 2.967274324621129, + "grad_norm": 2.1611461639404297, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7182495594024658, + "num_tokens": 699194222.0, + "step": 27020 + }, + { + "epoch": 2.9673841423237426, + "grad_norm": 2.0787715911865234, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7277435064315796, + "num_tokens": 699225044.0, + "step": 27021 + }, + { + "epoch": 2.9674939600263563, + "grad_norm": 2.2347347736358643, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7144357562065125, + "num_tokens": 699254793.0, + "step": 27022 + }, + { + "epoch": 2.9676037777289697, + "grad_norm": 2.546907424926758, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.731591522693634, + "num_tokens": 699276903.0, + "step": 27023 + }, + { + "epoch": 2.9677135954315834, + "grad_norm": 2.1370651721954346, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7423748970031738, + "num_tokens": 699306065.0, + "step": 27024 + }, + { + "epoch": 2.967823413134197, + "grad_norm": 2.749880075454712, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7285124063491821, + "num_tokens": 699329481.0, + "step": 27025 + }, + { + "epoch": 2.967933230836811, + "grad_norm": 2.11626935005188, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7122016549110413, + "num_tokens": 699361293.0, + "step": 27026 + }, + { + "epoch": 2.9680430485394247, + "grad_norm": 2.359816551208496, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7421860098838806, + "num_tokens": 699387539.0, + "step": 27027 + }, + { + "epoch": 2.968152866242038, + "grad_norm": 2.7433645725250244, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7450063228607178, + "num_tokens": 699407609.0, + "step": 27028 + }, + { + "epoch": 2.9682626839446518, + "grad_norm": 2.2571208477020264, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7237564921379089, + "num_tokens": 699435534.0, + "step": 27029 + }, + { + "epoch": 2.9683725016472655, + "grad_norm": 2.413684129714966, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.729857325553894, + "num_tokens": 699460880.0, + "step": 27030 + }, + { + "epoch": 2.9684823193498793, + "grad_norm": 2.5890278816223145, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.755583643913269, + "num_tokens": 699481884.0, + "step": 27031 + }, + { + "epoch": 2.968592137052493, + "grad_norm": 2.3945019245147705, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7342697381973267, + "num_tokens": 699506015.0, + "step": 27032 + }, + { + "epoch": 2.9687019547551063, + "grad_norm": 2.365814208984375, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7222403287887573, + "num_tokens": 699532009.0, + "step": 27033 + }, + { + "epoch": 2.96881177245772, + "grad_norm": 2.543220281600952, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7353039383888245, + "num_tokens": 699553030.0, + "step": 27034 + }, + { + "epoch": 2.968921590160334, + "grad_norm": 2.2571754455566406, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7424643039703369, + "num_tokens": 699579827.0, + "step": 27035 + }, + { + "epoch": 2.9690314078629476, + "grad_norm": 2.2623391151428223, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7284687757492065, + "num_tokens": 699606097.0, + "step": 27036 + }, + { + "epoch": 2.9691412255655614, + "grad_norm": 2.432474374771118, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7456995248794556, + "num_tokens": 699628139.0, + "step": 27037 + }, + { + "epoch": 2.9692510432681747, + "grad_norm": 2.254654884338379, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.726048469543457, + "num_tokens": 699656894.0, + "step": 27038 + }, + { + "epoch": 2.9693608609707884, + "grad_norm": 2.4974794387817383, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.724571943283081, + "num_tokens": 699682476.0, + "step": 27039 + }, + { + "epoch": 2.969470678673402, + "grad_norm": 2.3887767791748047, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7246869802474976, + "num_tokens": 699708794.0, + "step": 27040 + }, + { + "epoch": 2.969580496376016, + "grad_norm": 2.5524792671203613, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7459671497344971, + "num_tokens": 699731205.0, + "step": 27041 + }, + { + "epoch": 2.9696903140786297, + "grad_norm": 2.29856276512146, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7399636507034302, + "num_tokens": 699757450.0, + "step": 27042 + }, + { + "epoch": 2.969800131781243, + "grad_norm": 2.575852155685425, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7496377229690552, + "num_tokens": 699778921.0, + "step": 27043 + }, + { + "epoch": 2.969909949483857, + "grad_norm": 2.231867790222168, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.732284426689148, + "num_tokens": 699808168.0, + "step": 27044 + }, + { + "epoch": 2.9700197671864705, + "grad_norm": 2.205918073654175, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7292501330375671, + "num_tokens": 699838825.0, + "step": 27045 + }, + { + "epoch": 2.970129584889084, + "grad_norm": 2.2216649055480957, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7232156991958618, + "num_tokens": 699866348.0, + "step": 27046 + }, + { + "epoch": 2.970239402591698, + "grad_norm": 2.4507689476013184, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7454472780227661, + "num_tokens": 699889755.0, + "step": 27047 + }, + { + "epoch": 2.9703492202943114, + "grad_norm": 2.1546921730041504, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7100143432617188, + "num_tokens": 699922208.0, + "step": 27048 + }, + { + "epoch": 2.970459037996925, + "grad_norm": 2.8351056575775146, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7253366708755493, + "num_tokens": 699941048.0, + "step": 27049 + }, + { + "epoch": 2.970568855699539, + "grad_norm": 2.165417194366455, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7150336503982544, + "num_tokens": 699969241.0, + "step": 27050 + }, + { + "epoch": 2.970678673402152, + "grad_norm": 2.322873592376709, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7383173704147339, + "num_tokens": 699993630.0, + "step": 27051 + }, + { + "epoch": 2.970788491104766, + "grad_norm": 2.339569330215454, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7388995885848999, + "num_tokens": 700020713.0, + "step": 27052 + }, + { + "epoch": 2.9708983088073797, + "grad_norm": 2.6164588928222656, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7379165887832642, + "num_tokens": 700043333.0, + "step": 27053 + }, + { + "epoch": 2.9710081265099935, + "grad_norm": 2.315829277038574, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7220504283905029, + "num_tokens": 700071824.0, + "step": 27054 + }, + { + "epoch": 2.971117944212607, + "grad_norm": 2.309213638305664, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7504720687866211, + "num_tokens": 700096285.0, + "step": 27055 + }, + { + "epoch": 2.9712277619152205, + "grad_norm": 2.3577466011047363, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7264931201934814, + "num_tokens": 700122349.0, + "step": 27056 + }, + { + "epoch": 2.9713375796178343, + "grad_norm": 2.0957722663879395, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7352719902992249, + "num_tokens": 700152605.0, + "step": 27057 + }, + { + "epoch": 2.971447397320448, + "grad_norm": 2.5838944911956787, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7522299885749817, + "num_tokens": 700173611.0, + "step": 27058 + }, + { + "epoch": 2.971557215023062, + "grad_norm": 2.194260358810425, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7231025099754333, + "num_tokens": 700205386.0, + "step": 27059 + }, + { + "epoch": 2.9716670327256756, + "grad_norm": 2.1958444118499756, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7365256547927856, + "num_tokens": 700235046.0, + "step": 27060 + }, + { + "epoch": 2.971776850428289, + "grad_norm": 2.1736834049224854, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7270411252975464, + "num_tokens": 700265594.0, + "step": 27061 + }, + { + "epoch": 2.9718866681309026, + "grad_norm": 2.991346836090088, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7476721405982971, + "num_tokens": 700284713.0, + "step": 27062 + }, + { + "epoch": 2.9719964858335164, + "grad_norm": 2.4730007648468018, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7291383743286133, + "num_tokens": 700311304.0, + "step": 27063 + }, + { + "epoch": 2.97210630353613, + "grad_norm": 2.5968453884124756, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7356914281845093, + "num_tokens": 700334528.0, + "step": 27064 + }, + { + "epoch": 2.972216121238744, + "grad_norm": 2.1687583923339844, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7203744649887085, + "num_tokens": 700365145.0, + "step": 27065 + }, + { + "epoch": 2.972325938941357, + "grad_norm": 2.146066665649414, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7392327189445496, + "num_tokens": 700394174.0, + "step": 27066 + }, + { + "epoch": 2.972435756643971, + "grad_norm": 2.267235517501831, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7386297583580017, + "num_tokens": 700422187.0, + "step": 27067 + }, + { + "epoch": 2.9725455743465847, + "grad_norm": 2.4175541400909424, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7034717202186584, + "num_tokens": 700449559.0, + "step": 27068 + }, + { + "epoch": 2.9726553920491985, + "grad_norm": 2.319657802581787, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7210254669189453, + "num_tokens": 700479833.0, + "step": 27069 + }, + { + "epoch": 2.9727652097518122, + "grad_norm": 2.4576492309570312, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7420730590820312, + "num_tokens": 700501218.0, + "step": 27070 + }, + { + "epoch": 2.9728750274544256, + "grad_norm": 2.2143683433532715, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7033666372299194, + "num_tokens": 700529618.0, + "step": 27071 + }, + { + "epoch": 2.9729848451570393, + "grad_norm": 2.279428720474243, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7341204881668091, + "num_tokens": 700555811.0, + "step": 27072 + }, + { + "epoch": 2.973094662859653, + "grad_norm": 2.9165658950805664, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7683814167976379, + "num_tokens": 700574821.0, + "step": 27073 + }, + { + "epoch": 2.9732044805622664, + "grad_norm": 2.4627842903137207, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7219443917274475, + "num_tokens": 700599791.0, + "step": 27074 + }, + { + "epoch": 2.97331429826488, + "grad_norm": 2.3406145572662354, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7356964349746704, + "num_tokens": 700627048.0, + "step": 27075 + }, + { + "epoch": 2.973424115967494, + "grad_norm": 2.1442620754241943, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7171666622161865, + "num_tokens": 700656950.0, + "step": 27076 + }, + { + "epoch": 2.9735339336701077, + "grad_norm": 2.205275297164917, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7159130573272705, + "num_tokens": 700686909.0, + "step": 27077 + }, + { + "epoch": 2.9736437513727214, + "grad_norm": 2.474484920501709, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7515068054199219, + "num_tokens": 700711077.0, + "step": 27078 + }, + { + "epoch": 2.9737535690753347, + "grad_norm": 2.3452084064483643, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7339913249015808, + "num_tokens": 700737323.0, + "step": 27079 + }, + { + "epoch": 2.9738633867779485, + "grad_norm": 2.4820566177368164, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7256819009780884, + "num_tokens": 700763459.0, + "step": 27080 + }, + { + "epoch": 2.9739732044805622, + "grad_norm": 2.3768310546875, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7626920938491821, + "num_tokens": 700790020.0, + "step": 27081 + }, + { + "epoch": 2.974083022183176, + "grad_norm": 2.6894493103027344, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.76707923412323, + "num_tokens": 700810259.0, + "step": 27082 + }, + { + "epoch": 2.9741928398857898, + "grad_norm": 2.274409770965576, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7206491827964783, + "num_tokens": 700837021.0, + "step": 27083 + }, + { + "epoch": 2.974302657588403, + "grad_norm": 2.22395396232605, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7360885739326477, + "num_tokens": 700864837.0, + "step": 27084 + }, + { + "epoch": 2.974412475291017, + "grad_norm": 2.2120449542999268, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7320014238357544, + "num_tokens": 700893517.0, + "step": 27085 + }, + { + "epoch": 2.9745222929936306, + "grad_norm": 2.426945686340332, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7192322015762329, + "num_tokens": 700918292.0, + "step": 27086 + }, + { + "epoch": 2.9746321106962443, + "grad_norm": 2.385420560836792, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7359212040901184, + "num_tokens": 700942703.0, + "step": 27087 + }, + { + "epoch": 2.974741928398858, + "grad_norm": 2.5428988933563232, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7470344305038452, + "num_tokens": 700966432.0, + "step": 27088 + }, + { + "epoch": 2.9748517461014714, + "grad_norm": 2.2193076610565186, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7307894229888916, + "num_tokens": 700994539.0, + "step": 27089 + }, + { + "epoch": 2.974961563804085, + "grad_norm": 2.0485661029815674, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7217296361923218, + "num_tokens": 701028518.0, + "step": 27090 + }, + { + "epoch": 2.975071381506699, + "grad_norm": 2.30002498626709, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7383553981781006, + "num_tokens": 701053782.0, + "step": 27091 + }, + { + "epoch": 2.9751811992093127, + "grad_norm": 2.435598373413086, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7529281377792358, + "num_tokens": 701077793.0, + "step": 27092 + }, + { + "epoch": 2.9752910169119264, + "grad_norm": 2.360161781311035, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.7094492316246033, + "num_tokens": 701104306.0, + "step": 27093 + }, + { + "epoch": 2.9754008346145397, + "grad_norm": 2.300159454345703, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7411565780639648, + "num_tokens": 701129809.0, + "step": 27094 + }, + { + "epoch": 2.9755106523171535, + "grad_norm": 2.269534111022949, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7416499853134155, + "num_tokens": 701157533.0, + "step": 27095 + }, + { + "epoch": 2.9756204700197673, + "grad_norm": 2.150961399078369, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7299494743347168, + "num_tokens": 701188408.0, + "step": 27096 + }, + { + "epoch": 2.9757302877223806, + "grad_norm": 2.4680845737457275, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7221217751502991, + "num_tokens": 701212371.0, + "step": 27097 + }, + { + "epoch": 2.9758401054249948, + "grad_norm": 2.3821418285369873, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7507096529006958, + "num_tokens": 701235803.0, + "step": 27098 + }, + { + "epoch": 2.975949923127608, + "grad_norm": 2.314222574234009, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7227502465248108, + "num_tokens": 701263089.0, + "step": 27099 + }, + { + "epoch": 2.976059740830222, + "grad_norm": 2.367382287979126, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7230184674263, + "num_tokens": 701288344.0, + "step": 27100 + }, + { + "epoch": 2.9761695585328356, + "grad_norm": 2.323599100112915, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7260249853134155, + "num_tokens": 701314693.0, + "step": 27101 + }, + { + "epoch": 2.976279376235449, + "grad_norm": 2.7405788898468018, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7552733421325684, + "num_tokens": 701332861.0, + "step": 27102 + }, + { + "epoch": 2.9763891939380627, + "grad_norm": 2.301696300506592, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.710513710975647, + "num_tokens": 701362650.0, + "step": 27103 + }, + { + "epoch": 2.9764990116406764, + "grad_norm": 2.3982889652252197, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7227874398231506, + "num_tokens": 701387886.0, + "step": 27104 + }, + { + "epoch": 2.97660882934329, + "grad_norm": 2.419811964035034, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7279750108718872, + "num_tokens": 701411826.0, + "step": 27105 + }, + { + "epoch": 2.976718647045904, + "grad_norm": 2.3154921531677246, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.740166425704956, + "num_tokens": 701437492.0, + "step": 27106 + }, + { + "epoch": 2.9768284647485173, + "grad_norm": 2.173600435256958, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7285735011100769, + "num_tokens": 701467832.0, + "step": 27107 + }, + { + "epoch": 2.976938282451131, + "grad_norm": 2.322561740875244, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.72713702917099, + "num_tokens": 701495198.0, + "step": 27108 + }, + { + "epoch": 2.9770481001537448, + "grad_norm": 2.1865181922912598, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7221809029579163, + "num_tokens": 701523997.0, + "step": 27109 + }, + { + "epoch": 2.9771579178563585, + "grad_norm": 2.511796474456787, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7129241228103638, + "num_tokens": 701549938.0, + "step": 27110 + }, + { + "epoch": 2.9772677355589723, + "grad_norm": 2.309854745864868, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7266127467155457, + "num_tokens": 701576929.0, + "step": 27111 + }, + { + "epoch": 2.9773775532615856, + "grad_norm": 2.9714906215667725, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7331225275993347, + "num_tokens": 701594938.0, + "step": 27112 + }, + { + "epoch": 2.9774873709641994, + "grad_norm": 2.334073543548584, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7168088555335999, + "num_tokens": 701620664.0, + "step": 27113 + }, + { + "epoch": 2.977597188666813, + "grad_norm": 2.5862011909484863, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7499306797981262, + "num_tokens": 701642530.0, + "step": 27114 + }, + { + "epoch": 2.977707006369427, + "grad_norm": 2.3038227558135986, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7263069152832031, + "num_tokens": 701667475.0, + "step": 27115 + }, + { + "epoch": 2.9778168240720406, + "grad_norm": 2.7368600368499756, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7274494767189026, + "num_tokens": 701689185.0, + "step": 27116 + }, + { + "epoch": 2.977926641774654, + "grad_norm": 2.2594361305236816, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7539529800415039, + "num_tokens": 701717697.0, + "step": 27117 + }, + { + "epoch": 2.9780364594772677, + "grad_norm": 2.2336020469665527, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7462691068649292, + "num_tokens": 701743206.0, + "step": 27118 + }, + { + "epoch": 2.9781462771798815, + "grad_norm": 2.375087022781372, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7374067902565002, + "num_tokens": 701766654.0, + "step": 27119 + }, + { + "epoch": 2.978256094882495, + "grad_norm": 2.5425877571105957, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7240332961082458, + "num_tokens": 701789536.0, + "step": 27120 + }, + { + "epoch": 2.978365912585109, + "grad_norm": 2.4490740299224854, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7359346151351929, + "num_tokens": 701812744.0, + "step": 27121 + }, + { + "epoch": 2.9784757302877223, + "grad_norm": 2.405278444290161, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7473008632659912, + "num_tokens": 701836365.0, + "step": 27122 + }, + { + "epoch": 2.978585547990336, + "grad_norm": 2.727721929550171, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7377825975418091, + "num_tokens": 701857860.0, + "step": 27123 + }, + { + "epoch": 2.97869536569295, + "grad_norm": 2.3023128509521484, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7173090577125549, + "num_tokens": 701886383.0, + "step": 27124 + }, + { + "epoch": 2.978805183395563, + "grad_norm": 2.6478757858276367, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.770655632019043, + "num_tokens": 701907847.0, + "step": 27125 + }, + { + "epoch": 2.978915001098177, + "grad_norm": 2.355133533477783, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7351090312004089, + "num_tokens": 701936129.0, + "step": 27126 + }, + { + "epoch": 2.9790248188007906, + "grad_norm": 2.5185818672180176, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7297559380531311, + "num_tokens": 701959243.0, + "step": 27127 + }, + { + "epoch": 2.9791346365034044, + "grad_norm": 2.246896982192993, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7349253296852112, + "num_tokens": 701985609.0, + "step": 27128 + }, + { + "epoch": 2.979244454206018, + "grad_norm": 2.4468512535095215, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7244215607643127, + "num_tokens": 702010856.0, + "step": 27129 + }, + { + "epoch": 2.9793542719086314, + "grad_norm": 2.399073600769043, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7207117080688477, + "num_tokens": 702036322.0, + "step": 27130 + }, + { + "epoch": 2.979464089611245, + "grad_norm": 2.229524612426758, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7461248636245728, + "num_tokens": 702064103.0, + "step": 27131 + }, + { + "epoch": 2.979573907313859, + "grad_norm": 2.3592641353607178, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7368692755699158, + "num_tokens": 702089732.0, + "step": 27132 + }, + { + "epoch": 2.9796837250164727, + "grad_norm": 2.5191762447357178, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.759529173374176, + "num_tokens": 702111620.0, + "step": 27133 + }, + { + "epoch": 2.9797935427190865, + "grad_norm": 2.4569144248962402, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7062475681304932, + "num_tokens": 702136971.0, + "step": 27134 + }, + { + "epoch": 2.9799033604217, + "grad_norm": 2.1239469051361084, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7284731864929199, + "num_tokens": 702166116.0, + "step": 27135 + }, + { + "epoch": 2.9800131781243135, + "grad_norm": 3.240889072418213, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7630837559700012, + "num_tokens": 702180819.0, + "step": 27136 + }, + { + "epoch": 2.9801229958269273, + "grad_norm": 2.6370770931243896, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7415288090705872, + "num_tokens": 702201448.0, + "step": 27137 + }, + { + "epoch": 2.980232813529541, + "grad_norm": 2.467226028442383, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7330034971237183, + "num_tokens": 702225696.0, + "step": 27138 + }, + { + "epoch": 2.980342631232155, + "grad_norm": 2.409949779510498, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7311936616897583, + "num_tokens": 702252138.0, + "step": 27139 + }, + { + "epoch": 2.980452448934768, + "grad_norm": 2.5436153411865234, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7432734966278076, + "num_tokens": 702275638.0, + "step": 27140 + }, + { + "epoch": 2.980562266637382, + "grad_norm": 2.492697238922119, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7415065169334412, + "num_tokens": 702298503.0, + "step": 27141 + }, + { + "epoch": 2.9806720843399956, + "grad_norm": 2.648592948913574, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7536882162094116, + "num_tokens": 702321521.0, + "step": 27142 + }, + { + "epoch": 2.9807819020426094, + "grad_norm": 2.5104353427886963, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7218852043151855, + "num_tokens": 702343966.0, + "step": 27143 + }, + { + "epoch": 2.980891719745223, + "grad_norm": 2.611863374710083, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7231761813163757, + "num_tokens": 702366429.0, + "step": 27144 + }, + { + "epoch": 2.9810015374478365, + "grad_norm": 2.6313133239746094, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.732239842414856, + "num_tokens": 702388325.0, + "step": 27145 + }, + { + "epoch": 2.9811113551504502, + "grad_norm": 2.2354891300201416, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7175549864768982, + "num_tokens": 702417213.0, + "step": 27146 + }, + { + "epoch": 2.981221172853064, + "grad_norm": 2.4972095489501953, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7235779762268066, + "num_tokens": 702441237.0, + "step": 27147 + }, + { + "epoch": 2.9813309905556777, + "grad_norm": 2.061621904373169, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.707770049571991, + "num_tokens": 702474699.0, + "step": 27148 + }, + { + "epoch": 2.9814408082582915, + "grad_norm": 2.249579429626465, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7280411124229431, + "num_tokens": 702504887.0, + "step": 27149 + }, + { + "epoch": 2.981550625960905, + "grad_norm": 2.4835734367370605, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.715233325958252, + "num_tokens": 702528723.0, + "step": 27150 + }, + { + "epoch": 2.9816604436635186, + "grad_norm": 2.528273582458496, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7487248778343201, + "num_tokens": 702551150.0, + "step": 27151 + }, + { + "epoch": 2.9817702613661323, + "grad_norm": 2.4906013011932373, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7523419857025146, + "num_tokens": 702574087.0, + "step": 27152 + }, + { + "epoch": 2.9818800790687456, + "grad_norm": 2.425095796585083, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7225230932235718, + "num_tokens": 702601130.0, + "step": 27153 + }, + { + "epoch": 2.9819898967713594, + "grad_norm": 2.130589008331299, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7297994494438171, + "num_tokens": 702633442.0, + "step": 27154 + }, + { + "epoch": 2.982099714473973, + "grad_norm": 2.255657196044922, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7408685684204102, + "num_tokens": 702659621.0, + "step": 27155 + }, + { + "epoch": 2.982209532176587, + "grad_norm": 2.1109418869018555, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7356764078140259, + "num_tokens": 702687776.0, + "step": 27156 + }, + { + "epoch": 2.9823193498792007, + "grad_norm": 2.453967571258545, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7358304262161255, + "num_tokens": 702712274.0, + "step": 27157 + }, + { + "epoch": 2.982429167581814, + "grad_norm": 2.2216711044311523, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7349705696105957, + "num_tokens": 702740462.0, + "step": 27158 + }, + { + "epoch": 2.9825389852844277, + "grad_norm": 2.4372787475585938, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7408120632171631, + "num_tokens": 702764661.0, + "step": 27159 + }, + { + "epoch": 2.9826488029870415, + "grad_norm": 2.3704957962036133, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7300963401794434, + "num_tokens": 702792044.0, + "step": 27160 + }, + { + "epoch": 2.9827586206896552, + "grad_norm": 2.560687780380249, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7324832677841187, + "num_tokens": 702813905.0, + "step": 27161 + }, + { + "epoch": 2.982868438392269, + "grad_norm": 2.690509796142578, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7353752851486206, + "num_tokens": 702834398.0, + "step": 27162 + }, + { + "epoch": 2.9829782560948823, + "grad_norm": 2.426154136657715, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7217978835105896, + "num_tokens": 702860805.0, + "step": 27163 + }, + { + "epoch": 2.983088073797496, + "grad_norm": 2.431891918182373, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7348078489303589, + "num_tokens": 702887110.0, + "step": 27164 + }, + { + "epoch": 2.98319789150011, + "grad_norm": 2.324469566345215, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7208790183067322, + "num_tokens": 702914694.0, + "step": 27165 + }, + { + "epoch": 2.9833077092027236, + "grad_norm": 2.3326709270477295, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7180389165878296, + "num_tokens": 702941666.0, + "step": 27166 + }, + { + "epoch": 2.9834175269053373, + "grad_norm": 2.811492681503296, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.742257833480835, + "num_tokens": 702962865.0, + "step": 27167 + }, + { + "epoch": 2.9835273446079507, + "grad_norm": 2.163154125213623, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7453287839889526, + "num_tokens": 702991618.0, + "step": 27168 + }, + { + "epoch": 2.9836371623105644, + "grad_norm": 2.143479824066162, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7114322185516357, + "num_tokens": 703023298.0, + "step": 27169 + }, + { + "epoch": 2.983746980013178, + "grad_norm": 2.1697006225585938, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7227323055267334, + "num_tokens": 703053813.0, + "step": 27170 + }, + { + "epoch": 2.983856797715792, + "grad_norm": 2.982440710067749, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7525753378868103, + "num_tokens": 703072059.0, + "step": 27171 + }, + { + "epoch": 2.9839666154184057, + "grad_norm": 2.4933133125305176, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7220702171325684, + "num_tokens": 703096897.0, + "step": 27172 + }, + { + "epoch": 2.984076433121019, + "grad_norm": 2.670727252960205, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7287938594818115, + "num_tokens": 703117645.0, + "step": 27173 + }, + { + "epoch": 2.9841862508236328, + "grad_norm": 2.360311985015869, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7604058980941772, + "num_tokens": 703139908.0, + "step": 27174 + }, + { + "epoch": 2.9842960685262465, + "grad_norm": 2.290834665298462, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7533456087112427, + "num_tokens": 703163662.0, + "step": 27175 + }, + { + "epoch": 2.98440588622886, + "grad_norm": 2.346999406814575, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.742668628692627, + "num_tokens": 703188222.0, + "step": 27176 + }, + { + "epoch": 2.984515703931474, + "grad_norm": 2.3256449699401855, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7331681251525879, + "num_tokens": 703213448.0, + "step": 27177 + }, + { + "epoch": 2.9846255216340873, + "grad_norm": 2.318138599395752, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7367074489593506, + "num_tokens": 703239870.0, + "step": 27178 + }, + { + "epoch": 2.984735339336701, + "grad_norm": 2.5229759216308594, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.733723521232605, + "num_tokens": 703262075.0, + "step": 27179 + }, + { + "epoch": 2.984845157039315, + "grad_norm": 2.485456705093384, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7307778000831604, + "num_tokens": 703286704.0, + "step": 27180 + }, + { + "epoch": 2.984954974741928, + "grad_norm": 2.3010077476501465, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7106301188468933, + "num_tokens": 703315082.0, + "step": 27181 + }, + { + "epoch": 2.985064792444542, + "grad_norm": 2.3160157203674316, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7441306114196777, + "num_tokens": 703342324.0, + "step": 27182 + }, + { + "epoch": 2.9851746101471557, + "grad_norm": 2.15409779548645, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7295850515365601, + "num_tokens": 703373898.0, + "step": 27183 + }, + { + "epoch": 2.9852844278497694, + "grad_norm": 2.6317968368530273, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7456790208816528, + "num_tokens": 703394399.0, + "step": 27184 + }, + { + "epoch": 2.985394245552383, + "grad_norm": 2.5714805126190186, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.725121259689331, + "num_tokens": 703418105.0, + "step": 27185 + }, + { + "epoch": 2.9855040632549965, + "grad_norm": 2.2919979095458984, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7221460342407227, + "num_tokens": 703445569.0, + "step": 27186 + }, + { + "epoch": 2.9856138809576103, + "grad_norm": 2.667665481567383, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.726240873336792, + "num_tokens": 703466434.0, + "step": 27187 + }, + { + "epoch": 2.985723698660224, + "grad_norm": 2.343622922897339, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7355154752731323, + "num_tokens": 703493661.0, + "step": 27188 + }, + { + "epoch": 2.985833516362838, + "grad_norm": 2.6746573448181152, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.738081693649292, + "num_tokens": 703515129.0, + "step": 27189 + }, + { + "epoch": 2.9859433340654515, + "grad_norm": 2.9329748153686523, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7471640110015869, + "num_tokens": 703531781.0, + "step": 27190 + }, + { + "epoch": 2.986053151768065, + "grad_norm": 2.0899605751037598, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7137118577957153, + "num_tokens": 703564640.0, + "step": 27191 + }, + { + "epoch": 2.9861629694706786, + "grad_norm": 2.292783498764038, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7315586805343628, + "num_tokens": 703590692.0, + "step": 27192 + }, + { + "epoch": 2.9862727871732924, + "grad_norm": 2.467820882797241, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7282048463821411, + "num_tokens": 703614104.0, + "step": 27193 + }, + { + "epoch": 2.986382604875906, + "grad_norm": 2.1953787803649902, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7358890771865845, + "num_tokens": 703643107.0, + "step": 27194 + }, + { + "epoch": 2.98649242257852, + "grad_norm": 2.123250961303711, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7091819643974304, + "num_tokens": 703675675.0, + "step": 27195 + }, + { + "epoch": 2.986602240281133, + "grad_norm": 2.174175262451172, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7260158061981201, + "num_tokens": 703705056.0, + "step": 27196 + }, + { + "epoch": 2.986712057983747, + "grad_norm": 2.4103124141693115, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.759361982345581, + "num_tokens": 703728771.0, + "step": 27197 + }, + { + "epoch": 2.9868218756863607, + "grad_norm": 2.2023983001708984, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7342609167098999, + "num_tokens": 703757949.0, + "step": 27198 + }, + { + "epoch": 2.9869316933889745, + "grad_norm": 2.5016696453094482, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7401660084724426, + "num_tokens": 703781668.0, + "step": 27199 + }, + { + "epoch": 2.987041511091588, + "grad_norm": 2.2755565643310547, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7413840293884277, + "num_tokens": 703810232.0, + "step": 27200 + }, + { + "epoch": 2.9871513287942015, + "grad_norm": 2.1861331462860107, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7346856594085693, + "num_tokens": 703837505.0, + "step": 27201 + }, + { + "epoch": 2.9872611464968153, + "grad_norm": 2.293342351913452, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7251962423324585, + "num_tokens": 703864601.0, + "step": 27202 + }, + { + "epoch": 2.987370964199429, + "grad_norm": 2.3613598346710205, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6969942450523376, + "num_tokens": 703896134.0, + "step": 27203 + }, + { + "epoch": 2.9874807819020424, + "grad_norm": 2.6389689445495605, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7437004446983337, + "num_tokens": 703916331.0, + "step": 27204 + }, + { + "epoch": 2.987590599604656, + "grad_norm": 2.2575669288635254, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7263433933258057, + "num_tokens": 703944317.0, + "step": 27205 + }, + { + "epoch": 2.98770041730727, + "grad_norm": 2.575986385345459, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7370482087135315, + "num_tokens": 703965788.0, + "step": 27206 + }, + { + "epoch": 2.9878102350098836, + "grad_norm": 2.3055906295776367, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7331984639167786, + "num_tokens": 703991689.0, + "step": 27207 + }, + { + "epoch": 2.9879200527124974, + "grad_norm": 2.340843439102173, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7393007278442383, + "num_tokens": 704018020.0, + "step": 27208 + }, + { + "epoch": 2.9880298704151107, + "grad_norm": 2.5276710987091064, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7286084890365601, + "num_tokens": 704040366.0, + "step": 27209 + }, + { + "epoch": 2.9881396881177245, + "grad_norm": 2.366079330444336, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7309712767601013, + "num_tokens": 704064833.0, + "step": 27210 + }, + { + "epoch": 2.988249505820338, + "grad_norm": 2.6584932804107666, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7439397573471069, + "num_tokens": 704087321.0, + "step": 27211 + }, + { + "epoch": 2.988359323522952, + "grad_norm": 2.2911834716796875, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7067047357559204, + "num_tokens": 704115895.0, + "step": 27212 + }, + { + "epoch": 2.9884691412255657, + "grad_norm": 2.02117657661438, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7377409338951111, + "num_tokens": 704148074.0, + "step": 27213 + }, + { + "epoch": 2.988578958928179, + "grad_norm": 2.3417587280273438, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7236899137496948, + "num_tokens": 704175384.0, + "step": 27214 + }, + { + "epoch": 2.988688776630793, + "grad_norm": 2.4024434089660645, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7439825534820557, + "num_tokens": 704198495.0, + "step": 27215 + }, + { + "epoch": 2.9887985943334066, + "grad_norm": 2.2095136642456055, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7168523073196411, + "num_tokens": 704226538.0, + "step": 27216 + }, + { + "epoch": 2.9889084120360203, + "grad_norm": 2.625422954559326, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7537484765052795, + "num_tokens": 704248083.0, + "step": 27217 + }, + { + "epoch": 2.989018229738634, + "grad_norm": 2.621582508087158, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7547057867050171, + "num_tokens": 704267508.0, + "step": 27218 + }, + { + "epoch": 2.9891280474412474, + "grad_norm": 2.166614055633545, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.710006594657898, + "num_tokens": 704298993.0, + "step": 27219 + }, + { + "epoch": 2.989237865143861, + "grad_norm": 2.3609161376953125, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7425470352172852, + "num_tokens": 704323108.0, + "step": 27220 + }, + { + "epoch": 2.989347682846475, + "grad_norm": 2.732847213745117, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7347697019577026, + "num_tokens": 704343373.0, + "step": 27221 + }, + { + "epoch": 2.9894575005490887, + "grad_norm": 2.515995740890503, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7206968069076538, + "num_tokens": 704366538.0, + "step": 27222 + }, + { + "epoch": 2.9895673182517024, + "grad_norm": 2.682112693786621, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7258079051971436, + "num_tokens": 704390386.0, + "step": 27223 + }, + { + "epoch": 2.9896771359543157, + "grad_norm": 2.404269218444824, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7188989520072937, + "num_tokens": 704415032.0, + "step": 27224 + }, + { + "epoch": 2.9897869536569295, + "grad_norm": 2.2659919261932373, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.748438835144043, + "num_tokens": 704442716.0, + "step": 27225 + }, + { + "epoch": 2.9898967713595432, + "grad_norm": 2.3574259281158447, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7436376810073853, + "num_tokens": 704467676.0, + "step": 27226 + }, + { + "epoch": 2.9900065890621566, + "grad_norm": 2.579393148422241, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7402742505073547, + "num_tokens": 704488473.0, + "step": 27227 + }, + { + "epoch": 2.9901164067647708, + "grad_norm": 1.9950439929962158, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7276374101638794, + "num_tokens": 704520933.0, + "step": 27228 + }, + { + "epoch": 2.990226224467384, + "grad_norm": 2.1702568531036377, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7397667169570923, + "num_tokens": 704550513.0, + "step": 27229 + }, + { + "epoch": 2.990336042169998, + "grad_norm": 2.3839738368988037, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7129780054092407, + "num_tokens": 704577609.0, + "step": 27230 + }, + { + "epoch": 2.9904458598726116, + "grad_norm": 2.338531494140625, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7347285151481628, + "num_tokens": 704601933.0, + "step": 27231 + }, + { + "epoch": 2.990555677575225, + "grad_norm": 2.525463104248047, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7310987114906311, + "num_tokens": 704624988.0, + "step": 27232 + }, + { + "epoch": 2.9906654952778386, + "grad_norm": 2.2345094680786133, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7334551811218262, + "num_tokens": 704653433.0, + "step": 27233 + }, + { + "epoch": 2.9907753129804524, + "grad_norm": 2.12251353263855, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7367610931396484, + "num_tokens": 704681291.0, + "step": 27234 + }, + { + "epoch": 2.990885130683066, + "grad_norm": 2.593173027038574, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7453070878982544, + "num_tokens": 704704250.0, + "step": 27235 + }, + { + "epoch": 2.99099494838568, + "grad_norm": 2.485882043838501, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7514210343360901, + "num_tokens": 704729508.0, + "step": 27236 + }, + { + "epoch": 2.9911047660882932, + "grad_norm": 2.5140862464904785, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7369793653488159, + "num_tokens": 704753815.0, + "step": 27237 + }, + { + "epoch": 2.991214583790907, + "grad_norm": 2.231764316558838, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7074053287506104, + "num_tokens": 704785221.0, + "step": 27238 + }, + { + "epoch": 2.9913244014935207, + "grad_norm": 2.1436383724212646, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7246432304382324, + "num_tokens": 704815571.0, + "step": 27239 + }, + { + "epoch": 2.9914342191961345, + "grad_norm": 2.356483221054077, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7305342555046082, + "num_tokens": 704841008.0, + "step": 27240 + }, + { + "epoch": 2.9915440368987483, + "grad_norm": 2.303119659423828, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7353024482727051, + "num_tokens": 704867545.0, + "step": 27241 + }, + { + "epoch": 2.9916538546013616, + "grad_norm": 2.603811502456665, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7328641414642334, + "num_tokens": 704891252.0, + "step": 27242 + }, + { + "epoch": 2.9917636723039753, + "grad_norm": 2.643181800842285, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7438315749168396, + "num_tokens": 704910961.0, + "step": 27243 + }, + { + "epoch": 2.991873490006589, + "grad_norm": 2.058041572570801, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7096353769302368, + "num_tokens": 704943947.0, + "step": 27244 + }, + { + "epoch": 2.991983307709203, + "grad_norm": 2.2187178134918213, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.729847252368927, + "num_tokens": 704974192.0, + "step": 27245 + }, + { + "epoch": 2.9920931254118166, + "grad_norm": 2.609755277633667, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7448533177375793, + "num_tokens": 704995669.0, + "step": 27246 + }, + { + "epoch": 2.99220294311443, + "grad_norm": 2.375457763671875, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7383034825325012, + "num_tokens": 705020032.0, + "step": 27247 + }, + { + "epoch": 2.9923127608170437, + "grad_norm": 2.25052547454834, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7129210233688354, + "num_tokens": 705048849.0, + "step": 27248 + }, + { + "epoch": 2.9924225785196574, + "grad_norm": 2.286247730255127, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7364460229873657, + "num_tokens": 705077489.0, + "step": 27249 + }, + { + "epoch": 2.992532396222271, + "grad_norm": 2.1476094722747803, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7418574690818787, + "num_tokens": 705105991.0, + "step": 27250 + }, + { + "epoch": 2.992642213924885, + "grad_norm": 2.668872356414795, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7384660243988037, + "num_tokens": 705126974.0, + "step": 27251 + }, + { + "epoch": 2.9927520316274983, + "grad_norm": 2.3317501544952393, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7087677717208862, + "num_tokens": 705151444.0, + "step": 27252 + }, + { + "epoch": 2.992861849330112, + "grad_norm": 2.63083815574646, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.745672345161438, + "num_tokens": 705172830.0, + "step": 27253 + }, + { + "epoch": 2.9929716670327258, + "grad_norm": 2.235628128051758, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7286151647567749, + "num_tokens": 705202731.0, + "step": 27254 + }, + { + "epoch": 2.993081484735339, + "grad_norm": 2.691511631011963, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7360521554946899, + "num_tokens": 705223406.0, + "step": 27255 + }, + { + "epoch": 2.993191302437953, + "grad_norm": 2.210736036300659, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7181682586669922, + "num_tokens": 705252641.0, + "step": 27256 + }, + { + "epoch": 2.9933011201405666, + "grad_norm": 2.0712497234344482, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7259812355041504, + "num_tokens": 705283871.0, + "step": 27257 + }, + { + "epoch": 2.9934109378431804, + "grad_norm": 2.332075357437134, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7377619743347168, + "num_tokens": 705308264.0, + "step": 27258 + }, + { + "epoch": 2.993520755545794, + "grad_norm": 2.501284122467041, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7473945617675781, + "num_tokens": 705331376.0, + "step": 27259 + }, + { + "epoch": 2.9936305732484074, + "grad_norm": 2.326521873474121, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7027852535247803, + "num_tokens": 705359289.0, + "step": 27260 + }, + { + "epoch": 2.993740390951021, + "grad_norm": 2.281738042831421, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7215148210525513, + "num_tokens": 705386707.0, + "step": 27261 + }, + { + "epoch": 2.993850208653635, + "grad_norm": 2.1241095066070557, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7409014701843262, + "num_tokens": 705418513.0, + "step": 27262 + }, + { + "epoch": 2.9939600263562487, + "grad_norm": 2.094759464263916, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7462501525878906, + "num_tokens": 705447497.0, + "step": 27263 + }, + { + "epoch": 2.9940698440588625, + "grad_norm": 2.221757650375366, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.731672465801239, + "num_tokens": 705474844.0, + "step": 27264 + }, + { + "epoch": 2.9941796617614758, + "grad_norm": 2.562354564666748, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7568773031234741, + "num_tokens": 705496732.0, + "step": 27265 + }, + { + "epoch": 2.9942894794640895, + "grad_norm": 2.4706664085388184, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7384985685348511, + "num_tokens": 705519700.0, + "step": 27266 + }, + { + "epoch": 2.9943992971667033, + "grad_norm": 2.280787944793701, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7434736490249634, + "num_tokens": 705546443.0, + "step": 27267 + }, + { + "epoch": 2.994509114869317, + "grad_norm": 2.45967960357666, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7405083179473877, + "num_tokens": 705569904.0, + "step": 27268 + }, + { + "epoch": 2.994618932571931, + "grad_norm": 2.532248020172119, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7391827702522278, + "num_tokens": 705593737.0, + "step": 27269 + }, + { + "epoch": 2.994728750274544, + "grad_norm": 2.327258825302124, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7343443036079407, + "num_tokens": 705620490.0, + "step": 27270 + }, + { + "epoch": 2.994838567977158, + "grad_norm": 2.638505458831787, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7474147081375122, + "num_tokens": 705641312.0, + "step": 27271 + }, + { + "epoch": 2.9949483856797716, + "grad_norm": 2.427159070968628, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7427757978439331, + "num_tokens": 705668248.0, + "step": 27272 + }, + { + "epoch": 2.9950582033823854, + "grad_norm": 2.222205638885498, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7236063480377197, + "num_tokens": 705696773.0, + "step": 27273 + }, + { + "epoch": 2.995168021084999, + "grad_norm": 2.204392910003662, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7200072407722473, + "num_tokens": 705728765.0, + "step": 27274 + }, + { + "epoch": 2.9952778387876124, + "grad_norm": 2.4147746562957764, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7127256989479065, + "num_tokens": 705755618.0, + "step": 27275 + }, + { + "epoch": 2.995387656490226, + "grad_norm": 2.2864530086517334, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.726804256439209, + "num_tokens": 705784998.0, + "step": 27276 + }, + { + "epoch": 2.99549747419284, + "grad_norm": 2.5113391876220703, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7639422416687012, + "num_tokens": 705807142.0, + "step": 27277 + }, + { + "epoch": 2.9956072918954533, + "grad_norm": 2.2557151317596436, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7277177572250366, + "num_tokens": 705834156.0, + "step": 27278 + }, + { + "epoch": 2.9957171095980675, + "grad_norm": 2.375429630279541, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7301055788993835, + "num_tokens": 705858564.0, + "step": 27279 + }, + { + "epoch": 2.995826927300681, + "grad_norm": 2.2819535732269287, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7199236154556274, + "num_tokens": 705888824.0, + "step": 27280 + }, + { + "epoch": 2.9959367450032945, + "grad_norm": 2.4688730239868164, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7346156239509583, + "num_tokens": 705912030.0, + "step": 27281 + }, + { + "epoch": 2.9960465627059083, + "grad_norm": 2.396252155303955, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7278236150741577, + "num_tokens": 705937277.0, + "step": 27282 + }, + { + "epoch": 2.9961563804085216, + "grad_norm": 2.654712677001953, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7482222318649292, + "num_tokens": 705959095.0, + "step": 27283 + }, + { + "epoch": 2.9962661981111354, + "grad_norm": 2.6530704498291016, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7466514110565186, + "num_tokens": 705980298.0, + "step": 27284 + }, + { + "epoch": 2.996376015813749, + "grad_norm": 2.4219675064086914, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7602494955062866, + "num_tokens": 706002573.0, + "step": 27285 + }, + { + "epoch": 2.996485833516363, + "grad_norm": 2.557405948638916, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7261525392532349, + "num_tokens": 706024834.0, + "step": 27286 + }, + { + "epoch": 2.9965956512189766, + "grad_norm": 2.3112707138061523, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7153714299201965, + "num_tokens": 706050593.0, + "step": 27287 + }, + { + "epoch": 2.99670546892159, + "grad_norm": 2.2873380184173584, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.716247022151947, + "num_tokens": 706076878.0, + "step": 27288 + }, + { + "epoch": 2.9968152866242037, + "grad_norm": 2.594548225402832, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7590818405151367, + "num_tokens": 706098592.0, + "step": 27289 + }, + { + "epoch": 2.9969251043268175, + "grad_norm": 2.423290491104126, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7546408176422119, + "num_tokens": 706122101.0, + "step": 27290 + }, + { + "epoch": 2.9970349220294312, + "grad_norm": 2.2453131675720215, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7324837446212769, + "num_tokens": 706150293.0, + "step": 27291 + }, + { + "epoch": 2.997144739732045, + "grad_norm": 2.6411330699920654, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.757624089717865, + "num_tokens": 706172552.0, + "step": 27292 + }, + { + "epoch": 2.9972545574346583, + "grad_norm": 2.20867919921875, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7275633811950684, + "num_tokens": 706202391.0, + "step": 27293 + }, + { + "epoch": 2.997364375137272, + "grad_norm": 2.468785524368286, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7201601266860962, + "num_tokens": 706226644.0, + "step": 27294 + }, + { + "epoch": 2.997474192839886, + "grad_norm": 2.728994131088257, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7486217021942139, + "num_tokens": 706245912.0, + "step": 27295 + }, + { + "epoch": 2.9975840105424996, + "grad_norm": 2.6743080615997314, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7364174127578735, + "num_tokens": 706268033.0, + "step": 27296 + }, + { + "epoch": 2.9976938282451133, + "grad_norm": 2.5030934810638428, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7572765350341797, + "num_tokens": 706290252.0, + "step": 27297 + }, + { + "epoch": 2.9978036459477266, + "grad_norm": 2.1588287353515625, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7097402811050415, + "num_tokens": 706320660.0, + "step": 27298 + }, + { + "epoch": 2.9979134636503404, + "grad_norm": 2.5582540035247803, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7256938219070435, + "num_tokens": 706342544.0, + "step": 27299 + }, + { + "epoch": 2.998023281352954, + "grad_norm": 2.2954466342926025, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7444654703140259, + "num_tokens": 706369294.0, + "step": 27300 + }, + { + "epoch": 2.998133099055568, + "grad_norm": 2.499166488647461, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7702418565750122, + "num_tokens": 706392264.0, + "step": 27301 + }, + { + "epoch": 2.9982429167581817, + "grad_norm": 2.121795177459717, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7244774103164673, + "num_tokens": 706423052.0, + "step": 27302 + }, + { + "epoch": 2.998352734460795, + "grad_norm": 2.2926478385925293, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7560418844223022, + "num_tokens": 706448674.0, + "step": 27303 + }, + { + "epoch": 2.9984625521634087, + "grad_norm": 2.335726499557495, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7373168468475342, + "num_tokens": 706475897.0, + "step": 27304 + }, + { + "epoch": 2.9985723698660225, + "grad_norm": 1.97543466091156, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7444287538528442, + "num_tokens": 706508259.0, + "step": 27305 + }, + { + "epoch": 2.998682187568636, + "grad_norm": 2.4543066024780273, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7336589694023132, + "num_tokens": 706531279.0, + "step": 27306 + }, + { + "epoch": 2.9987920052712496, + "grad_norm": 2.304046630859375, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7214043140411377, + "num_tokens": 706557018.0, + "step": 27307 + }, + { + "epoch": 2.9989018229738633, + "grad_norm": 2.2494046688079834, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7255337238311768, + "num_tokens": 706587542.0, + "step": 27308 + }, + { + "epoch": 2.999011640676477, + "grad_norm": 2.8548338413238525, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.735161304473877, + "num_tokens": 706606653.0, + "step": 27309 + }, + { + "epoch": 2.999121458379091, + "grad_norm": 2.381063222885132, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7447463274002075, + "num_tokens": 706631700.0, + "step": 27310 + }, + { + "epoch": 2.999231276081704, + "grad_norm": 2.3544843196868896, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7403967380523682, + "num_tokens": 706656969.0, + "step": 27311 + }, + { + "epoch": 2.999341093784318, + "grad_norm": 2.49375057220459, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7315775156021118, + "num_tokens": 706681011.0, + "step": 27312 + }, + { + "epoch": 2.9994509114869317, + "grad_norm": 2.2204079627990723, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7368555068969727, + "num_tokens": 706707342.0, + "step": 27313 + }, + { + "epoch": 2.9995607291895454, + "grad_norm": 2.3707022666931152, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7386348843574524, + "num_tokens": 706731699.0, + "step": 27314 + }, + { + "epoch": 2.999670546892159, + "grad_norm": 2.4170238971710205, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7311004996299744, + "num_tokens": 706757635.0, + "step": 27315 + }, + { + "epoch": 2.9997803645947725, + "grad_norm": 2.1405928134918213, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7458480596542358, + "num_tokens": 706787813.0, + "step": 27316 + }, + { + "epoch": 2.9998901822973862, + "grad_norm": 2.1360666751861572, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7025198936462402, + "num_tokens": 706821902.0, + "step": 27317 + }, + { + "epoch": 3.0, + "grad_norm": 2.062453031539917, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7388681769371033, + "num_tokens": 706854136.0, + "step": 27318 + }, + { + "epoch": 3.0, + "step": 27318, + "total_flos": 3.1829322264618205e+19, + "train_loss": 0.922418457405818, + "train_runtime": 33639.8393, + "train_samples_per_second": 12.993, + "train_steps_per_second": 0.812 + } + ], + "logging_steps": 1, + "max_steps": 27318, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 2732, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 3.1829322264618205e+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..86ccfee --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:616fb977dbee4542f35ab3c4a0bf7aa81a6014835b843cee735300054456511a +size 13329